clinicopathologic study of 85 similarly treated patients with anaplastic astrocytic tumors

Clinicopathologic Study of 85 Similarly TreatedPatients with Anaplastic Astrocytic TumorsAn Analysis of DNA Content (Ploidy), Cellular Proliferation, and p53 Expression

Arie Perry, M.D.1

Robert B. Jenkins, M.D., Ph.D.2

Judith R. O’Fallon, Ph.D.2

Paul L. Schaefer, M.D.3

David W. Kimmel, M.D.2

Michelle R. Mahoney, M.S.2

Bernd W. Scheithauer, M.D.2

Sandra M. Smith2

Eunice M. Hill2

Thomas J. Sebo, M.D.2

Ralph Levitt, M.D.4

James Krook, M.D.5

Loren K. Tschetter, M.D.6

Roscoe F. Morton, M.D.7

Jan C. Buckner, M.D.2

1 Washington University School of Medicine, St.Louis, Missouri.

2 Mayo Clinic and Mayo Foundation, Rochester,Minnesota.

3 Toledo Community Hospital Oncology ProgramCCOP, Toledo, Ohio.

4 Meritcare Hospital CCOP, Fargo, North Dakota.

5 Duluth CCOP, Duluth, Minnesota.

6 Sioux Community Cancer Consortium, SiouxFalls, South Dakota.

7 Iowa Oncology Research Association CCOP, DesMoines, Iowa.

Supported in part by Public Health Service grantsCA-50905, CA-25224, CA-37404, CA-37417, CA-35269, CA-35103, CA-35101, CA-35113,CA-52352, CA-63849, CA-35195, CA-35272, CA-60276, and CA-35103 and by the Linse-BockFoundation.

This study was conducted as a collaborative trial ofthe North Central Cancer Treatment Group and theMayo Clinic. Additional participating institutionsincluded Cedar Rapids Oncology Project CCOP,Cedar Rapids, IA (Martin Wiesenfeld, M.D.); Mis-souri Valley Cancer Consortium, Omaha, NE(James A. Mailliard, M.D.); Billings Clinic, Billings,

MT (Larry P. Ebbert, M.D.); Carle Cancer CenterCCOP, Urbana, IL (Alan K. Hatfield, M.D.); GrandForks Clinic, Ltd., Grand Forks, ND (Daniel J.Walsh, M.D.); Ochsner CCOP, New Orleans, LA(Carl G. Kardinal, M.D.); Scottsdale CCOP, Scotts-dale, AZ (Richard Wheeler, M.D.); Siouxland He-matology-Oncology Associates, Sioux City, IA(John C. Michalak, M.D.); Illinois Oncology Re-search Association CCOP, Peoria, IL (John W. Ku-

gler, M.D.); and Quain and Ramstad Clinic, Bis-marck, ND (Delano M. Pfeifle, M.D.).

Address for reprints: Jan C. Buckner, M.D., Divi-sion of Medical Oncology, Mayo Clinic, 200 FirstStreet SW, Rochester, MN 55905.

Received December 9, 1998; revision receivedMarch 11, 1999; accepted March 11, 1999.

BACKGROUND. The biologic behavior of anaplastic (World Health Organization

Grade III) astrocytomas and oligoastrocytomas is highly variable, ranging from

rapid progression to prolonged survival. It is difficult to predict the outcome of an

individual patient based on morphology alone.

METHODS. To determine the prognostic value of commonly used clinicopathologic

markers, we reviewed our experience with 85 similarly treated patients enrolled in

3 North Central Cancer Treatment Group high grade glioma protocols. The pa-

thology was comprised exclusively of primary anaplastic astrocytic tumors (66

astrocytomas and 19 oligoastrocytomas). Variables examined included patient age,

morphologic type, preoperative performance score, extent of surgery, solitary

versus multiple mitoses, DNA flow cytometric and image morphometric parame-

ters, and expression of proliferating cell nuclear antigen, MIB-1, and p53 ex-

pression.

RESULTS. The study was comprised of 48 men and 37 women ranging in age from

14 –79 years (median age, 47 years). Overall survival ranged from ,1 month to .12

years (median, 21.6 months). Statistical analyses revealed that age accounted for

the majority of this extensive variability in survival. The median survival times were

65.5 months, 22.1 months, and 4.4 months, respectively, for the groups ,40 years,

40 –59 years, and $60 years, respectively (P , 0.0001). On univariate analyses,

aneuploidy by flow cytometry and a low performance score also predicted a better

survival (P values of 0.04 and 0.009, respectively). Statistical trends predicting a

better survival were observed for patients with a solitary mitosis and p53 immu-

nopositivity. However, only patient age remained significant in multivariate

models.

CONCLUSIONS. In a small but relatively uniformly treated cohort of patients with

anaplastic astrocytomas and oligoastrocytomas, patient age was associated

strongly and inversely with overall survival. Once patient age was taken into

account, the clinical and pathologic markers tested appeared to be of limited

prognostic value. Cancer 1999;86:672– 83. © 1999 American Cancer Society.

KEYWORDS: astrocytoma, glioma, prognosis, flow cytometry, immunohistochemis-try, proliferation, Ki-67, proliferating cell nuclear antigen, p53 protein.

672

© 1999 American Cancer Society

A lthough great strides have been made in our un-derstanding of glioma tumorigenesis and progres-

sion, our ability to predict behavior and providemeaningful therapy to the individual patient remainslimited. Whereas World Health Organization (WHO)Grade IV astrocytomas (glioblastoma multiforme), themost common and malignant form, are associatedwith an almost uniformly poor outcome, survival var-ies tremendously among patients with low and inter-mediate grade astrocytomas.1–5 As therapeutic optionsexpand, accurate patient stratification becomes criti-cal for optimal patient management. For example,more aggressive regimens with potential long termtoxicities could be reserved for those with otherwiseshort life expectancies and avoided or temporarilywithheld in those with longer life spans. In this regard,Grade III astrocytic tumors seem to be suited ideallyfor further patient stratification, because they range inclinical behavior from rapid progression over a fewmonths (similar to Grade IV tumors) to several years ofsurvival (similar to Grade II tumors). Due to theirrelative rarity compared with glioblastomas, few serieshave focused specifically on this subset of tumors.4 – 6

In this study, we assess the prognostic roles of variousancillary laboratory tests in a group of 85 patients withuniformly treated Grade III astrocytic neoplasms (66anaplastic astrocytomas and 19 anaplastic oligoastro-cytomas). We compare measurements of DNA ploidy,proliferative indices, and p53 protein expression withstandard clinicopathologic variables, such as patientage, preoperative performance status, extent of surgi-cal resection, solitary versus multiple mitotic figures,and the presence or absence of an oligodendroglialcomponent.

MATERIALS AND METHODSPatient/Tumor PopulationData were derived from 184 patients enrolled by 15North Central Cancer Treatment Group (NCCTG)member institutions (mostly community hospitals)into 3 NCCTG clinical trials designed to assess thera-pies for newly diagnosed high grade glioma. All had adiagnosis of anaplastic (St. Anne Mayo Grade 3) astro-cytoma or anaplastic (St. Anne Mayo Grade 3 or 4)oligoastrocytoma. Paraffin tissue blocks were receivedfor 108 of these patients (59%), 85 of which (46%)contained adequate tumor for further analysis. Eigh-teen of the patients (21%) underwent surgery at theMayo Clinic, and the rest of the tissue samples weresent for evaluation from other NCCTG member insti-tutions.

Eligibility for all 3 clinical trials required centralpathology review by one of the authors (B.W.S.). Caseswere classified and graded according to the revised

WHO7 and St. Anne Mayo8 schemes. For this study,anaplastic astrocytomas (AA) and anaplastic oligoas-trocytomas (AOA) were included, whereas pure oligo-dendrogliomas were ineligible. St. Anne Mayo Grade 3and 4 AOAs were accepted, because they both corre-sponded to WHO Grade 3. Pure astrocytomas wereexclusively St. Anne Mayo Grade 3, WHO Grade 3, withmitotic activity as the defining variable for both grad-ing schemes.

Preoperative performance status (PS) was scoredon a scale of 0 – 4, as described previously.9,10 In gen-eral, PS 0 5 fully active; PS 1 5 mildly restrictedphysical activity, fully ambulatory; PS 2 5 restrictedactivity, ambulatory .50% of the day; PS 3 5 confinedto bed or chair .50% of the day; and PS 4 5 totallyconfined to bed or chair.

Postoperative TherapyAll patients had a diagnosis of high grade glioma andwere treated according to one of three Phase III pro-spective clinical trials, each with two randomizedarms: NCCTG 79-72-51 (dates of enrollment: June 11,1980 to July 9, 1985),11 85-72-51 (dates of enrollment:June 19, 1985 to September 12, 1989),12 or 88-72-52(dates of enrollment: April 30, 1990 to August 22,1994).13 In each trial, patients were stratified prior torandomization by age, extent of surgery, histologicgrade of tumor, and performance score. All patientsreceived similar postsurgical treatment, consisting ofcranial irradiation (radiation therapy [RT]; 55– 65 grays[Gy] to the whole brain in 79-72-51, 60 Gy to theinvolved field in 85-72-51, or 65 Gy to the involvedfield in 88-72-52) and chemotherapy, usually a nitro-sourea (carmustine [BCNU] vs. dibromodulcitol in 79-72-51, BCNU vs. PCNU in 85-72-52, or BCNU vs.BCNU plus recombinant human interferon a-2a in88-72-51). Therefore, roughly half of all patients re-ceived RT and BCNU, whereas the other half receivedRT and alternate forms of chemotherapy. No differ-ence in survival between the two randomized treat-ment arms was identified in any of the studies. How-ever, survival improved over time between the firststudy and the latter two studies.

Immunohistochemical/Cell Image AnalysesOur methods of antigen retrieval and automated im-munohistochemistry on paraffin sections have beenreported previously.14 The monoclonal antibodiesused include MIB-1 (Ki-67 antigen; Immunotech,Westbrook, ME; 1:100 dilution), proliferating cell nu-clear antigen (PCNA) (PC10; Dako, Carpinteria, CA;1:7500 dilution), and p53 (DO-7; Dako; 1:200). Label-ing indices (LIs) were quantified from regions of max-imal nuclear staining on a CAS 200 cell analysis system

Anaplastic Astrocytoma Survival/Perry et al. 673

(Becton Dickinson, San Jose, CA). Approximately15–20 high power fields were analyzed per specimen,and the indices were expressed as a percentage ofnuclear area staining. The same cell analysis systemalso was used to determine DNA ploidy using feulgenstained sections and the Quantitative DNA analysissoftware (version 3.00). Cut-off values from our previ-ous studies were used for statistical analysis.14

DNA Flow CytometryDNA flow cytometry was performed as described pre-viously by using three 50-mm-thick paraffin sections.15

Specimens were evaluated on a FACScan analyzer(Becton-Dickinson), and the MODFIT software pro-gram (Verity, Topsham, ME) was used for debris andaggregate subtraction.16 Histograms were interpretedas diploid, tetraploid, or aneuploid as defined previ-ously, and the S-phase fraction was taken as the mea-surement of proliferative index. Similar to our priorstudy, this index was stratified as #6% and .6%.15

Specimens with either insufficient tissue in the block(e.g., stereotactic biopsy) or a coefficient of variation(CV) for the G0/G1 peak .10 were considered inde-terminate for all flow cytometric parameters.

Statistical MethodsFrequency distributions and summary statistics werecalculated for all tumor marker, histologic, and clinicalvariables. For several continuous variables (e.g., pa-tient age, PCNA, S-phase), categorical variables weredefined to create scientifically appropriate groups(e.g., patient age ,40 years, 40 –59 years, .60 years).Wilcoxon tests were used to investigate differences inthe distributions of continuous variables between var-ious subsets of patients as classified by categoricaldata (e.g., gender, DNA ploidy by flow cytometry).Cross tabulations were generated for categorical vari-ables, and Wilcoxon tests (for ordered variables) andPearson’s chi-square tests (for nonordered variables)were used to investigate the associations betweenthem. Kaplan–Meier methodology was used to esti-mate survival distributions, 95% confidence intervals(CI) were calculated for survival estimates reported,and two-sided log-rank (LR) tests were used to com-pare survival distributions between subsets of pa-tients. Cox proportional hazards regression model wasused to look for associations of survival with varioustumor marker, histologic, and clinical variables. Aboot-strap model selection procedure17 combiningboot-strap methodology18 with Cox model forward se-lection procedures was used to validate the Cox modelvariable selection.

RESULTSPatient Characteristics and Evaluation of ClinicalParametersThe study group consisted of 48 men and 37 womenranging in age from 14 years to 79 years (median, 47years) at the time of diagnosis. With the exception ofthe girl age 14 years with AA, all other patients wereadult (age .18 years). Sampling consisted of biopsyonly in 35% of patients, subtotal resection in 49% ofpatients, and gross total resection in 17% of patients.The qualifying diagnosis was AA in 66 patients andAOA in 19 patients.

Table 1 compares the clinical findings from the 85patients in the study group with the other 99 eligiblepatients who were not studied due to lack of availabletissue blocks or sufficient tissue in the blocks. Themajority of patients entered in the oldest (1979) pro-tocol were not studied, reflecting difficulties in retriev-ing tissue blocks from older cases. The patients en-tered on this protocol tended to be older and hadpoorer performance scores (data not shown). Thestudy group was significantly younger (P , 0.0001)and survived longer (P 5 0.003) than patients whosetumors were not analyzed. The extent of tumor resec-tion and of preoperative performance status were sim-ilar in the two groups.

The distributions of clinicopathologic variablesare summarized in Table 2 for the entire group of 85patients as well as for the subsets of 66 patients withAA and 19 patients with AOA. In general, the twohistologic groups were similar, although the patientswith AOA made up only 22% of all patients. For rea-sons that were not entirely clear, patients with AOA

TABLE 1Patient Population Differences: Tumor Analyzed versus Not Analyzed

Variable

Percentanalyzed(n 5 85)

Percent notanalyzed(n 5 99) P value

Protocol number 0.00779-72-51 20 8085-72-51 51 4988-72-52 51 49

Median age (yrs) 47.0 59.0 , 0.0001Performance status 0.67

0 24 301 50 412 17 173 10 11

Extent of resection 0.64Biopsy only 35 31Subtotal resection 49 46Gross total resection 17 22

Median survival (mos) 22 12 0.003

674 CANCER August 15, 1999 / Volume 86 / Number 4

were more likely to have gross total resection (P 50.013).

At the time of analysis, 68 of 85 patients (80%)with available and adequate tissue for this study haddied. Follow-up in the remaining 17 patients rangedfrom 38.4 months to 147.8 months. Overall survivalranged from ,1 month to .12 years, with a median of21.6 months (95% CI 5 13.7–34.4). Deaths occurred in82% of the 66 patients with AA and in 74% of the 19

patients with AOA. The median survivals were 21.9months (95% CI 5 13.7–36.4) and 14.3 months (95%CI 5 9.2–109.8) for the patients with AA and AOA,respectively, but this difference was not statisticallysignificant, as shown in Figure 1 (P 5 0.65; LR test).

Patient age was strongly associated with overallsurvival (Fig. 2, Table 3). Median survival times were65.5 months in patients age ,40 years, 22.1 monthsfor ages 40 –59 years, and 4.4 months for ages $60

TABLE 2Frequency Distributions of Clinicopathologic Variables in All 85 Patients by Histology

Baseline factor

AA (n 5 66) AOA (n 5 19)

P value

Total (n 5 85)

No. % No. % No. %

Age (yrs) 0.28,40 27 41 4 21 31 3640–60 24 36 9 47 33 39.60 15 23 6 32 21 25

Gender 0.89Female 29 44 8 42 37 44Male 37 56 11 58 48 56

Performance score 0.690 16 25 4 21 20 241 29 45 12 63 41 492–3 20 31 3 16 23 27

Extent of resection 0.0130Biopsy 28 43 1 5 29 35Subtotal 29 45 12 63 41 49Gross total 8 12 6 32 14 17

HistologyAstrocytoma 66 78Oligoastrocytoma 19 22

Mitoses 0.0710Singular 10 15 0 0 10 12Multiple 56 85 19 100 75 88

MIB-1 (%) 0.23#5 31 47 6 32 37 44.5 35 53 13 68 48 56

p53 (%) 0.88#10 30 45 9 47 39 46.10 36 55 10 53 46 54

PCNA (%) 0.23,20 41 62 11 58 52 6220–40 12 18 5 26 17 20.40 13 20 3 16 16 19

Ploidy (image analysis) 0.58Diploid 17 26 9 47 20 24Tetraploid 18 27 3 16 25 29Aneuploid 31 47 7 37 40 47

Ploidy (flow cytometry) 0.66Diploid 12 29 5 31 17 29Tetraploid 11 26 6 38 17 29Aneuploid 19 45 5 31 24 41

% S-phase (flow cytometry) 0.34,6 11 30 7 47 18 35$6 26 70 8 53 34 65

PCNA: proliferating cell nuclear antigen.


years (P , 0.0001). Roughly half of patients received apreoperative performance score of 1 (Table 3). Theremainder were split relatively evenly among favor-able scores (0) and poor scores (2–3). Overall survivalwas significantly and inversely proportional to theperformance score, with median survival times of 33.3months, 23.6 months, and 6.5 months for scores of 0,1, and 2–3, respectively (P 5 0.0089). Neither gendernor extent of surgical resection was associated withsurvival (Table 3).

Mitotic IndexA WHO Grade 3 designation was based on the findingof a solitary mitosis in 10 of 66 patients with AA (15%)and in none of the patients with AOA (Table 2). Mul-tiple mitoses were identified in the rest of the speci-mens. Patients with AAs that showed only 1 mitoticfigure had a median survival of 54.2 months comparedwith 18.2 months for those with multiple mitoses. Dueto the rarity of the former group, however, this survivaldifference did not reach statistical significance (P 50.14).

DNA Flow Cytometry/Ploidy by Image AnalysisDNA flow cytometry was evaluable in 58 (68%) pa-tients, including 42 patients with AA and 16 patientswith AOA. The success rate of performing flow cytom-etry from paraffin blocks, as expected, was highly as-sociated with extent of resection (i.e., specimen size).Results were indeterminate in 68% of biopsies, in 28%of subtotal resections, and in 13% of gross total resec-tions (P 5 0.002). There were associations betweenploidy and both patient age and overall survival. Themedian ages were 52 years, 43 years, and 41 years fordiploidy, tetraploidy, and aneuploidy, respectively. Al-though it was suggestive of greater aneuploidy in

young patients, this association did not quite reachstatistical significance (P 5 0.06). Conversely, mediansurvival times were 19.9 months, 24.2 months, and71.4 months for the diploid, tetraploid, and aneuploidgroups, respectively, by flow cytometry. This associa-tion of prolonged survival in patients with aneuploidneoplasms was statistically significant on univariateanalysis (P 5 0.04).

S-phase fractions ranged from 0% to 30%, with amean of 8.8% and median of 7.0%. A high S-phasefraction ($6%) was identified in 34 patients (65%) andwas not associated with survival.

DNA ploidy by image analysis of feulgen stainedsections was assessable in all 85 patients. The concor-dance rate for ploidy analysis by the two methods wasonly 62% (test of symmetry; P 5 0.29). The most com-mon discrepancies were diploid by flow cytometry,aneuploid by image analysis (6 patients); tetraploid byflow, aneuploid by image analysis (5 patients); andaneuploid by flow, tetraploid by image analysis (5patients). As in the case of flow cytometry, diploidy byimage analysis was associated with a greater medianage (52 years), although this did not reach statisticalsignificance (P 5 0.13). Women also were more likelyto have diploid gliomas (P 5 0.08). We found no as-sociation between ploidy by image analysis and sur-vival.

p53 ExpressionThe p53 LI ranged from 0.1% to 80.0%, with a mean of20.3% and a median of 13.0%. Forty-six tumors (54%)were considered immunopositive (LI . 10%). Therewas a trend toward decreased survival in p53 negativecases. The median survival time was only 10.4 monthsin this group compared with 29.3 months for p53positive tumors (P 5 0.07). No obvious association ofp53 expression and patient age was found.

FIGURE 2. Kaplan–Meier survival curve of 85 anaplastic astrocytic neo-

plasms based on patient age at diagnosis (P , 0.0001).

FIGURE 1. Kaplan–Meier survival curves of the 66 patients with anaplastic

astrocytoma (AA) versus the 19 patients with anaplastic oligoastrocytoma (AOA)

(P 5 0.65).


Proliferative Indices by ImmunohistochemistryThe MIB-1 LI ranged from 0% to 77.9%, with a mean of12.5% and a median of 5.8%. A high index (.5%) waspresent in 56% of specimens and was unassociatedwith overall survival in the group of 85 patients (Table3). However, analysis of the AA group revealed a trendtoward shorter survival in those with elevated indices(P 5 0.20). The PCNA LI ranged from 0.2% to 83.7%,

with a mean of 21.8% and a median of 11.8%. A highindex ($20%) was found in 39% of all cases with atrend toward decreased survival time (P 5 0.20). Op-posite to the findings with MIB-1 LI analysis, this trendwas weakened when assessing patients with AA alone.In general, PCNA also was associated with greatervariability in nuclear staining intensity, making inter-pretations more difficult than with MIB-1 LI.

TABLE 3Survival Estimates for 85 Patients

Baseline factor No. Deaths

Survival distributions

Median(months)

95% CI(months)

Log rankP value

Age (yrs) 0.0001,40 31 19 65.5 38.7–95.540–60 33 28 22.1 14.3–36.4.60 21 21 4.4 3.5–8.0

Gender 0.57Female 37 29 22.1 10.4–44.8Male 48 39 20.1 9.9–48.6

Extent of resection 0.93Biopsy 29 23 19.2 6.6–34.4Subtotal 41 34 21.7 11.9–55.9Gross total 14 10 20.1 9.8–59

Performance score 0.00890 20 13 33.3 10.4–NA1 42 33 23.6 14.3–59.02–3 22 21 6.5 4.0–19.2

Histology 0.65Astrocytoma 66 54 21.9 13.7–36.4Oligoastrocytoma 19 14 14.3 9.2–109.8

Mitoses 0.14Singular 10 7 54.2 32.3–88.1Multiple 75 61 17.9 9.9–23.0

MIB-1 (%) 0.31#5 37 27 23.0 7.9–65.5.5 48 41 15.1 9.8–36.4

p53 (%) 0.0658#10 39 34 10.4 6.3–21.7.10 46 34 29.3 18.5–64.3

PCNA (%) 0.13,20 52 41 25.3 18.5–55.920–40 17 13 19.9 6.6–64.3.40 16 14 9.1 4.1–21.7

Ploidy (image analysis) 0.25Diploid 20 18 15.8 4.4–22.1Tetraploid 25 17 36.4 9.2–59.0Aneuploid 40 33 21.6 10.4–64.3

Ploidy (flow cytometry) 0.0393Diploid 17 16 19.9 8.4–36.4Tetraploid 17 15 24.2 14.3–60.0Aneuploid 24 14 71.4 21.7–NA

% S-phase 0.86,6 18 14 22.0 14.3–71.4$6 34 25 38.7 18.5–65.5

95% CI: 95% confidence interval; NA: not available; PCNA: proliferating cell nuclear antigen.


Multivariate ModelingMultivariate analyses to identify prognostic factors forsurvival were performed as follows: First, 14 variableswere defined to quantify the 10 baseline factors listedin Table 3 for which data were available on all 85patients (e.g., ploidy and S-phase by flow cytometrywere excluded from this analysis). Then, Cox modelforward and backward selection procedures were ap-plied to identify which of the 14 variables were mostassociated strongly with survival in these 85 patients.The results are summarized in Table 4. Both the Coxforward selection model and the Cox backward selec-tion model thus calculated contained only the 2 agevariables. No strong association with survival could befound for any of the remaining 12 variables after ad-justment for the effect of patient age.

To validate the resulting variable selection, a boot-strap procedure was applied as follows: Five hundredrandom samples of size 85 were drawn with replace-ment from the original 85 patients, and the Cox for-ward model was generated for each of these 500 sam-ples to identify the variables that were associatedsignificantly (P , 0.05) with survival in that sample.Then, for each of the 14 variables, the percentage ofthe 500 Cox models that included that variable wascalculated (Table 4). In such boot-strap model valida-

tion procedures, variables must be included in ap-proximately 70% or more of the models generated inthe boot-strap samples to be considered validatedprognostic factors.17 By that rule, only 2 validatedprognostic factors were identified, i.e., the variable age.60 years, which was included in all 500 boot-strapmodels, and the variable age 40 – 60 years, which wasincluded in 67.6% of the boot-strap models. Thus, thevariables selected for the Cox model by using bothforward and backward selection procedures were val-idated by the boot-strap procedure.

DISCUSSIONClinical Prognostic FactorsPatient age is the prognostic variable that is reportedmost consistently for astrocytic neoplasms, regardlessof histologic grade.1– 4,6,14,15,19 –35 This was particularlytrue in our set of purely Grade III tumors, because thesurvival curves were strikingly different in the threepatient age groups. On multivariate analysis, no otherclinical or pathologic factor contributed significantlyany further variability in survival time as long as pa-tient age was in the model.

This study was not designed specifically to discernbiological mechanisms responsible for the impact ofpatient age on survival. However, the association of

TABLE 4Multivariate Cox Modeling Results for 85 Patients

Variable

Model selection procedureBootstrap validation

Forward BackwardInclusionpercent

Rank of inclusionpercentP value Step entered P value Step removed

Age (yrs)40–60 0.0122 2 0.0142 67.6 2.60 0.0001 1 0.0001 100.0 1

Male gender — — 0.96 1 10.4 11Extent of resection

Subtotala — — 0.31 10 15.2 8Gross totala — — 0.57 7 9.4 12

Performance score(continuous)a — — 0.52 8 28.8 4

MIB-1 .5%a — — 0.73 6 19.0 5p53 .10%a — — 0.37 9 16.8 7PCNA

20–40a — — 0.90 2 13.2 9.40a — — 0.14 11 18.2 6

Ploidy (image analysis)Tetraploida — — 0.75 4 11.6 10Aneuploida — — 0.85 3 8.4 13

AOA histologya — — 0.08 12 33.8 3Multiple mitosesa — — 0.73 5 8.0 14

PCNA: proliferating cell nuclear antigen; AOA: oglioastrocytoma.a These variables did not enter the forward model at the P 5 0.075 level.


patient age and ploidy anomalies is intriguing. Aneu-ploidy was associated with younger age and longersurvival, whereas diploidy was associated with olderage and shorter survival. It is possible that radiother-apeutic and chemotherapeutic treatments may bemore effective in aneuploid tumors, accounting forimproved survival. More likely, other unknown factorsaccount for differences in biologic behavior observedin younger patients. This has been suggested in pa-tients with glioblastoma: Older patients that presentde novo have tumors with amplification of the epider-mal growth factor receptor gene, whereas glioblas-tomas arising from malignant transformation of lowergrade tumors are associated with p53 mutations.36,37

We presently are evaluating our set of patients forspecific genetic alterations to assess the association ofsuch changes with patient age and survival.

Most recent studies of astrocytic neoplasms havefound that preoperative performance score was anindependent prognostic variable.1– 4,21,22,24 –30,32,35,37

Furthermore, performance score was significant onmultivariate analyses of the parent clinical trials,11–

13despite the subjectivity encountered in applying thisscore. In our subset of patients, there was a trendtoward longer survival with more favorable scores.This trend was overshadowed by the effects of patientage and the relatively small sample size; therefore,performance score did not enter as a significant vari-able on multivariate models.

Surprisingly, the extent of surgical resection hadno association with patient survival in our series, al-though gross total resection was not achieved fre-quently. Once again, the literature is divided onthis issue. Some studies have found that extent ofsurgical resection was an independent prognosticvariable,2,3,11–13,21,24,25,28,32,33 whereas others havenot.4,27,30,31,35,38 The subjectivity of this assessment iswell recognized, particularly with reference to infiltra-tive tumors, such as Grade III astrocytomas. We fur-ther recognize that interpretations regarding extent ofresection were less likely to be accurate in the earlierphases of this study, especially in those protocols per-formed prior to various neuroradiologic advances. An-other issue that must be considered is tissue sampling.Because 35% of our patients were diagnosed on biopsyonly, the risks of undersampling are substantial. Be-cause this study reflects the practices of both commu-nity and academic neurosurgical practices, it shouldbe regarded as a “realistic flaw” in the evaluation ofastrocytic neoplasms. Finally, the three cooperativetrials included a preponderance of glioblastomas,which may have influenced the identification of extentof resection as a clinical prognostic factor.

This study highlights the difficulties of performing

marker studies on tissue blocks collected retrospec-tively from patients entered on large cooperativegroup prospective trials. Tissue blocks often were dis-carded or misplaced by the member institutions, orthe tissue blocks did not have sufficient tumor mate-rial within them for our analyses. The patients withoutavailable tissue blocks or sufficient tumor materialwere older, died earlier, and were most likely to comefrom the oldest protocol (1979). Because pathologydepartments are required by statute to maintain theirtissue blocks for only a limited number of years, it wasnot surprising that we were unable to recover tissueblocks from the oldest protocol. Furthermore, patientsfrom the oldest protocol tended to be older and hadpoorer performance scores (data not shown). This isthe most likely explanation for the better survival ofthe patients we analyzed. In any case, had we beenable to collect additional tissue blocks, it is likely thatthe univariate and multivariate associations of patientage with survival would have been accentuated.

HistologyAlthough histologic variability was controlled partiallyin our study by including only Grade III gliomas, thecohort of tumors was not entirely uniform. For designreasons, both AAs and AOAs were included in theNCCTG protocols. Because differences in survivalhave been reported previously between patients withpure astrocytomas and those with oligoastrocyto-mas,6,39,42 we performed all of our analyses both onthe entire group of patients and on the group of pa-tients with AA alone. Tumors with a solitary mitoticfigure and frequent mitoses are classified into theGrade 3 category by both the revised WHO classifica-tion and the St. Anne-Mayo schema. However, thepresence of a single mitotic figure, particularly in alarge resection specimen, does not have the sameimplication as frequent mitoses.43 In support of thisviewpoint, we recently demonstrated that patientswith Grade 2 astrocytomas and Grade 3 astrocytomasassociated with a solitary mitosis have similar survivalrates.19 For this reason, we also compared the prog-nostic value of the presence of one mitosis versusmultiple mitoses in this data series.

In our NCCTG patient population, no significantdifferences were identified between patients with AAand patients with AOA, although the latter group wassmall. Similar to our previous study,19 the mediansurvival of AA patients with solitary mitoses was 54.2months, a survival time not unlike that expected forpatients with Grade II astrocytoma. However, becauseonly 10 patients displayed this feature, the survivaldifference did not reach statistical significance. It is ofparticular interest, however, that foci of tumor necro-


sis and/or endothelial hyperplasia were identified in 8of the patients with AA while they were being reeval-uated for mitotic index. These foci were found in thedeeper tissue sections obtained when the tissue blockswere recut. When we reanalyzed the survival of theremaining 58 patients with pure AA, the median sur-vival time increased to 29 months, but none of theconclusions regarding prognostic variables changed(data not shown). This finding emphasizes the gradingdifficulties encountered due to tumoral heterogeneityand stresses the need to evaluate tumor grade in thecontext of sampling adequacy and neuroimagingcharacteristics. Unfortunately, this information usu-ally is not available at the time of central pathologyreview for enrollment in large cooperative group treat-ment protocols. Tumoral heterogeneity is a well doc-umented problem in the study of gliomas and hasbeen demonstrated not only in terms of morphologybut also in terms of proliferation indices, ploidy,karyotype, growth factor and oncogene expressions,and resistance to various forms of therapy.44,45 Coonset al. have even demonstrated cytogenetic and ploidyvariations in a morphologically uniform tumor.45

DNA PloidyLike with several other variables, conclusions aboutploidy as a prognostic variable in astrocytic tumors aremixed and are sometimes contradictory. Whereas anumber of investigators found that aneuploidy wasassociated with shorter survival,33,34,46 – 48 others foundno association,49 and still others observed longer sur-vivals.15,50 In our series, aneuploidy was associatedwith younger patient age and longer survival time.Likewise, in a large study of astrocytic tumors, wefound that younger patients with glioblastomas livedlonger if their tumors were aneuploid by flow cytom-etry.15 Flow cytometry often was indeterminate in ourseries due to either insufficient tumor volume or aG0/G1 peak that was too wide to assess the tumorreliably. The success rate of flow cytometry in thisseries was inversely proportional to the extent of re-section (P 5 0.002). The high failure rate in biopsies,coupled with the fact that flow cytometry averages alltissue in the block (including the normal brain paren-chyma, which is almost invariably present in Grade IIIgliomas), makes this technique impractical. Althoughthese weaknesses generally are overcome by usingimage analysis for ploidy, our study found that thisparameter generated even weaker prognostic vari-ables. Because selectable yet smaller populations oftumor cells are assessed by image analysis, and be-cause clonal heterogeneity is common, perhaps it isnot surprising that the concordance for ploidy by us-ing these 2 techniques was only 60%.

Proliferative IndicesThe three proliferative indices assessed in our studywere MIB-1 (Ki-67) LI, PCNA LI, and S-phase fractionby flow cytometry. Although each one was analyzedboth as a continuous variable and using various cut-offs, we elected to report the data using the samecut-offs from our previous studies.14,15 Nevertheless,the conclusions did not seem to change significantlyregardless of the method chosen (data not shown).Most authors agree that MIB-1 LI is the most useful,most easily applied, and most easily interpreted ofthese techniques.19,22 However, whereas some authorsclaim that increased MIB-1 LI is an independent prog-nostic variable,19,21,38,51 others find it significant onlyin univariate analysis,14,22,23 because increased MIB-1LI is highly associated with tumor grade and patientage. Even with the latter associations, there is oftensignificant overlapping of values among astrocytomagrades.19 In addition, there are variations in techniqueand interpretation that make it difficult, if not impos-sible, to translate values from one laboratory to thenext. Therefore, considerable experience is requiredwithin a given laboratory to determine which valuesshould be considered elevated in which tumors. S-phase fractions suffer the same limitations discussedpreviously for flow cytometry performed on paraffinsections, whereas PCNA often is difficult to interpretdue to variations of nuclear staining intensity.

In our study, S-phase fraction was unassociatedwith survival, whereas MIB-1 LI and PCNA LI showedtrends that potentially could have been significantwith a larger number of cases. However, none of theseparameters was significant in multivariate models.This differs somewhat from the recent Mayo study byGiannini et al.19 In that series, MIB-1 LI was a highlysignificant prognostic variable in the group of patientswith Grade III astrocytomas and was independent ofpatient age on multivariate analysis. However, therewere many differences between that study and thisone. The series by Giannini et al. consisted of MayoClinic patients who were not treated uniformly andwho generally had a subtotal or gross total resection(47 of 50 patients) rather than biopsy alone. Also,special care was taken to perform the MIB-1 LI stain-ing in a single batch to reduce laboratory variability.The NCCTG patients in our study came from a varietyof institutions over a period of 18 years, with greatervariations in neurosurgical practices, variations in tis-sue fixation times and processing, higher risk of un-dersampling due to biopsy in only 35% of patients,and increased laboratory variation, because MIB-l LIimmunostaining was performed in three batches.Whereas the study by Giannini et al. represents ideal


laboratory practice and certainly is useful, in our opin-ion, the current study reflects more accurately theday-to-day clinical laboratory practice and, thus, ismore realistic clinically. We recommend the use ofMIB-1 LI in morphologically difficult tumors border-ing between Grades II and III.

p53 Protein ExpressionInactivation of the p53 gene is thought to represent anearly event in the tumorigenesis of diffuse astrocyto-mas, with mutations detectable in 30 – 40% of allgrades.52 The monoclonal antibody DO-7 recognizesboth wild type and mutant forms of the p53 protein.However, because wild type protein has a very shorthalf-life compared with the mutant form, the nuclearconcentrations of the former normally are below de-tection threshold using immunohistochemistry. Nev-ertheless, it has become clear that immunopositivitydoes not always equate with gene mutation, becausethe wild type protein can be stabilized via other mech-anisms. Ono et al. estimate that one-fourth of theirastrocytomas are p53 immunoreactive without an as-sociated gene mutation.52 In our study of patients withGrade III astrocytic tumors, p53 expression (LI . 10%)was demonstrable in 54% of patients, including themixed oligoastrocytomas. Similar to our prior study14

and to the literature in general, immunoreactivity wasnot prognostic.23,53–56 Some have claimed that thisstain may be used to distinguish neoplastic from re-active astrocytes.57 However, we would cautionagainst relying solely on this method, because, occa-sionally, we have seen positivity in nonneoplastic con-ditions (A.P. and B.W.S., personal observations).

CONCLUSIONSPatient age remains a powerful prognostic variable inpatients with Grade III astrocytic neoplasms and ac-counts for much of the wide variability associated withoverall survival. In our experience with patientstreated on NCCTG protocol, once patient age wastaken into account, no other clinical or pathologicvariables added significantly to the multivariate mod-els, including DNA ploidy, proliferation markers, andp53 expression. It is noteworthy that our study em-phasizes the difficulties of comparing data derivedfrom different patient populations and highlights theproblems of tumoral heterogeneity and adequate tu-mor sampling.

REFERENCES1. Schuurman PR, Troost D, Verbeeten B Jr., Bosch DA. Five-

year survival and clinical prognostic factors in progressivesupratentorial diffuse “low-grade” astrocytoma: a retrospec-tive analysis of 46 cases. Acta Neurochirurg 1997;139:2–7.

2. Nicolato A, Gerosa MA, Fina P, Iuzzolino P, Giorgiutti F,Bricolo A. Prognostic factors in low-grade supratentorialastrocytomas: a uni-multivariate statistical analysis in 76surgically treated adult patients. Surg Neurol 1995;44:208 –21.

3. Loiseau H, Bousquet P, Rivel J, Vital C, Kantor G, Rougier A,et al. Astrocytomes de bas grade sus-tentoriels de l’adulte.Facteurs pronostics et indications therapeutiques. A proposd’une serie de 141 patients. Neuro-Chirurgie 1995;41:38 –50.

4. Fischbach AJ, Martz KL, Nelson JS, Griffin TW, Chang CH,Horton J, et al. Long-term survival in treated anaplasticastrocytomas. A report of combined RTOG/ECOG studies.Am J Clin Oncol 1991;14:365–70.

5. Nakamura M, Konishi N, Tsunoda S, Hiasa Y, Tsuzuki T, InuiT, et al. Retinoblastoma protein expression and MIB-1 cor-relate with survival of patients with malignant astrocytoma.Cancer 1997;80:242–9.

6. Donahue B, Scott CB, Nelson JS, Rotman M, Murray KJ,Nelson DF, et al. Influence of an oligodendroglial compo-nent on the survival of patients with anaplastic astrocyto-mas: a report of Radiation Therapy Oncology Group 83-02.Int J Radiat Oncol Biol Phys 1997;911– 4.

7. Kleihues P, Burger PC, Scheithauer BW. Histological typingof tumours of the central nervous system. World HealthOrganization. New York: Springer-Verlag, 1993.

8. Daumas-Duport C, Scheithauer B, O’Fallon J, Kelly P. Grad-ing of astrocytomas: a simple and reproducible method.Cancer 1988;62:2152– 65.

9. Zubrod CG, Schneiderman M, Frei E III, Brindley C, GoldGL, Shnider B, et al. Appraisal of methods for the study ofchemotherapy of cancer in man: comparative therapeutictrial of nitrogen mustard and triethylene thiophosphor-amide. J Chron Dis 1960;7–33.

10. Oken MM, Creech RH, Tormey DC, Horton J, Davis TE,McFadden ET, et al. Toxicity and response criteria of theEastern Cooperative Oncology Group. Am J Clin Oncol 1982;5:649 –55.

11. Elliott TE, Dinapoli RP, O’Fallon JR, Krook JE, Earle JE,Morton RF, et al. Randomized trial of radiation therapy (RT)plus dibromodulcitol (DBD) versus RT plus BCNU in high-grade astrocytoma. J Neuro Oncol 1997;33:239 –50.

12. Dinapoli RP, Brown LD, Arusell RM, Earle JD, O’Fallon JR,Buckner JC, et al. Phase III comparative evaluation of PCNUand BCNU combined with radiation therapy for high-gradeglioma. J Clin Oncol 1993;11:1316 –21.

13. Buckner JC, Schomberg PJ, McGinnis WL, Cascino TL,Scheithauer BW, O’Fallon JR. Phase III study of radiationtherapy (RT) plus BCNU with or without recombinant in-terferon alpha (IFN) in high grade glioma: an NCCTG study.Proc Am Soc Clin Oncol 1997;16:385a(A1374).

14. Cunningham JM, Kimmel DW, Scheithauer BW, O’Fallon JR,Novotny PJ, Jenkins RB. Analysis of proliferation markersand p53 expression in gliomas of astrocytic origin: relation-ships and prognostic value. J Neurosurg 1997;86:121–30.

15. Ganju V, Jenkins RB, O’Fallon JR, Scheithauer BW, RansomDT, Katzmann JA, et al. Prognostic factors in gliomas. Amultivariate analysis of clinical, pathologic, flow cytometric,cytogenetic, and molecular markers. Cancer 1994;74:920 –7.

16. Dressler LG, Seemer L, Owens MA, Clark GM, McGuire WL.Evaluation of a modeling system for S-phase estimation inbreast cancer by flow cytometry. Cancer Res 1987;47:5294 –302.


17. Sauerbrei W, Schumacher M. A bootstrap resampling pro-cedure for model building: application to the Cox regressionmodel. Stat Med 1992;11:2093–109.

18. Efron B. Bootstrap methods: another look at the jackknife.Ann Stat 1979;7:1–26.

19. Giannini C, Scheithauer BW, Burger PC, Christensen MR,Wollan PC, Sebo TJ, et al. Cellular proliferation in pilocyticand diffuse astrocytomas: an MIB-1 labeling index study.J Neuropathol Exp Neurol 1999;58:46 –53.

20. Hsu DW, Louis DN, Efird JT, Hedley-Whyte ET. Use ofMIB-1 (Ki-67) immunoreactivity in differentiating Grade IIand Grade III gliomas. J Neuropathol Exp Neurol 1997;56:857– 65.

21. Schiffer D, Cavalla P, Chio A, Richiardi P, Giordana MT.Proliferative activity and prognosis of low-grade astrocyto-mas. J Neurooncol 1997;34:31–5.

22. McKeever PE, Ross DA, Strawderman MS, Brunberg JA,Greenberg HS, Junck L. A comparison of the predictivepower for survival in gliomas provided by MIB-1, bromode-oxyuridine and proliferating cell nuclear antigen with his-topathologic and clinical parameters. J Neuropathol ExpNeurol 1997;56:798 – 805.

23. Ellison DW, Steart PV, Bateman AC, Pickering RM, PalmerJD, Weller RO. Prognostic indicators in a range of astrocytictumours: an immunohistochemical study with Ki-67 andp53 antibodies. J Neurol Neurosurg Psychiatr 1995;59:413–9.

24. Winger MJ, MacDonald DR, Cairncross JG. Supratentorialanaplastic gliomas in adults. The prognostic importance ofextent of resection and prior low-grade glioma. J Neurosurg1989;71:487–93.

25. Vecht CJ, Avezaat CJJ, van Putten WLJ, Eijkenboom WMH,Stefanko SZ. The influence of the extent of surgery on theneurological function and survival in malignant glioma. Aretrospective analysis in 243 patients. J Neurol NeurosurgPsychiatr 1990;53:466 –71.

26. Phillips TL, Levin VA, Ahn DK, Gutin PH, Davis RL, WilsonCB, et al. Evaluation of bromodeoxyuridine in glioblastomamultiforme: a Northern California Cancer Center Phase IIstudy. Int J Radiat Oncol Biol Phys 1991;21:709 –14.

27. Duncan GG, Goodman GB, Ludgate CM, Rheaume DE. Thetreatment of adult supratentorial high grade astrocytomas.J Neurooncol 1992;13:63–72.

28. Simpson JR, Horton J, Scott C, Curran WJ, Rubin P, Fisch-bach J, et al. Influence of location and extent of surgicalresection on survival of patients with glioblastoma multi-forme: results of three consecutive Radiation Therapy On-cology Group (RTOG) clinical trials. Int J Radiat Oncol BiolPhys 1993;26:239 – 44.

29. Halperin EC, Gaspar L, Imperato J, Salter M, Herndon J,Dowling S. An analysis of radiotherapy data from the CNSCancer Consortium’s randomized prospective trial compar-ing AZQ to BCNU in the treatment of patients with primarymalignant brain tumors. Am J Clin Oncol 1993;16:277– 83.

30. Yaneva M, Arolski I, Iliev E, Zaprianov Z. Factors influencingprognosis of patients with CNS supratentorial astroglial tu-mors who underwent postoperative radiotherapy. FoliaMedica 1993;35:49 –54.

31. Sneed PK, Prados MD, McDermott MW, Larson DA, MalecMK, Lamborn KR, et al. Large effect of age on the survival ofpatients with glioblastoma treated with radiotherapy andbrachytherapy boost. Neurosurgery 1995;36:898 –903.

32. Gundersen S, Lote K, Hannisdal E. Prognostic factors forglioblastoma multiforme. Development of a prognostic in-dex. Acta Oncol 1996;35(Suppl 8):123–7.

33. Coons SW, Johnson PC, Pearl DK. Prognostic significance offlow cytometry deoxyribonucleic acid analysis of humanastrocytomas. Neurosurgery 1994;35:119 –26.

34. Varuch L, Nordenskuold B, Carstensen J, Enestrom S. Prog-nostic value of flow cytometry and correlation to someconventional prognostic factors: a retrospective study ofarchival specimens of 134 astrocytomas. J Neurosurg 1996;85:146 –51.

35. Kowalczuk A, Macdonald RL, Amidei C, Dohrmann G III,Erickson RK, Hekmatpanah J, et al. Quantitative imagingstudy of extent of surgical resection and prognosis of ma-lignant astrocytomas. Neurosurgery 1997;41:1028 –36.

36. von Deimling A, von Ammon K, Schoenfeld D, Wiestler OD,Seizinger BR, Louis DN. Subsets of glioblastoma multiformedefined by molecular genetic analysis. Brain Pathol 1993;3:19 –26.

37. Watanabe K, Tachibana O, Sata K, Yonekawa Y, Kleihues P,Ohgaki H. Overexpression of the EGF receptor and p53mutations are mutually exclusive in the evolution of primaryand secondary glioblastomas. Brain Pathol 1996;6:217–24.

38. Wakimoto H, Aoyagi M, Nakayama T, Nagashima G, Yam-mamoto S, Tamaki M, et al. Prognostic significance of Ki-67labeling indices obtained using MIB-1 monoclonal antibodyin patients with supratentorial astrocytomas. Cancer 1996;77:373– 80.

39. Landy HJ, Feun L, Schwade JG, Snodgrass S, Lu Y, GutmanF. Retreatment of intracranial gliomas. Southern Med J 1994;87:211– 4.

40. Leighton C, Fisher B, Bauman G, Depiero S, Stitt L, Mac-Donald D, et al. Supratentorial low-grade glioma in adults:an analysis of prognostic factors and timing of radiation.J Clin Oncol 1997;15:1294 –301.

41. Tamura M, Zama A, Kurihara H, Kano T, Imai H, Ishiuchi S,et al. Clinicohistological study of oligodendroglioma andoligoastrocytoma. Brain Tumor Pathol 1997;14:35–9.

42. Shaw EG, Scheithauer BW, O’Fallon JR. Supratentorial glio-mas: a comparative study by grade and histologic type.J Neurooncol 1997;31:273– 8.

43. Coons SW, Pearl DK. Mitosis identification in diffuse glio-mas. Implications for tumor grading. Cancer 1998;82:1550 –5.

44. Shapiro JR, Mehta BM, Ebrahim SAD, Scheck AC, Moots PL,Fiola MR. Tumor heterogeneity and intrinsically chemore-sistant subpopulations in freshly resected human malignantgliomas. In: Sudilovsky O, Pitot HC, Liotta LA, editors.Boundaries between promotion and progression during car-cinogenesis. New York: Plenum Publishing Corp., 1991:243–262.

45. Coons SW, Johnson PC, Shapiro JR. Cytogenetic and flowcytometry DNA analysis of regional heterogeneity in a lowgrade human glioma. Cancer Res 1995;55:1569 –77.

46. Cho KG, Nagashima T, Barnwell S, Hoshino T. Flow cyto-metric determination of modal DNA population in relationto proliferative potential of human intracranial neoplasms.J Neurosurg 1988;69:588 –92.

47. Nishizaki T, Orita T, Furutani Y, Ikeyama Y, Aoki H, Sasaki K.Flow-cytometric DNA analysis and immunohistochemicalmeasurement of Ki-67 labeling indices in human brain tu-mors. J Neurosurg 1989;70:379 – 84.

48. Danova M, Giaretti W, Merlo F, Mazzini G, Gaetani P, GeidoE, et al. Prognostic significance of nuclear DNA content inhuman neuroepithelial tumors. Int J Cancer 1991;48:663–7.


49. Jiminez O, Timms A, Quirke P, McLaughlin JE. Prognosis onmalignant glioma. A retrospective study of biopsy speci-mens by flow cytometry. Neuropathol Appl Neurobiol 1989;15:331– 8.

50. Mork SJ, Laerum OD. Modal DNA content of human intra-cranial neoplasms studied by flow cytometry. J Neurosurg1980;53:198 –204.

51. Kirkegaard LJ, DeRose PB, Yao B, Cohen C. Image cytomet-ric measurement of nuclear proliferation markers (MIB-1,PCNA) in astrocytomas. Prognostic significance. Am J ClinPathol 1998;109:69 –74.

52. Ono Y, Tamiya T, Ichikawa T, Matsumoto K, Furuta T,Ohmoto T, et al. Accumulation of wild-type p53 in astrocy-tomas is associated with increased p21 expression. ActaNeuropathol 1997;94:21–7.

53. Kordek R, Biernat W, Alwasiak J, Maculewicz R, Yanagiahara

R, Libersi PP. p53 protein and epidermal growth factor re-ceptor expression in human astrocytomas. J Neurooncol1995;26:11– 6.

54. Danks RA, Chopra G, Gonzales MF, Orian JM, Kaye AH. Aber-rant p53 expression does not correlate with the prognosis inanaplastic astrocytoma. Neurosurgery 1995;37:246–54.

55. al-Sarraj S, Bridges LR. p53 immunoreactivity in astrocyto-mas and its relationship to survival. Br J Neurosurg 1995;9:143–9.

56. Chozick BS, Pezzullo JC, Epstein MH, Finch PW. Prognosticimplications of p53 overexpression in supratentorial astro-cytic tumors. Neurosurgery 1994;35:831– 8.

57. Yaziji H, Massarani-Wafai R, Gujrati M, Kuhns JG, MartinAW, Parker JC Jr. Role of p53 immunohistochemistry indifferentiating reactive gliosis from malignant astrocytic le-sions. Am J Surg Pathol 1996;20:1086 –90.


clinicopathologic study of 85 similarly treated patients with anaplastic astrocytic tumors

Documents