formed in men with PSA 1.0 ng./ml. or greater (reference 21 inarticle). The authors note that the current challenge is to improvethe accuracy of cancer detection in the lower PSA ranges. Unneces-sary biopsies can be reduced by restricting the biopsy recommenda-tion to men at relatively higher risk in this generally low riskpopulation. This stratification can be achieved by combiningrisk factors, biochemical markers and statistical methods to providebetter discrimination of men more likely to have prostate cancer, asdiscussed in the article.

More frequent detection of prostate cancer in its curable stageswill require the use of lower PSA cutoffs and new screening modelsusing such cutoffs have the potential to detect more clinically impor-tant and potentially curable cancers that would not be detected withthe current practice. The use of lower PSA cutoffs will increase thechances for cure.

William J. CatalonaDivision of Urologic SurgeryWashington University School of MedicineSt. Louis, Missouri

REFERENCES

1. Catalona, W. J., Smith, D. S. and Ornstein, D. K.: Prostatecancer detection in men with serum PSA concentrations of 2.6to 4 ng/mL and benign prostate examination: enhancement ofspecificity with free PSA measurements. JAMA, 277: 1452,1997

2. Carter, H. B., Epstein, J. I. and Partin, A. W.: Influence of ageand prostate-specific antigen on the chance of curable prostatecancer among men with nonpalpable disease. Urology, 53: 126,1999

3. Catalona, W. J., Richie, J. P., Ahmann, F. R. et al: Comparison ofdigital rectal examination and serum prostate specific antigenin the early detection of prostate cancer: results of a multicenter clinical trial of 6,630 men. J Urol, 151: 1283, 1994

Screening for prostate cancer using serum PSA continues to becontroversial. While the American Urological Association and theAmerican Cancer Society recommend screening men older than 50years with digital rectal examination and serum PSA, the UnitedStates Preventive Services Task Force does not recommend thispractice. The American College of Physicians recommends neitherapproach but suggests that physicians should describe the potentialbenefits and known harms of screening, diagnosis and treatment. Inthis environment screening is considered positive if an abnormaldigital rectal examination is identified or serum PSA is greater than4.0 ng./ml. These varying recommendations reflect concerns aboutthe performance of serum PSA and digital rectal examination asscreening tests, and uncertainty surrounding the relative improve-ment in outcomes following aggressive interventions.

The authors have taken advantage of data collected as part ofERSPC to explore the performance of serum PSA and digital rectalexamination in detecting cancer among men with serum PSA 0.0 to4.0 ng./ml. They estimate that by using 4.0 ng./ml. as the cutoff todifferentiate normal from abnormal, physicians risk missing approx-imately 64% of potentially detectable cancers. Of these tumors ap-proximately half are potentially lethal because Gleason score is 7 orgreater and a significant majority (84%) are organ confined. Becausetumors with Gleason scores 7 or greater are rare in men with serumPSA less than 2.0 ng./ml., the authors argue against these menundergoing transrectal ultrasound and biopsy. Because a significantnumber of men with serum PSA greater than 4.0 ng./ml. who un-dergo prostate biopsy and subsequent surgery ultimately have ex-tracapsular disease, the authors argue that the appropriate windowof opportunity for screening for prostate cancer probably requiresidentifying disease earlier in its evolution when the serum PSA isbetween 2.0 and 4.0 ng./ml.

Although these results provide important new data concerning theperformance of serum PSA testing among men with low PSA, clini-cians should view the recommendations with some caution. Therationale for conducting screening trials is to determine whetherscreening for prostate cancer significantly lowers prostate cancermortality. Answers should be available from ERSPC and AmericanProstate, Lung, Colon and Ovarian Cancer Screening Project trialssometime in the next decade. Furthermore, data from the Surveil-lance, Epidemiology and End Results program should show a dra-matic decline in prostate cancer mortality in the next decade if

current screening efforts do result in early detection and cure. If datafrom these sources support screening for prostate cancer, the argu-ments will carry significant weight and justify lowering the currentPSA cutoff. If these trials prove negative or we fail to see a significantdecrease in prostate cancer mortality in the next decade, we face adilemma. Are current screening efforts and trials targeted at thewrong population? Do we miss the opportunity to cure some men bysetting the cutoff at 4.0 ng./ml. or are we saving a greater numberwho are not destined to benefit from intervention with its complica-tions, and emotional and financial costs? In the absence of data thatclearly demonstrate which patients benefit from aggressive interven-tion, performing biopsy in men with even lower serum PSA canpotentially cause more harm than good.

Ultimately questions surrounding the benefits of screening andtreatment are best answered by good scientific randomized trials.The authors have been instrumental in mounting an extraordinaryeffort to address the issue of prostate cancer screening using serumPSA testing. The task is daunting but the approach is clearly themost appropriate and definitive to answer the question. However,until results become available clinicians will continue to strugglewith how best to implement serum PSA and the many other newtests designed to identify localized prostate cancer. When makingrecommendations to patients clinicians need to clarify what we doand do not know about prostate cancer screening and treatment.Patients need to understand the series of events that frequentlybegins with prostate cancer screening with serum PSA. Only byunderstanding the potential risks and benefits can patients make aninformed choice concerning how best to proceed.

Peter C. AlbertsenDepartment of UrologyUniversity of Connecticut Health CenterFarmington, Connecticut

REPLY BY AUTHORS

Catalona compares the data of ERSPC to those produced in thelarge case study in St. Louis. It is interesting and noteworthy thatdata obtained in these 2 protocols are similar with respect to detec-tion rates and Gleason scores per PSA range. This similarity showsthat different types of recruitment (population registry versus ad-vertisement) seem to have no significant impact on either of theseparameters. Also, the comparison indicates that the Northern Euro-pean population of men recruited in The Netherlands is comparableper PSA range to a North American population. These importantfindings may facilitate interpretation of United States and Europeandata in the future.

The investigators of ERSPC Rotterdam section have a view ofrectal examination which is different from that of Catalona. In ourstudy digital rectal examination is exclusively considered in thecontext of screening. The clinical value of digital rectal examination,especially with respect to staging of the primary tumor, was notconsidered. The limited value of rectal examination for screening forprostate cancer has been more extensively discussed elsewhere (ref-erence 14 in article). The main criticism is that too many relevantcancers are missed in the PSA 3 to 3.9 and probably also in 2 to 2.9ng./ml. ranges. If screening ever becomes a health care policy, alter-native testing must be developed.

Albertsen addresses the controversy around screening for prostatecancer. We agree that caution must be used in the application ofmore aggressive biopsy indications in clinical practice. To our knowl-edge it is not known whether screening for prostate cancer lowersdisease related mortality, and what the cost is in terms of quality oflife and over treatment. However, we believe that the opportunitygiven by large randomized screening studies to improve screeningtechnology at least conceptually should be used. Even if these trialsare negative, it may be possible that an effect on prostate cancermortality is shown for select groups of men with clearly aggressivebut still locally confined disease. As the Surveillance, Epidemiologyand End Results Program statistics indicate, such differences may bevisible in much smaller populations and at a much earlier time thanthe planned 10 to 15 years of followup (reference 25 in article).

There is a need to improve screening strategies which providebetter selectivity (the right window of opportunity), improved sensi-tivity and better specificity (avoidance of unnecessary biopsies) forapplication to men who will be shown in the future to benefit fromscreening. Our report should be and has been correctly seen in thiscontext.

PROSTATE CANCER DETECTION AT LOW PROSTATE SPECIFIC ANTIGEN812