natural history and prognostic factors in localized

Natural history and prognostic factorsin localized prostate cancer

Avh_Ove Andre�n_K160308.indd 1 08-04-17 15.42.16

To Ulrika, Erik, Ina och Nils

”It’s a job that’s never started

that takes the longest to fi nish.”

J. R. R. Tolkien


Örebro Studies in Medicine 15

Ove Andrén

Natural history and prognostic factors

in localized prostate cancer


© Ove Andrén 2008

Title: Natural history and prognostic factorsin localized prostate cancer

Publisher: Örebro University 2008www.publications.oru.se

Editor: Heinz [email protected]

Printer: Intellecta DocuSys, V Frölunda 04/2008

issn 1642-4063isbn 978-91-7668-592-1


ABSTRACT

The natural history of localized prostate cancer is not fully understood. In most patients the tumor will never progress to a lethal disease, while a subset of patients will ultimately die of the disease. Effi cient tools to separate indolent from lethal disease is currently lacking which means that many patients will be offered treat-ment without any benefi t, but still be at risk of experiencing treatment related side effects.

The aims of these studies were to get more insight into the natural history of untreated localized prostate cancer, to assess the prognostic value of established clinical parameters such as Gleason score, nuclear grade and tumor volume and, moreover, some new prognostic markers Ki-67, AMACR and MUC-1. We also aimed to study time trends in the detection of incidental tumors in Sweden.

Patients with localized disease (n=223) and no initial treatment were followed for 21 years. Most patients had a favorable outcome. However, a subset of patients developed lethal disease even beyond 15 years of follow-up and these patients defi ne the group that may benefi t most from treatment with curative intent. Patients with poorly differentiated tumors experienced a 9 time higher risk of dying in prostate cancer.

The studies on prognostic markers are based on a cohort of patients (n=253) with incidental prostate cancer detected by transurethral resection for presumed benign hyperplasia. All patients were left without initial treatment. Gleason grade, nuclear grade and tumor volume turned all out to be independent prognostic factors. MUC-1, AMACR and Ki-67 also carried prognostic information. However, after adjustment for Gleason grade, nuclear grade and tumor volume only MUC-1 and AMACR remained as statistically signifi cant prognostic factors. When tested for sensitivity and specifi city they all failed and, consequently, they seem to be of less value in daily practice for cancelling an individual patient regarding the choice of treatment.

Time trends in incidental prostate tumors in Sweden were analyzed in a cohort of patients with prostate tumors detected by transurethral resection (TUR-P). Through linkage of the national registration number (NRN) with several registers, e.g. the Swedish Cancer Registry, the National Inpatient registry and the Cause of Death Registry we identifi ed, during the period 1970 through 2003, in total 23288 patients with incidental prostate cancer, who constituted the study group. As comparison group we choose all patients diagnosed with prostate cancer between 1970–2003 excluding those with incidental cancer, in total 112204 patients. Our result confi rms earlier fi ndings that there has been a dramatic change over time in incidence of incidental prostate cancers in Sweden, which parallels the introduction of prostate specifi c antigen. We also found that the cumulative incidence of prostate cancer death is high in the incidental group, opposing earlier fi ndings that incidental tumours are a non-lethal disease.

Keywords: Prostate cancer, Natural History, Survival, Prognostic factor, Tumor grade, TUR-P, Incidental, MUC-1, AMACR, Ki67, Tumor volume.


CONTENTS

LIST OF PAPERS ....................................................................................9

ABBREVIATIONS ...................................................................................10

INTRODUCTION ....................................................................................11

PROSTATE CANCER EPIDEMIOLOGY IN SWEDEN ......................11

PSA SCREENING .................................................................................12

TREATMENT OF LOCALIZED PROSTATE CANCER ....................13

PROSTATE CANCER DETECTED BY TRANSURETHRAL

RESECTION (TUR-P) ..........................................................................14

PROGNOSTIC AND PREDICTIVE FACTORS ................................... 15

PROGNOSTIC FACTORS USED TODAY ........................................... 15

Histological grade .................................................................................. 15

Local tumor stage ..................................................................................16

Prostate specifi c antigen (PSA) ...............................................................16

Age at diagnosis .....................................................................................16

Co-morbidity .........................................................................................17

NEW PROGNOSTIC MARKERS ........................................................17

Tumor proliferation markers ..................................................................17

AMACR ................................................................................................17

MUC-1 ..................................................................................................17

AIMS ......................................................................................................19

MATERIAL AND METHODS ..................................................................21

Paper I ....................................................................................................21

Paper II–IV ............................................................................................24

Paper V ..................................................................................................25

STATISTICAL METHODS ...................................................................27

Paper I ....................................................................................................27

Paper II–IV ............................................................................................27

Paper III .................................................................................................27

Paper V ..................................................................................................28


RESULTS ...............................................................................................29

Paper I ....................................................................................................29

Paper II ..................................................................................................37

Paper III .................................................................................................40

Paper IV .................................................................................................44

Paper V ..................................................................................................50

DISCUSSION . ........................................................................................5

Paper I ....................................................................................................5

Paper II .................................................................................................. 56

Paper III .................................................................................................

Paper IV .................................................................................................

Paper V ..................................................................................................6

CONCLUSIONS ...................................................................................

SUMMARY AND FUTURE ASPECTS ................................................

ACKNOWLEDGMENTS .....................................................................

67

REFERENCES ......................................................................................

69


5

5

59

162

65

17

9

LIST OF PAPERS

The thesis is based on the following papers, which will be referred to by their Roman numerals:

I. Jan-Erik Johansson, Ove Andrén, Swen-Olof Andersson, Paul W Dickman, Lars Holmberg, Anders Magnuson, Hans-Olov Adami. Natural history of early, localized prostate cancer. JAMA. 2004 Jun 9;291(22):2713-2719

II. Ove Andrén, Katja Fall, Lennart Franzén, Swen-Olof Andersson, Jan-Erik Johansson, Mark Rubin. How well does the Gleason score predict prostate cancer death? – A 20-year follow-up of a population-based cohort in Sweden. J Urol. 2006 Apr;175(4):1337-1340.

III. Mark Rubin MA, Bismar TA, Andrén O, Lorelei A Mucci, Shen R, Kim R, Ghosh D, Wei JT, Chinnaiyan AM. Adami HO, Kantoff PW, Johansson J-E. Decreased alpha-methylacyl CoA racemase expression in localized prostate cancer is associated with an increased rate of biochemical recurrence and cancer-specific death. Cancer Epidemiol

Biomarkers Prev. 2005 Jun; 14(6):1424-1432.

IV. Ove Andrén, Katja Fall, Sven-Olof Andersson, Mark A Rubin, Tarek A Bismar, Mats Karlsson, Jan-Erik Johansson, Lorelei A Mucci. MUC-1 gene is associated with prostate cancer death: – A 20 year follow-up of a population-based study in Sweden.Br J Cancer. 2007 Sep 17;97(6):730-734.

V. Ove Andrén, Hans Garmo, Lorelei Mucci, Swen-Olof Andersson, Jan-Erik Johansson and Katja Fall. Time trends and survival in men with incidental prostate cancer: -aregister-based study comprising 23,288 men diagnosed between 1970–2003 in Sweden,Submitted.


99

10

ABBREVIATIONS

ASAP atypical small acinar proliferationAMACR alpha methylacyl CoA racemaseBR Björn RisbergBRFS biochemical recurrence free survivalcDNA complementary deoxyribonucleic acidCI confidence intervalCT computer tomographyELISA enzyme linked immunosorbent assayHR hazard ratioICD international classification of diseasesHDR hospital discharge registerKi67 cell proliferation markerNRN national registration numberMIB-1 mindbomb homolog 1MR Mark RubinMRI magnetic resonance imagingMUC-1 mucin 1, anti-adhesion molecule OAE open adenoma enucleationPCa prostate cancerPSA prostate specific antigenROC receiver operating characteristicRT-PCR real time p chain reactionSD standard deviationSPCG Scandinavian prostate cancer groupTNM tumor nodule metastasisTUR-P transurethral resection of the prostateUSÖ university hospital of ÖrebroWHO world health organization


1010

11

INTRODUCTION

The appropriate management of localized prostate cancer is still unclear. Even though some tumors will progress to metastatic disease, most patients will not experience advanced disease nor need treatment. Indeed, several studies on the natural history of the localized disease, reveal that the vast majority of patients will not die of their disease1-3. Despite this knowledge, the number of patients treated for localized disease with curative intent has increased dramatically over the previous decades4. As a result, in current practice, prostate cancer patients are over treated and experience the consequences of treatment without any benefit5, 6.

Therefore, one of the most important areas in prostate cancer research today involves finding tools that distinguish indolent disease from aggressive disease. The prognostic factors used today are PSA, tumor stage7, grade and volume. Although the classification of tumor stage is dependent on the examiner, it has some prognostic value and is still a cornerstone in clinical practice. While tumor volume is a strong prognostic marker8, 9, it is hard to estimate for most tumors in the diagnostic setting10, 11. For a long time tumor grade, which was nuclear grade during the era of fine needle aspiration12, was the dominant grading system in Sweden. In the United States, where core biopsies were introduced earlier, the Gleason grading system13 has been the governing grading system and is the one supported by the World Health Organization(WHO). Even though these four factors are strong prognostic markers10, 14, they are not strong enough to predict the outcome of individual patients. New and more reliable prognostic markers are needed in determining the course of the disease and in deciding whether or not to treat the patient. Several studies to find new and more reliable markers11, 15,

16 have been performed, but no marker has proven to be a better predictor than tumor stage and grade.

PROSTATE CANCER EPIDEMIOLOGY IN SWEDEN

The incidence rate of prostate cancer has increased dramatically during the last twenty years (Fig. 1)4, 17, 18. The main reason for this is believed to be an increase in diagnostic activity. This view is partly supported by the fact that the incidence of T1c tumors, detected by elevated PSA, has seen the greatest increase. There have been smaller changes in incidence of clinically detected tumors, except for a decline in metastatic disease at diagnosis4. There are surprisingly large geographic differences in prostate cancer incidence; two geographically close counties can report large differences in prostate cancer incidence4, 19. The median age at diagnosis has decreased from 74 to 694. The number of patients treated with curative intent has increased dramatically during last 15 years4, 20. Even though we diagnose and treat more patients with localized tumors the number of men that die of prostate cancer has remained fairly unchanged.4


1111

12

PSA SCREENING

Prostate specific antigen (PSA) is a serine protease produced exclusively by the prostate. All prostatic diseases, including prostate cancer, may increase the serum levels of PSA. Increased levels of PSA in serum may be indicative of prostate cancer21. This knowledge has dramatically changed the management of prostate cancer, and increased PSA testing is the main factor for the enormous increase in prostate cancer incidence in Sweden over the past decades 4, 17, 18. PSA has been proposed as a screening tool for prostate cancer22, and two large clinical trials are ongoing to explore the effect of PSA screening; however, no results have been reported.


1212

13

Fig 1. Change in prostate cancer incidence and prostate cancer mortality in Sweden 1970–2006

0

50

100

150

200

250

300

1972

1975

1978

1981

1984

1987

1990

1993

1996

1999

2002

2005

Inc

ide

ns

pe

r1

00

00

0

Prostate cancer incidence

Prostate cancer mortality


1313

14

TREATMENT OF LOCALIZED PROSTATE CANCER

In the early eighties, patients with localized prostate cancer were usually left without initial treatment. The common opinion was that prostate cancer was an “old man’s” disease for which treatment was withheld until symptoms appeared. The treatment was mainly hormone manipulation, which dates back to late 1800, but which was first really understood in the forties by the work of Huggins23.

Since the early nineties there has been a dramatic shift in the treatment of localized disease4.In 2005, 5347 patients were treated with curative intent in Sweden as compared with 1069 patients in 1998. Curative treatment consists today of surgery or radiation therapy. The first surgical attempt to cure a patient from prostate cancer also dates back to the late 1800s, but it was not until the 1980s that the technique started to develop. In 1990, Dr. Patrick Walsh described in detail the procedure still used today24. This technique improved surgical outcome and quality of life, and has also been shown to improve cancer and overall survival compared to watchful waiting5. In addition, the technique of radiotherapy has improved during the last decade and today external multi-field techniques as well as brachytherapy are used. One reason behind the increase in number of patients treated with curative intent is the introduction of prostate specific antigen (PSA) in 198721. PSA screening has made it possible to find prostate cancer at an earlier stage, where the possibility to cure the patient is better. At the same time, PSA screening has led to considerable over diagnosis of cancers that otherwise would not have come to clinical attention25.

PROSTATE CANCER DETECTED BY TRANSURETHRAL RESECTION (TUR-P)

The clinical significance of incidental prostate cancer, detected by transurethral resection (TUR-P) or by open adenoma enucleation (OAE) for assumed benign hyperplasia, has been a matter of debate for years. One main question has been if these tumors represent a biological domain different from those detected by rectal palpation or by PSA tests. A common opinion has been that these tumors are harmless and could be left without immediate treatment, especially those with volume tumors less than 5% of the total amount of resected tissue (T1a) according to TNM classification 199226. Recent studies27-29 have shown that with increasing tumor volume, greater than 5% of total resected tissue (T1b), the course becomes increasingly unfavorable, comparable to palpable T2 tumors.

The resected tissue originates mainly from the transition zone of the prostate 30, 31. Tumors from this part of the prostate have been described as low grade with low potential of malignancy, and therefore are not believed to need treatment2, 32, 33. On the other hand, 8–37%of all incidental tumours34 ultimately will progress. Bostwick et al.34 suggested that TUR-Pdetected tumors may not exclusively have their origin in the transition zone, but instead might represent peripheral zone tumors that have grown into the transition zone. He further suggested that there may be a significant overlap between T1b, T1c and T2 tumors and that it is merely the location of the tumor in the prostate and the ability of the urologist to find it that ultimately determines the T-stage.Before the PSA era, the urologist had to rely mainly on rectal palpation and serum acid phosphates to evaluate if voiding symptoms were due to a benign condition or prostate cancer. A great proportion of these men with voiding symptoms were subject to TUR-P or


1414

15

OAE. Happaniemi et al. 35 showed that the introduction of TUR-P in the late seventies dramatically increased the number of incidental prostate cancer. The introduction of PSA revolutionized the diagnosis of prostate cancer and the management of men seeking urologists due to voiding symptoms. With the introduction of the PSA test, the frequency of incidental tumors decreased significantly (Kien et al.) 36. Currently in Sweden incidentally detected prostate tumors constitute around 10 % of all diagnosed PCa4.

PROGNOSTIC AND PREDICTIVE FACTORS

In the work of Humphrey, prognosis is defined as 37 “the prediction of future behavior of established malignancy, either in absence of or after application of therapy.” Factors used to predict prognosis fall into two groups38. Factors that provide information on the expected behavior of an untreated tumor are called prognostic factors, while those that provide information about the outcome of active therapy are called treatment-predictive. Factors that could improve the identification of tumors that will progress, or respond favorably to a certain therapy would facilitate choosing between treatment options. New prognostic factors mayalso increase the understanding of the pathogenesis of prostate cancer, and could identify new subgroups of tumors.

In the clinical practice we use prognostic factors on the group and the individual level. At the group level we have a larger cohort of patients whom we, for example, try to stratify into different risk groups. This perspective is often used in trials and in formulating general treatment guidelines. At the individual level, the physician is counseling one patient in making a decision on how to best treat his disease. In this latter situation, the patient is probably less interested to learn that he has a two times greater risk of progression, than in confidently knowing whether or not his tumor will progress to a lethal disease. Given this situation neither the patient nor the doctor is likely willing to take any risks. As a result, most patients opt for curative treatment even though very few of them will really benefit from the treatment5. To assess the value of a prognostic factor we need to calculate its sensitivity and specificity. The sensitivity is the probability that the disease will progress if the marker is positive. Specificity on the other hand is the probability that the prognostic factor is negative if the disease does not progress. The ultimate prognostic factor will have both a high sensitivity and a high specificity, but in certain occasions it may be enough to only have a high specificity or a high sensitivity.

PROGNOSTIC FACTORS USED TODAY

The prognostic factors that are used today for localized prostate cancer have not changed during recent years. The most used clinical prognostic factors are still tumor grade, local tumor stage, serum PSA, age and co-morbidity.

Histological grade

During the 20th century more then 40 different grading systems for prostate cancer were proposed37. Before the introduction of transrectal ultrasound and core biopsies in the late eighties, fine needle aspiration was the main tool used for diagnosis in Sweden. During this


1515

16

era nuclear grade was the grading system used in Sweden12. This is a three-tier grading system based on the morphology of the nuclei. Today tumor differentiation is based on the Gleason grading system39. This system grades the morphologic growth pattern of the tumor and divides them into five categories. The Gleason score is then constructed by adding the value of the most common pattern with the second most common, giving a Gleason score between 2 and 10. This grading system has proven to be a strong prognostic factor for both biochemical recurrence as well as prostate cancer death37.

Local tumor stage

The anatomic local spread of the tumor has been shown to be a strong prognostic factor in many types of cancer40. In prostate cancer this is judged by digital rectal examination7. This method is flawed by several limitations. Variability between observers, as well as the inaccuracy in determining capsule penetration, are the most important limitations, both of which are well documented41. Utilizing more objective staging modalities such as ultrasound, CT or MRI has so far not proven to enhance the accuracy of staging.

Prostate specific antigen (PSA)

Prostate specific antigen (PSA) is a glycoprotein characterized by Wang et al. in 197942. It is almost exclusively produced in the prostate gland. After secretion into the excretory ducts it rapidly forms complexes with alfa-1-antichymotrypsin and to a lesser extent with alfa-2-macroglobulin. Some 10–30% of the circulating PSA remains uncomplexed, so called “free” PSA. The main function of PSA is proteolysis of the semen coagulum to facilitate sperm motility43. Increased levels of PSA may be suggestive of prostate cancer. However, like all diseases affecting the prostate, benign prostatic hyperplasia may give rise to PSA levels44.Studies have also shown a strong correlation with patient age45, tumor volume and prognosis46. The sensitivity and specificity of PSA is low. However, several refinements of PSA screening have been proposed to enhance the predictability of PSA. The free PSA/total PSA ratio has shown to increase the specificity47. The PSA change (PSA velocity) before, as well as after diagnosis has also been suggested to carry prognostic information48. Despite these refinements, PSA screening has limitations for diagnosing of prostate cancer49.However, PSA has still proven to be a useful marker for monitoring the clinical course of the disease.

Age at diagnosis

Younger patients have not been shown to have a more aggressive disease50. However, they do have a greater risk of dying from prostate cancer as compared to older patients.51 This is mainly due to the fact that they have a longer life expectancy and a lower risk of dying of other causes. Thus, age has an impact on disease specific mortality in this slow growing disease. The impact of age is also seen in the results from the SPCG 4 trial5 where patients younger then 65 years seem to have the greatest advantage of surgery.


1616

17

Co-morbidity

Several studies have shown that co-morbidity is a strong prognostic factor in prostate cancer52. This is parallel to the discussion about age at diagnosis. A person with less co-morbidity has longer life expectancy and concurrently a higher risk that the prostate cancer will develop to a more advanced disease.

NEW PROGNOSTIC MARKERS

A number of new prognostic factors, including morphometric, immunophenotypic and genotypic have been discovered and put forward during the last decades15, 16. To prove the utility of these new markers they have to be tested in prospective, randomized, controlled clinical trials. Moreover they need to provide additive and independent value in multivariate analysis, beyond established prognostic factors, such as pathologic stage and Gleason score. So far none of these proposed new prognostic factors has fulfilled these criteria.

Tumor proliferation markers

There is convincing evidence that disturbance in the cell cycle machinery may contribute to the uncontrolled growth characteristic for tumor cells53. Immunohistochemical analysis using MIB-1/Ki67 antibody is very useful for measuring cell cycle progression in human tissue54.Proliferation rate, assessed by Ki-67 immunoreactivity, have further been associated with biochemical recurrence and prostate cancer death55-58.

AMACR

�-Methylacyl CoA racemase (AMACR) is a biomarker that was identified by both differential display and expression array analysis as a gene abundantly expressed in prostate cancer relative to benign prostate epithelium 59-61. In a meta-analysis of 4 cDNA expression array data sets, AMACR was one of the genes most consistently over expressed in prostate cancer62. AMACR is a peroxisomal and mitochondrial enzyme that plays an important role in bile acid biosynthesis and �-oxidation of branched-chain fatty acids through the interconversion of (R)- and (S)-2-methyl-branched-chain fatty acyl-CoA fragments63. Our group initially reported that AMACR expression is consistently lower both at the transcriptional (cDNA expression arrays and RT-PCR) and at the protein level (Western blot analysis and immunohistochemistry) in metastatic prostate cancer compared to localized prostate cancer 59, 64. More recently, a fluorescent-based measurement of AMACR in tissue samples confirmed these observations 65.

MUC-1

The mucin family of anti-adhesion molecules have been implicated in the biological behavior and progression of several types of cancer66, 67. The mucin, MUC-1, which is expressed at the apical cell surface of many normal secretory epithelial cells68, contains an extra cellular


1717

18

domain that extends above most other cell membrane-associated proteins69, 70. As such, MUC-1 has been suggested to prevent adhesion and to promote development of metastatic disease.In prostate cancer, over-expression of MUC-1 in tissue has been correlated with both higher Gleason grade and advanced tumor stage71. One study has, furthermore, suggested that MUC-1 expression may predict prostate cancer recurrence after prostatectomy72, although these results have been challenged by others73, 74. The disparate findings may in part be explained by the use of PSA-recurrence as a measure of outcome, since biochemical failure does not necessarily herald prostate cancer death75 .


1818

19

AIMS

The specific aims of the study were:

to study the natural history of prostate cancer, and evaluate the prognostic value of clinical parameters, in a population based cohort of patients with localized prostate cancer with an observation time of 20-years (paper I).

to study the prognostic value of Gleason score, nuclear grade, tumor extent and tumor proliferation using Ki67 (mib-1) as a proliferation marker among a cohort of 253 patients with untreated prostate cancer detected through TUR-P and with long follow up (paper II).

to study the prognostic value of AMACR expression in prostate tumors, and compare prostate cancer death and biochemical failure as endpoints (paper III).

to study the prognostic value of MUC-1 expression in prostate tumors in a cohort of 253 patients with untreated prostate cancer detected through TUR-P and with long follow up (paper IV).

to study the changes in the incidence and survival of TUR-P detected tumors in Sweden over time (paper V).


1919

21

MATERIAL AND METHODS

Paper I

The subjects comprised a population-based cohort of patients with early, initially untreated prostate cancer has previously been described in detail76. The TNM system of 1978 77 and the World Health Organization78 classification of malignant diseases were used. At the time of diagnosis, all patients underwent a clinical examination, excretory urography, chest radio-graphy, bone scan, skeletal radiography (if needed) and had routine blood samples taken. The nodal status was not known in any of the patients.

From March 1977 through February 1984, a total of 654 new cases of prostate cancer were diagnosed. Patients were given no initial treatment if the tumor growth was localized to the prostate gland. (T0-2) and no distant metastases were present (306 patients). The following restrictions were applied, however, among those with palpable tumors (T1-2). From March 1977 through February 1979, only patients with a highly differentiated tumor (grade I) were included in the untreated group. From March 1979 through the end of the recruitment period, patients under 75 years of age at diagnosis and with moderately or poorly differentiated tumors (grades II–III) were randomly allocated to receive local radiation (10 patients) or no treatment, and only the latter group was included in this cohort study. Patients older than 75 years of age were not treated and included in the study.

Among the 227 eligible patients, 4 (2 percent) were given initial treatment and had to be excluded from the analyses. The distribution of the study group of 223 patients by age, stage and grade at the time of diagnosis is shown in table 1. The mean age at diagnosis was 72 years (range 41–91 years). Altogether 106 (48 percent) were detected by histopathological examinations of specimens obtained at operations for suspected benign prostatic hyperplasia whilst the remaining 117 patients had a palpable clinical disease localized to the prostate gland. About 2/3 of patients had highly differentiated tumors whilst only 9 (4 percent) had a poorly differentiated tumor (Table 1).

All 223 patients were followed up from diagnosis until death or the end of the observation period (September 1st, 2001). No patient was lost to follow-up. Clinical examination, laboratory tests, and bone scans were carried out every six months during the first two years after diagnosis and subsequently once a year during the first 10 years of observation and thereafter at least once every second year. Those in whom the cancer progressed to sympto-matic disease were treated with exogenous estrogens or orchidectomy.

Local progression was defined as tumor growth through the prostate capsule (T3) as judged by digital rectal examination. Development of distant metastasis (M1) was classified as generalization.

During the six first years of follow-up, all patients who were still alive and consented had a new fine-needle biopsy every other year. We obtained such biopsies from altogether 178 of the 223 (80 percent) patients. Although this procedure has lower than 100 percent sensitivity, notably for impalpable tumors, remaining cancer growth was confirmed cytological in the


2121

22

majority of patients. Altogether 31 out of 178 (17 percent) patients showed evidence of dedifferentiation.

The medical records of all deceased patients were reviewed. In most instances, the cause of death was obvious on clinical grounds alone. Prostate cancer was recorded as the underlying cause of death, a contributory cause of death, or unrelated to death as described in detail in a previous report 76. If treatment of the prostate cancer was related to death prostate cancer was recorded as a contributory cause. As a validation, we compared our own classification of causes of death with those recorded in the Swedish Death Register. There was agreement in 90 percent of the patients and no evidence of systematic over or under ascertainment of prostate cancer as cause of death in our data.


2222

23

Table 1. Characteristics of 223 patients with early prostate cancer (T0-2, Nx, M0) who received no initial treatment, according to age, tumor stage and grade at time of diagnosis in 1977-1984. The number of patients with progression of the tumor manifested by local growth (T3) or distant metastases (M1) and the number and causes of death are shown.

Category Total number

T3 M1 Total (%)

Prostatic cancer (%)

Other cause1)

Age, years

-61 13 6 4 6 (46) 3 (23) 3 61-70 86 40 22 46 (53) 19 (22) 59

71-80 96 26 13 29 (30) 12 (12) 79

81- 28 8 0 8 (29) 1 (4) 27

Tumor stage2)

T01 72 14 11 17 (24) 10 (14) 55

T0d 34 14 8 17 (50) 8 (24) 26

T1-2 117 52 20 55 (47) 17 (15) 87

Grade3)

I 148 42 18 45 (30) 14 (9) 118

II 66 35 16 38 (58) 16 (24) 46

III 9 3 5 6 (67) 5 (56) 4

Total (%) 223 80 (36) 39 (17) 89 (40) 35 (16) 168 (75)

1) Three patients died of cardiovascular disease during treatment with estrogens.2) T0 = clinically occult, incidental;

T0 l =T0 pT loc; cancer <25% of the total specimen.T0d = T0pT diff; cancer >25% of the total specimen.

T1-2 = confined to prostate gland.T1 = nodule surrounded by normal prostatic tissue.T2 = large nodule or multiple nodules.T3 = localized to periprostatic area.This represents TNM classification from 1978 5. In the classification from 2000 T1 is considered no evidence of clinical disease and T2 is palpable disease.

3) I = highly, II = moderately, III = poorly differentiated.This refers to the World Health Organization classification of malignant disease 6. It could not directly be compared to Gleason grade and score but in some reports 1 grade I corresponds to Gleason score 2-4, grad II to Gleason 5-7 and grade III to Gleason score 8-10.


2323

24

Paper II–IV

The University Hospital in Örebro (USÖ) has a defined catchment area of about 190, 000 inhabitants. All patients with voiding symptoms and suspicion of prostate cancer were referred to the urological department at the University Hospital in Örebro. From March 1977through September 1991, a total of 1,230 patients were diagnosed with prostate cancer in the area. Of these, 252 patients were diagnosed through transurethral resection of the prostate (TUR-P) or transvesical adenoma enucleation(OAE). Histopathological material was missing in three patients and in nine patients there were no cancer found when reviewed by the pathologist. Thus, 240 subjects were included in the analyses, although tissue was only available for tissue micro array in 190 patients. Only patients without skeletal metastases at diagnosis were included in this study. The baseline evaluation at diagnosis included physical examination, chest radiography, bone scan and skeletal radiography (if needed). For staging and grading of the tumors the TNM-classification from 199226 were used. No initial treatment was offered to these patients with incidental prostate cancer. If the tumors progressed to symptomatic disease the patients were treated hormonally. The patients were followed until death from prostate cancer, or censored at time of other death or until end of observation period in September 2003. No patient was lost to follow-up. The mean follow-up period was 108 months (range 1 to 280 months). Follow up and cause of death was defined as in paper one.

In the analysis of AMACR we also used a cohort of patients with localized disease. This cohort consisted of 204 patients, who underwent radical retro pubic prostatectomy as a primary therapy. Clinical data regarding this cohort has been separately reported79, 80, and a summary of the patient demographics is presented in table 1. Disease progression was defined as a serum PSA increase to 0.2 ng./ml, or above after radical prostatectomy

Tumor specimens were fixed in buffered formalin and embedded in paraffin and the resected tissue chips were randomly divided and placed in 6 paraffin blocks if enough material, otherwise in as many as possible. Examination and Gleason grading39 was performed by one pathologist (MR). Grading according to WHO classification12 was also preformed by one pathologist (BR). We further assessed tumor volume by calculation of the ratio between the number of chips with cancer and the total number of chips (hereafter referred to as the ratio)8.The tissue micro arrays were assembled using the manual tissue arrayer (Beecher Instrument, Silver Spring, MD) as previously described81. A 4 �m section was cut and stained with standard biotin-avidin complex immunohistochemistry with MIB antibodies to evaluate Ki67 and (p504s, Zeta Co., Ca.) to evaluate AMACR and for MUC-1 (Mucin 1(VU4H5): sc 73–13) was used. Semi-automated, quantitative immunohistochemistry was undertaken using the Chromavision system, and protein intensity was measured on a scale from 0–255. In the AMACR study also a manual evaluation of the staining was preformed, by two study pathologists using a categorical scoring method ranging from negative to strong staining intensity as previously reported59.


2424

25

Paper V

In this study, we utilized nationwide data from several swedish health care registers. The use of the national registration number (NRN), a unique 10 digit identification number assigned to all Swedish residents, enables record linkages between the registers. There was almost no private institutional care available in Sweden during the study period, so essentially all men were referred to and treated at the main hospital in their county of residence and, therefore, the study can be considered population-based.

The National Inpatient Register

For primarily administrative purposes the National Board of Health and Welfare started to compile data for the National Inpatient Register in 1964 to capture hospitalizations. Registration increased over the years and was 60% in 1969, 85% in 1983, and in 1987 the coverage reached 100%. This register includes data on the NRN, discharge diagnoses (a maximum of 8 including up to 6 surgical procedures) for each hospitalization, and date of discharge. The 7th revision of the International Classification of Diseases (ICD-7) was used for coding diagnoses through 1968, the 8th revision (ICD-8) between 1969 and 1986, the 9th

revision (ICD-9) between 1987 and 1996, and the 10th revision (ICD-10) thereafter. Surgical procedures were coded according to the Swedish Classification of Operations and Major Procedures.

The Swedish Cancer Register and Cause of Death Register

The Swedish Cancer Registry captures all incident cancers in the population, including PCa. Mandatory reporting by clinicians and pathologists has contributed to the thoroughness of the Swedish Cancer Register and the coverage has exceeded 97% since 198482. The Cause of Death register was established in 1952, and the registries has captured virtually all deaths in Sweden since 1961. The most significant variables in this register are the main and contributory causes of death, date of death, notes on section, sex, age, and birthplace. Cancer diagnoses and causes of death are coded according to the International Classification of diseases and causes of deaths (ICD 7-10).

The Register of Population Changes

This register is based on the official Swedish Census data and includes all residents alive and living in Sweden at the end of each year.

Methods

We identified all men (N=76,778) in the Swedish Inpatient Register who had been discharged after TURP or OAE between 1970 and 2003. By linking the records of these men to the Cancer Registry, we identified all men who were incidentally diagnosed with PCa by TURP or OAE. The definition of incidentally detected PCa required the following criteria to be fulfilled: (1) the date of the PCa diagnosis had to follow the date of first TURP/OA admission, (2) the date of the PCa diagnosis had to be set within 14 days from discharge, (3) the diagnosis of PCa had to be histopathologically verified, and (4) the hospital stay had to be less than 60 days. A diagnosis of high-grade intraprostatic neoplasia (PIN) was not classified as PCa. In total 23,288 men with incidentally diagnosed PCa was identified.


2525

26

As a comparison group, we chose all men clinically diagnosed with PCa in the Swedish Cancer Register during the same time period and from the same counties. All men with PCa in the comparison group (N=112,204) were diagnosed on clinical grounds alone (i.e. by tissue biopsies or fine needle biopsies/cytology). By linking the records of these men to the records of those discharged for TURP/OAE we were able to separate out men with histologically confirmed PCa before surgery and men diagnosed later than 14 days after the TURP/OAE. We further excluded all autopsy detected PCa. In total, the study population, including those with both incidentally and clinically diagnosed PCa, comprised 135,492 men.


2626

27

STATISTICAL METHODS

Paper I

We estimated various measures of patient survival using the actuarial (life-table) method83.Cause-specific survival was estimated by considering only deaths due to prostate cancer as events of interest (deaths due to other causes were considered censored), observed survival by considering deaths due to any cause as events, and progression-free survival by considering progression as the event of interest. Relative survival was estimated, defined as the ratio of observed survival to the expected survival of a comparable group from the general population assumed to be free of prostate cancer. We estimated expected survival using the Hakulinen method84 based on Swedish population life tables stratified by age, gender, and calendar time. Prostate cancer-specific mortality rates was also calculated (deaths per 1,000 person years at risk) and associated 95 percent confidence intervals85. We estimated Poisson regression models85 to study the association between prostate cancer mortality and time since diagnosis while adjusting for age at diagnosis, stage, and grade. Relative survival was estimated using software developed at the Finnish Cancer Registry86. All other analyses were performed using State (Stata Corporation, College Station, Texas). All reported p-values are two-sided.

Paper II and IV

We used Cox regression models to examine the risk of dying in prostate cancer as a function of the various prognostic markers. The values of Ki-67 intensity were divided into quintiles based on the distribution among the patients, and the lowest quintile was used as the reference.

MUC-1 intensity was defined normal as the mean intensity in benign prostate tissue ± 0.25.standard deviations. Individuals whose MUC-1 tumor expression was within the normal range represented the reference group. Individuals whose tumor expression was above or below normal were so classified.

Relative hazard ratios were calculated before and after adjustment for all included variables. We tested for linear trend by constructing ordinal variables through assigning consecutive integers to consecutive levels of categorized variables. The Pearson correlation test and the chi-square test were applied to formally assess correlation.

Paper III

AMACR intensity readings were obtained for each of the TMA slides separately and were then normalized within each array before combining the data for analysis. Cut points for the AMACR intensity scores, were determined by determining the cut point that best differentiated PSA biochemical failure in the 204 patients from the surgical series. Kaplan-Meier estimates of the survival probabilities were computed and p-values from log-rank tests were obtained to assess the discriminative power of the dichotomized marker. The optimal cut point of dichotomization was then chosen to minimize the log-rank test p-values. A similar


2727

28

process was repeated for the Örebro watchful waiting cohort (n=240 cases) using cancer specific death as the endpoint.

We tested the optimal cut point derived using the surrogate endpoint (PSA failure) on the watchful waiting cohort to determine if it would predict a true endpoint (prostate cancer specific death). The cut point was then tested using prostate cancer specific death as the endpoint on the surgical series to see if it would predict PSA biochemical failure. Cox Hazards regression analysis was used to develop multivariate models taking into account other clinical parameters.

Investigators were blinded to all clinical outcome data. Statistical analysis for the surgical series and watchful waiting series were performed at the Statistical centers in Michigan and Sweden, respectively.

Paper V

The age-standardized incidence was calculated using the age distribution of the Swedish population on January 1, 2000; data were obtained from Statistics Sweden87. We calculated cumulative incidence curves to illustrate the risk of prostate cancer death while accounting for the competing risk of death from other causes88. All statistical calculations were performed using the statistical program package R89.


2828

29

RESULTS

Paper I

During a mean observation period of 21 years, 89 patients (40 percent) experienced progression of disease and of these 39 (17 percent of the entire cohort) developed generalized disease. A total of 203 patients (91 percent of the entire cohort) died during follow-up with prostate cancer considered the cause of death in 35 (16 percent of the entire cohort) (Table 1). Among patients who were 70 years or younger at diagnosis, 22 percent died from prostate cancer during follow-up whilst this proportion decreased markedly at higher ages. The proportion of patients dying from prostate cancer was strikingly similar among those with non-palpable (T0) tumors detected at transurethral resection (17 percent) and those with a palpable tumor (15 percent). In contrast, poor differentiation was a strong predictor of prostate cancer specific death (Table 1).

Although based on small numbers, the progression and mortality rates remained fairly constant during the first three 5-year periods following diagnosis (Table 2). Averaged over the first 15 years, the rate of progression to metastatic disease was 18 per 1000 person years (95 percent confidence interval: 13–25) and the prostate cancer mortality rate was 15 per 1000 person-years (95 percent confidence interval: 10–21). In contrast, an approximately three fold higher rate was found both for progression and death during follow-up beyond 15 years (Table 2). This increase was almost significant for progression (p=0.06) and statistically significant for death (p=0.01). Table 3 shows various measures of survival after 15 and 20 years of follow-up, respectively. During this 5-year period, the progression free survival decreased from 45.0 to 36.0 percent. The low and rapidly decreasing observed survival reflects chiefly the impact of causes of death other than prostate cancer. Most notably, however, we found a substantial decline by about 25 percentage points both in the relative and in the cause-specific-survival rate during the 5 years of follow-up. Figure 2 further illustrates how a gradual decline in relative and cause specific-survival seemingly dropped more rapidly after approximately 16 years of follow-up. This change seemed to affect tumors regardless of initial stage (Figure 3) and also to affect both tumors that were initially highly and moderately differentiated (Figure 4). The gloomy outlook among patients with poorly differentiated tumors became manifested already within the first 5 years of follow-up.

Prostate cancer mortality was slightly higher among patients diagnosed at age 70 or younger than among those diagnosed at older ages (Table 2). Strikingly similar mortality rates were found among patients who had localized non-palpable cancer compared with those who had a cancer in stage T1 or T2. In contrast, the mortality rate was 70 percent higher among men with a non-palpable diffuse cancer. With regard to differentiation, the mortality rate increased drastically from highly to poorly differentiated tumors (Table 2). Multivariable Poisson regression models were fitted in order to quantify the independent effects of follow-up time, age at diagnosis, grade and stage (Table 4). Our analyses showed a significant, about six fold, higher mortality rate after 15 years of follow-up compared with the first five years. The strong prognostic impact of grade, notably of poorly differentiated tumors, was also confirmed. In contrast, neither age at diagnosis nor stage of disease was significantly associated with risk of death due to prostate cancer. A separate model was fitted in which the event of interest was local progression. In this analysis we disregarded if and when regional and/or distant progression/metastases were ascertained. Except for age at diagnosis, the pattern for local


2929

30

progression was strikingly different from that of death due to prostate cancer (Table 4). Hence, the risk of local progression did not increase over follow-up time, and the association with grade was weak. Moreover, compared with T0l tumors, growth beyond the prostate capsule was two or three times more likely in patients with T0d and T1-2 tumors, respectively.


3030

31

Table 2. Rates per 1000 person-years with 95 percent confidence interval (CI) for progression to metastatic disease and death due to prostate cancer by years of follow-up, age, stage and grade at diagnosis

Progression Prostate cancer deathCategory Events Rate 95% CI Events Rate 95% CI

Follow-up 1)

0-4 18 20 13 - 32 11 12 7 - 225-9 9 15 8 - 30 11 18 10 - 3310-14 5 16 7 - 38 5 15 6 - 3615+ 7 41 20 - 87 8 44 22 - 88

Age, yrs- 70 26 24 16 - 35 22 19 13 - 29

71 - 13 15 9 - 26 13 15 8 – 25

Stage T01T0d

T1-2

11820

173119

9 - 30 15 - 62 12 - 30

10 817

152816

8 - 2814 - 5710 - 25

Grade III

III

18165

1329242

8 - 21 18 - 47101 - 581

1416 5

10 27194

6 - 1717 - 4481 - 466

Total 39 20 14-27 35 17 12-24

1) Numbers of patients alive after 5, 10 and 15 years of follow-up were 150, 91 and 49 respectively.


3131

32

Tab

le 3. Th

e 15 and

20 year progression

-free survival (P

FS

), observed

survival (O

S), relative su

rvival (RS

) and

cause-sp

ecific su

rvival (CS

) rates with

95% con

fiden

ce intervals (w

ithin

brack

ets), by stage an

d grad

e at diagn

osis.P

FS

OS

RS

CS

Category

15-year20-year

15-year20-year

15-year20-year

15-year20-year

Stage

T0 1

57.752.5

22.27.5

82.156.1

81.157.9

(38.9 – 76.5)(32.7 – 72.3)

(12.4 – 32,0)(-0.1 – 15.0)

(45.9 – 118.4)(0.6 – 112.8)

(66.5 – 95.6)(28.3 – 87.5)

T0 d

35.917.9

14.70

63.60

69.746.4

(15.1 – 56.5)(-9.5 – 45.3)

(2.6 – 26.8)

(11.1 – 116.1)

(50.1 – 89.2(6.3 – 86.5)

T1-2

38.433.1

23.110.3

75.261.5

80.356.9

(26.8 – 50.1)(20.9 – 45.4)

(15.3 – 30.9)(4.2 – 16.3)

(49.8 – 100.6)(25.2 – 97.9)

(69.8– 90.8)

(34.9 – 78.9)

Grade

I56.0

46.024.3

9.783.7

63.488.9

71.8(44.4 – 67.6)

(30.9 – 61.0)(17.2 – 31.4)

(4.3 – 15.1)(59.4 – 108.0)

(28.1 – 98.8)(81.4 – 96.3)

(54.9 – 88.7) II

28.824.3

18.23.5

64.723.4

64.522.1

(15.2 – 42.3)(10.2 – 38.4)

(8.7 – 27.7)(-1.8 – 8.7)

(30.9 – 98.4)(-12.0 – 58.7)

(47.2 – 81.8)(-7.5 – 51.7)

III

All

15.6*(-12.8 – 43.9)

45.036.0

0

21.5

07.5

0

75.9

0

49.9

28.6*(5.6 – 62.7)

78.754.5

Patients

(35.7 – 54.3)(24.2 – 47.9)

(16.0– 27.0)

(3.5 – 11.4)(56.5 – 95.3)

(23.4 – 76.4)(70.8 – 86.7)

(37.6 – 71.4)

Fo

otn

ote

: * afte

r 7 y

ear.


3232

33

Table 4. Multivariable* relative risks (RR) with 95% confidence intervals (CI) of death from prostate cancer and of local progression (growth through the prostate capsule) in relation to follow-up time, age at diagnosis, tumor grade and tumor stage.

Death Local Progression

Category RR 95% CI RR 95% CI

Follow-up

0-4 1.0 Ref 1.0 Ref

5-9 2.2 0.9-5.5 0.6 0.3-1.1

10-14 2.0 0.6-6.3 0.6 0.3-1.3

15+ 6.4 2.3-17.8 0.8 0.3-2.1

Age at diagnosis

<70 1.0 Ref 1.0 Ref

70+ 0.7 0.3-1.6 0.9 0.6-1.5

Grade

I 1.0 Ref 1.0 Ref

II 3.4 1.6-7.3 2.5 1.6-4.0

III 46.6 12.3-177.4 3.3 0.9-11.9

Stage

T01 1.0 Ref 1.0 Ref

T0d 0.7 0.2-2.1 2.0 0.9-4.4

T1-2 0.7 0.3-1.6 2.7 1.5-4.9

* Estimated using Poisson regression where each factor is simultaneously adjusted for all other factors.


3333

34

Figure 2. Observed survival (OS), cause-specific

survival (CS), and relative survival (RS) in a cohort of

223 patients with prostate cancer.


3434

35

Figure 3. Cause-specific survival by stage of disease at diagnosis:

T0 loc, T0 diff and T1+T2.


3535

36

Figure 4. Cause-specific survival by tumor grade at time of diagnosis:

Grade I, highly


3636

37

Paper II

The baseline characteristics of the 240 patients and distribution according to tumor stage, grade of differentiation, Gleason score, Ki67 expression and tumor volume are shown in table 5. Of the 240 patients 118 (49.2%) patients were in stage T1a and 122 (50.8%) in stage T1b. The mean age at diagnosis was 73 years. Eighty-four percent (n=167) of the cases had either Gleason score 6 or 7 and only 4% (n=11) had a Gleason score below 6. 175 patients (74%) had nuclear grade 1. By the end of the observation period in September 2003, a total of 207 of the 240 patients (86%) (Table 5) had died. Prostate cancer was a more common cause of death in patients 70 years of age and younger in comparison with those over 70 years of age.

The age-adjusted risk of dying in prostate cancer with respect to grade of differentiation, Ki67, Gleason score and tumor volume is shown in table 2 (Table 6). The risks tended to rise with increasing Gleason score, nuclear grade and tumor volume (p for trend<0.05). There was a 9-fold statistically significant increased risk for those with the highest Gleason score (8–9) compared to those with the lowest (4–6). Among patients with Gleason score 7 a Gleason 3 predominant pattern (3+4) was not significantly different from a Gleason 4 predominant pattern (4+3) (data not shown). Nuclear grade 3 tumors hade an almost eleven fold higher risk to die in prostate cancer compared with tumors with nuclear grade 1. After adjustment, cases with T1b tumors (ratio > 5%) were 2.5 times more likely to die of their disease than cases with T1a tumors (HR=2.5; 95% CI, 1.1–5.7).

In the multivariate model Gleason score, nuclear grade and tumor volume remained statistically significant independent prognostic markers of prostate cancer death. Ki67 was highly correlated with Gleason grade, and the trend of increasing mortality rates with rising values of Ki67, that was observed in the univariate analysis, disappeared after adjustment for the other variables.

The sensitivity and specificity of the Gleason grading system at different cut-off points is presented in table 7. Setting the cut-off value to �7 resulted in a sensitivity of 0.64 and a specificity of 0.66. As the cut-off was increased by one step, the corresponding values fell to 0.33 and increased to 0.92, respectively. Moreover, the positive predictive value of Gleason �7 was 0.29 and the negative predictive value 0.89. With a cut-off of 8, the corresponding values were 0.48 and 0.87, respectively.


3737

38

Table 5. Characteristics of 240 patients with incidental prostate cancer (T1a-b, Nx, M0) who received no initial treatment, according to age, nuclear grade, Gleason score, tumor volume and Ki67 expression at time of diagnosis.

TotalN

Prostate cancer deaths

Deaths by

other cause

Alive MeanSurvival(Month)

MinimumSurvival(Month)

MaximumSurvival(Month)

Age < 70 80 16 42 22 138 16 280 > 70 160 26 123 11 90 1 272

Nuclear-grade* 1 175 20 124 31 98 1 280 2 45 13 31 1 94 4 229 3 16 8 7 1 59 1 158

Gleason Score 4 4 - 2 2 132 56 167 5 7 3 3 1 122 5 235 6 136 12 99 25 118 1 284 7 64 13 46 5 96 1 230 8 26 13 13 0 73 4 229 9 3 1 2 0 49 17 85

Ratio chips with cancer (%)** �5 118 9 87 22 102 1 237 6-25 92 20 61 11 97 1 280 26-50 13 6 7 0 85 17 153 >50 17 7 10 0 54 4 143

Ki67 (mean) Q1 (101) 40 6 24 10 123 6 280 Q2 (112) 36 5 26 5 105 10 273 Q3 (117) 39 10 26 3 94 1 228 Q4 (122) 37 7 26 4 96 4 212 Q5 (129) 38 9 25 4 98 8 211*Only 236 patients where graded according to nuclear grade.** T1a has a ratio � 5% T1b has a ratio > 5%


3838

39

Table 6. Hazard Ratio (HR) and 95% Confidence Interval (CI) of death from prostate cancer in relation to age, nuclear grade, Gleason score, tumour volume and Ki67 expression.

HRa 95 % CI HRb 95% CIAge <70 1.0 1.0 >70 1.5 0.8-2.8 1.4 0.6-3.2

Nuclear grade I 1.0 1.0 II 3.5 1.7-7.1 1.4 0.5-3.5 III 10.7 4.5-25.4 3.4 1.1-10.7p ( trend) 0.03

Gleason score 4-6 1.0 1.0 7 2.5 1.2-5.3 1.5 0.5-4.2 8-9 9.5 4.5-20.2 3.3 1.0-11.6p ( trend) 0.04

Ratio chips with cancer (%) �5 1.0 1.0 6-25 3.0 1.3-6.6 1.9 0.6-5.8 25-50 10.3 3.5-30.5 3.6 0.8-15.7 50 16.8 6.0-47.5 5.2 1.2-22.8p(trend) <0.01

Ki67 Q1 (medelvärde för varje Q) 1.0 1.0 Q2 1.2 0.4-4.2 1.2 0.3-4.0 Q3 2.6 0.9-7.6 1.4 0.4-4.9 Q4 1.9 0.6-6.0 0.9 0.2-3.2 Q5p(trend)

2.5 0.9-7.4 1.3 0.4-4.20.86

aage-adjusted badjusted for all other variables

Table 7. Sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) of Gleason grading in predicting prostate cancer death.

Gleason Score* Sensitivity Specificity PPV NPV

� 6 0.92 0.04 0.17 0.73

� 7 0.64 0.66 0.29 0.89

� 8 0.33 0.92 0.48 0.87

*cut-off point


3939

40

Paper III

AMACR protein expression was evaluated manually by the study pathologist and graded on a 4-tiered scale, demonstrated a significant difference in intensity between prostate cancer (mean score=3.14/4) and benign prostate epithelium (mean score=1.3/4). Using these categorical scores for AMACR, no significant associations between AMACR intensity and biochemical failure were observed in the surgical series consistent with previous observations59.

Using the semi outomated analysis, in the surgical series using biochemical failure as the endpoint, lower AMACR intensity was associated with a worse outcome (Tables 8 and 9). We established a dichotomous cut point for AMACR intensity of –1.11 SD (i.e. samples with minimum AMACR intensity of –1.11 SD below that of mean), wherein, 37,5% of patients with AMACR intensity scores below the cut point progressed as compared to 14,5% ofpatients with AMACR intensity scores above the cut point (P=0.0002). This univariate association can be visually appreciated by Kaplan-Meier analysis (Figure 5). Using this AMACR cut point in multivariate analysis, patients with AMACR expression levels below the threshold were at a significantly higher risk of developing PSA recurrence (HR=2.12; 95% CI 1.04–4.32) after adjusting for pre-operative PSA, Gleason score and surgical margin status.

Doing the same procedure in the watchfull waiting cohort using prostate cancer death as endpoint. We determined an optimal cutpoint of 0.18 SD (HR=4.06; 95% CI 1.82–9.06). As demonstrated in table 10 and figure 8, a significant cut point was identified in multivariable analysis that takes patient age, Gleason score, and tumor stage into account. Although this cut point is different from the one derived using biochemical failure, we also observe that lower AMACR intensity is associated with worse outcome, in this case cancer specific death.

Using the surgical cutpoint –1.1SD, AMACR intensity did not predict prostate cancer specific survival. In contrast, there was evidence that AMACR levels below the cutpoint determined by prostate cancer death was significantly associated with time to biochemical failure at the univariate level as demonstrated by the Kaplan-Meier analysis (Figure 6). On multivariate analysis, AMACR expression, Gleason score, surgical margin status, and pre-treatment PSA were all independently associated with PSA biochemical failure (HR= 3.67; 95% CI 1.25–10.78).


4040

41

Table 8. Univariate Cox Regression Analysis for Associations between AMACR intensity and Prostate Specific Biochemical FailureFactor P-value Hazard Ratio 95% CI for

HRAMACR(min>-1.11 vs. <=-1.11) 0.0003 0.34 0.19 0.60

Gleason Score �7 vs. <7) 0.0003 4.36 1.96 9.73

SM (Positive vs. Negative) <.0001 3.61 2.26 5.76

Path stage <.0001 3.01 2.09 4.34

ln(PSA) <.0001 2.41 1.72 3.38

Tumor Size (>2 vs.�2 cm) <.0001 3.49 1.91 6.37

DRE (Positive vs. Negative) 0.03 1.90 1.07 3.36

Gland Weight 0.94 1.00 0.99 1.02

Age 0.61 1.01 0.97 1.05

EPE <.0001 2.79 1.97 3.94

Table 9. Multivariate Cox model: Independent associations of AMACR,Gleason score, PSA and positive margin with PSA Recurrence after Radical Prostatectomy

Parameter P-valueHazard Ratio

95% CI for HR

AMACR(min>-1.11 vs. <=-1.11) 0.001 0.36 0.20 0.66

Gleason(>=7 vs. <=6) 0.04 2.35 1.02 5.40

SM (Positive vs. Negative) 0.001 2.78 1.53 5.03

ln(PSA) 0.001 1.89 1.31 2.74


4141

42

Tabel 10.

Note: D

ata in multivariate m

odel are adjusted for AM

AC

R cutpoint, age, T

stage and Gleason score

Tab

el 10. Un

ivariate and

mu

ltivariate Cox m

odels: association

betw

een A

MA

CR

and

prostate can

cer-sp

ecific death

in w

atchfu

ll waitin

g cohort, Ö

rebro 1977-2003

Variable

Censored

Pc-death

Unadjusted

Multivariate adjusted

(n=152)

(n=36)

HR

95% C

IP

HR

95% C

IP

AM

AC

R<

0.18 (low)

60.169.4

1.400.68-2.83

0.364.06

1.82-9.060.0006

>0.18 (high)

39.930.6

RE

FR

EF

Age (per year)

74.1y72.5y

1.030.99-1.08

0.171.01

0.97-1.060.58

T stage

T1a

46.113.9

5.121.98-13.21

0.00073.61

1.30-9.970.013

T1b

53.9.86.1

RE

FR

EF

Geason score

8-108.5

33.410.45

4.55-24.0<

0.000113.03

4.88-34.8<

0.00017

26.933.3

2.841.27-6.38

0.0112.70

1.12-6.520.027

5-662.7

33.3R

EF

RE

F


4242

43

Figur 5.

Fig 6.


4343

44

Paper IV

The baseline characteristics of the 195 patients are presented in table 11. Mean MUC-1expression for all 195 patients was 107.3 (range 95–179, SD 10.2). Of the 43 patients that had a MUC-1 intensity close to normal tissue (102.5–106) (Fig. 7 A-B) three (7%) died of prostate cancer, compared with 34 (23%) of the 152 patients that deviated from the normal MUC-1intensity (Fig. 7 C-D). As illustrated in table 12, there was no correlation between Gleason score and MUC-1 intensity (p-value 0.8), while there was a tendency to correlation between tumor extent and MUC-1 (p-value 0.08). The age-adjusted risk of dying of prostate cancer with respect to MUC-1, Gleason score and tumor extent is presented in table 12. The risk of dying of prostate cancer was four times higher among those with a higher [HR 3.9 (95% CI, 1.1–14)] or lower [HR 3.8 (95%CI, 1.1–13)] MUC-1 expression than among those with a MUC-1expression within the normal range (Fig 8). After adjusting for tumor extent and Gleason score, the effect of MUC-1 was even stronger [HR 5.1(95%CI, 1.4–18)] and [HR 4.5 (95%CI, 1.3–15)], respectively indicating that MUC-1 predicts prostate cancer death independently of clinical parameters (Fig 9).

We further cross-classified participants on MUC-1 and Gleason score. The group with Gleason score � 7 and MUC-1 lower or higher then normal had a 17 [HR 17.1 (95%CI, 2.3–128)] times higher risk of prostate cancer death compared with tumors with Gleason score < 7 and normal MUC-1 intensity (Table 14). We further assessed the ability of MUC-1 expression (deviating from the normal range) to correctly classify prostate cancer cases as indolent or lethal (defined as progressing to metastases and/or death). The sensitivity for MUC-1 as predictor of lethal prostate cancer was 0.91 while the specificity was 0.25 (Table 15). When combined with information on Gleason score, the specificity increased to 0.75 but the sensitivity decreased to 0.56.


4444

45

Table 11. Characteristics of 195 patients with incidental prostate cancer (T1a-b, Nx,M0) who received no initial treatment, according to age, Gleason score, tumor extent, and MUC-1expression at time of diagnosis 1977-1991

N Prostate cancer deaths

Deaths by other

causes

Alive Meansurvival

Minimumsurvival

Maximumsurvival

Age <70 56 13 27 16 138 16 280 > 70 139 24 105 10 90 1 272Gleason score 4 3 - 1 2 132 56 167 5 7 3 3 1 122 5 235 6 107 10 77 20 118 1 284 7 53 12 38 3 96 1 230 8 22 11 11 0 73 4 229 9 3 1 2 0 49 17 85Percentage of chips with cancer <5% 80 6 58 16 102 1 237 6-25% 85 18 57 10 97 1 280 26-50% 13 6 7 0 85 17 153 >50% 17 7 10 0 54 4 143MUC-1 Normal 43 3 35 5 118 4 284 Low 68 16 38 14 108 1 280 High 84 18 59 7 100 1 238

Table 12. Correlation between MUC-1 intensity expression and age, Gleason score and tumor volume

MUC-1 intensityFactor Normal

N (%)Low

N (%)High

N (%)p-value

Age<70 10 (23) 25 (37) 21 (25)Age>70 33 (77) 43 (63) 63 (75)

0.188

Gleason 4-6 25 (58) 41 (60) 51(61)Gleason 7 13 (30) 20 (29) 20 (24)Gleason 8-9 5 (12) 7 (10) 13(16)

0.826

Percent chips< 5% 15 (35) 23 (34) 42 (50)Percent chips 5-25% 23 (54) 33 (49) 29 (35)Percent chips 25-50% 0 (0) 5 (7) 8 (10)Percent chips >50% 5 (12) 7 (10) 5 (6)

0.085


4545

46

Table 13. Hazard Ratio (HR) and 95% Confidence Interval (CI) for prostate cancer death in relation to protein expression of MUC-1 in tumor tissue from patients with localized prostate cancer.

Na Prostate cancer deaths

Crude HR (95% CI)

HRb (95% CI)

MUC-1 (intensity) Normal (102.5-106) 43 3 1.0 1.0 Low (<102.5) 68 16 3.9 (1.1-14) 5.1 (1.4-18) High (>105.5) 84 18 3.8 (1.1-13) 4.5 (1.3-15)a A total of 195 patients were assayed for MUC-1. b Adjusted for age, Gleason score, and tumor extent.

Table 14. Hazard ratio (95% CI) of prostate cancer death associated with MUC-intensity and Gleason score, cross classified.

Normal MUC-intensity Aberrant MUC-intensityPC death/ Total N

HR (95% CI)PC death/ Total N

HR (95% CI)Gleason < 7 1/25

Ref.12/92

3.8 (0.5-29)Gleason � 7 2/18

3.8(0.3-43)22/60

17.1 (2.3-128)

Table 15. Sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) of Gleason grade and MUC-1 intensity in predicting prostate cancer death.Gleason Score and MUC-1intensity

Sensitivity Specificity PPV NPV

Aberrant MUC-1 intensity 0.91 0.25 0.22 0.93

Gleason >6 and Aberrant MUC-1 intensity

0.59 0.76 0.37 0.89


4646

47

Figure 7. Tissue micro array analysis of MUC-1 immunohistochemistry:

selected images of TMA cores representing normal, high, and low MUC-1

intensity

Fig A, B. Normal MUC-1 intensity

Fig C, D. High (C) and Low (D) MUC-1 intensity

A B

C D


4747

48

Figure 8.


4848

49


Figure 9

4949

50

Paper V

Background data obtained through computerized registry linkage are shown in table 16. Most patients with incidental PCa are older patients, about 50% are 75 years or older (mean age 74.4) and somewhat older than patients in the comparison group, i.e. men diagnosed on clinical grounds alone. The mean age at incidental PC diagnosis increased slightly over time from 72.5 in 1970–79 to 75.2 years in the last calendar period 1999–2003. The age standardized incidence of incidental PCa peaked in the early nineties (Fig 10) and successively declined during subsequent years, reaching 12% during the period of 1999–2003. We observed a marked change in hospitalization time over calendar time. Initially, in the early seventies, when the transurethral technique was introduced, the hospitalization time was comparably long, lasting about 2–3 weeks. After a continuously and dramatic decrease the hospitalization time leveled off at 4–5 days during the last 5-year period. During the entire study period, 1970–2003, the age standardized PCa incidence rate raised continuously and increased almost 15 times from 1973 to the mid eighties. The upward trend appeared to be driven by the increased frequency of TUR-P, but the incidence rate rose continuously even after the TUR-P frequency had leveled off.

At the end of 2003, 6300 patients (27.1%) with incidental PCa had died due to PCa and a majority of them (50,9%) had died of intercurrent causes, while 5138 patients (22,1%) were still alive. A somewhat higher death rate due to PCa was found in the comparison group consisting of clinical detected PCa (33.2%). The age standardized cumulative incidence of PCa death is demonstrated in fig 11. The corresponding curves by time period and by mode of detection are shown in fig 12. We observed a clear decreasing trend over time after diagnosis, but the prostate cancer specific mortality decreased both according to mode of detection and calendar time. The mortality was higher among those with incidentally detected tumors compared to those in the comparison group. However, it should be noted that although the prostate cancer specific mortality among those with incidental PCa leveled off by time, it was still 30% after 15 years of follow up after diagnosis.


5151

49

Figur 10. Age-standardized prostate cancer incidence and relation to history of TUR-P, in Sweden between 1970 and 2003.

1970 1975 1980 1985 1990 1995 2000 2005

Year

0

50

100

150

200

250

Age

sta

ndar

dize

d in

cide

nce

of

pro

stat

e ca

ncer

(pe

r 10

0 00

0 m

en)

TURP-detected

Not TURP detected

All PC


5052

Tabel 16. Characteristics of 23288 prostate cancer patients diagnosed through or hospitalized for TUR-P between 1970 and 2003 in Sweden

TURP-hospitalisations

TURP-diagnosed patients PC-comparisons

N % N % N %

Age

<55 708 (0.9) 123 (0.5) 2462 (2.2)

55-64 8931 (11.6) 2157 (9.3) 17740 (15.8)

65-74 30854 (40.2) 9085 (39.) 43060 (38.4)

75-84 31163 (40.6) 10114 (43.4) 40769 (36.3)

85+ 5122 (6.7) 1809 (7.8) 8173 (7.3)

Years

1970-79 12324 (16.1) 2627 (11.3) 13045 (11.6)

1980-86 20793 (27.1) 6693 (28.7) 17116 (15.3)

1987-90 14409 (18.8) 4961 (21.3) 13311 (11.9)

1991-93 10106 (13.2) 3349 (14.4) 12480 (11.1)

1994-98 9790 (12.8) 2795 (12.0) 20270 (18.1)

1999-2003 9356 (12.2) 2863 (12.3) 35982 (32.1)

Hospitalisation time

0-3 days 17978 (23.4) 5680 (24.4) NA NA

4-7 days 34758 (45.3) 10997 (47.2) NA NA

8-14 days 14964 (19.5) 4175 (17.9) NA NA

2-4 weeks 6655 (8.7) 1844 (7.9) NA NA

4+ weeks 2423 (3.2) 592 (2.5) NA NA

Nr TURP-treatments

0 NA NA 0 (0.0) 82389 (73.4)

1 NA NA 18007 (77.3) 23689 (21.1)

2 NA NA 4118 (17.7) 4719 (4.2)

3 NA NA 864 (3.7) 1062 (0.9)

4+ NA NA 299 (1.3) 345 (0.3)

Status at end of follow up 20031231

Alive NA NA 5138 (22.1) 39527 (35.2)

Dead from other causes NA NA 11850 (50.9) 35469 (31.6)

PC-death NA NA 6300 (27.1) 37208 (33.2)

Follow up

Time in years , mean (sd) NA NA 6.1 (4.7) 4.3 (4.0)


53

Figur 11. Cumulative incidence of prostate cancer deaths in Sweden between 1970 and 2003, in relation to history of TUR-P.

0 2 4 6 8 10 12 14 16

Time (years)

0

0.1

0.2

0.3

0.4

0.5

Cum

ulat

ive

inci

denc

e o

f pr

ost

ate

can

cer

dea

thTURP-detected ( n = 23288 )

Not TURP-detected ( n = 112204 )


54

53

DISCUSSION

Paper I

Although our cohort of patients with early stage, initially untreated, prostate cancer has been previously followed-up in great detail during on average 15 years 76, the additional six years included in this analysis revealed an unexpected change in prognostic outlook; the cause-specific survival rate dropped by almost 25 percentage points reflecting an approximate threefold increase in prostate cancer mortality rate compared to the first 15 years of follow-up. This change occurred consistently across stage and grade except for poorly differentiated cancers in which excess mortality becomes manifest already during early follow-up. We were unable to conceive any bias that could have spuriously generated these recent findings. Indeed, the internal validity of our population-based study should be high because we achieved complete follow-up and used standardized procedures for clinical examination, ascertainment of disease progression and classification of death. Moreover, the slight difference between cause-specific and relative survival estimates was largely consistent over time. This argues against any shifting criteria for classification of cause of death since the relative survival rate reflects excess mortality (compared to mortality in the general population) and is thus unaffected by any subjective judgment. Hence, chance is the only realistic alternative to a real deterioration in prognosis after long-term follow-up and the level of statistical significance argues against this explanation.

If our data reflect a real phenomenon, they would imply that the probability of progression from localized and indolent to metastatic mortal disease increases markedly after long-term follow-up. This progression is not restricted to cancers diagnosed due to clinical symptoms, but includes also tumors detected incidentally at transurethral resection due to presumed benign prostatic hyperplasia. Our survival data – supported by biopsies during follow-up –would further imply that these latter lesions are either incompletely removed or multifocal with malignant clones left at a transurethral resection. Contrary to emerging views 90, 91 our data also suggest that metastases may arise as a consequence of late mutations rather than being determined already by the early mechanisms of malignant transformation. According to a rival interpretation, the phenomenon we observed reflects transformation of new, more aggressive cancer clones rather than progression of those initially detected. Empirical testing of these complementary, but not mutually exclusive, theories seems difficult. It may be difficult to validate our survival data in any new cohort study of watchful waiting since aggressive treatment of localized prostate cancer has become more routine now than 25 years ago when we started to assemble our cohort 92-95. Indeed, it has been estimated that about 60,000 men undergo radical prostatectomy yearly in the United States alone and the number performed annually in England increased nearly 20-fold between 1991 and 1999 96. This development towards treatment with a curative intent may further accelerate following recent documentation that radical prostatectomy reduces prostate cancer mortality by about 50 percent 97. Hence, support for our findings has to be found chiefly in existing studies of watchful waiting. Other such cohorts are, however, few and small, none of them are population-based and prospective, 98-101 and virtually no follow-up data are available beyond 15 years after diagnosis. Within these constraints, experience during the first 10 years after diagnosis is strikingly similar in existing cohorts of patients with early stage prostate cancer left without initial treatment with a favorable course of the disease for men with highly or moderately differentiated tumors 3.


54 55

54

From a public health perspective, implications of late progression from early stage to mortal disease may not be dramatic because without PSA testing, average age at diagnosis of prostate cancer is so high that competing causes of death predominate (Table 1). Although it is well established 1, 50 that an excess death rate continues long-term in population-based cohorts of prostate cancer patients, these data do not enable distinction of deaths generated by patients initially diagnosed with localized disease. In our entire cohort, 25 out of 223 (11 percent) patients died from prostate cancer within 15 years of diagnosis, and an additional 10 during subsequent follow-up until a time when only 9 percent of all patients in the cohort were still alive and therefore at risk of progression. Assuming that radical prostatectomy prevents approximately 50 percent of prostate cancer deaths 97, about 18 patients in our entire cohort (8 percent) – that is 0.5 x 35 – would have experienced a survival benefit whilst the remaining 205 would not. However, among elderly men, reducing the risk of death from prostate cancer by a certain amount may have limited impact on their overall survival.

Our data may be important for counseling and clinical management of individual patients.First, postponement of death is not the only treatment objective since local progression may create substantial suffering. Indeed, many of our patients experienced symptomatic local growth without generalized disease (Table 1) requiring treatment with estrogens or orchidectomy. Obviously, radical prostatectomy is a major procedure with substantial side-effects, chiefly impotence and incontinence102, 103. Because these complications are surprisingly well tolerated 6 many patients may prefer a radical prostatectomy even if prolonged survival is an uncertain consequence. Our data may be particularly relevant to otherwise healthy men diagnosed with prostate cancer at an early age. If such patients are in their 60s or younger, disease progression occurring after 15 or more years may be a real concern arguing for early local treatment with a curative intent. In patients with a PSA detected cancer 22, such counseling is, however, complicated by the fact that a lead time that cannot be individually determined, has to be added to the approximately 15 years that may precede more rapid tumor progression. One important and complicated question is how the findings of this study relate to the current era when many patients are detected by means of PSA testing. The results are directly relevant for patients with clinical disease diagnosed before the PSA era and also to preclinical disease detected at transurethral resection for presumed benign prostatic hyperplasia. Indeed, as shown in Table 4, these two categories of patients experienced similar risk of dying from prostate cancer. The natural history we have described, reflect also what would happen among PSA detected cancer if the lead-time couldbe accommodated and the patients were left without early therapeutic intervention. However, a substantial proportion of PSA diagnosed cancers represents over detection of sub clinical disease. These cancers would never have surfaced clinically during the subject’s lifetime –either because they are indolent or because death occurs from competing causes before clinical manifestation of the malignancy. By definition, these cancers do not generate any mortality.

In conclusion, our data indicate that the probability of progression to a more aggressive and lethal phenotype may increase after long-term follow-up of prostate cancers that are diagnosed at an early stage and initially left without treatment. These findings provide no information on the possible benefit of early PSA testing followed by aggressive treatment in men with sub clinical disease. However, they argue for early radical treatment of patients with clinical disease and a long-life expectancy. Not only would such surgical intervention potentially prevent deaths; it would also convey prevention from disability caused by local tumor growth.


5556

55

Paper II

In this population-based cohort of incidentally discovered localized prostate cancer cases (T1a-b, Nx, M0) left without initial treatment, our main focus was evaluation of clinical determinants of prostate cancer death. Deriving from a follow-up of more than 20 years, our results confirm the findings of earlier studies that Gleason score is a predictor of prostate cancer death 32, 98, 104, 105. However, the predictive value of the Gleason grading is noticeably low.

The study cohort is well characterized, carefully followed during an extended period of time and no patients have been lost to follow up. Validation of the classification of causes of death against the Swedish Death Register showed a correlation exceeding 90%, in our previous study76. The population-based study design with inclusion of all consecutive cases with newly diagnosed, incidentally discovered, prostate cancer from a strictly defined catchment area, further alleviates concerns of selection bias. Our results should therefore be both valid and generalizable.

Like earlier longitudinal cohort studies using prostate cancer death as outcome we found Gleason and nuclear grade to be a determinant of prostate cancer death98, 104-106. There are today many articles like the one of Roehl et al106 that uses biochemical recurrence-free survival (bRFS) as end point. Using bRFS as surrogate endpoint for prostate cancer specific death has been debated107, 108, and as long as there is no clear evidence that there is a good correlation between bRFS and prostate cancer specific death it is difficult to compare studies with different end points. However, the accuracy of using the Gleason grading system to correctly classify patients according to their risk of dying in prostate cancer was rather poor. As a consequence, the positive predictive value of the Gleason grading system is expected to be low. In these data, even a high set cut-point (Gleason score �8) could only forecast a 50% probability of dying in prostate cancer. Moreover, it appears that the evaluation according tothe Gleason system has shifted towards intermediary values over time and in Sweden, more than 70% of the localized tumors are presently graded as Gleason 6 or 7 4 (Fig.13). In the present data more than 80% were diagnosed with Gleason 6 or 7. A grading system that categorizes a majority of the cases into one group with highly variable outcomes is obviously insufficient. In view of the increasing diagnostic activities in the Western world and the resulting over-treatment of indolent prostate cancers, the need for other or additional prognostic markers becomes even more evident.

The previous findings 55, 58, 109 of Ki67 as a prognostic marker for prostate cancer outcome could not be confirmed in these data. Possible explanations may be the comparably long follow-up time of the present study or the absence of more advanced tumors. In another follow-up study of tumors detected through transurethral resection (TUR-P), Ki-67 was found to be a prognostic marker of prostate cancer death, but this study included patients with more advanced disease (T1-T3, M0–M1)58. Increased Ki-67 expression may precede the onset of more advanced disease, such as local infiltration and metastases, which may explain the difference between the study results. Comparisons with studies using PSA-recurrence as outcome are furthermore difficult55, 109, because of limitations with surrogate end-points and the potential effects of surgical intervention.

Furthermore, the present data confirm the previous findings of tumor volume at diagnosis as an important predictor of outcome 8, 110. Yet, in clinical practice the value of tumor volume is limited since it is difficult to assess other than through TUR-P.


56 57

56

In summary, these data draw attention to the apparent limitations of the most frequently used prognostic tools and highlight the need for improvement in this field. Considering the increasing numbers of early detected prostate cancers and the consequent over-treatment of many benign tumors, further efforts in finding other, or complementary, indicators of prostate cancer outcome are warranted.

Fig 13. Gleasonscore in localized tumours (T0-T2,Nx,M0) in patients diagnosed in Sweden 1996 and 2002. Data from the Swedish Nationel Prostate Cancer register.

Gleasonscore 1996 och 2002 in localized tumours

0

5

10

15

20

25

30

35

40

45

50

1996 2002

Year

Pe

rce

nta

ge

2 3 4

5 6 7

8 9 10


5758

57

Paper III

This is the first demonstration that AMACR protein expression levels are associated with long-term clinical outcome. Our original observations suggested that AMACR expression decreases with prostate cancer progression59, 64. However, standard immunohistochemical evaluation failed to identify clinically significant associations between AMACR expression and biochemical failure 59, 111, 112. Using a more sensitive technique to measure AMACR protein expression, we were able to identify an AMACR cut point that was associated with a three times greater chance of biochemical recurrence and a three to seven greater risk of developing cancer specific death. These observations were independent of other conventional clinical and pathologic parameters in multivariable analysis.

The paradigm for developing prostate cancer biomarkers has focused on using PSA biochemical failure as the surrogate endpoint for the development of metastatic disease and cancer specific death. Studies based on surrogate endpoints are inherently less reliable than studies with clinical endpoints such as cancer specific death113. PSA failure is considered a surrogate for the progression of metastatic disease and prostate cancer specific death114, 115. Indeed, studies by Pound et al. support the potential role of PSA failure as a clinically relevant surrogate endpoint115. For example, of 304 men who develop a PSA failure following radical prostatectomy for clinically localized disease, 103 (34%) developed metastatic disease with a mean actuarial time to metastases of 8 years following the initial PSA elevation. The Pound study also demonstrated that time from initial diagnosis to PSA failure and PSA doubling time predicted metastatic disease. These finding suggest that biochemical failure per se is not a complete surrogate for clinical outcomes, in particular death, and should take into account PSA kinetics as well116. Additional support that biochemical failure may not be an optimal endpoint was recently reported by D’Amico et al.117. They evaluated surrogate endpoints for prostate cancer-specific mortality in two multi-institutional databases of over 8669 patients with prostate cancer treated with surgery (5918 men) or radiation (2751 men) The post-treatment PSA doubling time was statistically significantly associated with time to prostate cancer-specific mortality and with time to all-cause mortality.

As the current study demonstrates, the optimal definition of biochemical failure is still controversial and may be critical for the development of prostate cancer biomarkers. We observed that the AMACR cutpoints differed depending on whether it was derived from the PSA failure or cancer specific death. Biochemical failure may be due to other causes unrelated to the biology of prostate cancer such as local recurrence due to positive surgical margins. Moreover, given the slow progression of prostate cancer, even eight years may not be sufficient to identify all of the patients who will die from prostate cancer. Therefore censoring underestimates the number of men who will experience prostate cancer progression. The work from Pound et al. and D’Amico et al. would suggest that the surrogate endpoint needs further consideration.

The use of prostate cancer specific death, to develop a biomarker test is likely a more valid strategy. In the current study, using cancer specific death to determine the AMACR cutoff yielded a different cutoff from that derived from the surrogate endpoint. Using this revised cutoff on the patients treated for clinically localized disease, AMACR expression independently predicted PSA relapse even accounting for pre-treatment PSA, surgical margin status, and Gleason score. The high-risk group included 81% of the patients in this surgical cohort suggesting that using PSA recurrence as the surrogate endpoint, we are missing some patients who would have progressed. This finding is not what one would anticipate if


58 59

58

biochemical failure also included “false positives” due to other cases such as surgical technique. If these data are confirmed then risk of progression given sufficient time is not being adequately measured using one elevation of PSA following treatment. The most recent review of the entire Örebro watchful waiting series identified a significant increase in prostate cancer specific mortality beyond 15 years of follow up118. This would also further support the need for better prognostic markers. In the randomized Swedish study, survival benefits were observed for surgery over watchful waiting but the absolute reduction in prostate cancer mortality was small 97. Longer follow up might demonstrate even greater benefits from localized therapy give the significant percentage of men who developed metastatic disease over the 8 years of follow up97.

This study also highlights one of the technical limitations of prior biomarker work. Most studies to date, including our own work, have used manual evaluation dividing immunostaining results into a small number of categories. Although reasonable repro-ducibility may be achieved using a 4-tiered (i.e., negative, weak, moderate, or strong), 3-tiered or dichotomous scale, the range of expression is compressed and the ability of the pathologist to distinguish, for example, various shades of moderate staining intensity is not possible. The current study illustrates how critical this might be in the development of prostate cancer biomarkers. Several studies on AMACR failed to identify an association with decreased expression and prostate cancer progression 59, 60, 64, 111, 112. Our initial observations suggested that AMACR expression decreased with cancer progression as both at the transcriptional level and protein level AMACR was lower in metastatic prostate samples 59, 64. However, using manual standard pathology review, no clinical associations were discovered. More recent work using a fluorescence-based approach confirmed lower AMACR expression in metastatic samples. This finding suggested that a quantitative method might be required to identify a critical cutoff to distinguish clinically localized tumors with subtle differences in AMACR expression levels 65. The current study confirmed this hypothesis. No significant cutoff could be determined using a standard evaluation of immunohistochemical markers by the pathologist. In contrast, the use of a continuous scale of protein measurement allowed us to identify a critical cutoff. This cutoff could then be translated into a specific test using an intensity level that was not appreciated despite numerous studies on AMACR to date.

The use of this semi-automated quantitative method for biomarker analysis is promising. However, the clinical application of using such systems still in its early phases, and the results need to be validated and tested in different cohorts and standardization needs to be worked out. Specifically, can we determine a cut point that would be useful to apply to prostate needle biopsies in the clinical assessment of men with clinically localized prostate cancer? One future approach would be to evaluate prostate needle biopsies for AMACR staining intensity and determine if this information added to currently employed clinical nomograms. As recently described by Kattan et al., one could explore to see if the addition of the biomarker, in this case AMACR intensity, improves the ROC area under the curve as compared to a traditional nomogram 119, 120.

The current study now extends the potential utility of AMACR as prostate cancer biomarker. The AMACR (p504s) antibody is gaining wide acceptance in clinical practices as an adjuvant tool in the work up diagnostically challenging prostate lesions referred to as atypical small

acinar proliferation (ASAP) or atypical suspicious for cancer59, 60, 64, 111, 112, 121-126.

Pathologists can now use AMACR (p504s) in combination with a basal cell marker to help make a more definitive diagnosis. We also recently observed a measurable humoral response directed against the AMACR protein127. Using an ELISA assay, this response could be


5960

59

measured in serum samples from men with known prostate cancer but not age-matched controls. This early work suggests that AMACR expression, despite the fact that it is not a secreted protein, may potentially become a clinically useful serum test. A second tissue based assay can also detect enzymatic activity in prostate needle biopsy samples and may represent another means of determining AMACR protein expression in a more reproducibly 128.

In summary, we demonstrated for the first time the AMACR expression is associated with prostate cancer progression after examining tumors from over 440 men diagnosed with clinically localized prostate cancer. Selection of the optimal test cut point came from analysis using prostate specific death and not a surrogate endpoint, suggesting that the definition of the surrogate endpoints, such a PSA biochemical failure need to be further evaluated. Future work will need to determine if this AMACR tissue test can be applied to prostate needle biopsies in a prospective manner to predict the risk of recurrence.

Paper IV

In this population-based cohort of men with localized prostate cancer cases (T1a-b, Nx, M0), we found that MUC-1 expression at diagnosis was a predictor of prostate cancer death. Individuals with dysregulated (either over- or under) MUC-1 expression had a 4–5-fold increased risk of dying of prostate cancer, independent of clinical parameters. We further examined the accuracy of using MUC-1 as a test for prostate cancer progression and found that using MUC-1 alone resulted in a reasonable sensitivity but poor specificity. The addition of information on Gleason Score improved the specificity of the MUC-1 biomarker, but at the cost of a lower sensitivity. Still, these data suggest that MUC-1 may be promising to include in a panel of molecular markers to distinguish aggressive disease from indolent at diagnosis.

These data are in line with the accumulating evidence of the role of MUC-1 in the cell- cell interaction. On one hand, over-expression of MUC-1 has been shown to increase the metastatic potential of the cancer cells129-131. On the other, Kontani et al.132 suggested that the loss of MUC-1 expression or modulation of its antigenicity might cause cancer cells to be unresponsive to the effect of cytotoxic T-lymphocytes. These competing mechanisms provide possible explanations of our findings that both over and under expression of MUC-1 increase the risk of prostate cancer progression.

In this study, we could not confirm the findings of Kirchenbaum et al.71 of a correlation between MUC-1 expression and tumor differentiation. Differences in the assessment of MUC-1 expression may explain the diverging results however; while we evaluated staining intensity, Kirchenbaum et al examined staining patterns (apical and diffuse). Our data furthermore confirm the results of Lapointe et al.72 demonstrating that MUC-1 is an independent prognostic marker that adds prognostic information over and above known risk factors of grade and stage.

The strengths of the study include the long-term and complete follow-up of more than 20 years. A unique feature is that the cohort includes a sizeable number of prostate cancer deaths, which allowed for us to evaluate an important clinical outcome. The fact that the cases had not received any curative treatment further enabled us to explore the effect of dysregulated MUC-1 expression in the natural course of the disease. We employed high-density tissue micro arrays, which represent an efficient approach for immunohistochemical analysis that also reduces potential batch-to-batch variation in staining. Moreover, the Chromavision


60 61

60

system provided an automated assessment of protein intensity on a continuous scale. Our cases were diagnosed prior to introduction of PSA screening in the population, and thus PSA levels at diagnosis were not available. While controlling for PSA levels could attenuate the association between MUC-1 and prostate cancer death, it is unlikely that it would fully explain the relationship.

In summary, these data show that MUC-1 is an independent prognostic marker for prostate cancer death. Furthermore, although based on small numbers, our data suggest that MUC-1expression together with Gleason grade provide substantial information to distinguish prostate cancer outcomes.

As the accuracy of MUC-1 alone or together with the Gleason Score in predicting prostate cancer progression was low, we conclude that its use as a single biomarker in clinical decision making is limited. Yet, we believe that alterations in MUC-1 expression may be useful as part of a composite set of biomarkers in accurately predicting prostate cancer outcome.

Paper V

In this large, population-based study of incidental PCa we observed changes over time in the incidence of incidental PCa in Sweden. We also found significant prostate-specific mortality among men with PCa diagnosed by TURP or by other modes of detection. These data do not fit earlier descriptions of incidental PCa as a non-lethal disease32, 133.

The introduction of TURP in the late 1970s increased not only the incidence of incidental tumours as earlier described 35, but also the overall incidence of prostate cancer. TURP specimens mainly derive from the transition zone of the prostate 30, 31. Tumours originating from this part of the prostate have been described as low grade with low potential of malignancy, and therefore not considered in need of treatment2, 32, 33. On the other hand, 8-37% of all incidental tumours will ultimately progress34. Chun et al134 demonstrated in a study of radical prostatectomy specimens that there is no difference in time to biochemical failure between transition zone tumours and tumours that derive from the peripheral zone. Bostwick et al34 suggested that TURP detected tumours do not exclusively have their origin in the transition zone, but may instead represent peripheral zone tumours that have grown into the transition zone. Thus, there may be a significant overlap between T1b, T1c, and T2 tumours. In the end, it might be the tumor location within the prostate gland and the chance the urologist has of finding it that ultimately determines the T stage.

Before the PSA era, the clinical decision of whether or not presentation with voiding symptoms was indicative of a benign condition or PCa was mainly based on rectal palpation and measurements of acid phosphates. A substantial proportion of men with voiding symptoms were treated by TURP or OAE. Our data confirm earlier observations of a declining incidence of incidental tumours after the introduction of PSA screening36. In Sweden, incidental PCa currently constitutes around 10% of all diagnosed PCa4. We further observed that the cumulative incidence of PCa mortality seemed to decrease over time; this may either describe an increase in lead time after the introduction of PSA, or that tumours detected with TURP today represent a different biological domain then earlier incidental tumours.


6162

61

This large, register-based study with essentially complete follow-up allowed a comprehensive examination of incidence and mortality trends in Sweden over the last three decades. Since health care is free to all Swedish residents, selection forces associated with the TURP procedure should be minimal. A limitation, shared by all register-based studies, though, is the quality of available data. The accuracy of the official causes of death data, with regard to prostate cancer, should however be high135.

Most studies of cancer mortality among incidental tumours are based on patients diagnosed during the 1970s and 1980s. Our study indicates that those findings are not representative of incidental tumours of today. Indeed, the incidental tumors of earlier decades are more likely to represent today’s T1c and T2 tumours. Our data support the thesis that the T-classification system describes availability of tools for the urologist rather than the biology of the tumour34.Thus, new knowledge is needed to characterize incidental tumours detected today.


62 63

63

CONCLUSIONS

Paper I

Although most prostate cancers diagnosed at an early stage have an indolent course, local tumor progression and aggressive metastatic disease may develop in the long term follow up. These findings would argue for early radical treatment of clinical disease, notably among patients with an estimated life expectancy exceeding 15 years. Tumor grade is a prognostic factor, where poorly differentiated tumors experienced a 9 time higher risk of dying in prostate cancer.

Paper II

Although a high Gleason score, tumor volume and high nuclear grade are determinans of prostate cancer death, their positive predictive value are relatively low. Further efforts in finding other, or complementary, indicators of prostate cancer outcome are thus needed. The proliferation marker Ki67 is closely related to Gleason grade and do not add prognostic information.

Paper III

AMACR expression is associated with prostate cancer progression. It also highlights the fact that surrogate endpoint, such a PSA biochemical failure is doubtful, and need to be further evaluated.

Paper IV

MUC-1 is an independent prognostic marker for prostate cancer death. The usefulness of MUC-1 in clinical decision-making is limited, at least when used as a single determinant of prostate cancer progression.

Paper V

There has been a dramatic change over time in the incidence of incidental prostate cancers in Sweden, which parallels the introduction of prostate specific antigen. We also found that the cumulative incidence of prostate cancer death is high in the incidental group, opposing that incidental tumours are a non-lethal disease. Further we found that the cumulative incidence of prostate cancer death decreased over time.


6365

65

SUMMARY AND FUTURE ASPECTS

The result of our study on the natural history of localized prostate cancer with high to intermediate grade of differentiation indicates that in most cases, the course of the disease is favorable and most men die with and not from their prostate cancer. However, it seems that asubset of tumors still progress and become lethal as long as 15 years after diagnosis, including a proportion of T1a or low grade tumors. Although our study does not provide any explanation behind these remarkable phenomena, one might speculate that it could reflect a transformation of new, more aggressive cancer clones or progression of those initially detected.

Our studies further show that tumor characteristics, such as tumor volume, nuclear grade and Gleason score are all strong and independent prognostic factors. Regarding Gleason grading we observe a shift in grading towards Gleason grades 6 and 7 over time. Indeed, now more than 70% of all prostate tumors are judged to fall into these two groups, which obviously reduces the predictive value of this grading system in current clinical practice.

We also found that Ki-67, MUC-1 and AMACR expression in tumors harbors prognostic information. Both MUC-1 and AMACR, but not Ki-67, were found to have independent prognostic value when other clinical parameters such as stage, grade and tumor volume were taken into account. However, possibly the most important lesson from the studies of these markers is that risk estimates for a specific tumor marker alone is not sufficient to assess its clinical value. More important is to analyze its accuracy in terms of sensitivity and specificity, preferably as the area under the curve of a Receiver Operating Characteristic (ROC). This is especially important if we want to use a prognostic or predictive tumor marker in daily practice, where the patient demands information regarding the prognosis and we have to council him regarding the best form of treatment.

The results from our study on trends over time indicate that many of the tumors we classified as T1A and T1B in the eighties resemble those we today classify as T1C. Since most studies on tumors detected by TUR-P were performed during the seventies and eighties they cannot be viewed as biologically similar to those we currently diagnose by TUR-P. More studies regarding T1 A-C tumors are needed to clarify their biological behavior. During the last decade an enormous body of work has emerged on potential prognostic factors15, 16, 136, 137. So far none has shown to be superior to those we already use in clinical practice today. However, there is intense research ongoing in this field.

Why have none of the proposed tumor markers turned out to be useful in daily practice? One simple reason is that when it comes to sensitivity and specificity they have all failed. Moreover, as our knowledge about molecular genetics and the complexity of cellular mechanism have increased enormously during recent years, it does not seem reasonable to hope for a single tumor marker that would predict the course of an individual tumor as there appear to be a large number of important pathways cooperating in networks. Consequently, we hope to identify sets of genes that carry prognostic and predictive information. We have used this strategy in a large collaboration group and identified a 9-138 and a 12-gene model (in press) where we included the strongest prognostic genes, but so far the results have been discouraging. In fact, none of these models were superior to the Gleason grading alone in predicting outcome.


6567

66

A new approach in the search for new genes carrying prognostic information has been the utilization of microarray16 analysis in defining gene expression profiles co-segregating with poor clinical outcome. In breast cancer, for example, a 70-gene model has been found to adequately identify those women that will benefit from adjuvant treatment after surgery139, 140.In a collaborative work we have used this approach in a large cohort of patients with prostate cancer with extended follow-up and with disease specific death as endpoint. So far the results have been discouraging. This may have several explanations. First, prostate cancer is a multifocal and heterogeneous disease in a majority of cases (> 80%),141-145 which confers sampling difficulties, in identifying the index tumor, which ultimately will define the course. Second, due to the genetic instability and mutations occurring in the tumor over time, the hypothesis that there is a genetic profile of an index tumor identified at diagnosis that can predict the outcome might be wrong. This would imply that the prognostic information is not available at diagnosis. If so, this would be discouraging for the effort to find new prognostic markers. Nonetheless, there is an urgent need for more knowledge in this field. It is quiet clear that this knowledge only can be obtained within the context of large collaborative groups of multidisciplinary nature consisting of urologists, pathologists, bio-informatics, statisticians, genetics, and molecular genetics. Moreover, these types of studies require large, well-characterized cohorts of molecular genetics among patients with long follow-up.


6668

67

ACKNOWLEDGMENTS

I wish to express sincere gratitude to everyone involved in these investigations.

I would like to extend special thanks to:

Jan-Erik Johansson, for being a trusting tutor, guiding me in to the field of prostate cancer research, and always being there when needed.

Swen-Olof Andersson, my co-author and tutor who have spent a lot of time revising my manuscripts. Also for thought full discussions and for being a wonderful travel companion.

Anders Kristoffersson, for long and stimulating lunch talks, and introducing me in to the field of urology.

Katja Fall, for introducing me in to the field of epidemiology, and for long and fruit full conversations during our strolls in Boston and always guiding me home.

Mark Rubin, for introducing me in to the field of modern lab technology, and for teaching me prostate pathology during our endless reviews of materials.

Lorelei Mucci, for all the linguistic help with my manuscript and for being such an enthusiastic and helpful collaborator.

My fellow doctorial student Gunnel Andersson, for fruit full discussions and learning me all about collars.

Mats, Aleksandra and Gisela at the pathology department, for good advice in the field of pathology and excellent support with all my requests for blocks and slides.

All my colleagues at the Department of Urology, Örebro University Hospital for your patience and help with our clinical work while I was preoccupied with this thesis.

Irene, for excellent secretarial service.

Lennart Franzén, for introducing me in to the world of pathology

Hans Garmo, for invaluable statistical help and fruit full collaboration.

Bibbi och Göran, for having the best accommodation in Örebro.

All my research colleagues in Boston for fun and inspiring collaborations

All my research colleagues at Mark Rubins lab at Cornell University for fun and inspiring collaborations

All my colleagues in Norrbotten for their support and patience with my research during my time in Norrbotten.


6769

68

Erik, Ina and Nils, my children, for love and tolerance and for steadily reminding me that life is more than urology.

Ulrika my dearest wife for patience, love and support

The studies were supported by grants from:

Örebro County Council Research CommitteeNorrbotten County Council Research CommitteeNyckelfondenSwedish Cancer SocietyNationell institute of health research foundation.


6870

69

REFERENCES

1. Adolfsson, J., Oksanen, H., Salo, J. O. et al.: Localized prostate cancer and 30 years of follow-up in a population-based setting. Prostate Cancer Prostatic Dis, 3: 37, 2000

2. Albertsen, P. C., Hanley, J. A., Fine, J.: 20-year outcomes following conservative management of clinically localized prostate cancer. JAMA, 293: 2095, 2005

3. Chodak, G. W., Thisted, R. A., Gerber, G. S. et al.: Results of conservative management of clinically localized prostate cancer. N Engl J Med, 330: 242, 1994

4. Regionalt Onkologiskt Centrum ROC: Prostatacancer. Primärregistrering 1997-2001: Regionalt Onkologiskt Centrum, vol. 2004, 2003

5. Bill-Axelson, A., Holmberg, L., Ruutu, M. et al.: Radical prostatectomy versus watchful waiting in early prostate cancer. N Engl J Med, 352: 1977, 2005

6. Steineck, G., Helgesen, F., Adolfsson, J. et al.: Quality of life after radical prostatectomy or watchful waiting. N Engl J Med, 347: 790, 2002

7. Hendrick, M. J.: WHO International histological classification of tumors of domestic animals. Washington: Armed Forces Institute of Pathology, p. 64, 1998

8. Humphrey, P., Vollmer, R. T.: The ratio of prostate chips with cancer: a new measure of tumor extent and its relationship to grade and prognosis. Hum Pathol, 19: 411, 1988

9. Ackerman, D. A., Barry, J. M., Wicklund, R. A. et al.: Analysis of risk factors associated with prostate cancer extension to the surgical margin and pelvic node metastasis at radical prostatectomy. J Urol, 150: 1845, 1993

10. Bostwick, D. G., Grignon, D. J., Hammond, M. E. et al.: Prognostic factors in prostate cancer. College of American Pathologists Consensus Statement 1999. Arch Pathol Lab Med, 124: 995, 2000

11. Montironi, R., Mazzucchelli, R., Scarpelli, M. et al.: Prostate carcinoma I: prognostic factors in radical prostatectomy specimens and pelvic lymph nodes. BJU Int, 97: 485, 2006

12. Mostofi, F. K., Sesterhenn, I., Sobin, L. H.: Histological typing of prostate tumours. Geneva,, pp. 26 , [16] pl.-bl., 1980

13. Gleason, D. F.: Histologic grade, clinical stage, and patient age in prostate cancer. NCI Monogr: 15, 1988

14. Partin, A. W., Kattan, M. W., Subong, E. N. et al.: Combination of prostate-specific antigen, clinical stage, and Gleason score to predict pathological stage of localized prostate cancer. A multi-institutional update. JAMA, 277: 1445, 1997

15. Chakravarti, A., Zhai, G. G.: Molecular and genetic prognostic factors of prostate cancer. World J Urol, 21: 265, 2003

16. Quinn, D. I., Henshall, S. M., Sutherland, R. L.: Molecular markers of prostate cancer outcome. Eur J Cancer, 41: 858, 2005

17. Sandblom, G., Mattsson, E., Nilsson, J. et al.: Prostate cancer registration in four Swedish regions 1996-- differences in incidence, age structure and management. Scand J Urol Nephrol, 33: 306, 1999

18. Varenhorst, E., Garmo, H., Holmberg, L. et al.: The National Prostate Cancer Register in Sweden 1998-2002: trends in incidence, treatment and survival. Scand J Urol Nephrol, 39: 117, 2005

19. Stattin, P., Johansson, R., Lodnert, R. et al.: Geographical variation in incidence of prostate cancer in Sweden. Scand J Urol Nephrol, 39: 372, 2005


6971

70

20. Varenhorst, E., Garmo, H., Holmberg, L. et al.: [Prostatic cancer in Sweden 1998-2003. Drastic change of the disease panorama]. Lakartidningen, 103: 285, 2006

21. Stamey, T. A., Yang, N., Hay, A. R. et al.: Prostate-specific antigen as a serum marker for adenocarcinoma of the prostate. N Engl J Med, 317: 909, 1987

22. Catalona, W. J., Smith, D. S., Ratliff, T. L. et al.: Measurement of prostate-specific antigen in serum as a screening test for prostate cancer. N Engl J Med, 324: 1156, 1991

23. Huggins, C., Hodges, C. V.: Studies on prostatic cancer: I. The effect of castration, of estrogen and of androgen injection on serum phosphatases in metastatic carcinoma of the prostate. 1941. J Urol, 168: 9, 2002

24. Walsh, P. C., Quinlan, D. M., Morton, R. A. et al.: Radical retropubic prostatectomy. Improved anastomosis and urinary continence. Urol Clin North Am, 17: 679, 1990

25. Etzioni, R., Penson, D. F., Legler, J. M. et al.: Overdiagnosis due to prostate-specific antigen screening: lessons from U.S. prostate cancer incidence trends. J Natl Cancer Inst, 94: 981, 2002

26. Hermanek, P., Sobin, L. H., International Union against Cancer: TNM classification of malignant tumours, 4. ed. Berlin ; New York: Springer-Vlg., p. 217, 1992

27. Cuzick, J., Fisher, G., Kattan, M. W. et al.: Long-term outcome among men with conservatively treated localised prostate cancer. Br J Cancer, 95: 1186, 2006

28. Robinson, D., Aus, G., Bak, J. et al.: Long-term follow-up of conservatively managed incidental carcinoma of the prostate: a multivariate analysis of prognostic factors. Scand J Urol Nephrol, 41: 103, 2007

29. Andren, O., Fall, K., Franzen, L. et al.: How well does the Gleason score predict prostate cancer death? A 20-year followup of a population based cohort in sweden. J Urol, 175: 1337, 2006

30. McNeal, J. E., Price, H. M., Redwine, E. A. et al.: Stage A versus stage B adenocarcinoma of the prostate: morphological comparison and biological significance. J Urol, 139: 61, 1988

31. McNeal, J. E., Redwine, E. A., Freiha, F. S. et al.: Zonal distribution of prostatic adenocarcinoma. Correlation with histologic pattern and direction of spread. Am JSurg Pathol, 12: 897, 1988

32. Cantrell, B. B., DeKlerk, D. P., Eggleston, J. C. et al.: Pathological factors that influence prognosis in stage A prostatic cancer: the influence of extent versus grade. J Urol, 125: 516, 1981

33. Epstein, J. I., Carmichael, M. J., Partin, A. W. et al.: Small high grade adenocarcinoma of the prostate in radical prostatectomy specimens performed for nonpalpable disease: pathogenetic and clinical implications. J Urol, 151: 1587, 1994

34. Bostwick, D. G.: The pathology of incidental carcinoma. Cancer Surv, 23: 7, 199535. Haapiainen, R.: Latent (pTO) prostatic carcinoma--a retrospective study of frequency

and natural history. Ann Chir Gynaecol, 75: 172, 198636. Mai, K. T., Isotalo, P. A., Green, J. et al.: Incidental prostatic adenocarcinomas and

putative premalignant lesions in TURP specimens collected before and after the introduction of prostrate-specific antigen screening. Arch Pathol Lab Med, 124: 1454, 2000

37. Humphrey, P. A.: Gleason grading and prognostic factors in carcinoma of the prostate. Mod Pathol, 17: 292, 2004

38. Hayes, D. F., Bast, R. C., Desch, C. E. et al.: Tumor marker utility grading system: a framework to evaluate clinical utility of tumor markers. J Natl Cancer Inst, 88: 1456, 1996


7072

71

39. Gleason, D. F.: Classification of prostatic carcinomas. Cancer Chemother Rep, 50:125, 1966

40. Fielding, L. P., Fenoglio-Preiser, C. M., Freedman, L. S.: The future of prognostic factors in outcome prediction for patients with cancer. Cancer, 70: 2367, 1992

41. Hamdy, F. C.: Prognostic and predictive factors in prostate cancer. Cancer Treat Rev, 27: 143, 2001

42. Wang, M. C., Valenzuela, L. A., Murphy, G. P. et al.: Purification of a human prostate specific antigen. Invest Urol, 17: 159, 1979

43. Balk, S. P., Ko, Y. J., Bubley, G. J.: Biology of prostate-specific antigen. J Clin Oncol, 21: 383, 2003

44. Carter, H. B.: Assessing risk: does this patient have prostate cancer? J Natl Cancer Inst, 98: 506, 2006

45. Oesterling, J. E., Jacobsen, S. J., Chute, C. G. et al.: Serum prostate-specific antigen in a community-based population of healthy men. Establishment of age-specific reference ranges. JAMA, 270: 860, 1993

46. Freedland, S. J., Hotaling, J. M., Fitzsimons, N. J. et al.: PSA in the New Millennium: A Powerful Predictor of Prostate Cancer Prognosis and Radical Prostatectomy Outcomes - Results from the SEARCH Database. Eur Urol, 2007

47. Catalona, W. J., Smith, D. S., Ornstein, D. K.: Prostate cancer detection in men with serum PSA concentrations of 2.6 to 4.0 ng/mL and benign prostate examination. Enhancement of specificity with free PSA measurements. JAMA, 277: 1452, 1997

48. D'Amico, A. V., Chen, M. H., Roehl, K. A. et al.: Preoperative PSA velocity and the risk of death from prostate cancer after radical prostatectomy. N Engl J Med, 351:125, 2004

49. Fall, K., Garmo, H., Andren, O. et al.: Prostate-specific antigen levels as a predictor of lethal prostate cancer. J Natl Cancer Inst, 99: 526, 2007

50. Helgesen, F., Holmberg, L., Johansson, J. E. et al.: Trends in prostate cancer survival in Sweden, 1960 through 1988: evidence of increasing diagnosis of nonlethal tumors. J Natl Cancer Inst, 88: 1216, 1996

51. Gronberg, H., Damber, J. E., Jonsson, H. et al.: Patient age as a prognostic factor in prostate cancer. J Urol, 152: 892, 1994

52. Albertsen, P. C., Fryback, D. G., Storer, B. E. et al.: The impact of co-morbidity on life expectancy among men with localized prostate cancer. J Urol, 156: 127, 1996

53. Sherr, C. J.: Cancer cell cycles. Science, 274: 1672, 199654. Scalzo, D. A., Kallakury, B. V., Gaddipati, R. V. et al.: Cell proliferation rate by MIB-

1 immunohistochemistry predicts postradiation recurrence in prostatic adenocarcinomas. Am J Clin Pathol, 109: 163, 1998

55. Bubendorf, L., Sauter, G., Moch, H. et al.: Ki67 labelling index: an independent predictor of progression in prostate cancer treated by radical prostatectomy. J Pathol, 178: 437, 1996

56. Bubendorf, L., Tapia, C., Gasser, T. C. et al.: Ki67 labeling index in core needle biopsies independently predicts tumor-specific survival in prostate cancer. Hum Pathol, 29: 949, 1998

57. Khoo, V. S., Pollack, A., Cowen, D. et al.: Relationship of Ki-67 labeling index to DNA-ploidy, S-phase fraction, and outcome in prostate cancer treated with radiotherapy. Prostate, 41: 166, 1999

58. Stattin, P., Damber, J. E., Karlberg, L. et al.: Cell proliferation assessed by Ki-67immunoreactivity on formalin fixed tissues is a predictive factor for survival in prostate cancer. J Urol, 157: 219, 1997


7173

72

59. Rubin, M. A., Zhou, M., Dhanasekaran, S. M. et al.: alpha-Methylacyl coenzyme A racemase as a tissue biomarker for prostate cancer. JAMA, 287: 1662, 2002

60. Luo, J., Zha, S., Gage, W. R. et al.: alpha-Methylacyl-CoA Racemase: A New Molecular Marker for Prostate Cancer. Cancer Res, 62: 2220, 2002

61. Xu, J., Stolk, J. A., Zhang, X. et al.: Identification of differentially expressed genes in human prostate cancer using subtraction and microarray. Cancer Res, 60: 1677, 2000

62. Rhodes, D. R., Barrette, T. R., Rubin, M. A. et al.: Meta-analysis of microarrays: interstudy validation of gene expression profiles reveals pathway dysregulation in prostate cancer. Cancer Res, 62: 4427, 2002

63. Ferdinandusse, S., Denis, S., L, I. J. et al.: Subcellular localization and physiological role of alpha-methylacyl-CoA racemase. J Lipid Res, 41: 1890, 2000

64. Kuefer, R., Varambally, S., Zhou, M. et al.: alpha-Methylacyl-CoA racemase: expression levels of this novel cancer biomarker depend on tumor differentiation. Am J Pathol, 161: 841, 2002

65. Rubin, M. A., Zerkowski, M. P., Camp, R. L. et al.: Quantitative Determination of Expression of the Prostate Cancer Protein alpha-Methylacyl-CoA Racemase Using Automated Quantitative Analysis (AQUA): A Novel Paradigm for Automated and Continuous Biomarker Measurements. Am J Pathol, 164: 831, 2004

66. Gendler, S. J., Spicer, A. P.: Epithelial mucin genes. Annu Rev Physiol, 57: 607, 199567. Bresalier, R. S., Niv, Y., Byrd, J. C. et al.: Mucin production by human colonic

carcinoma cells correlates with their metastatic potential in animal models of colon cancer metastasis. J Clin Invest, 87: 1037, 1991

68. Ho, S. B., Niehans, G. A., Lyftogt, C. et al.: Heterogeneity of mucin gene expression in normal and neoplastic tissues. Cancer Res, 53: 641, 1993

69. Hilkens, J., Vos, H. L., Wesseling, J. et al.: Is episialin/MUC1 involved in breast cancer progression? Cancer Lett, 90: 27, 1995

70. Hilkens, J., Ligtenberg, M. J., Vos, H. L. et al.: Cell membrane-associated mucins and their adhesion-modulating property. Trends Biochem Sci, 17: 359, 1992

71. Kirschenbaum, A., Itzkowitz, S. H., Wang, J. P. et al.: MUC1 Expression in Prostate Carcinoma: Correlation with Grade and Stage. Mol Urol, 3: 163, 1999

72. Lapointe, J., Li, C., Higgins, J. P. et al.: Gene expression profiling identifies clinically relevant subtypes of prostate cancer. Proc Natl Acad Sci U S A, 101: 811, 2004

73. O'Connor J, C., Julian, J., Lim, S. D. et al.: MUC1 expression in human prostate cancer cell lines and primary tumors. Prostate Cancer Prostatic Dis, 2004

74. Zellweger, T., Ninck, C., Bloch, M. et al.: Expression patterns of potential therapeutic targets in prostate cancer. Int J Cancer, 113: 619, 2005

75. Jhaveri, F. M., Zippe, C. D., Klein, E. A. et al.: Biochemical failure does not predict overall survival after radical prostatectomy for localized prostate cancer: 10-year results. Urology, 54: 884, 1999

76. Johansson, J. E., Holmberg, L., Johansson, S. et al.: Fifteen-year survival in prostate cancer. A prospective, population-based study in Sweden. Jama, 277: 467, 1997

77. Cancer, I. U. A.: TNM Classification of Malignant Tumors. Revised 3rd ed., 197878. Organization, W. H.: International Classification of Tumors. Geneva, Switzerland,

198079. Nelson, C. P., Rubin, M. A., Strawderman, M. et al.: Preoperative parameters for

predicting early prostate cancer recurrence after radical prostatectomy. Urology, 59:740, 2002

80. Nelson, C. P., Dunn, R. L., Wei, J. T. et al.: Contemporary preoperative parameters predict cancer-free survival after radical prostatectomy: a tool to facilitate treatment decisions. Urol Oncol, 21: 213, 2003


7274

73

81. Perrone, E. E., Theoharis, C., Mucci, N. R. et al.: Tissue microarray assessment of prostate cancer tumor proliferation in African- American and white men. J Natl Cancer Inst, 92: 937, 2000

82. Mattsson, B., Wallgren, A.: Completeness of the Swedish Cancer Register. Non-notified cancer cases recorded on death certificates in 1978. Acta Radiol Oncol, 23:305, 1984

83. Cutler, S. J., Ederer, F.: Maximum utilization of the life table method in analyzing survival. J Chronic Dis, 8: 699, 1958

84. Hakulinen, T.: On long-term relative survival rates. J Chronic Dis, 30: 431, 197785. Clayton D, H. M.: Statistical Models in Epidemiology-. Oxford, England: Oxford

University Press, 199386. Hakulinen, T., Abeywickrama, K. H.: A computer program package for relative

survival analysis. Comput Programs Biomed, 19: 197, 198587. Statistiska Centralbyrån (Statistics Sweden): Statistical database. Stockholm: SCB,

vol. 200888. Kalbfleisch, J. D., Prentice, R. L.: The statistical analysis of failure time data, 2nd ed.

Hoboken, N.J: Wiley-Interscience, 200289. Ihaka, R., Gentleman, R.: R: a language for data analysis and graphics. Journal of

Computational and Graphical Statistics, 5: 299, 199690. Ramaswamy, S., Ross, K. N., Lander, E. S. et al.: A molecular signature of metastasis

in primary solid tumors. Nat Genet, 33: 49, 200391. Bernards, R., Weinberg, R. A.: A progression puzzle. Nature, 418: 823, 200292. Walsh, P. C., Lepor, H.: The role of radical prostatectomy in the management of

prostatic cancer. Cancer, 60: 526, 198793. Gerber, G. S., Thisted, R. A., Scardino, P. T. et al.: Results of radical prostatectomy in

men with clinically localized prostate cancer. Jama, 276: 615, 199694. Blasko, J. C., Grimm, P. D., Sylsvester, J. E. et al.: The role of external beam

radiotherapy with I-125/Pd-103 brachytherapy for prostate carcinoma. Radiother Oncol, 57: 273, 2000

95. Hanks, G. E., Hanlon, A. L., Pinover, W. H. et al.: Dose selection for prostate cancer patients based on dose comparison and dose response studies. Int J Radiat Oncol Biol Phys, 46: 823, 2000

96. Oliver, S. E., Donovan, J. L., Peters, T. J. et al.: Recent trends in the use of radical prostatectomy in England: the epidemiology of diffusion. BJU Int, 91: 331, 2003

97. Holmberg, L., Bill-Axelson, A., Helgesen, F. et al.: A randomized trial comparing radical prostatectomy with watchful waiting in early prostate cancer. N Engl J Med, 347: 781, 2002

98. Albertsen, P. C., Fryback, D. G., Storer, B. E. et al.: Long-term survival among men with conservatively treated localized prostate cancer. Jama, 274: 626, 1995

99. Aus, G., Hugosson, J., Norlen, L.: Long-term survival and mortality in prostate cancer treated with noncurative intent. J Urol, 154: 460, 1995

100. Borre, M., Nerstrom, B., Overgaard, J.: The natural history of prostate carcinoma based on a Danish population treated with no intent to cure. Cancer, 80: 917, 1997

101. Adolfsson, J., Steineck, G., Hedlund, P. O.: Deferred treatment of clinically localized low-grade prostate cancer: actual 10-year and projected 15-year follow-up of the Karolinska series. Urology, 50: 722, 1997

102. Litwin, M. S., Hays, R. D., Fink, A. et al.: Quality-of-life outcomes in men treated for localized prostate cancer. Jama, 273: 129, 1995


7375

74

103. Bishoff, J. T., Motley, G., Optenberg, S. A. et al.: Incidence of fecal and urinary incontinence following radical perineal and retropubic prostatectomy in a national population. J Urol, 160: 454, 1998

104. Egevad, L., Granfors, T., Karlberg, L. et al.: Prognostic value of the Gleason score in prostate cancer. BJU Int, 89: 538, 2002

105. Partin, A. W., Walsh, A. C., Pitcock, R. V. et al.: A comparison of nuclear morphometry and Gleason grade as a predictor of prognosis in stage A2 prostate cancer: a critical analysis. J Urol, 142: 1254, 1989

106. Roehl, K. A., Han, M., Ramos, C. G. et al.: Cancer progression and survival rates following anatomical radical retropubic prostatectomy in 3,478 consecutive patients: long-term results. J Urol, 172: 910, 2004

107. Sandler, H. M., DeSilvio, M. L.: Surrogate end points for prostate cancer: what is prostate-specific antigen telling us? J Natl Cancer Inst, 95: 1352, 2003

108. Kwan, W., Pickles, T.: In regard to Kupelian et al.: impact of biochemical failure on overall survival after radiation therapy for localized prostate cancer in the PSA era. IJROBP 2002;52:704-711. Int J Radiat Oncol Biol Phys, 54: 1577, 2002

109. Cowen, D., Troncoso, P., Khoo, V. S. et al.: Ki-67 staining is an independent correlate of biochemical failure in prostate cancer treated with radiotherapy. Clin Cancer Res, 8: 1148, 2002

110. Lowe, B. A., Listrom, M. B.: Incidental carcinoma of the prostate: an analysis of the predictors of progression. J Urol, 140: 1340, 1988

111. Beach, R., Gown, A. M., De Peralta-Venturina, M. N. et al.: P504S immunohistochemical detection in 405 prostatic specimens including 376 18-gauge needle biopsies. Am J Surg Pathol, 26: 1588, 2002

112. Jiang, Z., Woda, B. A., Rock, K. L. et al.: P504S: a new molecular marker for the detection of prostate carcinoma. Am J Surg Pathol, 25: 1397, 2001

113. Schatzkin, A., Gail, M.: The promise and peril of surrogate end points in cancer research. Nat Rev Cancer, 2: 19, 2002

114. Stamey, T. A., McNeal, J. E., Yemoto, C. M. et al.: Biological determinants of cancer progression in men with prostate cancer. Jama, 281: 1395, 1999

115. Pound, C. R., Partin, A. W., Eisenberger, M. A. et al.: Natural history of progression after PSA elevation following radical prostatectomy. Jama, 281: 1591, 1999

116. Swindle, P. W., Kattan, M. W., Scardino, P. T.: Markers and meaning of primary treatment failure. Urol Clin North Am, 30: 377, 2003

117. D'Amico, A. V., Moul, J. W., Carroll, P. R. et al.: Surrogate end point for prostate cancer-specific mortality after radical prostatectomy or radiation therapy. J Natl Cancer Inst, 95: 1376, 2003

118. Johansson, J. E., Andren, O., Andersson, S. O. et al.: Natural history of early, localized prostate cancer. JAMA, 291: 2713, 2004

119. Kattan, M. W., Shariat, S. F., Andrews, B. et al.: The addition of interleukin-6 soluble receptor and transforming growth factor beta1 improves a preoperative nomogram for predicting biochemical progression in patients with clinically localized prostate cancer. J Clin Oncol, 21: 3573, 2003

120. Kattan, M. W.: Judging new markers by their ability to improve predictive accuracy. J Natl Cancer Inst, 95: 634, 2003

121. Epstein, J. I.: Diagnosis and reporting of limited adenocarcinoma of the prostate on needle biopsy. Mod Pathol, 2004

122. Zhou, M., Jiang, Z., Epstein, J. I.: Expression and diagnostic utility of alpha-methylacyl-CoA-racemase (P504S) in foamy gland and pseudohyperplastic prostate cancer. Am J Surg Pathol, 27: 772, 2003


7476

75

123. Magi-Galluzzi, C., Luo, J., Isaacs, W. B. et al.: Alpha-methylacyl-CoA racemase: a variably sensitive immunohistochemical marker for the diagnosis of small prostate cancer foci on needle biopsy. Am J Surg Pathol, 27: 1128, 2003

124. Kunju, L. P., Rubin, M. A., Chinnaiyan, A. M. et al.: Diagnostic usefulness of monoclonal antibody P504S in the workup of atypical prostatic glandular proliferations. Am J Clin Pathol, 120: 737, 2003

125. Yang, X. J., Wu, C. L., Woda, B. A. et al.: Expression of alpha-Methylacyl-CoA racemase (P504S) in atypical adenomatous hyperplasia of the prostate. Am J Surg Pathol, 26: 921, 2002

126. Jiang, Z., Wu, C. L., Woda, B. A. et al.: P504S/alpha-methylacyl-CoA racemase: a useful marker for diagnosis of small foci of prostatic carcinoma on needle biopsy. Am J Surg Pathol, 26: 1169, 2002

127. Sreekumar, A., Laxman, B., Rhodes, D. R. et al.: Humoral immune response to alpha-methylacyl-CoA racemase and prostate cancer. J Natl Cancer Inst, 96: 834, 2004

128. Kumar-Sinha, C., Shah, R. B., Laxman, B. et al.: Elevated alpha-methylacyl-CoA racemase enzymatic activity in prostate cancer. Am J Pathol, 164: 787, 2004

129. Schut, I. C., Waterfall, P. M., Ross, M. et al.: MUC1 expression, splice variant and short form transcription (MUC1/Z, MUC1/Y) in prostate cell lines and tissue. BJU Int, 91: 278, 2003

130. Huang, D. M., Guh, J. H., Chueh, S. C. et al.: Modulation of anti-adhesion molecule MUC-1 is associated with arctiin-induced growth inhibition in PC-3 cells. Prostate, 59: 260, 2004

131. Evangelou, A., Letarte, M., Marks, A. et al.: Androgen modulation of adhesion and antiadhesion molecules in PC-3 prostate cancer cells expressing androgen receptor. Endocrinology, 143: 3897, 2002

132. Kontani, K., Taguchi, O., Narita, T. et al.: Modulation of MUC1 mucin as an escape mechanism of breast cancer cells from autologous cytotoxic T-lymphocytes. Br J Cancer, 84: 1258, 2001

133. Epstein, J. I., Paull, G., Eggleston, J. C. et al.: Prognosis of untreated stage A1 prostatic carcinoma: a study of 94 cases with extended followup. J Urol, 136: 837, 1986

134. Chun, F. K., Briganti, A., Jeldres, C. et al.: Zonal origin of localized prostate cancer does not affect the rate of biochemical recurrence after radical prostatectomy. Eur Urol, 51: 949, 2007

135. Fall, K., Strömberg, F., Rosell, J. et al.: Reliability of death certificates in prostate cancer patients. Scand J Urol Nephrol, in press 2008

136. Kumar-Sinha, C., Chinnaiyan, A. M.: Molecular markers to identify patients at risk for recurrence after primary treatment for prostate cancer. Urology, 1: 19, 2003

137. Verhagen, P. C., Tilanus, M. G., de Weger, R. A. et al.: Prognostic factors in localised prostate cancer with emphasis on the application of molecular techniques. Eur Urol, 41: 363, 2002

138. Mucci, L. A., Pawitan, Y., Demichelis, F. et al.: Nine-gene molecular signature is not associated with prostate cancer death in a watchful waiting cohort. Cancer Epidemiol Biomarkers Prev, 17: 249, 2008

139. Nuyten, D. S., Kreike, B., Hart, A. A. et al.: Predicting a local recurrence after breast-conserving therapy by gene expression profiling. Breast Cancer Res, 8: R62, 2006

140. Nuyten, D. S., van de Vijver, M. J.: Gene expression signatures to predict the development of metastasis in breast cancer. Breast Dis, 26: 149, 2006


7577

76

141. Bostwick, D. G., Shan, A., Qian, J. et al.: Independent origin of multiple foci of prostatic intraepithelial neoplasia: comparison with matched foci of prostate carcinoma. Cancer, 83: 1995, 1998

142. Cheng, L., Song, S. Y., Pretlow, T. G. et al.: Evidence of independent origin of multiple tumors from patients with prostate cancer. J Natl Cancer Inst, 90: 233, 1998

143. Greene, D. R., Wheeler, T. M., Egawa, S. et al.: A comparison of the morphological features of cancer arising in the transition zone and in the peripheral zone of the prostate. J Urol, 146: 1069, 1991

144. Macintosh, C. A., Stower, M., Reid, N. et al.: Precise microdissection of human prostate cancers reveals genotypic heterogeneity. Cancer Res, 58: 23, 1998

145. Ruijter, E. T., Miller, G. J., van de Kaa, C. A. et al.: Molecular analysis of multifocal prostate cancer lesions. J Pathol, 188: 271, 1999


7678

115

PAPER I

Klunserna Study 05-10-25, 10.22115

PAP

ER I

ORIGINAL CONTRIBUTION

Natural History of Early, LocalizedProstate CancerJan-Erik Johansson, MD, PhDOve Andren, MDSwen-Olof Andersson, MD, PhDPaul W. Dickman, PhDLars Holmberg, MD, PhDAnders Magnuson, BScHans-Olov Adami, MD, PhD

WITHOUT UNDERSTAND-ing the natural history ofprostate cancer diag-nosed at an early, local-

ized stage, patient counseling and clini-cal management are difficult. Thechallenge is to maximize the possibili-ties for survival without extensive over-treatment. Even without initial treat-ment, only a small proportion of allpatients with cancer diagnosed at anearly clinical stage die fromprostate can-cer within 10 to 15 years following di-agnosis.1-3 However, to our knowledge,no study has hitherto adequately ana-lyzedwhether patientswho escapedme-tastasis and death during those 10 to 15yearswithout treatment continue tohavean indolent, nonfatal disease course orwhether in the long term, tumor pro-gression takes amore aggressive course.Recently, a randomized trial4 demon-strated that radical prostatectomy mayfurther reduce the low-death rate in earlyprostate cancer by approximately 50%.Because it takes several years after op-eration for this benefit to emerge, age atdiagnosis, comorbidity that influenceslife expectancy, and long-term naturalhistory will determine the potential ad-vantage with radical primary treat-ment.

This study focuses on informationthat aids clinical decision making,namely the association between prog-nostic factors available at diagnosis andthe long-term natural history in pa-tients without initial treatment. Suchknowledge can help us understandwhether there aremenwithprostate can-cer and a long life expectancy in whomearly radical treatment might be justi-fied despite the fact that they have fa-vorableprognostic signs seen fromaper-spective of 5 to 10 years of follow-up.We studied these issues prospectively inthe largest population-based cohort ever

impaneled to analyze survival follow-ing watchful waiting of patients withearly prostate cancer. Complete fol-

Author Affiliations: Departments ofUrology andClini-cal Medicine (Drs Johansson, Andren, and Andersson)and Statistical Unit, Center for Clinical Research (MrMagnuson), OrebroUniversity Hospital, andCenter forAssessment ofMedical Technology (Drs Johansson andAndersson), Orebro, Sweden; Department of MedicalEpidemiologyandBiostatistics,Karolinska Institute, Stock-holm, Sweden (DrsDickman andAdami); RegionalOn-cologic Center, University Hospital, Uppsala, Sweden(Dr Holmberg); and Department of Epidemiology andthe Harvard Center for Cancer Prevention, HarvardSchool of Public Health, Boston, Mass (Dr Adami).CorrespondingAuthor: Jan-Erik Johansson,MD, PhD,Department of Urology, Orebro University Hospital,SE-70185 Orebro, Sweden ([email protected]).

Context Among men with early prostate cancer, the natural history without initialtherapy determines the potential for survival benefit following radical local treatment.However, little is known about disease progression andmortality beyond 10 to 15 yearsof watchful waiting.

Objective To examine the long-term natural history of untreated, early stage pros-tatic cancer.

Design Population-based, cohort study with a mean observation period of 21 years.

Setting Regionally well-defined catchment area in central Sweden (recruitmentMarch1977 through February 1984).

Patients A consecutive sample of 223 patients (98% of all eligible) with early-stage(T0-T2 NX M0 classification), initially untreated prostatic cancer. Patients with tumorprogression were hormonally treated (either by orchiectomy or estrogens) if they hadsymptoms.

Main Outcome Measures Progression-free, cause-specific, and overall survival.

Results After complete follow-up, 39 (17%) of all patients experienced generalizeddisease. Most cancers had an indolent course during the first 10 to 15 years. How-ever, further follow-up from 15 (when 49 patients were still alive) to 20 years re-vealed a substantial decrease in cumulative progression-free survival (from 45.0% to36.0%), survival without metastases (from 76.9% to 51.2%), and prostate cancer–specific survival (from 78.7% to 54.4%). The prostate cancer mortality rate increasedfrom 15 per 1000 person-years (95% confidence interval, 10-21) during the first 15years to 44 per 1000 person-years (95% confidence interval, 22-88) beyond 15 yearsof follow-up (P=.01).

Conclusion Although most prostate cancers diagnosed at an early stage have anindolent course, local tumor progression and aggressive metastatic disease may de-velop in the long term. These findings would support early radical treatment, notablyamong patients with an estimated life expectancy exceeding 15 years.JAMA. 2004;291:2713-2719 www.jama.com

For editorial comment see p 2757.

©2004 American Medical Association. All rights reserved. (Reprinted) JAMA, June 9, 2004—Vol 291, No. 22 2713

on February 4, 2008 www.jama.comDownloaded from


low-up of this cohort has now beenachieved during an average of 21 years,andonly 9%of the patients are still alive.

METHODSPatients

The patients comprised a population-based cohort of patients with early, ini-tially untreated prostate cancer as pre-viously described in detail.3 The TNMsystem5 and theWorld Health Organi-zation6 classification of malignant dis-eases were used. At the time of diag-nosis, all patients underwent a clinicalexamination, excretory urography,chest radiography, bone scan, and skel-etal radiography (if needed) and hadroutine blood samples taken. The nodalstatus was not known for any of the pa-tients. Prostate-specific antigen (PSA)testingwas not available, andno screen-ing activities for prostate cancer tookplace during the period when this co-hort was recruited.FromMarch 1977 through February

1984, a total of 654 new cases of pros-

tate cancer were diagnosed among resi-dents intheregionallywell-definedcatch-ment area in central Sweden. Patientswere given no initial treatment if thetumor growthwas localized to the pros-tate gland as judged by digital rectalexamination (T0-T2) and no distantmetastases were present (306 patients).This was in accordance with the stan-dard management at the time in Swe-den. The following restrictions wereapplied,however, among thosewithpal-pable tumors(T1-T2).FromMarch1977through February 1979, only patientswithahighlydifferentiated tumor(grade1)were included in theuntreatedgroup.FromMarch1979throughtheendof therecruitment period, patients youngerthan75years atdiagnosis andwithmod-erately or poorly differentiated tumors(grades 2-3)were randomly allocated toreceive local radiation (10 patients) orno treatment, and only the latter groupwas included in this cohort study.Patients75yearsorolderwerenot treatedand included in the study.

Among the 227 eligible patients, 4(2%) were given initial treatment andhad to be excluded from the analyses.The distribution of the study group of223 patients by age, stage, and grade atthe time of diagnosis is shown inTABLE 1. Themean age at diagnosis was72 years (range, 41-91 years). Alto-gether 106 (48%) cases were detectedby histopathologic examinations ofspecimens obtained at operations forsuspected benign prostatic hyperpla-sia. The remaining 117 patients had apalpable clinical disease localized to theprostate gland. A review by an experi-enced histopathologist confirmed theinitial diagnosis in all cohort mem-bers. Approximately two thirds of pa-tients had highly differentiated tu-mors, whereas only 9 (4%) had a poorlydifferentiated tumor (Table 1).

ProceduresAll 223 patients were followed up fromdiagnosis until death or the end of theobservation period (September 1,2001). No patient was lost to follow-up. Clinical examination, laboratorytests, and bone scans were performedevery 6months during the first 2 yearsafter diagnosis and subsequently oncea year during the first 10 years of ob-servation and thereafter at least once ev-ery second year. Those in whom thecancer progressed to symptomatic dis-ease were treated with exogenous es-trogens or orchidectomy.Local progression was defined as tu-

mor growth through the prostate cap-sule (T3) as judged by digital rectal ex-amination. Development of distantmetastasis (M1) was classified as gen-eralization. If both local progression andmetastatic disease were present, the pa-tient was classified as having general-ized disease.During the first 6 years of follow-

up, all patients whowere still alive andconsented underwent a new fine-needle biopsy every other year.We ob-tained such biopsy specimens from178(80%)of the 223patients. Although thisprocedure has lower than 100% sensi-tivity, notably for impalpable tumors,remaining cancer growth was con-

Table 1. Characteristics of 223 Patients With Early Prostate Cancer (T0-2 NX M0) WhoReceived No Initial Treatment According to Age, Tumor Stage, and Grade at Time ofDiagnosis in 1977-1984*

CategoryTotal No.of Patients

Progression, No. (%)

Cause of Death,No. (%)

T3 M1 TotalProstaticCancer

OtherCause†

Age, y�61 13 6 (46) 4 (31) 6 (46) 3 (23) 3 (23)

61-70 86 40 (46) 22 (26) 46 (53) 19 (22) 59 (69)

71-80 96 26 (27) 13 (14) 29 (30) 12 (12) 79 (82)

�81 28 8 (29) 0 (0) 8 (29) 1 (4) 27 (96)

Tumor stage‡T01 72 14 (19) 11 (15) 17 (24) 10 (14) 55 (76)

T0d 34 14 (41) 8 (24) 17 (50) 8 (24) 26 (76)

T1-T2 117 52 (44) 20 (47) 55 (47) 17 (15) 87 (74)

Grade§1 148 42 (28) 18 (12) 45 (30) 14 (9) 118 (80)

2 66 35 (53) 16 (24) 38 (58) 16 (24) 46 (70)

3 9 3 (33) 5 (56) 6 (67) 5 (56) 4 (44)

Total, No. (%) 223 80 (36) 39 (17) 89 (40) 35 (16) 168 (75)

*The number of patients with progression of the tumor manifested by local growth (T3) or distant metastases (M1) andthe number and causes of death are shown.

†Three patients died of cardiovascular disease during treatment with estrogens.‡T0 indicates clinically occult, incidental; T01, T0pT localized (cancer �25% of the total specimen); T0d, T0pT diffuse

(cancer �25% of the total specimen); T1-2, confined to prostate gland (T1, nodule surrounded by normal prostatictissue; T2, large nodule or multiple nodules; and T3, localized to periprostatic area). These grades correspond to theTNM classification from 1978.5 In the classification from 2002, T1 is considered no evidence of clinical disease andT2 is considered palpable disease.

§Grade 1 indicates highly differentiated; grade 2, moderately differentiated; and grade 3, poorly differentiated. Thesegrades refer to the World Health Organization classification of malignant diseases.6 They are not directly translatableto the Gleason grading system. However, in an earlier report,1 grade 1 was compared with Gleason score 2 to 4,grade 2 with Gleason score 5 to 7, and grade 3 with Gleason score 8 to 10.

EARLY, LOCALIZED PROSTATE CANCER

2714 JAMA, June 9, 2004—Vol 291, No. 22 (Reprinted) ©2004 American Medical Association. All rights reserved.



PAP

ER I

firmed cytologically in most patients:45 (73%) among those with T01 dis-ease, 24 (92%) among those with T0ddisease, and 90 (100%) among thosewith T1-2 disease. Altogether 31 (17%)of 178 patients showed evidence of de-differentiation. Twenty-one patients(18%) changed from high to moderatedifferentiation, 7 (13%) from moder-ate to low differentiation, and 3 (3%)from high to low differentiation.The medical records of all deceased

patients were reviewed. In most in-stances, the cause of death determinedin real time was obvious on clinicalgrounds alone. An autopsy was per-formed if the cause of death was notclear. Prostate cancerwas recordedas theunderlying cause of death, a contribu-tory causeof death, orunrelated todeathas described in detail in a previous re-port.3 If treatment of the prostate can-cer was related to death (chiefly due tocardiovascular complications follow-ing estrogen administration), prostatecancer was recorded as a contributorycause. As a validation,we compared ourownclassificationof causesof deathwiththose recorded in the Swedish DeathRegister. This information was ob-tained through record linkage betweenour study cohort and the SwedishDeathRegister based on the individuallyunique national registration number as-signed to all Swedish residents. Therewas agreement in 90% of the patientsandno evidenceof systematic overascer-tainmentorunderascertainmentof pros-tate cancer as cause of death in our data.Althoughbasedon small numbers, therewas no evidence that the disagreementwas larger in older than in younger pa-tients.

Statistical MethodsWe estimated various measures of pa-tient survival using the actuarial (life-table) method.7 Cause-specific sur-vival was estimated by considering onlydeaths due to prostate cancer as eventsof interest (deaths due to other causeswere considered censored), observedsurvival by considering deaths due toany cause as events, and progression-free survival by considering progres-

sion as the event of interest. We alsoestimated relative survival, defined asthe ratio of observed survival to the ex-pected survival of a comparable groupfrom the general population assumedto be free of prostate cancer. We esti-mated expected survival using theHakulinen method8 based on Swedishpopulation life tables stratified by age,sex, and calendar time. We also calcu-lated prostate cancer–specific mortal-ity rates (deaths per 1000 person-years at risk) and associated 95%confidence intervals (CIs).9 We esti-mated Poisson regression models9 tostudy the association between pros-tate cancermortality and time since di-agnosis while adjusting for age at di-agnosis, stage, and grade. Relativesurvival was estimated using softwaredeveloped at the Finnish Cancer Reg-istry.10 All other analyses were per-formed using Stata statistical software(Stata Corporation, College Station,Tex). All reported P values are 2-sided.Statistical significance was P�.05.

RESULTSOverall Findings

During amean observation period of 21years, 89 patients (40%) experiencedprogression of disease, and of these 39

(17% of the entire cohort) developedgeneralized disease. A total of 203 pa-tients (91% of the entire cohort) diedduring follow-up, with prostate can-cer considered the cause of death in 35(16% of the entire cohort; Table 1).Among patients who were 70 years oryounger at diagnosis, 22 (22%) diedfrom prostate cancer during follow-up, whereas this proportion de-creased markedly at higher ages. Theproportion of patients dying frompros-tate cancerwas strikingly similar amongthosewith nonpalpable (T0) tumors de-tected at transurethral resection (18 pa-tients [17%]) and those with a pal-pable tumor (17 patients [15%]). Incontrast, poor differentiation was astrong predictor of prostate cancer–specific death (Table 1).

Progression and Survival RatesAlthough based on small numbers, theprogression and mortality rates re-mained fairly constant during the firstthree 5-year periods following diagno-sis (TABLE 2). Averaged over the first15 years, the rate of progression tometastatic disease was 18 per 1000 per-son-years (95%CI, 13-25) and the pros-tate cancer mortality rate was 15 per1000 person-years (95%CI, 10-21). In

Table 2. Rates per 1000 Person-Years for Progression to Metastatic Disease and Death Dueto Prostate Cancer by Years of Follow-up, Age, Stage, and Grade at Diagnosis*

Category

Progression Prostate Cancer Death

No. of Events Rate (95% CI) No. of Events Rate (95% CI)

Follow-up, y*0-4 18 20 (13-32) 11 12 (7-22)

5-9 9 15 (8-30) 11 18 (10-33)

10-14 5 16 (7-38) 5 15 (6-36)

�15 7 41 (20-87) 8 44 (22-88)

Age, y�70 26 24 (16-35) 22 19 (13-29)

�71 13 15 (9-26) 13 15 (8-25)

StageT01 11 17 (9-30) 10 15 (8-28)

T0d 8 31 (15-62) 8 28 (14-57)

T1-2 20 19 (12-30) 17 16 (10-25)

Grade1 18 13 (8-21) 14 10 (6-17)

2 16 29 (18-47) 16 27 (17-44)

3 5 242 (101-581) 5 194 (81-466)

Total 39/223 (17) 20 (14-27) 35/223 (16) 17 (12-24)Abbreviation: CI, confidence interval.*Numbers of patients alive after 5 years were 150; 91 after 10 years; and 48 after 15 years.





contrast, an approximately 3-foldhigherratewas found both for progression anddeath during follow-up beyond15 years(Table 2). This increasewas almost sta-tistically significant for progression(P=.06) and statistically significant fordeath (P=.01).TABLE 3 shows various measures of

survival after 15 and 20 years of follow-up. During this 5-year period, the pro-gression-free survival among all pa-tients decreased from 45.0% to 36.0%.The low and rapidly decreasing ob-served survival reflects chiefly the im-pact of causes of death other than pros-tate cancer. Most notably, however, wefound a substantial decline by approxi-mately 25 percentage points in both therelative and the cause-specific survivalrate during the last 5 years of follow-

up. FIGURE 1 further illustrates how agradual decline in relative and cause-specific survival seemingly occurredmore rapidly after approximately 16years of follow-up. This change seemedto affect tumors regardless of initial stageand also to affect tumors that were bothinitially highly andmoderately differen-tiated (FIGURE 2). The gloomy outlookamong patients with poorly differenti-ated tumors becamemanifested alreadywithin the first 5 years of follow-up.Prostate cancermortalitywas slightly

higher among patients whose cancerwas diagnosed at 70 years or youngerthan among thosewhose cancerwas di-agnosed at older ages (Table 2). Strik-ingly similarmortality rates were foundamong patients who had localized non-palpable cancer compared with those

who had a cancer in stage T1 or T2. Incontrast, the mortality rate was 70%higher among men with a nonpal-pable diffuse cancer.With regard to dif-ferentiation, the mortality rate in-creaseddrastically fromhighly to poorlydifferentiated tumors (Table 2).

Multivariable AnalysesMultivariable Poisson regressionmod-els were fitted to quantify the indepen-dent effects of follow-up time, age at di-agnosis, grade, and stage (TABLE4).Ouranalyses showed a significant (approxi-mately 6-fold) higher mortality rate af-ter 15 years of follow-up comparedwiththe first 5 years. The strong prognos-tic impact of grade, notably of poorlydifferentiated tumors, was also con-firmed. In contrast, neither age at di-agnosis nor stage of disease was sig-nificantly associated with risk of deathdue to prostate cancer. Althoughwehadlimited power to test interaction, therisk of death due to prostate cancer af-ter 15 or more years of follow-up com-pared with 0 to 14 years was similaramong patients with cancer diag-nosed before (relative risk, 4.1) and af-ter (relative risk, 3.1) 70 years of age.A separatemodel was fitted in which

the event of interest was local progres-sion. In this analysis, we disregarded ifand when regional and/or distant pro-gression or metastases were ascer-tained. Except for age at diagnosis, thepattern for local progression was strik-ingly different from that of death dueto prostate cancer (Table 4). Hence, the

Figure 1. Survival of Prostate Cancer Patients (n = 223)

No. at Risk 223 150 48 991

Cause-SpecificSurvivalRelative SurvivalObserved Survival

100

90

80

70

60

50

40

30

20

10

0

Time Since Diagnosis, y

Sur

viva

l, %

205 10 15

Table 3. The 15- and 20-Year Progression-Free Survival, Observed Survival, Relative Survival, and Cause-Specific Survival Rates by Stage andGrade at Diagnosis*

Category

Percentage of Survival (95% Confidence Interval)

Progression Free Observed Relative Cause Specific

15 Years 20 Years 15 Years 20 Years 15 Years 20 Years 15 Years 20 Years

StageT01 57.7 (38.9 to 76.5) 52.5 (32.7 to 72.3) 22.2 (12.4 to 32.0) 7.5 (−0.1 to 15.0) 82.1 (45.9 to 118.4) 56.1 (0.6 to 112.8) 81.1 (66.5 to 5.6) 57.9 (28.3 to 87.5)

T0d 35.9 (15.1 to 56.5) 17.9 (−9.5 to 45.3) 14.7 (2.6 to 26.8) 0 63.6 (11.1 to 116.1) 0 69.7 (50.1 to 89.2) 46.4 (6.3 to 86.5)

T1-2 38.4 (26.8 to 50.1) 33.1 (20.9 to 45.4) 23.1 (15.3 to 30.9) 10.3 (4.2 to 16.3) 75.2 (49.8 to 100.6) 61.5 (25.2 to 97.9) 80.3 (69.8 to 90.8) 56.9 (34.9 to 78.9)

Grade1 56.0 (44.4 to 67.6) 46.0 (30.9 to 61.0) 24.3 (17.2 to 31.4) 9.7 (4.3 to 15.1) 83.7 (59.4 to 108.0) 63.4 (28.1 to 98.8) 88.9 (81.4 to 96.3) 71.8 (54.9 to 88.7)

2 28.8 (15.2 to 42.3) 24.3 (10.2 to 38.4) 18.2 (8.7 to 27.7) 3.5 (−1.8 to 8.7) 64.7 (30.9 to 98.4) 23.4 (−12.0 to 58.7) 64.5 (47.2 to 81.8) 22.1 (−7.5 to 51.7)

3 15.6 (−12.8 to 43.9)† 0 0 0 0 0 28.6 (5.6 to 62.7)† 0

All patients 45.0 (35.7 to 54.3) 36.0 (24.2 to 47.9) 21.5 (16.0 to 27.0) 7.5 (3.5 to 11.4) 75.9 (56.5 to 95.3) 49.9 (23.4 to 76.4) 78.7 (70.8 to 86.7) 54.5 (37.6 to 71.4)

*For definitions of tumor stages and grades, see Table 1 footnotes.†After 7 years.





PAP

ER I

risk of local progression did not in-crease over follow-up time, and the as-sociation with grade was weak. More-over, compared with T01 tumors,growth beyond the prostate capsulewas2 or 3 timesmore likely in patients withT0d and T1-2 tumors, respectively.

COMMENTAlthough our cohort of patients withearly stage, initially untreated pros-tate cancer has been previously fol-lowed up in great detail during an av-erage of 15 years,3 the additional 6 yearsincluded in this analysis revealed anun-expected change in prognostic out-look; the cause-specific survival rate de-creased by almost 25 percentage points,reflecting an approximate 3-fold in-crease in prostate cancer mortality ratecompared with the first 15 years of fol-low-up. This change occurred consis-tently across stage and grade except forpoorly differentiated cancers in whichexcess mortality becomes manifest al-ready during early follow-up.Wewereunable to conceive of any bias thatcould have spuriously generated theserecent findings. Indeed, the internal va-lidity of our population-based studyshould be high because we achieved

complete follow-up and used standard-ized procedures for clinical examina-tion, ascertainment of disease progres-sion, and classification of death.Moreover, the slight difference be-tween cause-specific and relative sur-vival estimates were largely consistentover time. This argues against any shift-ing criteria for classification of cause ofdeath, since the relative survival rate re-flects excess mortality (compared withmortality in the general population) andis thus unaffected by any subjectivejudgment. Prostate cancer mortalityrates were mirrored closely by rates ofdisease progression to metastatic dis-ease. Hence, chance is the only realis-tic alternative to a real deterioration inprognosis after long-term follow-up,and the level of statistical significanceargues against this explanation.If our data reflect a real phenom-

enon, they would imply that the prob-ability of progression from localized andindolent tometastaticmortal disease in-creasesmarkedly after long-term follow-up. This progression is not restrictedto cancers diagnosed due to clinicalsymptoms but includes also tumors de-tected incidentally at transurethral re-section due to presumed benign pros-

tatic hyperplasia. Our survival data,supported by biopsy specimens takenduring follow-up, would further im-

Table 4. Multivariable Relative Risks ofDeath From Prostate Cancer and LocalProgression (Growth Through the ProstateCapsule) in Relation to Follow-up Time, Ageat Diagnosis, Tumor Grade, and TumorStage*

Category

Relative Risk(95% Confidence Interval)

DeathLocal

Progression

Follow-up, y0-4 1.0 1.0

5-9 2.2 (0.9-5.5) 0.6 (0.3-1.1)

10-14 2.0 (0.6-6.3) 0.6 (0.3-1.3)

�15 6.4 (2.3-17.8) 0.8 (0.3-2.1)

Age atdiagnosis, y

�70 1.0 1.0

�70 0.7 (0.3-1.6) 0.9 (0.6-1.5)

Grade1 1.0 1.0

2 3.4 (1.6-7.3) 2.5 (1.6-4.0)

3 46.6 (12.3-177.4) 3.3 (0.9-11.9)

StageT01 1.0 1.0

T0d 0.7 (0.2-2.1) 2.0 (0.9-4.4)

T1-2 0.7 (0.3-1.6) 2.7 (1.5-4.9)

*Estimated using Poisson regression where each factoris simultaneously adjusted for all other factors. For defi-nitions of tumor stage and grade, see Table 1 footnotes.

Figure 2. Cause-Specific Survival by Stage of Disease and Tumor Grade at Diagnosis

100

90

80

70

60

50

40

30

20

10

0 0

No. at Risk

T0 Localized Tumor 72 52 16 231

T0 Diffuse 34 24 5 012

T1 + T2 Tumor 117 74 27 748


Survival by Tumor Stage at DiagnosisS

urvi

val,

%

T0 Localized TumorT1 + T2 Palpable TumorT0 Diffuse Tumor

Grade 1Tumor Grade

Grade 2Grade 3

No. at Risk

205 10 15


205 10 15

Grade 1 148 103 36 862

Grade 2 66 45 12 129

Grade 3 9 2 0 00

Survival by Tumor Grade at Diagnosis

Disease stages were T0 localized, T0 diffuse, and T1-T2. Grades were as follows: grade 1, highly differentiated; grade 2, moderately differentiated; and grade 3, poorlydifferentiated.





ply that these latter lesions are eitherincompletely removed or multifocalwith malignant clones left at a trans-urethral resection. Contrary to emerg-ing views,11,12 our data also suggest thatmetastases may arise as a conse-quence of late mutations rather thanbeing determined already by the earlymechanisms of malignant transforma-tion. According to a rival interpreta-tion, the phenomenonwe observed re-flects transformation of new, moreaggressive cancer clones rather thanprogression of those initially detected.Empirical testing of these complemen-tary, but not mutually exclusive, theo-ries seems difficult.Itmay be difficult to validate our sur-

vival data in any new cohort study ofwatchful waiting since aggressive treat-ment of localized prostate cancer hasbecome more routine now than it was25 years ago when we started to as-semble our cohort.13-16 Indeed, it hasbeen estimated that approximately60000men undergo radical prostatec-tomy yearly in the United States alone,and the number performed annually inEngland increased nearly 20-fold be-tween 1991 and 1999.17 This develop-ment toward treatment with a cura-tive intent may further acceleratefollowing recent documentation thatradical prostatectomy reduces pros-tate cancer mortality by approxi-mately 50%.4 Hence, support for ourfindings has to be found chiefly in ex-isting studies ofwatchfulwaiting.Othersuch cohorts are, however, few andsmall, none of them are populationbased and prospective,2,18-20 and virtu-ally no follow-up data are available be-yond 15 years after diagnosis. Withinthese constraints, experience during thefirst 10 years after diagnosis is strik-ingly similar in existing cohorts of pa-tients with early stage prostate cancerleft without initial treatment, with a fa-vorable course of the disease for menwith highly or moderately differenti-ated tumors.1

Fromapublic health perspective, im-plications of late progression from earlystage to mortal disease may not be sig-nificant because without PSA testing,

average age at diagnosis of prostate can-cer is so high that competing causes ofdeath predominate (Table 1). Al-though it is well established21,22 that anexcess death rate continues long termin population-based cohorts of pros-tate cancer patients, these data do notenable distinction of deaths generatedby patients initially diagnosed as hav-ing localized disease. In our entire co-hort, 25 (11%)of 223patients died fromprostate cancer within 15 years ofdiagnosis and an additional 10 duringsubsequent follow-upuntil a timewhenonly 9% of all patients in the cohortwere still alive and therefore at risk ofprogression. Assuming that radicalprostatectomy prevents approxi-mately 50% of prostate cancer deaths,4

approximately 18 patients (8%) in ourentire cohort (that is, 0.5�35) wouldhave experienced a survival benefit,whereas the remaining 205 would not.However, among elderly men, reduc-ing the risk of death from prostate can-cer by a certain amount may have lim-ited impact on their overall survival.Our datamay be important for coun-

seling and clinical management of in-dividual patients. Postponement ofdeath is not the only treatment objec-tive because local progressionmay cre-ate substantial suffering. Indeed, manyof our patients experienced symptom-atic local growth without generalizeddisease (Table 1), requiring treatmentwith estrogens or orchidectomy. Ob-viously, radical prostatectomy is a ma-jor procedure with substantial ad-verse effects, chiefly impotence andincontinence.23,24 Because these com-plications are surprisingly well toler-ated,25 many patientsmay prefer a radi-cal prostatectomy even if prolongedsurvival is an uncertain consequence.Our data may be particularly relevantto otherwise healthymen diagnosed ashaving prostate cancer at an early age.If such patients are in their 60s oryounger, disease progression that oc-curs after 15 ormore yearsmay be a realconcern, arguing for early local treat-ment with a curative intent. In pa-tients with a PSA-detected cancer,26

such counseling is, however, compli-

cated by the fact that a lead time thatcannot be individually determined hasto be added to the approximately 15years that may precede more rapid tu-mor progression.One important and complicated

question is how the findings of thisstudy relate to the current era whenmany patients are detected bymeans ofPSA testing. The results are directly rel-evant for patients with clinical diseasediagnosed before the PSA era and alsoto preclinical disease detected at trans-urethral resection for presumedbenignprostatic hyperplasia. Indeed, as showninTable 4, these 2 categories of patientsexperienced similar risk of dying fromprostate cancer. The natural historywehave described reflects alsowhatwouldhappen among PSA-detected cancer ifthe lead time could be accommodatedand the patients were left withoutearly therapeutic intervention. How-ever, a substantial proportion of PSA-diagnosed cancers represents overde-tection of subclinical disease. Thesecancerswouldneverhave surfacedclini-cally during thepatient’s lifetime, eitherbecause they are indolent or becausedeath occurs from competing causesbefore clinical manifestation of themalignancy. By definition, these can-cers do not generate any mortality.In conclusion, our data indicate that

the probability of progression to amoreaggressive and lethal phenotypemay in-crease after long-term follow-upof pros-tate cancers that are diagnosed at anearly stage and initially left withouttreatment. These findings argue forearly radical treatment of patients withlong life expectancy. Not only wouldsuch surgical intervention potentiallyprevent deaths, it would also conveyprevention fromdisability caused by lo-cal tumor growth.

Author Contributions: Dr Johansson, as principal in-vestigator, had full access to all of the data in this studyand takes responsibility for the integrity of these dataand the accuracy of the data analysis.Study concept and design: Johansson, Andersson,Adami.Acquisition of data: Johansson, Andren, Andersson.Analysis and interpretation of data: Johansson,Andersson, Dickman, Holmberg, Magnuson, Adami.Drafting of the manuscript: Johansson, Andren,Andersson, Magnuson, Adami.Critical revision of the manuscript for important in-





PAP

ER I

tellectual content: Johansson, Andren, Andersson,Dickman, Holmberg, AdamiStatistical expertise: Dickman, Magnuson.Obtained funding: Johansson.Administrative, technical, or material support:Johansson, Andren, Andersson, Adami.Study supervision: Adami.Funding/Support: This studywas supported by grantsfrom the Orebro County Council Research Commit-tee, the Orebro University Hospital Research Foun-dation, Obrebro, Sweden, and the Swedish CancerSociety.Disclaimer: Study sponsors had no role in data Col-lection, analysis, manuscript preparation, or authori-zation for publication.Acknowledgment: We thank the Orebro CountyCouncil Research Committee, University Hospital Re-search Foundation, and the Swedish Cancer Society.

REFERENCES

1. Chodak GW, Thisted RA, Gerber GS, et al. Resultsof conservativemanagement of clinically localized pros-tate cancer. N Engl J Med. 1994;330:242-248.2. Adolfsson J, Steineck G, Hedlund PO. Deferredtreatment of clinically localized low-grade prostate can-cer: actual 10-year and projected 15-year follow-upof the Karolinska series. Urology. 1997;50:722-726.3. Johansson JE, Holmberg L, Johansson S, Berg-strom R, Adami HO. Fifteen-year survival in prostatecancer: a prospective, population-based study in Swe-den. JAMA. 1997;277:467-471.4. Holmberg L, Bill-Axelson A, Helgesen F, et al. Arandomized trial comparing radical prostatectomywithwatchfulwaiting in early prostate cancer.NEngl JMed.2002;347:781-789.5. TNM Classification of Malignant Tumors. Re-

vised 3rd ed. Geneva, Switzerland: International UnionAgainst Cancer; 1978.6. World Health Organization. International Histo-logical Classification of Tumors. Vol 22. Geneva, Swit-zerland: World Health Organization; 1980.7. Cutler SJ, Ederer F. Maximum utilization of the lifetablemethod in analyzing survival. J ChronicDis. 1958;8:699-712.8. Hakulinen T. On long-term relative survival rates.J Chronic Dis. 1977;30:431-443.9. Clayton D, Hills M. Statistical Models in Epidemi-ology. Oxford, England:OxfordUniversity Press; 1993.10. Hakulinen T, Abeywickrama K. A computer pro-gram package for relative survival analysis. ComputPrograms Biomed. 1985;19:197-207.11. Ramaswamy S, Ross KN, Lander ES, Golub TR. Amolecular signature of metastasis in primary solid tu-mors. Nat Genet. 2003;33:49-54.12. Bernards R, Weinberg RA. A progression puzzle.Nature. 2002;418:823.13. Walsh PC, Lepor H. The role of radical prosta-tectomy in the management of prostatic cancer. Can-cer. 1987;60(3 suppl):526-537.14. Gerber GS, Thisted RA, Scardino PT, et al.Results of radical prostatectomy in men with clini-cally localized prostate cancer. JAMA. 1996;276:615-619.15. Blasko JC, Grimm PD, Sylsvester JE, CavanaghW.The role of external beam radiotherapy with I-125/Pd-103 brachytherapy for prostate carcinoma. Ra-diother Oncol. 2000;57:273-278.16. Hanks GE, Hanlon AL, PinoverWH, Horwitz EM,Price RA, Schultheiss T. Dose selection for prostate can-cer patients based on dose comparison and dose re-sponse studies. Int J Radiat Oncol Biol Phys. 2000;46:823-832.17. Oliver SE, Donovan JL, Peters TJ, Frankel S,

Hamdy FC, Neal DE. Recent trends in the use ofradical prostatectomy in England: the epidemiologyof diffusion. BJU Int. 2003;91:331-336.18. Albertsen PC, Fryback DG, Storer BE, Kolon TF,Fine J. Long-term survival among men with conser-vatively treated localized prostate cancer. JAMA. 1995;274:626-631.19. Aus G, Hugosson J, Norlen L. Long-term survivaland mortality in prostate cancer treated with noncu-rative intent. J Urol. 1995;154(2 pt 1):460-465.20. Borre M, Nerstrom B, Overgaard J. The naturalhistory of prostate carcinoma based on aDanish popu-lation treated with no intent to cure. Cancer. 1997;80:917-928.21. Helgesen F, Holmberg L, Johansson JE, Berg-strom R, Adami HO. Trends in prostate cancer sur-vival in Sweden, 1960 through 1988: evidence of in-creasing diagnosis of nonlethal tumors. J Natl CancerInst. 1996;88:1216-1221.22. Adolfsson J, Oksanen H, Salo JO, Steineck G. Lo-calized prostate cancer and 30 years of follow-up ina population-based setting. Prostate Cancer Pros-tatic Dis. 2000;3:37-42.23. LitwinMS, Hays RD, Fink A, et al. Quality-of-lifeoutcomes in men treated for localized prostate can-cer. JAMA. 1995;273:129-135.24. Bishoff JT, Motley G, Optenberg SA, et al. Inci-dence of fecal and urinary incontinence following radi-cal perineal and retropubic prostatectomy in a na-tional population. J Urol. 1998;160:454-458.25. Steineck G, Helgesen F, Adolfsson J, et al. Qual-ity of life after radical prostatectomy or watchful wait-ing. N Engl J Med. 2002;347:790-796.26. Catalona WJ, Smith DS, Ratliff TL, et al. Mea-surement of prostate-specific antigen in serum as ascreening test for prostate cancer.N Engl JMed. 1991;324:1156-1161.





115

PAPER II


PA

PE

IR

I

How Well Does the Gleason Score Predict Prostate CancerDeath? A 20-Year Followup of a Population Based Cohort in SwedenOve Andrén,* Katja Fall, Lennart Franzén, Swen-Olof Andersson, Jan-Erik Johansson andMark A. RubinFrom the Departments of Urology (OA, SOA, JEJ) and Pathology (LF), Örebro University Hospital, Örebro, Department of MedicalEpidemiology, Karolinska Institutet, Stockholm (KF), Sweden, and Department of Pathology, Brigham and Women’s Hospital, Boston,Massachusetts (MAR)

Purpose: Adenocarcinoma of the prostate is the most common cancer among men in Western countries. Although theprognostic heterogeneity of prostate cancer is enormous, clinically insignificant aggressive prostate cancers cannot be reliablydistinguished. Therefore, identifying prognostic factors is increasingly important, notably among men diagnosed withlocalized prostate cancer, because many of them may not require aggressive treatment.Materials and Methods: We analyzed a population based cohort of 253 men with early stage (T1a-b, Nx, M0) initiallyuntreated prostate cancer diagnosed between 1977 and 1991, before PSA screening was available. Tissue samples wereavailable for 240 patients diagnosed with transurethral resection. During complete followup through September 2003,standardized criteria were used to classify histopathological characteristics, progression and causes of death.Results: Higher Gleason grade, higher nuclear grade and larger tumor volume were independent predictors of death inprostate cancer with monotonous and statistically significant trends (p �0.05). In contrast, the level of Ki-67 – stronglycorrelated to Gleason score – was not an independent predictor of prostate cancer death. Given a Gleason score of 7 or greater,the probability of dying of prostate cancer was 29%. The corresponding predictive value for Gleason score 8 or greater was48%.Conclusions: Although a high Gleason score is a determinant of prostate cancer death, its PPV is relatively low. Thus,further efforts in finding other or complementary indicators of prostate cancer outcome are needed.

Key Words: prostatic neoplasms, prognosis, neoplasm staging

Adenocarcinoma of the prostate has emerged as one ofthe most common cancer forms in the Western world.Although some patients with prostate cancer experi-

ence a rapidly progressing or lethal disease, the majority ofpatients diagnosed at an early stage have a long expectedsurvival.1 While treatment such as radical prostatectomyhas been demonstrated as effective in reducing prostatecancer mortality, it carries with it side effects that mayimpair quality of life. Furthermore, it appears that a signif-icant proportion of men would have survived their cancereven without treatment.2 Not withstanding the apparentrisks of over treatment, the number of younger men under-going radical treatment is increasing. In Sweden the num-bers have increased from 1,107 in 1998 to 2,213 in 2002.3

Wider indications for biopsy testing, as proposed by some,3

would further increase this number. Therefore, identifica-tion of reliable predictors for the outcomes of this unpredict-able and highly variable disease is critical.

In addition to age and comorbidity4 at diagnosis, clinicalmanagement is mainly based on classification of tumor stage(TNM),5 grading (Gleason)6 and PSA. Tumor volume and

proliferation rate, assessed by Ki-67 immunoreactivity, havefurther been associated with prostate cancer death.7,8 Glea-son score is currently known as the most informative pre-dictor of prostate cancer outcome, especially for low and highgrade tumors. The prognostic value, however, seems lowerfor the intermediate group (Gleason 6 to 7), to which themajority of tumors belong.9 It has been suggested that theorder of the 2 values comprising the Gleason score is ofimportance,10 indicating that predominance of Gleason 4 (4� 3) would have a worse prognosis than predominance ofGleason 3 (3 � 4).

Extension of a previously described population based co-hort of untreated prostate cancer patients from Örebro, Swe-den, allowed us to examine patients with early localizedincidental tumors, diagnosed through TURP, to evaluate theinfluence of tumor grade, predominance pattern, volume,and proliferation rate (Ki-67) on prostate cancer death.

MATERIALS AND METHODS

The University Hospital in Örebro has a defined catchmentarea of about 190,000 inhabitants. All patients with voidingsymptoms and suspicion of prostate cancer were referred to theurological department at the hospital for diagnosis, treatmentand followup. From March 1977 through September 1991, atotal of 1,230 patients were diagnosed with prostate cancer inthe area. Of these, 252 patients were diagnosed through TURP

Submitted for publication May 18, 2005.Supported by grants from the Örebro County Research Founda-

tion, County of Norbotten Research Foundation, Swedish CancerSociety, and Maud and Birger Gustavsson Research Foundation.

* Correspondence: Department of Urology, Örebro UniversityHospital, SE-701 85 Örebro, Sweden (e-mail: [email protected]).

0022-5347/06/1754-1337/0 Vol. 175, 1337-1340, April 2006THE JOURNAL OF UROLOGY® Printed in U.S.A.Copyright © 2006 by AMERICAN UROLOGICAL ASSOCIATION DOI:10.1016/S0022-5347(05)00734-2

1337


or transvesical adenoma enucleation. Histopathological mate-rial was missing in 3 patients and in 9 patients there were nocancer found when reviewed by the pathologist. Thus, 240subjects were included in the analyses, although tissue wasonly available for tissue microarray in 190 patients. The caseswithout enough tissue to be included in the array were smallerin size but had the same distribution of Gleason score andprostate cancer death. Only patients without skeletal metas-tases at diagnosis were included in this study. At the time,there were no private urologists practicing in the area, and noscreening for prostate cancer took place during the study pe-riod. The baseline evaluation at diagnosis included physicalexamination, chest radiography, bone scan and skeletal radi-ography (if needed). For staging and grading of the tumors theTNM classification from 1992 were used.5 Nodal staging wasnot performed.

TreatmentNo initial treatment was offered to these patients with inci-dental prostate cancer. If the tumors progressed to symp-tomatic disease the patients were treated hormonally,typically with exogenous estrogens, gonadotropin-releasinghormone analogues, antiandrogen or bilateral orchiectomy.

FollowupThe patients were followed until death from prostate cancer,or censored at time of other death or until end of observationperiod in September 2003. No patient was lost to followup.The mean followup period was 108 months (range 1 to 280).A bone scan was performed every 6 to 12 months during thefirst 5 years, every second year thereafter. The medicalrecords of all deceased patients were reviewed. In mostinstances, the cause of death was obvious on clinical groundsalone, but autopsy was performed if the cause of death wasnot clear. Prostate cancer was recorded as the underlyingcause of death, a contributory cause of death or unrelated todeath. If treatment of the prostate cancer was related todeath (eg cardiovascular complications following estrogenadministration) prostate cancer was recorded as a contribu-tory cause.

Histopathological AnalysesTumor specimens were fixed in buffered formalin and em-bedded in paraffin and the resected tissue chips were ran-domly divided and placed in 6 paraffin blocks if enoughmaterial, otherwise in as many as possible. The averagenumber of paraffin blocks per case was 3.2. Grading accord-ing to nuclear grade11 was performed by 1 pathologist (BR).All specimens were reevaluated and graded according to theGleason6 in 2003 (MR). We further assessed tumor volumeby calculation of the ratio between the number of chips withcancer and the total number of chips (hereafter referred toas the ratio).7 The tissue micro arrays were assembled usingthe manual tissue arrayer (Beecher Instrument, SilverSpring, Maryland) as preciously described.12 Tissue coresfrom circled areas were targeted for transfer to recipientarray blocks. One to 2 replicate tissue cores were sampledfrom each patient sample. Selection of tumor was based onareas most representative of overall tumor Gleason grade.The 0.6 mm diameter tissue microarrays cores were eachspaced at 0.8 mm from core center to core center. Afterconstruction, 4 �m sections were cut and stained with he-

matoxylin and eosin on the initial slides to verify the histo-logical diagnosis. An additional 4 �m section was cut andstained with standard biotin-avidin complex immunohisto-chemistry with MIB antibodies to evaluate Ki-67.

Statistical MethodsWe used Cox regression models to examine the risk of dyingin prostate cancer as a function of various clinical parame-ters as well as Ki-67 levels. The values of Ki-67 intensitywere divided into quintiles based on the distribution amongthe patients, and the lowest quintile was used as the refer-ence. Relative hazard ratios were calculated before and afteradjustment for all included variables. We tested for lineartrend by constructing ordinal variables through assigningconsecutive integers to consecutive levels of categorizedvariables. The Pearson correlation test and the chi-squaretest were applied to formally assess correlation.

RESULTS

The baseline characteristics of the 240 patients and distri-bution according to tumor stage, nuclear grade, Gleasonscore, Ki-67 expression and tumor volume are shown intable 1. Of the 240 patients 118 (49.2%) patients were instage T1a and 122 (50.8%) in stage T1b. Mean age at diag-nosis was 73 years. Of the cases 84% (167) had Gleason score6 or 7 and only 4% (11) had a Gleason score below 6. 175patients (74%) had nuclear grade 1. Most tumors (49.2% or118) were small with a ratio less than 5%. By the end of theobservation period in September 2003, 207 of the 240 pa-tients (86%) (table 1) had died. Prostate cancer was a morecommon cause of death in patients 70 years of age andyounger in comparison with those older than 70 years, 20%(16) and 16% (26), respectively.

The age adjusted risk of dying in prostate cancer withrespect to nuclear grade, Ki-67, Gleason score and tumorvolume is shown in table 2. The risks tended to rise withincreasing Gleason score, nuclear grade and tumor volume(p for trends �0.05). There was a 9-fold statistically signif-icant increased risk for those with the highest Gleasonscore8,9 compared to those with the lowest.4–6 Although theanalysis was somewhat hampered by small numbers, therisk associated with a Gleason 3 predominant pattern (3 �4) did not seem different from that of a Gleason 4 predomi-nant pattern (4� 3) (data not shown). Nuclear grade III wasassociated with an almost 11-fold risk increase comparedwith nuclear grade I.

In the multivariate model Gleason score, nuclear grade, andtumor volume remained statistically significant independentprognostic markers of prostate cancer death. Ki-67 was highlycorrelated with Gleason grade, and the trend of increasingmortality rates with rising values of Ki-67, that was observedin the univariate analysis, disappeared after adjustment forthe other variables. Cases with T1b tumors (ratio greater than5%) were 2.5 times more likely to die of their disease than caseswith T1a tumors (HR 2.5, 95% CI 1.1-5.7)

In table 3 the sensitivity and specificity of the Gleasongrading system at different cutoff points are presented. Set-ting the cutoff value at 7 or greater resulted in a sensitivityof 0.64 and a specificity of 0.66. As the cutoff was increasedby 1 step, the corresponding values decreased to 0.33 andincreased to 0.92, respectively. Moreover, the PPV of Glea-son 7 or greater was 0.29 and the NPV was 0.89. With a

GLEASON SCORE FOR PREDICTING PROSTATE CANCER DEATH1338


PA

PE

IR

I

cutoff of 8 the corresponding values were 0.48 and 0.87,respectively.

DISCUSSION

In this population based cohort of incidentally discoveredlocalized prostate cancer cases (T1a-b, Nx, M0) left withoutinitial treatment, our main focus was evaluation of clinicaldeterminants of prostate cancer death. Deriving from a fol-lowup of more than 20 years, our results confirm the findings

of earlier studies that Gleason score is a predictor of prostatecancer death.13–15 However, the predictive value of the Glea-son grading is noticeably low.

The study cohort is well characterized, carefully followedduring an extended period of time and no patients have beenlost to followup. Validation of the classification of causes ofdeath against the Swedish Death Register showed a corre-lation greater than 90%, in our previous study.16 The popu-lation based study design with inclusion of all consecutivecases with newly diagnosed, incidentally discovered, pros-tate cancer from a strictly defined catchment area, furtheralleviates concerns of selection bias. The method of casefinding twenty years ago differs from contemporary prac-tices, particularly in areas where PSA testing is intensive. Itmay be argued that tumors detected in such settings maydiffer biologically from those included in our study, and thatthe generalizability of our results is limited. However, to ourknowledge there are few data to support such biologicaldifferences and we believe that our results should be validand generalizable.

Like earlier longitudinal cohort studies using prostate can-cer death as outcome we found Gleason score and nucleargrade to be determinant of prostate cancer death.13–15,17 Inrecent years there have been many reports like that of Roehl etal in which bRFS is used as the end point.17 The use of bRFSas surrogate end point for prostate cancer specific death has

TABLE 1. Characteristics at diagnosis of 240 patients with incidental prostate cancer (T1a-b, Nx, M0) who received no initial treatment

Total No.No. Prostate Ca

DeathsNo. Deaths ByOther Cause No. Alive

Mean MosSurvival

Min MosSurvival

Max MosSurvival

Age:Younger than 70 80 16 42 22 138 16 280Older than 70 160 26 123 11 90 1 272

Nuclear grade:*1 175 20 124 31 98 1 2802 45 13 31 1 94 4 2293 16 8 7 1 59 1 158

Gleason score:4 4 - 2 2 132 56 1675 7 3 3 1 122 5 2356 136 12 99 25 118 1 2847 64 13 46 5 96 1 2308 26 13 13 0 73 4 2299 3 1 2 0 49 17 85

Ratio chips with Ca (%):†5 or less 118 9 87 22 102 1 2376–25 92 20 61 11 97 1 28026–50 13 6 7 0 85 17 153Greater than 50 17 7 10 0 54 4 143

Ki67:Q1 (101) 40 6 24 10 123 6 280Q2 (112) 36 5 26 5 105 10 273Q3 (117) 39 10 26 3 94 1 228Q4 (122) 37 7 26 4 96 4 212Q5 (129) 38 9 25 4 98 8 211

* Only 236 patients had disease graded according to nuclear grade.† T1a has a ratio 5% or less. T1b has a ratio greater than 5%.

TABLE 2. Hazard ratio and 95% confidence interval of deathfrom prostate cancer

HR* 95% CI HR† 95% CI

Age:Younger than 70 1.0 1.0Older than 70 1.5 0.8–2.8 1.4 0.6–3.2

Nuclear grade:I 1.0 1.0II 3.5 1.7–7.1 1.4 0.5–3.5III 10.7 4.5–25.4 3.4 1.1–10.7

p (trend) 0.03Gleason score:

4–6 1.0 1.07 2.5 1.2–5.3 1.5 0.5–4.28–9 9.5 4.5–20.2 3.3 1.0–11.6

p (trend) 0.04Ratio chips with Ca (%):

5 or less 1.0 1.06–25 3.0 1.3–6.6 1.9 0.6–5.825–50 10.3 3.5–30.5 3.6 0.8–15.750 16.8 6.0–47.5 5.2 1.2–22.8

p (trend) �0.01Ki67:

Q1 1.0 1.0Q2 1.2 0.4–4.2 1.2 0.3–4.0Q3 2.6 0.9–7.6 1.4 0.4–4.9Q4 1.9 0.6–6.0 0.9 0.2–3.2Q5 2.5 0.9–7.4 1.3 0.4–4.2

p (trend) 0.86

* Age adjusted.† Adjusted for all other variables.

TABLE 3. Sensitivity, specificity, positive predictive value andnegative predictive value of Gleason grading in predicting

prostate cancer death

Gleason Score* Sensitivity Specificity PPV NPV

6 or Greater 0.92 0.04 0.17 0.737 or Greater 0.64 0.66 0.29 0.898 or Greater 0.33 0.92 0.48 0.87

* Cutoff point.

GLEASON SCORE FOR PREDICTING PROSTATE CANCER DEATH 1339


been debated,18,19 the absence of clear evidence that there is agood correlation between bRFS and prostate cancer specificdeath makes comparisons of studies with different end pointsdifficult. The accuracy of the Gleason grading system to cor-rectly classify patients according to their risk of dying of pros-tate cancer in this study was rather poor, and as aconsequence, the PPV of the Gleason grading system is ex-pected to be low. In these data, even a high set cut point(Gleason score 8 or greater) could only forecast a 50% proba-bility of dying in prostate cancer. Moreover, it appears that theevaluation according to the Gleason system has shifted towardintermediary values over time and in Sweden, more than 70%of the localized tumors are presently graded as Gleason 6 or 7.9

In the present data, more than 80% were diagnosed with Glea-son 6 or 7. A grading system that categorizes a majority of thecases into 1 group with highly variable outcomes is obviouslyinsufficient. In view of the increasing diagnostic activities inthe Western world and the resulting over treatment of indolentprostate cancers, the need for other or additional prognosticmarkers becomes even more evident.

The previous findings8 of Ki-67 as a prognostic marker forprostate cancer outcome could not be confirmed in thesedata. Possible explanations may be the comparably longfollowup time of the present study or the absence of moreadvanced tumors. In another followup study of tumors de-tected through TURP, Ki-67 was found to be a prognosticmarker of prostate cancer death, but this study includedpatients with more advanced disease (T1-T3, M0-M1).8 In-creased Ki-67 expression may precede the onset of moreadvanced disease, such as local infiltration and metastases,which may explain the difference between the study results.Comparisons with studies using PSA recurrence as outcomeare difficult, because of limitations with surrogate endpoints and the potential effects of surgical intervention.

The present data confirm the previous findings of tumorvolume at diagnosis as an important predictor of out-come.7,20 Interestingly, in these data the cut point of prog-nostic interest appeared to be higher (25% cancer) than thatused presently in the TNM classification (5% cancer).

CONCLUSIONS

These data draw attention to the apparent limitations of themost frequently used prognostic tools and highlight the needfor improvement in this field. Considering the increasingnumbers of early detected prostate cancers and the conse-quent over treatment of many benign tumors, further effortsin finding other, or complementary, indicators of prostatecancer outcome are warranted.

ACKNOWLEDGMENTS

Björn Risberg graded the material according to nuclear grade.

Abbreviations and Acronyms

bRFS � biochemical recurrence-free survivalNPV � negative predictive valuePPV � positive predictive valuePSA � prostate specific antigen

TURP � transurethral resection of the prostate

1. Johansson, J. E., Andren, O., Andersson, S. O., Dickman, P. W.,Holmberg, L., Magnuson, A. et al: Natural history of early,localized prostate cancer. JAMA, 291: 2713, 2004

2. Bill-Axelson, A., Holmberg, L., Ruutu, M., Haggman, M.,Andersson, S. O., Bratell, S. et al: Radical prostatectomyversus watchful waiting in early prostate cancer. N EnglJ Med, 352: 1977, 2005

3. Carter, H. B.: Prostate cancers in men with low PSA levels—must we find them? N Engl J Med, 350: 2292, 2004

4. Albertsen, P. C., Fryback, D. G., Storer, B. E., Kolon, T. F.and Fine, J.: The impact of co-morbidity on life expectancyamong men with localized prostate cancer. J Urol, 156:127, 1996

5. Hermanek, P. and Sobin, L. H.: TNM classification of MalignantTumours, 4th ed. New York: Springer-Verlag, 1987

6. Gleason, D. F.: Histologic grade, clinical stage, and patient agein prostate cancer. NCI Monogr, 7: 15, 1988

7. Humphrey, P. and Vollmer, R. T.: The ratio of prostate chipswith cancer. a new measure of tumor extent and its relation-ship to grade and prognosis. Hum Pathol, 19: 411, 1988

8. Stattin, P., Damber, J.-E., Karlberg, L. and Bergh, A.: Cellproliferation assessed by Ki-67 immunoreactivity on formalinfixed tissues is a predictive factor for survival in prostatecancer. J Urol, 157: 219, 1997

9. Regionalt Onkologiskt Centrum ROC. Prostatacancer. Primär-registrering 1997-2001. In: Regionalt Onkologiskt Centrum,2003

10. Egevad, L., Granfors, T., Karlberg, L., Bergh, A. and Stattin, P.:Percent Gleason grade 4/5 as prognostic factor in prostatecancer diagnosed at transurethral resection. J Urol, 168: 509,2002

11. Mostofi, F. K., Sesterhenn, I. A. and Sobin, L. H.: HistologicalTyping of Prostate Tumours. Geneva: World Health Organi-zation, 1980

12. Perrone, E. E., Theoharis, C., Mucci, N. R., Hayasaka, S., Tay-lor, J. M., Cooney, K. A. et al: Tissue microarray assessmentof prostate cancer tumor proliferation in African-Americanand white men. J Natl Cancer Inst, 92: 937, 2000

13. Egevad, L., Granfors, T., Karlberg, L., Bergh, A. and Stattin, P.:Prognostic value of the Gleason score in prostate cancer. BJUInt, 89: 538, 2002

14. Partin, A. W., Walsh, A. C., Pitcock, R. V., Mohler, J. L., Ep-stein, J. I. and Coffey, D. S.: A comparison of nuclear mor-phometry and Gleason grade as a predictor of prognosis instage A2 prostate cancer. a critical analysis. J Urol, 142:1254, 1989

15. Albertsen, P. C., Fryback, D. G., Storer, B. E., Kolon, T. F. andFine, J.: Long-term survival among men with conservativelytreated localized prostate cancer. JAMA, 274: 626, 1995

16. Johansson, J. E., Holmberg, L., Johansson, S., Bergstrom, R.and Adami, H. O.: Fifteen-year survival in prostate cancer. Aprospective, population-based study in Sweden. JAMA, 277:467, 1997

17. Roehl, K. A., Han, M., Ramos, C. G., Antenor, J. A. V. andCatalona, W. J.: Cancer progression and survival rates fol-lowing anatomical radical retropubic prostatectomy in 3,478consecutive patients. long-term results. J Urol, 172: 910,2004

18. Sandler, H. M. and DeSilvio, M. L.: Surrogate end points forprostate cancer: what is prostate-specific antigen telling us?J Natl Cancer Inst, 95: 1352, 2003

19. Kwan, W., Pickles, T., Duncan, G., Liu, M., Agranovich, A.,Berthelet, E. et al: PSA failure and the risk of death inprostate cancer patients treated with radiotherapy. Int JRadiat Oncol Biol Phys, 60: 1040, 2004

20. Lowe, B. A. and Listrom, M. B.: Incidental carcinoma of theprostate: an analysis of the predictors of progression. J Urol,140: 1340, 1988

GLEASON SCORE FOR PREDICTING PROSTATE CANCER DEATH1340


115

PAPER III


PAP

ER III

Decreased A-Methylacyl CoA Racemase Expression inLocalized Prostate Cancer is Associated with anIncreased Rate of Biochemical Recurrence andCancer-Specific Death

Mark A. Rubin,1,2,4 Tarek A. Bismar,1,2 Ove Andren,9 Lorelei Mucci,2,3 Robert Kim,1

Ronglai Shen,5 Debashis Ghosh,5 John T. Wei,6,8 Arul M. Chinnaiyan,6,7,8

Hans-Olov Adami,3,10 Philip W. Kantoff,2,4 and Jan-Erik Johansson9

1Department of Pathology, Brigham and Women’s Hospital, 2Harvard Medical School, 3Harvard School of Public Health,and 4Dana-Farber Harvard Cancer Center, Boston, Massachusetts; 5Department of Biostatistics, University ofMichigan School of Public Health; Departments of 6Urology and 7Pathology, 8University of MichiganMedical School, Ann Arbor, Michigan; 9Department of Urology, Orebro University Hospital,Orebro, Sweden; and 10Department of Medical Epidemiology and Biostatistics,Karolinska Institute, Stockholm, Sweden

Abstract

A-Methylacyl CoA racemase (AMACR) is overexpressed inprostate cancer relative to benign prostatic tissue. AMACRexpression is highest in localized prostate cancer anddecreases in metastatic prostate cancer. Herein, we exploredthe use of AMACR as a biomarker for aggressive prostatecancer. AMACR protein expression was determined byimmunohistochemistry using an image analysis system ontwo localized prostate cancer cohorts consisting of 204 mentreated by radical prostatectomy and 188 men followedexpectantly. The end points for the cohorts were time toprostate-specific antigen (PSA) failure (i.e., elevation >0.2 ng/mL) and time to prostate cancer death in the watchfulwaiting cohort. Using a regression tree method, optimalAMACR protein expression cutpoints were determined tobest differentiate prostate cancer outcome in each of thecohorts separately. Cox proportional hazard models were

then employed to examine the effect of the AMACR cutpointon prostate cancer outcome, and adjusted for clinicalvariables. Lower AMACR tissue expression was associatedwith worse prostate cancer outcome, independent of clinicalvariables (hazard ratio, 3.7 for PSA failure; P = 0.018; hazardratio, 4.1 for prostate cancer death, P = 0.0006). Among thosewith both low AMACR expression and high Gleason score,the risk of prostate cancer death was 18-fold higher (P =0.006). The AMACR cutpoint developed using prostatecancer–specific death as the end point predicted PSA failuresindependent of Gleason score, PSA, and margin status. Thisis the first study to show that AMACR expression issignificantly associated with prostate cancer progressionand suggests that not all surrogate end points may be optimalto define biomarkers of aggressive prostate cancer. (CancerEpidemiol Biomarkers Prev 2005;14(6):1424–32)

Introduction

The prevalence of pathologic prostate cancer is extremely highand increases with age. Pathologic prostate cancer is seen inautopsy series in men in their 20s and 30s and increases to>80% in men over the age of 70 (1). Prostate-specific antigen(PSA) testing has facilitated the detection of prostate cancer,which may have contributed to the decline in mortality fromprostate cancer over the past few years (2, 3). However, PSAmay also have introduced an increase in the detection ofclinically insignificant disease (4-6). Moreover, the relationshipbetween pretherapy clinical variables and outcomes havebecome obscured in part due to an eroding relationshipbetween serum PSA and the abundance of cancer and/or

higher grade cancer in the gland. Given the incidence andprevalence of prostate cancer, the ease of diagnosis, the agingof the population, and the morbidity of treatment, the ability todistinguish aggressive versus indolent forms of prostate canceris critical.

Clinicians treating prostate cancer patients can assess therisk that a prostate cancer poses to the patient by usingpretreatment nomograms that have been developed andvalidated to predict prostate cancer recurrence after treatmentfor localized disease (7-10). These nomograms account forserum PSA levels, prostate needle biopsy Gleason score, andclinical stage. At best, these clinical nomograms have an areaunder the receiver operating characteristic curve of 0.75.Kattan et al. have recently shown that the addition of twoserum markers (IL6SR and TGF-h1) improve the area underthe receiver operating characteristic curve to 0.83, suggestingthat molecular markers in combination with clinical variablescould improve on existing predictive nomograms in theirability to predict recurrence after local therapy (11). Molecularbiomarkers are being characterized in order to help refinethese clinical nomograms, and moreover, to help distinguishaggressive from indolent prostate cancer in an effort to sparemen from unnecessary treatment (12, 13).

a-Methylacyl CoA racemase (AMACR) is a biomarker thatwas identified by both differential display and expressionarray analysis as a gene abundantly expressed in prostatecancer relative to benign prostate epithelium (14-16). In ametaanalysis of four cDNA expression array data sets,

Cancer Epidemiology, Biomarkers & Prevention1424

Cancer Epidemiol Biomarkers Prev 2005;14(6). June 2005

Received 11/4/04; revised 2/9/05; accepted 3/21/05.

Grant support: Specialized Program of Research Excellence for Prostate Cancer, NationalCancer Institute grant P50CA90381 (P.W. Kantoff and M.A. Rubin), P50CA69568 (M.A. Rubinand A.M. Chinnaiyan), National Cancer Institute grant CA 97063 (A.M. Chinnaiyan andM.A. Rubin), R01AG21404 (M.A. Rubin) and the American Cancer Society RSG-02-179-MGO(A.M. Chinnaiyan and M.A. Rubin), the Swedish Cancer Society (O. Andren and J-E.Johansson), and by GMP companies in accordance with Partners Health Organization (BWH)and the University of Michigan.

The costs of publication of this article were defrayed in part by the payment of page charges.This article must therefore be hereby marked advertisement in accordance with 18 U.S.C.Section 1734 solely to indicate this fact.

Note: This represents an original research study that was presented in part at the NationalCancer Institute – sponsored Specialized Program of Research Excellence meeting in Baltimore,MD, July 2004.

Requests for reprints: Mark A. Rubin, Department of Pathology (Amory 3-195), Brigham andWomen’s Hospital, Harvard Medical School, 75 Francis Street, Boston, MA 02115.Phone: 617-525-6747; Fax: 617-264-5169. E-mail: [email protected]

Copyright D 2005 American Association for Cancer Research.


AMACR was one of the genes most consistently overexpressedin prostate cancer (17). AMACR is a peroxisomal andmitochondrial enzyme that plays an important role in bileacid biosynthesis and h-oxidation of branched-chain fattyacids through the interconversion of (R)- and (S)-2-methyl-branched-chain fatty acyl-CoA fragments (18). Our groupinitially reported that AMACR expression is consistently lowerboth at the transcriptional (cDNA expression arrays and RT-PCR) and at the protein level (Western blot analysis andimmunohistochemistry) in metastatic prostate cancer com-pared with localized prostate cancer (14, 19). More recently, afluorescent-based measurement of AMACR in tissue samplesconfirmed these observations (20).

In the current study, we describe the development of aquantitative AMACR protein expression test to determine therisk of progression for men with clinically localized prostatecancer and for men treated for localized prostate cancer.

Materials and Methods

Patient Cohorts

Surgical Cohort. This cohort consisted of 204 patients fromthe University of Michigan (Ann Arbor, MI), who underwentradical retro pubic prostatectomy as a primary therapy (i.e., nopreceding hormonal or radiation therapy) for clinicallylocalized prostate cancer between 1994 and 1998. Clinical andpathologic data for all patients were acquired with approvalfrom the Institutional Review Board at the University ofMichigan. Clinical data regarding this cohort has beenreported separately (21, 22). Disease progression was definedas a serum PSA increase >0.2 ng/mL after radical prostatec-tomy. Patients were considered censored if they have not had aPSA biochemical failure at the last follow-up time evaluatedfor that individual. Mean follow-up time in the cohort was4.8 (2.4) years. In this cohort, 60% of cases were Gleason 7 ormore. Mean age at diagnosis was 60 years.

Watchful Waiting Cohort . This cohort is the largest population-based watchful waiting cohort, and consists of patients fromOrebro, Sweden with clinically localized prostate cancer, whounderwent watchful waiting. This cohort, initially described in1989 (23), consists of men who all presented with voidingsymptoms referred to the urology department to rule out thediagnosis of prostate carcinoma. From March 1977 throughSeptember 1991, 1,230 patients were diagnosed with prostatecancer in Orebro County. Among these, 253 were diagnosedthrough transurethral resection of the prostate and theserepresent the study base for the watchful waiting cohort. Nonewere diagnosed by PSA screening. For the current investigation,cases were excluded due to insufficient amount of tumor (n = 39),inadequate immunohistochemistry (n = 13), inability to confirmthe original diagnosis of cancer (n = 9), or initially presenting forcystoprostatectomy due to bladder cancer (n = 1). Thus, datafrom 188 watchful waiting cases were included in this study.

The baseline evaluation of these patients at diagnosisincluded physical examination, chest radiography, i.v. pyelo-gram, bone scan, and skeletal radiography (if needed). Lymphnode staging was not done. In accordance with standardpractices at that time in Orebro, these patients were initiallyfollowed expectantly (‘‘watchful waiting’’). Patients weretreated with androgen deprivation therapy only if theyexhibited symptoms. Patient follow-up included clinicalexaminations, laboratory tests, and bone scans every 6 monthsduring the first 2 years following the initial prostate cancerdiagnosis and subsequently every 2 years. Medical records ofall deceased patients have been reviewed to determine cause ofdeath. As a validation, the classification of cause of death wascompared with that recorded in the Swedish Death Register.Thus far, agreement on cause of death has been >90%, with no

evidence of systematic over- or underestimation of prostatecancer as cause of death. Through March 2003, with up to 23years of follow-up, 36 (19.2%) of the patients in this cohort diedof prostate cancer. The remaining patients are consideredcensored, having either died of other causes (126 or 67.0%) orwere still alive without disease at time of last follow-up (26 or13.8%). No patients have been lost to follow-up.

In order to ensure a uniform review of the pathology, one ofthe study pathologists reviewed all cases from both series.Uniform pathology review included Gleason grading, anestimate of overall tumor involvement (tumor burden pertissue samples evaluated), and tumor type (peripheral zoneversus transition zone). Although there are no strict criteria fordistinguishing a transition zone tumor from a peripheral zonetumor that has invaded the transition zone, we defined thetransition zone tumors for the sake of this analysis as tumorswith Gleason score of 6 and below with a well-circumscribedgrowth pattern. For staging and grading of the tumors, theTNM classification from 1992 (6) and the WHO classification(8) were used. Of the 188 patients in the watchful waitingcohort, 75 (39.9%) were stage T1a and 113 (60.1%) were foundto have T1b. The mean age at diagnosis was 73 years.

Tissue Microarray Construction. The tissue microarrays(TMA) from both patient cohorts were assembled using themanual tissue arrayer (Beecher Instruments, Silver Spring, MD)as previously described (24). Tissue cores from circled areaswere targeted for transfer to the recipient array blocks. Three tofive replicate tissue cores were sampled from each patientsample. In all cases, the dominant prostate cancer nodule or thenodule with the highest Gleason pattern was sampled for theTMA. The 0.6-mm diameter TMA cores were each spaced at 0.8mm from core-center to core-center. Six TMA blocks with anaverage of 480 cores per block were used for this study. Allblocks contained benign prostate tissue as well as prostatecancer. Each block was assembled without prior knowledge ofassociated clinical or pathology staging information. Afterconstruction, 4-Am sections were cut and stained with H&E onthe initial slides to verify the histologic diagnosis. All data ismaintained on a relational database as previously described (25).

Immunohistochemistry. Pretreatment conditions and incu-bations were worked out for AMACR immunostaining using acommercially available monoclonal antibody directed againstAMACR (p504s, Zeta Co., Sierra Madre, CA). Pretreatmentincluded placing the slide in a pH 6.0 citrate buffer andmicrowaving for 30 minutes. Primary p504s antibodies wereincubated for 40 minutes at room temperature. Secondary anti-mouse antibodies applied for 30 minutes and the enzymaticreaction was completed using a streptavidin biotin detectionkit (Dako Developing System, Dako, Carpinteria, CA) for 5minutes. Optimal primary antibody concentration was deter-mined by serial dilutions, optimizing for maximal signalwithout background immunostaining.

Manual Scoring of AMACR. All TMA cores were assigneda diagnosis (i.e., benign, atrophy, PIN, or prostate cancer) bythe two study pathologists. Prostate cancer samples wereonly included in the analysis if both reviewers agreed that itwas cancer. All manual scoring was done on an Internet-based image evaluation tool that employs zoomable TMAimages generated by the BLISS Imaging System (Bacus Lab,Lombard, IL). The AMACR protein expression was evaluatedusing a categorical scoring method ranging from negative tostrong staining intensity as previously reported (14).

Semiautomated Quantitative Image Analysis of AMACR.A semiautomated quantitative image analysis system, ACIS II(Chromavision, San Juan Capistrano, CA), was used toevaluate the same TMA slides from both cohorts. The ACISII device consists of a microscope with a computer-controlledmechanical stage. Proprietary software is used to detect the

Cancer Epidemiology, Biomarkers & Prevention 1425



PAP

ER III

brown stain intensity of the chromogen used for theimmunohistochemical analysis and compares this value toblue counterstain used as background. Theoretical intensitylevels range from 0 to 255 chromogen intensity units. In pilotexperiments for this study, the reproducibility of the ACIS IIsystem was tested and confirmed by scoring several TMAs onseparate occasions. The correlation coefficient for these experi-ments was r2 = 0.973. Because of tissue heterogeneity, one ofthe study pathologist digitally circled the areas of histologi-cally recognizable prostate cancer using the ACIS II softwarefor each TMA core. This process ensured that AMACRintensity measurements were from prostate cancer tissue onlyand not the surrounding benign glands or stroma.

Statistical Analysis. AMACR intensity readings wereobtained for each of the TMA slides separately and were thennormalized within each array before combining the data foranalysis. After several pilot studies (data not shown), wedetermined that normalization of the data was necessary.Despite using the same protocol for immunohistochemicalstaining, experiment-to-experiment variation was observed.Therefore, we normalized the AMACR intensity readings foreach TMA core on a given array prior to merging all of the datafor the final analysis. Critical for the normalization process wasthe presence of approximately equally distributed numbers ofnormal and cancer samples on each TMA. AMACR stainingintensity readings for each TMA core from a given array weresubtracted by the mean intensity for that same array and thendivided by the SD:

intensitynormalizedij ¼

intensityij �mean ðintensityiÞSD ðintensityiÞ

where j = 1, . . ., ni (ni is the total number of cores on TMAi). Asa result, each of the normalized arrays has mean score of 0 andSD equal to 1. Data were then combined using this normalizedscale.

In order to determine an optimal cutpoint for AMACR, weused a modification of the method of regression trees (26)applied to censored data (27). The regression tree method is anestimation procedure that selects a cutpoint for AMACR basedon optimizing a discriminating measure using the censored

failure time outcome. The method employs a likelihoodcriterion to optimize the cutpoint, and assumes that the costof a false-positive and false-negative are equal. We further didan adjusted analysis for determining a cutpoint, whichinvolves obtaining Martingale residuals (28) at the first stageby adjusting for potential confounders and then applying theregression tree algorithm in order to find a cutpoint. Theadjusted method allowed for the cutpoint to be determinedaccounting for clinical variables.

The cutpoints for the AMACR intensity scores had atheoretical range between 0 and 255 intensity units. Usingthe regression tree method, we determined the cutpoint thatbest differentiated PSA biochemical failure in the 204 patientsfrom the PSA-screened surgical series. A similar process wasrepeated for the Orebro watchful waiting cohort (n = 188 cases)using cancer-specific death as the end point.

Once the cutpoints were determined for each cohort, wethen applied the cutpoint to the other cohort. For example, wetested the optimal cutpoint derived using the surrogate endpoint (PSA failure) on the watchful waiting cohort todetermine if it would predict a true end point (prostatecancer–specific death). We then tested the cutpoint derivedusing prostate cancer–specific death as the end point on thesurgical series to see if it would predict PSA biochemicalfailure. We further employed Cox proportional hazardsregression analysis to examine the association between theAMACR cutpoint and time to prostate cancer outcome, takinginto account other clinical variables.

Results

Manual Evaluation of AMACR. AMACR protein expres-sion was evaluated manually by the study pathologist andgraded on a four-tiered scale. We found a significantdifference in intensity between prostate cancer (mean score =3.14/4) and benign prostate epithelium (mean score = 1.3/4)with a mean difference of 1.84 (ANOVA post hoc Scheffeanalysis, P < 0.00001). In the surgical series, no significantassociations between AMACR intensity scores and biochem-ical failure were observed, consistent with previous observa-tions (14).

Table 1. AMACR tissue expression and prostate cancer patient demographics among men (n= 204) with clinically localizedprostate cancer treated by radical prostatectomy

Mean (range) AMACR AMACR expression (tertiles)

1 2 3

�1.8 (�4.3 to �1.2) �0.77 (�1.20 to �0.28) +0.40 (�0.24 to +1.9)

CharacteristicMean age (y) 60.6 60.6 59.4

Gleason score (%)4/5 6.0 1.4 3.06 (3+3) 31.3 42.9 35.87 (3+4) 43.3 40.0 47.87 (4+3) 13.4 12.9 10.58/9 6.0 2.9 3.0

Mean preoperative PSA 9.3 7.5 8.7Mean tumor weight (g) 52.7 52.4 49.2Surgical margin status

Negative 67.2 72.9 71.6Positive 32.8 27.1 28.4

Extraprostatic extension (%) 25.4 21.4 23.9Abnormal digital

rectal examination (%)41.8 42.9 43.3

PSA recurrence duringfollow-up (%)

37.3 15.7 17.9

NOTE: Surgical margin status, tumor size, digital rectal examination results, and extraprostatic extension, by two-sample t test for pretreatment PSA serum levels,age, and gland weight.

AMACR Expression Associated with Prostate Cancer Progression1426



Semiautomated Quantitative AMACR Expression Analysis

The Surgery Cohort with Time to PSA Recurrence. In Table 1,we present clinical characteristics of the surgery cohort inrelation to tertiles of AMACR expression. There were onlysmall differences in clinical characteristics comparing menwith lower and higher AMACR tissue expression. In contrast,men with the lowest levels of AMACR expression were morethan twice as likely to experience PSA recurrence duringfollow-up compared to those with higher levels.

Using the adjusted regression tree method, we established adichotomous cutpoint for AMACR intensity of 1.11 SD (i.e.,samples with minimum AMACR intensity of 1.11 SD belowthat of the mean), wherein 37.5% of patients with AMACRintensity scores below the cutpoint had PSA biochemicalfailure compared with 14.5% of patients with AMACRintensity scores above the cutpoint (P = 0.0002). This uni-variate association can be visually appreciated by Kaplan-Meier analysis (Fig. 1). Tables 2 and 3 present the univariateand multivariate associations, respectively, between AMACR

expression, clinical characteristics and time to PSA recurrence.In multivariate analysis, patients with AMACR expressionlevels below the threshold were at a significantly higher risk ofdeveloping PSA recurrence [P = 0.04, hazard ratio (HR) 2.12;95% confidence interval (CI), 1.04-4.32] after adjusting forpreoperative PSA, Gleason score, and surgical margin status.In complimentary analyses, we found that the likelihood ofPSA recurrence increased 60% with each additional SDdecrease in AMACR expression (P = 0.004). Comparing thelowest to highest tertiles of AMACR expression, the HR (95%CI) was 2.80 (1.37-5.72).

The Watchful Waiting Cohort with Time to Prostate CancerDeath. In the watchful waiting cohort, AMACR tissueexpression correlated with clinical characteristics (Table 4).Cases with lowest AMACR expression were also more likely tobe clinically stage T1b and to have Gleason > 7.

Using the adjusted regression tree method, we determinedan optimal cutpoint of 0.18 SD using time to prostate cancerdeath as the outcome. The univariate and multivariate analysis

Figure 1. A. Screen shot of the automated cellular imaging system (ACIS II) software interface demonstrating a low-power view of multipleTMA samples immunostained for AMACR (left) and a higher power view of two TMA cores (right). B. The AMACR intensity for each TMAcore was determined after the area of the tissue containing prostate cancer was circled digitally (data not shown). Using the normalized AMACRcutoff of �1.11, examples of three high and low AMACR expressing prostate cancers as determined by the ACIS system are presented withtheir raw intensity scores. C. As described in Materials and Methods, the raw AMACR intensity readings required normalization due toexperiment-to-experiment variation (‘‘chip to chip’’ variation) despite the same immunohistochemical protocol. Two histograms show thedistribution of scores for the raw (left, original scale) and normalized AMACR analysis. The majority of cases had AMACR expression valueswithin a 2 SD range. D. Lower AMACR intensity was associated with prostate cancer progression as determined by PSA biochemical failure.PSA biochemical recurrence occurred in 30 of 80 (37.5%) men with AMACR scores below the cutpoint of �1.11 as compared with 18 of 124(14.5%) with AMACR scores above the cutpoint (log-rank P value of P < 0.001).




PAP

ER III

for the association between the +0.18 SD AMACR cutpoint andprostate cancer–specific survival are presented in Table 5. Asshown in Table 5 and Fig. 2, there was a significant associationbetween the AMACR cutpoint and prostate cancer death in themultivariable analysis. Adjusting for clinical characteristics,the HR for prostate cancer death after >20 years of follow-upwas 4.06 (95% CI 1.82-9.06; P = 0.0006) comparing those withlow AMACR levels to those with high levels. The effect ofAMACR was independent of age at diagnosis, Gleason score,and tumor stage as evidenced in the multivariable analysis.The association between the AMACR cutpoint and prostatecancer death was constant across follow-up time (dataavailable upon request). In complimentary analyses, weobserved that the likelihood of PSA recurrence increased50% with each additional SD decrease in AMACR expression(P = 0.023). Men in the lowest tertile of AMACR expressionwere 3-fold more likely to die of prostate cancer over follow-upcompared with those in the highest tertile.

Cross-validation of the AMACR Cutpoints. After develop-ment of an optimal AMACR expression cutpoint to identifymen at highest risk of developing PSA biochemical recurrence,we sought to explore if this cutpoint developed with PSArelapses following radical prostatectomy also predicted cancer-specific death among patients left without curative treatmentby applying the �1.11 SD cutoff on the watchful waitingcohort. Similarly, we examined the AMACR cutpoint deter-mined in the watchful waiting cohort (0.18 SD) to the surgerycohort (Table 6). Using the surgical cutpoint of �1.11 SD,AMACR intensity did not predict prostate cancer–specificsurvival in univariate or multivariate analysis. The multivar-iate adjusted HR for prostate cancer death after 20 years offollow-up was 1.09 (95% CI, 0.33-3.62; P = 0.89). In contrast,there was evidence that AMACR levels below the cutpointdetermined by prostate cancer death was significantly associ-ated with time to PSA biochemical failure at the univariatelevel as shown by the Kaplan-Meier analysis (Fig. 2) Onmultivariate analysis, AMACR expression <0.18 SD cutpointwas a significant predictor of biochemical failure, independentof Gleason score, surgical margin status, and pretreatmentPSA, with a HR of 3.67 (95% CI, 1.25-10.78; P = 0.018; Table 6).

Lymph node status, tumor stage, and seminal vesicle invasionwere associated with a 50% greater likelihood of PSA failurebut were not significant predictors of outcome in the finalmultivariate model. As this was a PSA-screened population,only 4 of 204 had positive lymph nodes and 10 of 204 hadseminal vesicle invasion. After excluding these cases, therewas no significant difference in the HRs for either cutpoint(data not shown).

Cross-classifying AMACR Expression and Gleason Score.The application of a biomarker in the clinical setting wouldlikely accommodate clinical characteristics of patients as wellin assessing prostate cancer prognosis. In Table 7, we examinethe joint association between AMACR expression and Gleasonscore on prostate cancer outcome in the two cohorts. For thisanalysis, we rely on the AMACR cutpoint of 0.18 SD.Compared to those with ‘‘better’’ biomarker and clinicalmeasures (i.e., high AMACR expression and low Gleasonscore), those with both low AMACR expression and highGleason score had an almost four times higher risk of PSAbiochemical failure. In the watchful waiting cohort, however,individuals with the ‘‘poorer’’ measures had an 18-fold higherrisk of prostate cancer death (P = 0.006). In fact, among the 12prostate cancer deaths with Gleason score of 6 or less, usingthe AMACR cutpoint appropriately predicted 11 as deaths.

Table 2. Univariate Cox regression analysis for associationswith prostate-specific biochemical failure

Variables Univariate

Censored(n = 156)

Recurred(n = 48)

HR 95% CIfor HR

P

AMACR<�1.11 (low) 32.1% 62.5% 2.89 1.47-5.68 0.002z�1.11 (high) 67.9% 37.5% REF

Age, per year 60 y 61 y 1.01 0.97-1.05 0.61Gleason score

8-9 1.9 10.5 10.7 3.40-33.9 <0.00017 50.0 75.0 4.04 1.79-9.08 0.0007<7 48.0 24.6 REF

ln(PSA), perunit increase

7.2 ng/mL 12.7 ng/mL 2.41 1.72-3.38 <0.0001

Tumor size (cm)>2 89.7% 66.7% 3.49 1.91-6.37 <0.0001V2 10.3% 33.3% REF

Surgical margin statusPositive 79.5% 41.7% 3.61 2.26-5.76 <0.0001Negative 20.5% 58.3% REF

Extraprostatic extensionExtension 84.6% 50.0% 2.79 1.97-3.94 <0.0001No extension 15.4% 50.0% REF

Digit rectal examPositive 61.5% 43.8% 1.90 1.07-3.36 0.03Negative 38.5% 56.3% REF

NOTE: ln(PSA), the natural logarithm of pretreatment PSA.

Table 3. Multivariate Cox regression model: independentassociations with PSA recurrence for men with clinicallylocalized prostate cancer following radical prostatectomy

Variable Multivariate adjusted

HR 95% CI P

AMACR<�1.11 (low) 2.12 1.04-4.32 0.039z�1.11 (high) REF

Gleason score8+ 2.97 0.44-20.0 0.267 3.31 0.94-11.6 0.065-6 REF

Surgical margin statusPositive 2.92 1.68-5.08 0.0001Negative REF

ln(PSA) 1.70 1.09-2.67 0.020

NOTE: Data in multivariate model are adjusted for AMACR cutpoint, Gleasonscore, surgical margin status, and natural log of PSA.

Table 4. AMACR tissue expression and clinical character-istics, Orebro watchful waiting cohort, Sweden

Mean (range)AMACR

AMACR expression (tertiles)

1 2 3

�1.0(�1.5 to �0.6)

�0.15(�0.56 to +0.27)

+1.2(+0.28 to +3.8)

CharacteristicsMean age, years 72.9 74.2 74.1

T stage (%)T1a 47.6 46.8 25.4T1b 52.4 53.2 74.6

Gleason score (%)4/5 4.8 8.1 3.26 (3+3) 60.3 61.3 38.17 (3+4) 12.7 14.5 30.27 (4+3) 7.9 8.1 11.18/9 14.3 8.1 17.5

Status in 2003 (%)Prostate cancer death 19.1 21.0 17.5Other death 68.3 62.9 69.8Alive 12.7 16.1 12.7




Furthermore, among the four prostate cancer deaths withGleason score 6 or less and tumor stage T1a, all were correctlypredicted as death. These data indicate that the AMACRbiomarker, in combination with clinical variables, can sub-stantially predict prostate cancer mortality.

Discussion

This is the first demonstration that AMACR protein expressionlevels are associated with long-term clinical outcome inearly prostate cancer. Our original observations suggestedthat AMACR expression decreases with prostate cancerprogression (14, 19). However, standard immunohistochemicalevaluation failed to identify clinically significant associa-tions between AMACR expression and biochemical failure(14, 29, 30). Using a more sensitive technique to measureAMACR protein expression, we were able to identify anAMACR cutpoint that was associated with a three timesgreater chance of biochemical recurrence and a three toseven greater risk of developing cancer-specific death. Theseobservations were independent of conventional clinical andpathologic variables in multivariable analysis.

The paradigm for developing prostate cancer biomarkershas focused on using PSA biochemical failure as the surrogateend point for the development of metastatic disease andcancer-specific death. Studies based on surrogate end pointsare inherently less reliable than studies with clinical end pointsthat therapeutic intervention aims to prevent or delay, such ascancer-specific death (31). PSA failure is considered a

surrogate for the progression of metastatic disease andprostate cancer–specific death (32, 33). Indeed, studies byPound et al. support the potential role of PSA failure as aclinically relevant surrogate end point (33). For example, of 304men who develop a PSA biochemical failure following radicalprostatectomy for clinically localized disease, 103 (34%)developed metastatic disease with a mean actuarial time tometastases of 8 years following the initial PSA elevation. ThePound study also showed that time from initial diagnosis tobiochemical failure and PSA doubling time predicted meta-static disease. These findings suggest that biochemical failureper se is not a complete surrogate for clinical outcomes, inparticular death, and should take into account PSA kinetics aswell (34). Additional support that biochemical failure may notbe an optimal end point was recently reported by D’Amicoet al. (35). They evaluated surrogate end points for prostatecancer-specific mortality in two multi-institutional databasesof over 8,669 patients with prostate cancer treated with surgery(5,918 men) or radiation (2,751 men). The posttreatment PSAdoubling time was significantly associated with time toprostate cancer–specific mortality and with time to all-causemortality.

The current study is consistent with the view that PSAfailure may not be the optimal end point to use to predictprostate cancer–specific death and thus not optimal for thedevelopment of prostate cancer biomarkers. We observed thatthe AMACR cutpoints differed depending on whether it wasderived from PSA failure or cancer-specific death. Biochemicalfailure may be due to causes unrelated to the biology ofprostate cancer such as local recurrence due to positive

Table 5. Univariate and multivariate Cox models: association between AMACR and prostate cancer–specific death inwatchful waiting cohort, Orebro 1977-2003

Variable Censored(n = 152)

Prostate cancerdeath (n = 36)

Unadjusted Multivariate adjusted

HR 95% CI P HR 95% CI P

AMACR<0.18 (low) 60.1 69.4 1.40 0.68-2.83 0.36 4.06 1.82-9.06 0.0006z0.18 (high) 39.9 30.6 REF REF

Age (per year) 74.1 y 72.5 y 1.03 0.99-1.08 0.17 1.01 0.97-1.06 0.58T stage

T1b 46.1 13.9 5.12 1.98-13.21 0.0007 3.61 1.30-9.97 0.013T1a 53.9 86.1 REF REF

Gleason score8+ 8.5 33.4 10.45 4.55-24.0 <0.0001 13.03 4.88-34.8 <0.00017 26.9 33.3 2.84 1.27-6.38 0.011 2.70 1.12-6.52 0.0275-6 62.7 33.3 REF REF

NOTE: Data in multivariate model are adjusted for AMACR cutpoint, age, T stage, and Gleason score.

Figure 2. A. Lower AMACR intensity is associated with prostate cancer–specific survival in a watchful waiting cohort with up to 20 years ofclinical follow-up (see tables) using a cutpoint of 0.18. This histogram shows the distribution of AMACR intensity units for the tumors from thesurgical series. B. Using the continuous values, one appreciates lower expression in the 48 cases with a PSA biochemical failure versus thepatients without disease progression (error bars with 95% CI). C. The Kaplan-Meier curve shows a PSA recurrence–free survival advantagewith patients whose tumor had lower AMACR protein expression using the 0.18 cutpoint (see text for details as to how this cutpoint wasdetermined).




PAP

ER III

surgical margins. Moreover, given the slow progression ofprostate cancer, even 8 years may not be sufficient to identifyall patients who will die from prostate cancer. Therefore,censoring underestimates the number of men who willultimately experience prostate cancer progression. The workfrom Pound et al. and D’Amico et al. would suggest that thesurrogate end point needs further consideration.

However, we also need to be cautious as there are otherpossible explanations as to why the cutpoint derived bybiochemical failure did not predict prostate cancer–specificdeath. There are several important differences between the twopopulations including eras collected (i.e., pre- and post-PSAscreening era), populations (i.e., Swedish versus U.S.), andtreatment (i.e., watchful waiting versus surgery). Theoretically,a clinical trial such as the Scandinavian Prostate Cancer GroupIV study (36), where patients were randomized to be eitherfollowed expectantly or treated by surgery, would beextremely useful in the development of prostate cancerbiomarkers to predict cancer-specific death. Given the impor-tant need for the development of uniformly collected cohorts,there is currently an important effort by the National CancerInstitute–supported prostate cancer Specialized Programs ofResearch Excellence groups to develop such resources.

The use of prostate cancer–specific death to develop abiomarker test is likely a more valid strategy. In the currentstudy, using cancer-specific death to determine the AMACRcutoff yielded a different cutoff from that derived from thesurrogate end point. Using this revised cutoff on the patientstreated for clinically localized disease, AMACR expressionindependently predicted PSA relapse, even accounting forpretreatment PSA, surgical margin status, and Gleason score.The high-risk group included 81% of the patients in thissurgical cohort suggesting that using PSA recurrence as thesurrogate end point, we are missing some patients who wouldhave progressed. This finding is not what one would anticipateif biochemical failure also included ‘‘false-positives’’ due toother cases such as surgical technique. If these data areconfirmed, then risk of progression given sufficient time is notbeing adequately measured using one elevation of PSAfollowing treatment. The most recent review of the entireOrebro watchful waiting series identified a significant increasein prostate cancer–specific mortality beyond 15 years offollow-up (37). This would also further support the need forbetter prognostic markers. In the randomized Swedish study,

survival benefits were observed for surgery over watchfulwaiting but the absolute reduction in prostate cancer mortalitywas small (36). Longer follow-up might show even greaterbenefits from localized therapy given the significant percent-age of men who developed metastatic disease over the 8 yearsof follow-up (36).

This study also highlights one of the technical limitationsof prior biomarker work. Most studies to date, including ourown work, used manual evaluation dividing immunostain-ing results into a small number of categories. Althoughreasonable reproducibility may be achieved using a four-tiered (i.e., negative, weak, moderate, or strong), three-tieredor dichotomous scale, the range of expression is compressedand the ability of the pathologist to distinguish, for example,various shades of moderate staining intensity is not possible.The current study illustrates how critical this might be in thedevelopment of prostate cancer biomarkers (Fig. 3). Severalstudies on AMACR failed to identify an association withdecreased expression and prostate cancer progression (14,19, 29, 30, 38). Our initial observations suggested thatAMACR expression decreased with cancer progression asboth at the transcriptional level and protein level AMACRwas lower in metastatic prostate samples (14, 19). However,using manual standard pathology review, no clinicalassociations were discovered. More recent work using afluorescence-based approach confirmed lower AMACR ex-pression in metastatic samples. This finding suggests that aquantitative method might be required to identify a criticalcutoff to distinguish clinically localized tumors with subtledifferences in AMACR expression levels (20). The currentstudy supports this hypothesis. No significant cutoff couldbe determined using a standard evaluation of immunohis-tochemical markers by the pathologist. In contrast, the use ofa continuous scale of protein measurement allowed us toidentify a critical cutoff. This cutoff could then be translatedinto a specific test using an intensity level that was notappreciated despite numerous studies on AMACR to date.

The use of this semiautomated quantitative method forbiomarker analysis is promising. However, the clinicalapplication of such systems is still in its early phases. Resultsneed to be validated and tested in different cohorts andstandardization needs to be worked out. Specifically, can wedetermine a cutpoint that would be useful to apply to prostateneedle biopsies in the clinical assessment of men with clinically

Table 6. Effect of derived AMACR cutpoints in the watchful waiting cohort with prostate cancer death as outcomecompared with Michigan cohort with biochemical recurrence as outcome

AMACR cutpoint Surgical cohort PSA biochemical failure Watchful waiting cohort (Prostate Cancer Death)

HR* (95% CI) P HR* (95% CI) P

Prostate cancer deathLow (<0.18) 3.67 (1.25-10.78) 0.018 4.06 (1.82-9.06) 0.0006High (z0.18) REF REF

Biochemical failureLow (<�1.11) 2.12 (1.04-4.32) 0.039 1.09 (0.33-3.62) 0.89High (z�1.11) REF REF

*Data were adjusted for Gleason score; surgical margin status, and the natural log of pretreatment serum PSA.

Table 7. HR (95% CI) of prostate cancer outcome associated with AMACR expression and Gleason score, cross-classified

Surgical cohort, PSA biochemical failure Watchful waiting cohort, prostate cancer death

Low AMACR High AMACR Low AMACR High AMACR

Gleason z 7 3.9 (0.53-29.2) 1.7 (0.21-14.5) 18.0 (2.3-140.8) 4.9 (0.63-37.9)Gleason < 7 1.2 (0.12-10.9) REF 6.4 (0.8-51.2) REF

Note: Data are adjusted for surgical margins and PSA (surgical cohort) or age and T stage (watchful waiting cohort). AMACR cutpoint = 0.18.




localized prostate cancer? One future approach would be toevaluate prostate needle biopsies for AMACR stainingintensity and determine if this information adds to currentlyemployed clinical nomograms. As recently described byKattan et al., one could explore if the addition of AMACRintensity improves the receiver operating characteristic areaunder the curve as compared with a traditional nomogram(11, 39). However, as seen in this current study using aregression tree method to identify the optimal cutpoints,different populations may end up having different significantcutpoints after adjusting for different covariates. Therefore,predictive models using molecular data will be no differentthan standard nomograms as both approaches require care-fully defining the clinical cohort.

Another reasonable approach to developing cutpoint is tomake them cohort-specific biomarkers. The biomarkers couldtheoretically be tested in half of the cohorts and validated inthe remaining cases. If the testing phase took into account theclinical variables, one might predict that the model wouldstand a better chance of being predictive in the validation setwhere clinical variables would also be required to show thatthe molecular biomarker improves over the current clinicalmodel.

One can also use a metric called the concordance index todetermine how much a predictive model with the molecularbiomarker improves over and beyond that with the clinicalmodels alone. The concordance index is the probability that,given two randomly selected patients, the patient with theworse outcome is, in fact, predicted to have a worseoutcome (40). This measure ranges from 0.5 (i.e., chance)to 1.0 (perfect ability to rank patients). This step is important

because the ultimate goal of identifying molecular signaturesfor prostate cancer outcome is to improve the current clinicalmodels (39). If the molecular signature is only informative inthat it predicts the clinical markers, then the signature willnot be useful in the clinical setting. One limitation of thisstudy was that the relative small size of the two groupsprecluded dividing them up for testing and validation sets.We did, however, try a leave-one-out cross-validatedanalysis in which each sample in the study was held out,and various cutpoints of AMACR were considered. Thepredictive power of AMACR was assessed using the C-index(41) with the response being time to biochemical failure inthe Michigan cohort and time to death in the Orebro cohort.The cross-validated values of the C-index for the twocohorts, across a variety of cutpoints, ranged between 0.5and 0.6.

The current study now extends the potential utility ofAMACR as a prostate cancer biomarker. The AMACR (p504s)antibody is gaining wide acceptance in clinical practices as anadjuvant tool in the workup of diagnostically challengingprostate lesions referred to as atypical small acinar proliferationor atypical suspicious for cancer (14, 19, 29, 38, 42-48).Pathologists can now use AMACR (p504s) in combinationwith a basal cell marker to help make a more definitivediagnosis. We also recently observed a measurable humoralresponse directed against the AMACR protein (49). Using anELISA assay, this response could be measured in serumsamples from men with known prostate cancer but not in age-matched controls. This early work suggests that AMACRexpression, despite the fact that it is not a secreted protein, maypotentially become a clinically useful serum test. A second

Figure 3. Prostate cancer biomarker development requires selecting an end point for the study. After initial diagnosis, several treatment optionsare available for men with clinically localized prostate cancer such as radiation therapy, surgery, or observation (also known as expectant therapyor watchful waiting). In most cases, prostate cancer biomarkers are developed using samples from prostatectomy series as this provides the bestsource of tumor tissue. Patients are clinically followed and the development of either a PSA elevation following surgery, the development ofmetastatic disease or cancer-specific death are all end points used for the development of prostate cancer biomarkers. However, most studies usePSA biochemical failure (surrogate end point) as the end point given the long time required to observe clinical events. It is assumed thatbiochemical failure is a surrogate end point for the development of these clinical events. However, the current study suggests that the surrogateend point may not be sufficient to define cutpoints for predicting cancer-specific death. Specifically, the AMACR cutpoint used to predict PSAbiochemical failure did not predict cancer-specific death, whereas the AMACR cutpoint used for predicting cancer-specific death identified adistinct cutpoint that predicted biochemical failure in the surgical cohort.




PAP

ER III

tissue-based assay can also detect enzymatic activity inprostate needle biopsy samples. This assay may representanother means of determining AMACR protein expressionmore reproducibly (50).

In summary, we showed for the first time that the AMACRexpression is associated with prostate cancer progression afterexamining tumors from >440 men diagnosed with clinicallylocalized prostate cancer. Selection of the optimal test cutpointcame from analysis using prostate-specific death and nota surrogate end point, suggesting that the definition of thesurrogate end points, such a PSA biochemical failure need tobe further evaluated. Future work will need to determine ifthis AMACR tissue test can be applied to prostate needlebiopsies in a prospective manner to predict the risk ofrecurrence.

AcknowledgmentsWe thank Julia Elvin for constructive comments during the develop-ment of this project and Martina Storz-Schweizer and Lela Schumacherfor their technical assistance.

References1. Sakr WA, Sarkar FH, Sreepathi P, Drozdowicz S, Crissman JD. Measurement

of cellular proliferation in human prostate by AgNOR, PCNA, and SPF.Prostate 1993;22:147–54.

2. Weinmann S, Richert-Boe K, Glass AG, Weiss NS. Prostate cancer screeningand mortality: a case-control study (United States). Cancer Causes Control2004;15:133–8.

3. Labrie F, Candas B, Cusan L, et al. Screening decreases prostate cancermortality: 11-year follow-up of the 1988 Quebec prospective randomizedcontrolled trial. Prostate 2004;59:311–8.

4. Stamey TA, Johnstone IM, McNeal JE, Lu AY, Yemoto CM. Preoperativeserum prostate specific antigen levels between 2 and 22 ng/ml correlatepoorly with post-radical prostatectomy cancer morphology: prostate specificantigen cure rates appear constant between 2 and 9 ng/ml. J Urol 2002;167:103 –11.

5. Klotz L. Active surveillance: an individualized approach to early prostatecancer. BJU Int 2003;92:657.

6. Barry MJ. Clinical practice. Prostate-specific-antigen testing for earlydiagnosis of prostate cancer. N Engl J Med 2001;344:1373–7.

7. Partin AW, Kattan MW, Subong EN, et al. Combination of prostate-specificantigen, clinical stage, and Gleason score to predict pathological stage oflocalized prostate cancer. A multi-institutional update. JAMA 1997;277:1445– 51.

8. Kattan MW, Eastham JA, Stapleton AM, Wheeler TM, Scardino PT. Apreoperative nomogram for disease recurrence following radical prosta-tectomy for prostate cancer. J Natl Cancer Inst 1998;90:766– 71.

9. Kattan MW, Zelefsky MJ, Kupelian PA, Scardino PT, Fuks Z, Leibel SA.Pretreatment nomogram for predicting the outcome of three-dimensionalconformal radiotherapy in prostate cancer. J Clin Oncol 2000;18:3352 –9.

10. Kattan MW, Potters L, Blasko JC, et al. Pretreatment nomogram forpredicting freedom from recurrence after permanent prostate brachytherapyin prostate cancer. Urology 2001;58:393– 9.

11. Kattan MW, Shariat SF, Andrews B, et al. The addition of interleukin-6soluble receptor and transforming growth factor h1 improves a preoperativenomogram for predicting biochemical progression in patients with clinicallylocalized prostate cancer. J Clin Oncol 2003;21:3573– 9.

12. Varambally S, Dhanasekaran SM, Zhou M, et al. The polycomb groupprotein EZH2 is involved in progression of prostate cancer. Nature 2002;419:624– 9.

13. Rhodes DR, Sanda MG, Otte AP, Chinnaiyan AM, Rubin MA. Multiplexbiomarker approach for determining risk of prostate-specific antigen-defined recurrence of prostate cancer. J Natl Cancer Inst 2003;95:661– 8.

14. Rubin MA, Zhou M, Dhanasekaran SM, et al. a-Methylacyl coenzyme Aracemase as a tissue biomarker for prostate cancer. JAMA 2002;287:1662 –70.

15. Luo J, Zha S, Gage WR, et al. a-Methylacyl-CoA racemase: a new molecularmarker for prostate cancer. Cancer Res 2002;62:2220 –6.

16. Xu J, Stolk JA, Zhang X, et al. Identification of differentially expressed genesin human prostate cancer using subtraction and microarray. Cancer Res2000;60:1677–82.

17. Rhodes DR, Barrette TR, Rubin MA, Ghosh D, Chinnaiyan AM. Meta-analysis of microarrays: interstudy validation of gene expression profilesreveals pathway dysregulation in prostate cancer. Cancer Res 2002;62:4427– 33.

18. Ferdinandusse S, Denis S, L IJ, Dacremont G, Waterham HR, Wanders RJ.Subcellular localization and physiological role of a-methylacyl-CoA race-mase. J Lipid Res 2000;41:1890 –6.

19. Kuefer R, Varambally S, Zhou M, et al. a-Methylacyl-CoA racemase:expression levels of this novel cancer biomarker depend on tumordifferentiation. Am J Pathol 2002;161:841 –8.

20. Rubin MA, Zerkowski MP, Camp RL, et al. Quantitative determination ofexpression of the prostate cancer protein a-methylacyl-CoA racemase usingautomated quantitative analysis (AQUA): a novel paradigm for automatedand continuous biomarker measurements. Am J Pathol 2004;164:831 –40.

21. Nelson CP, Dunn RL, Wei JT, Rubin MA, Montie JE, Sanda MG.Contemporary preoperative parameters predict cancer-free survival afterradical prostatectomy: a tool to facilitate treatment decisions. Urol Oncol2003;21:213–8.

22. Nelson CP, Rubin MA, Strawderman M, Montie JE, Sanda MG. Preoperativeparameters for predicting early prostate cancer recurrence after radicalprostatectomy [Discussion 5–6]. Urology 2002;59:740 –5.

23. Johansson JE, Adami HO, Andersson SO, Bergstrom R, Krusemo UB, KraazW. Natural history of localised prostatic cancer. A population-based studyin 223 untreated patients. Lancet 1989;1:799 –803.

24. Rubin MA, Dunn R, Strawderman M, Pienta KJ. Tissue microarray samplingstrategy for prostate cancer biomarker analysis. Am J Surg Pathol 2002;26:312 –9.

25. Manley S, Mucci NR, De Marzo AM, Rubin MA. Relational databasestructure to manage high-density tissue microarray data and images forpathology studies focusing on clinical outcome: the prostate specializedprogram of research excellence model. Am J Pathol 2001;159:837 –43.

26. Breiman L, Friedman J, Olshen R, Stone C. Classification and regressiontrees. In: Breiman L, editor. Kluwer Academic Publishers; 1984. p. 368.

27. LeBlanc M, Crowley J. Relative risk trees for censored survival data.Biometrics 1992;48:411 –25.

28. Therneau T, Grambsch P, Fleming T. Martingale residuals for survivalmodels. Biometrika 1990;77:147–60.

29. Beach R, Gown AM, De Peralta-Venturina MN, et al. P504S immunohisto-chemical detection in 405 prostatic specimens including 376 18-gauge needlebiopsies. Am J Surg Pathol 2002;26:1588–96.

30. Jiang Z, Woda BA, Rock KL, et al. P504S: a new molecular marker for thedetection of prostate carcinoma. Am J Surg Pathol 2001;25:1397 –404.

31. Schatzkin A, Gail M. The promise and peril of surrogate end points in cancerresearch. Nat Rev Cancer 2002;2:19 – 27.

32. Stamey TA, McNeal JE, Yemoto CM, Sigal BM, Johnstone IM. Biologicaldeterminants of cancer progression in men with prostate cancer. JAMA1999;281:1395–400.

33. Pound CR, Partin AW, Eisenberger MA, Chan DW, Pearson JD, Walsh PC.Natural history of progression after PSA elevation following radicalprostatectomy. JAMA 1999;281:1591 –7.

34. Swindle PW, Kattan MW, Scardino PT. Markers and meaning of primarytreatment failure. Urol Clin North Am 2003;30:377–401.

35. D’Amico AV, Moul JW, Carroll PR, Sun L, Lubeck D, Chen MH. Surrogateend point for prostate cancer-specific mortality after radical prostatectomyor radiation therapy. J Natl Cancer Inst 2003;95:1376 –83.

36. Holmberg L, Bill-Axelson A, Helgesen F, et al. A randomized trialcomparing radical prostatectomy with watchful waiting in early prostatecancer. N Engl J Med 2002;347:781 –9.

37. Johansson JE, Andren O, Andersson SO, et al. Natural history of early,localized prostate cancer. JAMA 2004;291:2713–9.

38. Luo JH, Yu YP, Cieply K, et al. Gene expression analysis of prostate cancers.Mol Carcinog 2002;33:25 –35.

39. Kattan MW. Judging new markers by their ability to improve predictiveaccuracy. J Natl Cancer Inst 2003;95:634 –5.

40. Harrell FE Jr, Califf RM, Pryor DB, Lee KL, Rosati RA. Evaluating the yieldof medical tests. JAMA 1982;247:2543–6.

41. Harrell FE Jr, Lee KL, Califf RM, Pryor DB, Rosati RA. Regressionmodelling strategies for improved prognostic prediction. Stat Med 1984;3:143 –52.

42. Epstein JI. Diagnosis and reporting of limited adenocarcinoma of theprostate on needle biopsy. Mod Pathol 2004;17:307–15.

43. Zhou M, Jiang Z, Epstein JI. Expression and diagnostic utility ofa-methylacyl-CoA-racemase (P504S) in foamy gland and pseudohy-perplastic prostate cancer. Am J Surg Pathol 2003;27:772–8.

44. Magi-Galluzzi C, Luo J, Isaacs WB, Hicks JL, de Marzo AM, Epstein JI.a-Methylacyl-CoA racemase: a variably sensitive immunohistochemicalmarker for the diagnosis of small prostate cancer foci on needle biopsy.Am J Surg Pathol 2003;27:1128–33.

45. Kunju LP, Rubin MA, Chinnaiyan AM, Shah RB. Diagnostic usefulness ofmonoclonal antibody P504S in the workup of atypical prostatic glandularproliferations. Am J Clin Pathol 2003;120:737 –45.

46. Yang XJ, Wu CL, Woda BA, et al. Expression of a-methylacyl-CoA racemase(P504S) in atypical adenomatous hyperplasia of the prostate. Am J SurgPathol 2002;26:921–5.

47. Jiang Z, Wu CL, Woda BA, et al. P504S/a-methylacyl-CoA racemase: auseful marker for diagnosis of small foci of prostatic carcinoma on needlebiopsy. Am J Surg Pathol 2002;26:1169 –74.

48. Kang D, Jiang H, Wu Q, Pestka S, Fisher PB. Cloning and characterization ofhuman ubiquitin-processing protease-43 from terminally differentiatedhuman melanoma cells using a rapid subtraction hybridization protocolRaSH. Gene 2001;267:233– 42.

49. Sreekumar A, Laxman B, Rhodes DR, et al. Humoral immune responseto a-methylacyl-CoA racemase and prostate cancer. J Natl Cancer Inst2004;96:834–43.

50. Kumar-Sinha C, Shah RB, Laxman B, et al. Elevated a-methylacyl-CoAracemase enzymatic activity in prostate cancer. Am J Pathol 2004;164:787 –93.




115

PAPER IV


PAP

ER IV

MUC-1 gene is associated with prostate cancer death: a 20-yearfollow-up of a population-based study in Sweden

O Andren*,1, K Fall2, S-O Andersson1, MA Rubin3, TA Bismar3, M Karlsson4, J-E Johansson1 and LA Mucci5,6

1Department of Urology, Orebro University Hospital, Orebro 701 85, Sweden; 2Department of Medical Epidemiology and Biostatistics, KarolinskaInstitutet, Stockholm 171 77, Sweden; 3Department of Pathology, Brigham and Womens Hospital, Boston, MA 02215, USA; 4Department of Pathology,Orebro University Hospital, Orebro 701 85, Sweden; 5Department of Medicine, Channing Laboratory, Harvard Medical School/Brigham and Women’sHospital, Boston, MA 02215, USA; 6Department of Epidemiology, Harvard School of Public Health, Boston, MA 02215, USA

Anti-adhesion mucins have proven to play an important part in the biology of several types of cancer. Therefore, we test thehypothesis that altered expression of MUC-1 is associated with prostate cancer progression. We retrieved archival tumour tissuefrom a population-based cohort of 195 men with localised prostate cancer (T1a-b, Nx, M0) that has been followed for up to 20 yearswith watchful waiting. Semi-automated, quantitative immunohistochemistry was undertaken to evaluate MUC-1 expression. Wemodelled prostate cancer-specific death as a function of MUC-1 levels accounting for age, Gleason grade and tumour extent, andcalculated age-adjusted and multivariate adjusted hazard ratios (HR). Men that had tumours with an MUC-intensity lower or higherthan normal tissue had a higher risk of dying in prostate cancer, independent of tumour extent and Gleason score (HR 5.1 and 4.5,respectively). Adjustment for Gleason grade and tumour stage did not alter the results. Men with a Gleason score X7 and MUC-1deviating from the normal had a 17 (RR¼ 17.1 95% confidence interval¼ 2.3–128) times higher risk to die in prostate cancercompared with men with Gleason scoreo7 and normal MUC-1 intensity. In summary, our data show that MUC-1 is an independentprognostic marker for prostate cancer death.British Journal of Cancer (2007) 97, 730–734. doi:10.1038/sj.bjc.6603944 www.bjcancer.comPublished online 28 August 2007& 2007 Cancer Research UK

Keywords: prostate cancer; MUC-1; population-based; prognostic marker; twenty year follow-up; anti-adhesion

��

Cancer metastasis involves the dysregulation of a complexinterplay of multiple pathways. One critical step towards meta-static potential involves the detachment of tumour cells from thesurrounding environment. Anti-adhesion molecules inhibit thecell–cell and cell–extra-cellular matrix interactions, and maypromote development of metastatic disease by down regulatingcellular adhesion.

The mucin family of anti-adhesion molecules have beenimplicated in the biological behaviour and progression of severaltypes of cancer (Bresalier et al, 1991; Gendler and Spicer, 1995).The mucin MUC-1, which is expressed at the apical cell surface ofmany normal secretory epithelial cells (Ho et al, 1993), contains anextra-cellular domain that extends above most other cellmembrane-associated proteins (Hilkens et al, 1992; Hilkens et al,1995). As such, MUC-1 has been suggested to prevent adhesion andto promote development of metastatic disease. In prostate cancer,overexpression of MUC-1 in tissue has been correlated both withhigher Gleason grade and advanced tumour stage (Kirschenbaumet al, 1999). One study has, furthermore, suggested that MUC-1expression may predict prostate cancer recurrence after prosta-tectomy (Lapointe et al, 2004), although these results have beenchallenged by others (O’Connor et al, 2005; Zellweger et al, 2005).

The disparate findings may in part be explained by the use of PSA-recurrence as a measure of outcome, as biochemical failure doesnot necessarily herald prostate cancer death (Jhaveri et al, 1999).

In the present study, we test the hypothesis that altered tumourexpression of MUC-1 is associated with prostate cancer death. Wenest the study in a population-based cohort of men with localisedprostate cancer who have been followed prospectively for morethan 20 years.

MATERIALS AND METHODS

The study population is a cohort of all cases of early prostatecancer (T1a-b, Nx, M0) diagnosed at the Orebro UniversityHospital, Sweden between 1977 and 1991 by transurethralresection of the prostate or transvesical adenoma enucleation forsymptomatic benign prostatic hyperplasia (Andren et al, 2006). Asno private institutional care was available in the region at the timeand as the population was required to seek care within their countyof residence, the cohort can be considered population-based.Baseline evaluation at diagnosis included physical examination,chest radiography, bone scan and skeletal radiography (if needed).Nodal staging was not carried out. In accordance with standardpractice at that time in Orebro, these patients were initiallyfollowed expectantly (‘watchful waiting’) (Johansson et al, 2004).Patients received clinical exams, laboratory tests and bone scans

Received 2 April 2007; revised 29 June 2007; accepted 20 July 2007;published online 28 August 2007

*Correspondence: Dr O Andren; E-mail: [email protected]

British Journal of Cancer (2007) 97, 730 – 734

& 2007 Cancer Research UK All rights reserved 0007 – 0920/07 $30.00

www.bjcancer.com

Clin

icalStu

die

s


every 6 months during the first 2 years after diagnosis andsubsequently at 12-month intervals. Patients that developedmetastases, as judged by bone scan, were treated with androgendeprivation therapy if they exhibited symptoms.

Cause of death in the cohort was determined by review ofmedical records by the study investigators. An autopsy wasperformed if the cause of death was not clear. A validation studyregarding cause of death compared with the Swedish DeathRegister showed greater than 90% concordance, with no systematicunder- or over-reporting of any cause of death. Follow-up of thecohort with respect to mortality was 100%, and no patients werelost to follow up.

We retrieved archival tissue specimens (formalin-fixed paraffinembedded) and H&E slides from all 240 cases in the cohort, and wehad sufficient tumour tissue available for a total of 195 cases.Histological examination was performed by one pathologist(MAR) for Gleason grade who also assessed tumour extent bycalculating the ratio of the number of chips with cancer and thetotal number of chips (Humphrey and Vollmer, 1988). High-density tissue micro arrays (TMAs) were assembled from the TUR-P specimens using the manual tissue arrayer (Beecher Instrument,Silver Spring, MD, USA) as described previously (Perrone et al,2000). Two representative cores were taken from each tumourspecimen. Benign tissue was also included on the TMA. Afterconstruction, a 4 mm section was cut and stained with standardbiotin–avidin complex immunohistochemistry antibodies (Mucin1(VU4H5): sc 73–13) to evaluate MUC-1. Semi-automated,quantitative immunohistochemistry was undertaken using theChromavision system, and protein intensity was measured on ascale from 0 to 255 (Figure 1A–D).

Through March 2003, with up to 23 years of follow-up(mean¼ 9, median¼ 8), 37 (19.0%) patients in this cohort haddied of prostate cancer. The remaining patients were consideredcensored having either died of other causes (132 or 67.7%) or werestill alive without disease at time of last follow-up (26 or 13.8%).We estimated person-time in the cohort as time between date ofdiagnosis to cancer death, or censored at death from other cause orend of follow-up (October 2003). We used Cox regression to modeltime to prostate cancer death as a function of MUC-1 levelsaccounting for age, Gleason grade and tumour extent. For eachindividual, we calculated the mean intensity across the two cores.We defined normal MUC-1 intensity as the mean intensity inbenign prostate tissue70.25 standard deviations (s.d.). Individualswhose MUC-1 tumour expression was within the normal rangerepresented the reference group. Individuals whose tumourexpression was above or below normal were so classified. UsingCox-regression, we calculated age-adjusted and multivariateadjusted hazard ratios (HR). We tested for linear trend forcontinuous variables using the McPhearson’s test. Furthermore, wecalculated the sensitivity and specificity of MUC-1 proteinexpression as a predictor of prostate cancer death to explore itsusefulness as a biomarker of prostate cancer outcomes.

RESULTS

The baseline characteristics of the 195 patients are presented inTable 1. Mean MUC-1 expression for all 195 patients was 107.3(range 95–179, s.d. 10.2). Of the 43 patients that had an MUC-1intensity close to normal tissue (Figures 1 and 2) (102.5–106),

Figure 1 Tissue micro array analysis of MUC-1 immunohistochemistry: selected images of TMA cores representing normal, high and low MUC-1 intensity(A, B). Normal MUC-1 intensity (C, D). High (C) and Low (D) MUC-1 intensity.

MUC-1 gene is associated with prostate cancer death

O Andren et al

731

British Journal of Cancer (2007) 97(6), 730 – 734& 2007 Cancer Research UK

ClinicalStu

die

s


PAP

ER IV

three (7%) died of prostate cancer, compared with 34 (23%) of the152 patients that deviated from the normal MUC-1 intensity(Figure 2 and 3). As illustrated in Table 2, there was no correlationbetween Gleason score and MUC-1 intensity (P-value 0.8), whereasthere was a tendency to correlation between tumour extent andMUC-1 (P-value 0.08).

The age-adjusted risk of dying of prostate cancer with respect toMUC-1, Gleason score and tumour extent is presented in Table 3.The risk of dying of prostate cancer was four times higher amongthose with a higher (3.9 (95% confidence interval (CI)¼ 1.1–14))or lower (3.8 (95%CI¼ 1.1–13)) MUC-1 expression than amongthose with an MUC-1 expression within the normal range. Afteradjusting for tumour extent and Gleason score, the effect ofMUC-1 was even stronger (HR 5.1 (95%CI¼ 1.4–18) and 4.5(95%CI¼ 1.3–15), respectively), indicating that MUC-1 predictsprostate cancer death independently of clinical parameters.

Table 1 Characteristics of 195 patients with incidental prostate cancer (T1a-b, Nx, M0) who received no initial treatment, according to age, Gleasonscore, tumour extent and MUC-1 expression at time of diagnosis 1977–1991

NProstate cancer

deathsDeaths by other

causes AliveMean survival

(month)Minimum

survival (month)Maximum

survival (month)

Ageo70 56 13 27 16 138 16 280470 139 24 105 10 90 1 272

Gleason score4 3 — 1 2 132 56 1675 7 3 3 1 122 5 2356 107 10 77 20 118 1 2847 53 12 38 3 96 1 2308 22 11 11 0 73 4 2299 3 1 2 0 49 17 85

Percentage of chips with cancero5% 80 6 58 16 102 1 2376–25% 85 18 57 10 97 1 28026–50% 13 6 7 0 85 17 153450% 17 7 10 0 54 4 143

MUC-1Normal 43 3 35 5 118 4 284Low 68 16 38 14 108 1 280High 84 18 59 7 100 1 238

0 50 100 150 200 250 300Months

0.0

0.2

0.4

0.6

0.8

1.0

Cum

ulat

ive

surv

ival

MUC intensityHigh MUC

Low MUC

Normal MUC

High MUC-censored

Low MUC-censoredNormal MUC-censored

Figure 2 Survival curves of 195 patients according to high-, low- andnormal MUC-1 intensity.

0 50 100 150 200 250 300Months

0.0

0.2

0.4

0.6

0.8

1.0

Cum

ulat

ive

surv

ival

Combination of Gleasonand MUC intensity

High or low MUC,Gleason<7High or low MUC,Gleason>6

Normal MUC,Gleason<7Normal MUC,Gleason>6

High or low MUC,Gleason<7-censored

High or low MUC,Gleason>6-censored

Normal MUC,Gleason<7-censoredNormal MUC,Gleason>6-censored

Figure 3 Survival curves of 195 patients according to combination ofGleason score and MUC-1 intensity.

Table 2 Correlation between MUC-1 intensity expression and age,Gleason score and tumour volume

MUC-1 intensity

Factor Normal N (%) Low N (%) High N (%) P-value

Age o70 10 (23) 25 (37) 21 (25) 0.188Age 470 33 (77) 43 (63) 63 (75)Gleason 4–6 25 (58) 41 (60) 51(61) 0.826Gleason 7 13 (30) 20 (29) 20 (24)Gleason 8–9 5 (12) 7 (10) 13(16)Percent chips o5% 15 (35) 23 (34) 42 (50) 0.085Percent chips 5–25% 23 (54) 33 (49) 29 (35)Percent chips 25–50% 0 (0) 5 (7) 8 (10)Percent chips 450% 5 (12) 7 (10) 5 (6)


O Andren et al

732

British Journal of Cancer (2007) 97(6), 730 – 734 & 2007 Cancer Research UK

Clin

icalStu

die

s


We further cross-classified participants on MUC-1 and Gleasonscore. The group with Gleason score X7 and MUC-1 lower orhigher than normal had a 17 (HR¼ 17.1 (95%CI¼ 2.3–128)) timeshigher risk of prostate cancer death compared with tumours withGleason score o7 and normal MUC-1 intensity (Table 4).

We further assessed the ability of MUC-1 expression (deviatingfrom the normal range) to correctly classify prostate cancercases as indolent or lethal (defined as progressing to metastasesand/or death). The sensitivity for MUC-1 as predictor oflethal prostate cancer was 0.91, whereas the specificity was0.25 (Table 5). When combined with information on Gleasonscore, the specificity increased to 0.75 but the sensitivity decreasedto 0.56.

DISCUSSION

In this population-based cohort of men with localised prostatecancer cases (T1a-b, Nx, M0), we found that MUC-1 expression atdiagnosis was a predictor of prostate cancer death. Individualswith dysregulated (either over or under) MUC-1 expression had afour- to five-fold increased risk of dying of prostate cancer,independent of clinical parameters. We further examined theaccuracy of using MUC-1 as a test for prostate cancer progressionand found that using MUC-1 alone resulted in a reasonable

sensitivity but poor specificity. The addition of information onGleason Score improved the specificity of the MUC-1 biomarker,but at the cost of a lower sensitivity. Still, these data suggest thatMUC-1 may be promising to include in a panel of molecularmarkers to distinguish aggressive disease from indolent atdiagnosis.

These data are in line with the accumulating evidence of the roleof MUC-1 in the cell–cell interaction. On the one hand,overexpression of MUC-1 has been shown to increase themetastatic potential of the cancer cells (Evangelou et al, 2002;Schut et al, 2003; Huang et al, 2004). On the other, Kontani et al(2001) suggested that the loss of MUC-1 expression or modulationof its antigenicity might cause cancer cells to be unresponsive tothe effect of cytotoxic T lymphocytes. These competing mechan-isms provide possible explanations for our findings that both over-and underexpressions of MUC-1 increases the risk of prostatecancer progression.

In this study, we could not confirm the findings ofKirschenbaum et al (1999) of a correlation between MUC-1expression and tumour differentiation. Differences in the assess-ment of MUC-1 expression may explain the diverging resultshowever; although we evaluated staining intensity, Kirchenbaumet al examined staining patterns (apical and diffuse). Our datafurthermore confirm the results of Lapointe et al (2004)demonstrating that MUC-1 is an independent prognostic markerthat adds prognostic information over and above known riskfactors of grade and stage.

The strengths of the study include the long-term and completefollow-up of more than 20 years. A unique feature is that thecohort includes a sizeable number of prostate cancer deaths, whichallowed for us to evaluate an important clinical outcome. The factthat the cases had not received any curative treatment furtherenabled us to explore the effect of dysregulated MUC-1 expressionin the natural course of the disease. We employed high-densityTMAs, which represent an efficient approach for immuno-histochemical analysis, that also reduce potential batch-to-batchvariation in staining. Moreover, the Chromavision systemprovided an automated assessment of protein intensity on acontinuous scale. Our cases were diagnosed before introduction ofPSA screening in the population, and thus PSA levels at diagnosiswere not available. Although controlling PSA levels could attenuatethe association between MUC-1 and prostate cancer death, it isunlikely that it would fully explain the relationship.

In summary, these data show that MUC-1 is an independentprognostic marker for prostate cancer death. Furthermore,although based on small numbers, our data suggest that MUC-1expression together with Gleason grade provide substantialinformation to distinguish prostate cancer outcomes.

As the accuracy of MUC-1 alone or together with the GleasonScore in predicting prostate cancer progression was low, weconclude that its use as a single biomarker in clinical decisionmaking is limited. Yet, we believe that alterations in MUC-1expression may be useful as part of a composite set of biomarkersin accurately predicting prostate cancer outcome.

ACKNOWLEDGEMENTS

This research was supported by grants from Orebro CountyResearch Foundation, County of Norbotten Research Foundation,Swedish Cancer Society and the Maud and Birger GustavssonResearch Foundation. LAM is supported in part by a traininggrant from the National Cancer Institute/National Institutesof Health (Program for Training in Cancer EpidemiologyT32 CA009001) and a Career Development Award from the DF/HCC Prostate Cancer SPORE. KF is partially supported by aPostdoctoral Traineeship Award from the US DOD (nr W81XWH-05-1-0612).

Table 3 HR and 95% CI for prostate cancer death in relation to proteinexpression of MUC-1 in tumour tissue from patients with localised prostatecancer

NaProstate cancer

deathsCrude HR(95%CI)

HRb

(95%CI)

MUC-1 (intensity)Normal (102.5–

106)43 3 1.0 1.0

Low (o102.5) 68 16 3.9 (1.1–4) 5.1 (1.4–18)High (4105.5) 84 18 3.8 (1.1–13) 4.5 (1.3–15)

CI¼ confidence interval; HR¼ hazard ratio. aA total of 195 patients were assayed forMUC-1. bAdjusted for age, Gleason score and tumour extent.

Table 4 Hazard ratio (95% CI) of prostate cancer death associated withMUC-intensity and Gleason score, cross classified

Normal MUC-intensity Aberrant MUC-intensity

PC death/total N HR(95%CI)

PC death/total N HR(95%CI)

Gleason o7 1/25 Ref. 12/92 3.8 (0.5–29)Gleason X7 2/18 3.8(0.3–43) 22/60 17.1 (2.3–28)

Table 5 Sensitivity, specificity, PPV and NPV of Gleason grade andMUC-1 intensity in predicting prostate cancer death

Gleason score andMUC-1 intensity Sensitivity Specificity PPV NPV

Aberrant MUC-1intensity

0.91 0.25 0.22 0.93

Gleason 46 andaberrant MUC-1intensity

0.59 0.76 0.37 0.89

NPV¼ negative predictive value; PPV¼ positive predictive value.


O Andren et al

733

British Journal of Cancer (2007) 97(6), 730 – 734& 2007 Cancer Research UK

ClinicalStu

die

s


PAP

ER IV

REFERENCES

Andren O, Fall K, Franzen L, Andersson SO, Johansson JE, Rubin MA (2006)How well does the Gleason score predict prostate cancer death? A 20-yearfollow-up of a population based cohort in Sweden. J Urol 175: 1337–1340

Bresalier RS, Niv Y, Byrd JC, Duh QY, Toribara NW, Rockwell RW, DahiyaR, Kim YS (1991) Mucin production by human colonic carcinoma cellscorrelates with their metastatic potential in animal models of coloncancer metastasis. J Clin Invest 87: 1037–1045

Evangelou A, Letarte M, Marks A, Brown TJ (2002) Androgen modulationof adhesion and antiadhesion molecules in PC-3 prostate cancer cellsexpressing androgen receptor. Endocrinology 143: 3897–3904

Gendler SJ, Spicer AP (1995) Epithelial mucin genes. Annu Rev Physiol 57:607–634

Hilkens J, Ligtenberg MJ, Vos HL, Litvinov SV (1992) Cell membrane-associated mucins and their adhesion-modulating property. TrendsBiochem Sci 17: 359–363

Hilkens J, Vos HL, Wesseling J, Boer M, Storm J, van der Valk S, Calafat J,Patriarca C (1995) Is episialin/MUC1 involved in breast cancerprogression? Cancer Lett 90: 27–33

Ho SB, Niehans GA, Lyftogt C, Yan PS, Cherwitz DL, Gum ET, Dahiya R,Kim YS (1993) Heterogeneity of mucin gene expression in normal andneoplastic tissues. Cancer Res 53: 641–651

Huang DM, Guh JH, Chueh SC, Teng CM (2004) Modulation of anti-adhesion molecule MUC-1 is associated with arctiin-induced growthinhibition in PC-3 cells. Prostate 59: 260–267

Humphrey P, Vollmer RT (1988) The ratio of prostate chips with cancer: anew measure of tumor extent and its relationship to grade and prognosis.Hum Pathol 19: 411–418

Jhaveri FM, Zippe CD, Klein EA, Kupelian PA (1999) Biochemical failuredoes not predict overall survival after radical prostatectomy for localizedprostate cancer: 10-year results. Urology 54: 884–890

Johansson JE, Andren O, Andersson SO, Dickman PW, Holmberg L,Magnuson A, Adami HO (2004) Natural history of early, localizedprostate cancer. JAMA 291: 2713–2719

Kirschenbaum A, Itzkowitz SH, Wang JP, Yao S, Eliashvili M, Levine AC(1999) MUC1 expression in prostate carcinoma: correlation with gradeand stage. Mol Urol 3: 163–168

Kontani K, Taguchi O, Narita T, Izawa M, Hiraiwa N, Zenita K, Takeuchi T,Murai H, Miura S, Kannagi R (2001) Modulation of MUC1 mucin asan escape mechanism of breast cancer cells from autologous cytotoxicT-lymphocytes. Br J Cancer 84: 1258–1264

Lapointe J, Li C, Higgins JP, van de Rijn M, Bair E, Montgomery K, FerrariM, Egevad L, Rayford W, Bergerheim U, Ekman P, DeMarzo AM,Tibshirani R, Botstein D, Brown PO, Brooks JD, Pollack JR (2004) Geneexpression profiling identifies clinically relevant subtypes of prostatecancer. Proc Natl Acad Sci USA 101: 811–816

O’Connor JC, Julian J, Lim SD, Carson DD (2005) MUC1 expression inhuman prostate cancer cell lines and primary tumors. Prostate CancerProstatic Dis 2(1): 36–44

Perrone EE, Theoharis C, Mucci NR, Hayasaka S, Taylor JM, Cooney KA,Rubin MA (2000) Tissue microarray assessment of prostate cancer tumorproliferation in African-American and white men. J Natl Cancer Inst 92:937–939

Schut IC, Waterfall PM, Ross M, O’ullivan C, Miller WR, Habib FK,Bayne CW (2003) MUC1 expression, splice variant and short formtranscription (MUC1/Z, MUC1/Y) in prostate cell lines and tissue. BJUInt 91: 278–283

Zellweger T, Ninck C, Bloch M, Mirlacher M, Koivisto PA, Helin HJ,Mihatsch MJ, Gasser TC, Bubendorf L (2005) Expression patternsof potential therapeutic targets in prostate cancer. Int J Cancer 113:619–628


O Andren et al

734

British Journal of Cancer (2007) 97(6), 730 – 734 & 2007 Cancer Research UK

Clin

icalStu

die

s


115

PAPER V


PAP

ER V

Time trends and survival among men diagnosed with incidental prostate

cancer in Sweden – a register-based study between 1970 and 2003

Ove Andrèn1, Hans Garmo3, Lorelei Mucci4,5, Swen-Olof Andersson1, Jan-Erik

Johansson1, and Katja Fall2

1 Department of Urology Örebro University Hospital,701 85 Örebro, Sweden

2 Department of Medical Epidemiology and Biostatistics, Karolinska Institutet,171 77 Stockholm, Sweden

3 Regional Oncologic Center, Uppsala University Hospital, Uppsala, Sweden

4 Channing Laboratory, Harvard Medical School/Brigham and Women’s Hospital, Boston MA

5Department of Epidemiology, Harvard School of Public Health, Boston M

Corresponding author:

Ove Andrén, M.D.

Department of Urology

Örebro University Hospital

Email: [email protected]


Abstract

Background: The clinical significance of incidental prostate cancer (PCa) has been

debated during recent years. One main question has been whether or not these

tumors represent a biological domain different from those clinically detected through

PSA screening or digital rectal examination. In this study we took advantage of

several Swedish registers to study trends and survival in patients with incidental PCa.

Methods: Using the national registration number, we identified a study group of

patients diagnosed with PCa between 1970 and 2003 incidentally by TURP

(N=23,288) as well as a comparison group consisting of all other men diagnosed with

PCa (n=112,204) during the same period. We followed the men prospectively for

cancer-specific and total mortality.

Results: The incidence of incidental PCa increased during 1970 to 1990, and has

since declined. The cumulative incidence of cancer mortality among men diagnosed

with incidental PCa was 27.1% compared to 33.2% among those with clinically

detected disease between 1970 and 2003.

Conclusions: There has been a dramatic change over time in the incidence of

incidental prostate cancers in Sweden. We also found that the cumulative incidence

of prostate cancer mortality is high in the incidental group.

Our results do not fit earlier descriptions of incidental PCa as a non-lethal disease.

The data suggest that most studies on incidental PCa, which are based on men

diagnosed in the 1970s and 1980s, involve tumors that we today would classify as

T1c and T2 tumors. Our findings support Bostwicks thesis that T-classification

primarily reflects mode and tools of detection rather than variation in tumor biology. In

conclusion, new studies are needed to gain knowledge of the incidental tumors we

find today.


PAP

ER V

Introduction

The clinical significance of incidental prostate cancer (PCa) detected by transurethral

resection (TURP) or by open adenoma enucleation (OAE) for assumed benign

hyperplasia has been a matter of debate1-3. The main question is whether or not

these tumours represent a biological domain different from those clinically detected

by rectal palpation or by PSA tests. It is commonly believed that incidental tumors,

especially those of small volume4 (<5% of the total amount of resected tissue; T1a

tumors), are harmless and can be left without immediate treatment5. However, recent

studies6-8 have demonstrated that with increasing tumor volume (>5% of totally

resected tissue;T1b tumors), the clinical course of PCa becomes gradually more

unfavourable, and is comparable to that of palpable T2 tumors.

We undertook a large, population-based study to more fully characterize incidental

PCa in order to understand whether its biological behaviour differs from other types

of PCa. Using the Swedish Inpatient Register and the Swedish Cancer Register we

were able to examine trends over time and survival among patients with incidental

PCa in Sweden between 1970 and 2005.


Materials and methods

In this study, we utilized nationwide data from several swedish health care registers.

The use of the national registration number (NRN), a unique 10 digit identification

number assigned to all Swedish residents, enables record linkages between the

registers. There was almost no private institutional care available in Sweden during

the study period, so essentially all men were referred to and treated at the main

hospital in their county of residence and, therefore, the study can be considered

population-based.

The National Inpatient Register

For primarily administrative purposes the National Board of Health and Welfare

started to compile data for the National Inpatient Register in 1964 to capture

hospitalizations. Registration increased over the years and was 60% in 1969, 85% in

1983, and in 1987 the coverage reached 100%. This register includes data on the

NRN, discharge diagnoses (a maximum of 8 including up to 6 surgical procedures)

for each hospitalization, and date of discharge. The 7th revision of the International

Classification of Diseases (ICD-7) was used for coding diagnoses through 1968, the

8th revision (ICD-8) between 1969 and 1986, the 9th revision (ICD-9) between 1987

and 1996, and the 10th revision (ICD-10) thereafter. Surgical procedures were coded

according to the Swedish Classification of Operations and Major Procedures.

The Swedish Cancer Register and Cause of Death Register

The Swedish Cancer Registry captures all incident cancers in the population,

including PCa. Mandatory reporting by clinicians and pathologists has contributed to


PAP

ER V

the thoroughness of the Swedish Cancer Register and the coverage has exceeded

97% since 19849. The Cause of Death register was established in 1952, and the

registries has captured virtually all deaths in Sweden since 1961. The most

significant variables in this register are the main and contributory causes of death,

date of death, notes on section, sex, age, and birthplace. Cancer diagnoses and

causes of death are coded according to the International Classification of diseases

and causes of deaths (ICD 7-10).

The Register of Population Changes

This register is based on the official Swedish Census data and includes all residents

alive and living in Sweden at the end of each year.

Methods

We identified all men (N=76,778) in the Swedish Inpatient Register who had been

discharged after TURP or OAE between 1970 and 2003. By linking the records of

these men to the Cancer Registry, we identified all men who were incidentally

diagnosed with PCa by TURP or OAE. The definition of incidentally detected PCa

required the following criteria to be fulfilled: (1) the date of the PCa diagnosis had to

follow the date of first TURP/OA admission, (2) the date of the PCa diagnosis had to

be set within 14 days from discharge, (3) the diagnosis of PCa had to be

histopathologically verified, and (4) the hospital stay had to be less than 60 days. A

diagnosis of high-grade intraprostatic neoplasia (PIN) was not classified as PCa. In

total 23,288 men with incidentally diagnosed PCa was identified.


As a comparison group, we chose all men clinically diagnosed with PCa in the

Swedish Cancer Register during the same time period and from the same counties.

All men with PCa in the comparison group (N=112,204) were diagnosed on clinical

grounds alone (i.e. by tissue biopsies or fine needle biopsies/cytology). By linking the

records of these men to the records of those discharged for TURP/OAE we were able

to separate out men with histologically confirmed PCa before surgery and men

diagnosed later than 14 days after the TURP/OAE. We further excluded all autopsy

detected PCa. In total, the study population, including those with both incidentally

and clinically diagnosed PCa, comprised 135,492 men.

Statistical Methods

The age-standardized incidence rates of PCa were calculated using the age

distribution of the Swedish population on January 1, 2000, which was obtained from

Statistics Sweden10. The men were followed prospectively from date of cancer

diagnosis to death from cancer, or censored from other causes of death or end of

follow-up (Dec 31, 2003). Information on cause of death was obtained through

linkage with the Cause of Death Registry and the ICD codes were used to identify

PCa (ICD10 = C619) as the main cause of death. We calculated cumulative

incidence curves to illustrate the risk of PCa death while accounting for the

competing risk of death from other causes11. All statistical calculations were

performed using the statistical program package R12.


PAP

ER V

Results

Clinical characteristics among the men with PCa are presented in table 1. Most

patients with incidental PCa are diagnosed age 65 and older, with half age 75 years

or older (mean age 74.4). Men with incidental PCa are somewhat older than men

diagnosed on clinical grounds alone. The mean age at PCa diagnosis increased

slightly over time from 72.5 in 1970-79 to 75.2 in the last calendar period 1999-2003

(Fig 1).

The age-standardised incidence of incidental PCa peaked in the early 1990s (Fig 3)

and successively declined during subsequent years, reaching 12% during the period

of 1999-2003. During the entire study period, 1970-2003, the total PCa incidence rate

increased continuously and during the mid 1980s was almost 15 times higher than in

1973. The upward trend appeared to be driven by the increased frequency of TURP,

but the incidence rate continued to rise even after the TURP frequency had levelled

off.

We observed a marked decrease in hospitalisation time over calendar time. Initially,

in the early 70s, when the transurethral technique was introduced, the hospitalisation

time was comparably long, lasting about 2-3 weeks. After a continuous and dramatic

decrease, the hospitalization time levelled off to 4-5 days in the last 5-year period

(Fig 2).

Among the men with incidental PCa, 6,300 had died of PCa, 11,850 had died of

intercurrent causes, and 5,138 patients were still alive at the end of 2003. Accounting

for competing causes, the cumulative incidence of PCa mortality during the study


period was 27.1%. This compares to 33.2% among the comparison group consisting

of clinically detected PCa. The age standardised cumulative incidence of PCa

mortality comparing incidental and clinical PCa is shown in Fig 4. The corresponding

incidence curves by time period and by mode of detection are shown in Fig 5. We

observed a clear decreasing trend over time after diagnosis, but the prostate cancer

specific mortality decreased both according to mode of detection and calendar time.

The mortality was lower among those with incidentally detected tumors compared to

those in the comparison group. However, it should be noted that although the

prostate cancer specific mortality among those with incidental PCa levelled off by

time, it was still 30% 15 years post diagnosis.


PAP

ER V

Discussion

In this large, population-based study of incidental PCa we observed changes over

time in the incidence of incidental PCa in Sweden. We also found significant prostate-

specific mortality among men with PCa diagnosed by TURP or by other modes of

detection. These data do not fit earlier descriptions of incidental PCa as a non-lethal

disease3, 5.

The introduction of TURP in the late 1970s increased not only the incidence of

incidental tumours as earlier described 13, but also the overall incidence of prostate

cancer. TURP specimens mainly derive from the transition zone of the prostate 14, 15.

Tumours originating from this part of the prostate have been described as low grade

with low potential of malignancy, and therefore not considered in need of treatment5,

16, 17. On the other hand, 8-37% of all incidental tumours will ultimately progress2.

Chun et al18 demonstrated in a study of radical prostatectomy specimens that there is

no difference in time to biochemical failure between transition zone tumours and

tumours that derive from the peripheral zone. Bostwick et al2 suggested that TURP

detected tumours do not exclusively have their origin in the transition zone, but may

instead represent peripheral zone tumours that have grown into the transition zone.

Thus, there may be a significant overlap between T1b, T1c, and T2 tumours. In the

end, it might be the tumor location within the prostate gland and the chance the

urologist has of finding it that ultimately determines the T stage.


Before the PSA era, the clinical decision of whether or not presentation with voiding

symptoms was indicative of a benign condition or PCa was mainly based on rectal

palpation and measurements of acid phosphates. A substantial proportion of men

with voiding symptoms were treated by TURP or OAE. Our data confirm earlier

observations of a declining incidence of incidental tumours after the introduction of

PSA screening19. In Sweden, incidental PCa currently constitutes around 10% of all

diagnosed PCa20. We further observed that the cumulative incidence of PCa mortality

seemed to decrease over time; this may either describe an increase in lead time after

the introduction of PSA, or that tumours detected with TURP today represent a

different biological domain then earlier incidental tumours.

This large, register-based study with essentially complete follow-up allowed a

comprehensive examination of incidence and mortality trends in Sweden over the

last three decades. Since health care is free to all Swedish residents, selection forces

associated with the TURP procedure should be minimal. A limitation, shared by all

register-based studies, though, is the quality of available data. The accuracy of the

official causes of death data, with regard to prostate cancer, should however be

high21.

Most studies of cancer mortality among incidental tumours are based on patients

diagnosed during the 1970s and 1980s. Our study indicates that those findings are

not representative of incidental tumours of today. Indeed, the incidental tumors of

earlier decades are more likely to represent today’s T1c and T2 tumours. Our data

support the thesis that the T-classification system describes availability of tools for


PAP

ER V

the urologist rather than the biology of the tumour2. Thus, new knowledge is needed

to characterize incidental tumours detected today.


References

1. Epstein, J. I.: Cancer detected incidental to simple prostatectomy (stage A1). J Cell Biochem Suppl, 16H: 78, 1992

2. Bostwick, D. G.: The pathology of incidental carcinoma. Cancer Surv, 23: 7, 19953. Epstein, J. I., Paull, G., Eggleston, J. C. et al.: Prognosis of untreated stage A1

prostatic carcinoma: a study of 94 cases with extended followup. J Urol, 136: 837, 1986

4. Hermanek, P., Sobin, L. H., International Union against Cancer: TNM classification of malignant tumours, 4. ed. Berlin ; New York: Springer-Vlg., p. 217, 1992

5. Cantrell, B. B., DeKlerk, D. P., Eggleston, J. C. et al.: Pathological factors that influence prognosis in stage A prostatic cancer: the influence of extent versus grade. J Urol, 125: 516, 1981

6. Cuzick, J., Fisher, G., Kattan, M. W. et al.: Long-term outcome among men with conservatively treated localised prostate cancer. Br J Cancer, 95: 1186, 2006

7. Robinson, D., Aus, G., Bak, J. et al.: Long-term follow-up of conservatively managed incidental carcinoma of the prostate: a multivariate analysis of prognostic factors. Scand J Urol Nephrol, 41: 103, 2007

8. Andren, O., Fall, K., Franzen, L. et al.: How well does the Gleason score predict prostate cancer death? A 20-year followup of a population based cohort in sweden. J Urol, 175: 1337, 2006

9. Mattsson, B., Wallgren, A.: Completeness of the Swedish Cancer Register. Non-notified cancer cases recorded on death certificates in 1978. Acta Radiol Oncol, 23:305, 1984

10. Statistiska Centralbyrån (Statistics Sweden): Statistical database. Stockholm: SCB, vol. 2008

11. Kalbfleisch, J. D., Prentice, R. L.: The statistical analysis of failure time data, 2nd ed. Hoboken, N.J: Wiley-Interscience, 2002

12. Ihaka, R., Gentleman, R.: R: a language for data analysis and graphics. Journal of Computational and Graphical Statistics, 5: 299, 1996

13. Haapiainen, R.: Latent (pTO) prostatic carcinoma--a retrospective study of frequency and natural history. Ann Chir Gynaecol, 75: 172, 1986

14. McNeal, J. E., Price, H. M., Redwine, E. A. et al.: Stage A versus stage B adenocarcinoma of the prostate: morphological comparison and biological significance. J Urol, 139: 61, 1988

15. McNeal, J. E., Redwine, E. A., Freiha, F. S. et al.: Zonal distribution of prostatic adenocarcinoma. Correlation with histologic pattern and direction of spread. Am J Surg Pathol, 12: 897, 1988

16. Epstein, J. I., Carmichael, M. J., Partin, A. W. et al.: Small high grade adenocarcinoma of the prostate in radical prostatectomy specimens performed for nonpalpable disease: pathogenetic and clinical implications. J Urol, 151: 1587, 1994

17. Albertsen, P. C., Hanley, J. A., Fine, J.: 20-year outcomes following conservative management of clinically localized prostate cancer. JAMA, 293: 2095, 2005

18. Chun, F. K., Briganti, A., Jeldres, C. et al.: Zonal origin of localized prostate cancer does not affect the rate of biochemical recurrence after radical prostatectomy. Eur Urol, 51: 949, 2007

19. Mai, K. T., Isotalo, P. A., Green, J. et al.: Incidental prostatic adenocarcinomas and putative premalignant lesions in TURP specimens collected before and after the introduction of prostrate-specific antigen screening. Arch Pathol Lab Med, 124: 1454, 2000


PAP

ER V

20. Regionalt Onkologiskt Centrum ROC. Prostatacancer. Primärregistrering 1997-2001: Regionalt Onkologiskt Centrum, vol. 2004, 2003

21. Fall, K., Strömberg, F., Rosell, J. et al. Reliability of death certificates in prostate cancer patients. Scand J Urol Nephrol, in press 2008


Table 1. Characteristics of prostate cancer patients diagnosed by or hospitalized for TUR-P between 1970 and 2003 in Sweden compared to other PCa patients.

TUR-P-diagnosedpatients

TUR-P-hospitalisations

PCa-comparisons

N % N % N %

Age

<55 123 (0.5) 708 (0.9) 2462 (2.2)

55-64 2157 (9.3) 8931 (11.6) 17740 (15.8)

65-74 9085 (39.0) 30854 (40.2) 43060 (38.4)

75-84 10114 (43.4) 31163 (40.6) 40769 (36.3)

85+ 1809 (7.8) 5122 (6.7) 8173 (7.3)

Years

1970-79 2627 (11.3) 12324 (16.1) 13045 (11.6)

1980-86 6693 (28.7) 20793 (27.1) 17116 (15.3)

1987-90 4961 (21.3) 14409 (18.8) 13311 (11.9)

1991-93 3349 (14.4) 10106 (13.2) 12480 (11.1)

1994-98 2795 (12.0) 9790 (12.8) 20270 (18.1)

1999-2003 2863 (12.3) 9356 (12.2) 35982 (32.1)

Hospitalisation time

0-3 days 5680 (24.4) 17978 (23.4) NA NA

4-7 days 10997 (47.2) 34758 (45.3) NA NA

8-14 days 4175 (17.9) 14964 (19.5) NA NA

2-4 weeks 1844 (7.9) 6655 (8.7) NA NA

4+ weeks 592 (2.5) 2423 (3.2) NA NA

N TURP-treatments

0 0 (0.0) NA NA 82389 (73.4)

1 18007 (77.3) NA NA 23689 (21.1)

2 4118 (17.7) NA NA 4719 (4.2)

3 864 (3.7) NA NA 1062 (0.9)

4+ 299 (1.3) NA NA 345 (0.3)

Status at end of follow up

(Dec. 31, 2003)

Alive 5138 (22.1) NA NA 39527 (35.2)

Dead from other causes 11850 (50.9) NA NA 35469 (31.6)

PCa-death 6300 (27.1) NA NA 37208 (33.2)

Follow up

Time in years , mean (sd) 6.1 (4.7) NA NA 4.3 (4.0)


PAP

ER V

Figure 1

1970 1975 1980 1985 1990 1995 2000 2005

Year

65

70

75

80M

ean

age


Figure 2

1970 1975 1980 1985 1990 1995 2000 2005

Year

0

5

10

15

20

25

Mea

n ho

spit

alis

atio

n ti

me


PAP

ER V

Figure 3

1970 1975 1980 1985 1990 1995 2000 2005

Year

0

50

100

150

200

250

Age

sta

ndar

dize

d in

cide

nce

of

pro

stat

e ca

ncer

(pe

r 10

0 00

0 m

en)

TURP-detected

Not TURP detected

All PC


Figure 4

0 2 4 6 8 10 12 14 16

Time (years)

0

0.1

0.2

0.3

0.4

0.5

Cum

ulat

ive

inci

denc

e o

f pr

ost

ate

can

cer

dea

th

TURP-detected ( n = 23288 )

Not TURP-detected ( n = 112204 )


PAP

ER V


Figure 5

Figure legends

Figure 1. Mean age at cancer diagnosis among TURP/OAE detected PCa) in Sweden between 1970 and 2003.

Figure 2. Mean duration of hospitalization for TURP procedure in Sweden between 1970 and 2003.

Figure 3. Trends in the age-standardized incidence of TURP-detected and total prostate cancer in Sweden between 1970 and 2003

Figure 4. Cumulative incidence of prostate cancer-specific mortality in Sweden between 1970 and 2003 among men diagnosed incidentally by TURP or clinically detected.

Figure 5. Cumulative incidence of prostate cancer deaths by calendar time period among patients diagnosed at TURP (left panel) and among patients not detected at TURP (right panel).


Publikationer i serien Örebro Studies in Medicine

1. Bergemalm, Per-Olof (2004). Audiologic and cognitive long-term sequelae from closed head injury.

2. Jansson, Kjell (2004). Intraperitoneal Microdialysis.Technique and Results.

3. Windahl, Torgny (2004). Clinical aspects of laser treatment of lichen sclerosus and squamous cell carcinoma of the penis.

4. Carlsson, Per-Inge (2004). Hearing impairment and deafness.Genetic and environmental factors – interactions – consequences.A clinical audiological approach.

5. Wågsäter, Dick (2005). CXCL16 and CD137 in Atherosclerosis.

6. Jatta, Ken (2006). Infl ammation in Atherosclerosis.

7. Dreifaldt, Ann Charlotte (2006). Epidemiological Aspects onMalignant Diseases in Childhood.

8. Jurstrand, Margaretha (2006). Detection of Chlamydia trachomatis and Mycoplasma genitalium by genetic and serological methods.

9. Norén, Torbjörn (2006). Clostridium diffi cile, epidemiology and antibiotic resistance.

10. Anderzén Carlsson, Agneta (2007). Children with Cancer – Focusing on their Fear and on how their Fear is Handled.

11. Ocaya, Pauline (2007). Retinoid metabolism and signalling invascular smooth muscle cells.

12. Nilsson, Andreas (2008). Physical activity assessed by accelerometry in children.

13. Eliasson, Henrik (2008). Tularemia – epidemiological, clinical and diagnostic aspects.

14. Walldén, Jakob (2008). The infl uence of opioids on gastric function: experimental and clinical studies.

15. Andrén, Ove (2008). Natural history and prognostic factors inlocalized prostate cancer.


natural history and prognostic factors in localized

Documents