Prostata-shg.de

Recommended Prostate-Specific Antigen Testing Intervals for the Detection of Curable Prostate Cancer H. Ballentine Carter, MD; Jonathan I. Epstein, MD; Daniel W. Chan, PhD; James L. Fozard, PhD; Jay D. Pearson, PhD Objective. p=m- Toevaluate prostate-specific antigen (PSA) testing intervals that curable cancer with less frequent test¬ maintain the detection of curable cancer and reduce unnecessary testing.
ing. Thus, a reduction of testing neces¬ Design and Patients. p=m- Historicalprospective study of serial PSA measure- sarily requires evidence that the pro¬ ments at 2- and 4-year intervals from frozen serum samples of 40 men who even- posed testing interval will not result in tually developed prostate a lost opportunity for cure of cancer. To cancer and 272 men without prostate cancer who were identify those men who a prospective aging study (Gerontology Research Center of the Na- yearly PSA testing, we evaluated PSA tional Institute on Aging, the Baltimore Longitudinal Study of Aging) and the case testing intervals that would maintain series of 389 consecutive men treated surgically for nonpalpable prostate cancer.
the detection of curable cancer while Main Outcome Measures. p=m- Probabilityof a PSA conversion to 4.1 to 5.0 ng/ reducing the number of unnecessary mL and to greater than 5.0 ng/mL at 2 and 4 years and probability of detecting cur- able prostate cancer by age and PSA level.
Results. p=m- Whenthe pretreatment PSA level was less than or equal to 4.0 ng/ mL, nonpalpable prostate cancers were highly likely (34/36, 94%) to be curable or capsular penetration with Gleason score <7 and negative mar- gins), and the majority (25/36,69%) were small cancers (confined tumor m=le 0.5cm3 Baltimore Longitudinal Study ofAg¬ with no Gleason pattern 4 or 5). When the pretreatment PSA level was greater than ing (BLSA) Database.—The BLSA is 4.0 ng/mL and less than an ongoing, long-term prospective study or equal to 5.0 ng/mL, cancers were highly likely to be cur- able (32/36,89%), and aging conducted by the National In¬ a minority were small cancers (12/36,33%). When the pre- stitute on Aging, Bethesda, Md, which treatment PSA level was greater than 5.0 ng/mL, 96 (30%) of 317 cancers were has been described previously.3 Since noncurable. The PSA conversion (for cancer cases) to a level at which cure is less the inception of the BLSA in 1958, a likely (>5.0 ng/mL) is rare (0%) after 2 or 4 years when the initial PSA is less than total of 1527 men and 654 women have 2.0 ng/mL. PSA conversion to a range at which cancers are likely to be curable and participated in the study for varying less likely to be small (4.1-5.0 ng/mL) is rare after 2 years (0%-4%) when the base- lengths oftime. Participants in the study line PSA level is less than 2.0 ng/mL but common when the baseline PSA level is return for follow-up visits at approxi¬ between 2.1 and 3.0 ng/mL (27%) or 3.1 and 4.0 ng/mL (36%).
mate 2-year intervals.
Conclusions. p=m- Thesedata suggest that for Serum PSA levels have been men with no cancer suspected on digital rectal examination, on a total of 681 men aged 55 a PSA level of 4.0 to 5.0 ng/mL is an acceptable range for maintaining the detection of curable prostate years and older either at the time of cancer and a 2-year PSA testing routine subject visits (since 1991) or us¬ interval is not likely to miss a curable prostate cancer when the initial PSA level is ing a frozen serum bank for retrospec¬ less than 2.0 ng/mL. Recognizing that 70% of a screened population between the tive samples. All PSA measurements ages of 50 years and 70 years have PSA levels less than 2.0 ng/mL, elimination performed using a monoclonal of annual PSA testing for these men would result in large health care cost savings.
immunoradiometric assay (Tandem-R, JAMA. 1997;277:1456-1460 Hybritech Ine, San Diego, Calif). The stability of PSA in these frozen serum samples stored at -70°C has been pre¬ THE USE of prostate-specific antigen ficacy of early detection and treatment.1 viously described.4 After excluding (PSA) testing in asymptomatic men for Despite this controversy, PSA testing individuals whose medical history of detection of prostate cancer is contro¬ is being widely used in the United States prostate disease could not be verified, versial due to the lack of evidence from for the early diagnosis of prostate can¬ individuals with a prior history of pros¬ randomized trials demonstrating the ef- cer, and a recommendation for yearly tate surgery, and individuals being PSA testing of men at greatest risk has treated with finasteride (Prosear, Merck, been made.2 Thus, even in the absence Whitehouse Station, NJ), 77 men with Urology (Drs Carter and support a policy recom¬ prostate cancer and 517 men with no Pearson) and Pathology (Drs Epstein and Chan), Johns mendation for PSÄ screening, it is im¬ evidence of prostate cancer remained in Hopkins University School of Medicine, James Buchanan Brady Urological Institute, Johns Hopkins portant to determine ifunnecessary PSA the study group.
Hospital, Baltimore, Md; the Gerontology Research tests can be reduced by testing lower- To examine the probability of PSA Center, National Institute on Aging, Baltimore, Md (Drs risk men on a less frequent basis.
increases of different sizes at 2- and Fozard and Pearson); and the Department of Epidemi- The frequency of screening for any 4-year intervals, we identified all men ology, Merck Research Laboratories, Blue Bell, Pa (Dr cancer is a trade-off between unneces¬ who had 1 or more pairs of PSA mea¬ Reprints: H. Ballentine Carter, MD, Department of sary testing and the downstream effects surements taken between IV2 and 2V2 Urology, 403 Marburg, Johns Hopkins Hospital, 600 N false-positive tests as a result of fre¬ apart. This led to the final study quent screening and the risk of missing groups of 40 cancer cases (median age at diagnosis, 73 years; range, 54-87 years) coat of the seminal vesicles. Tumor vol¬ PSA level was 2.0 to 2.9 ng/mL or 3.0 to and 272 men with no evidence of pros¬ ume was calculated using a computer- 3.9 ng/mL (effect size index >0.4).
tate cancer (median age at most recent assisted image analysis system, as pre¬ Logistic regression analyses were BLSA visit, 69 years; range, 55-95 years).
viously described.9 used to model the relationship of age The clinical stage (TNM) among the can¬ and initial PSA level to the probability cer cases was Tl in 9 subjects, T2 in 7, of a PSA conversion and the relation¬ T3 in 3, Ml in 1, and unknown in 20 of Nonpalpable Tumors ship of age and PSA level at time of subjects. The Gleason score was 6 or prostatectomy to pathologic stage. Re¬ less for 17 of the cancer cases, 7 or more pathologically on a continuum (catego¬ stricted cubic splines12 were used to as¬ for 11 cases, and unknown for 12 cases.
ries A to D), from smaller tumors (cat¬ sess for nonlinear relationships with the Among the 272 men without prostate egories A and B) that may pose little predictor variables. Examination of the cancer, 36 had a simple prostatectomy threat to the life of some individuals residuals indicated an adequate fit for for benign prostatic hyperplasia and had (potentially unimportant) to larger and the final models. Power analysis showed no cancer in the final pathology speci¬ more extensive tumors (categories C and that the logistic regression analysis had men, 56 had 1 or more negative prostate D) that may be life threatening.6 Cat¬ a power greater than 0.80 with an biopsies, 156 had no suspicion of cancer egory A tumors were smaller than 0.2 level of .05 to detect a relative risk of 2.0 by digital rectal examination (DRE) per¬ cm3, organ-confined without seminal or greater for PSA conversion to 4.1 to formed by 1 examiner (H.B.C.), and none vesicle or lymph node involvement, and 5.0 ng/mL or a relative risk of 2.5 or had a PSA level above the age-specific without the presence of Gleason pattern greater for a PSA conversion to greater 4 or 5. Category tumors were be¬ than 5.0 ng/mL. Power analysis showed Nonpalpable Tumors (Surgical Da¬ tween 0.2 and 0.5 cm3, organ-confined that the logistic regression analysis had tabase).—The probability of detecting without seminal vesicle or lymph node a power greater than 0.95 with an a curable prostate cancer at a given pre¬ involvement, and without the presence level of .05 to detect a relative risk of 1.5 treatment serum PSA level was deter¬ of Gleason pattern 4 or 5. Category C or greater for curable tumors.
mined from 389 men with nonpalpable tumors were confined, 0.5 cm3 or larger, The logistic regression analyses were prostate cancer who underwent radical or with capsular penetration of low grade performed using S-Plus for Windows, prostatectomy at Johns Hopkins Hos¬ (Gleason score <7) with negative mar¬ and all other analyses were performed pital between 1989 and 1994. These 389 gins and no involvement of the seminal using SAS. Two-tailed tests with an cases were selected from 398 men with vesicles and lymph nodes. Category D level of .05 were used for all statistical nonpalpable cancer who had step-sec¬ tumors had capsular penetration ofhigh tion analysis of prostate specimens; 157 grade (Gleason score a 7) or established consecutive cases treated surgically be¬ (nonfocal) capsular penetration with tween 1989 and 1992 that were the sub¬ positive margins or involvement of semi¬ Nonpalpable Cancers a previous report,6 and 232 of 241 nal vesicles or lymph nodes.
(Surgical Database) cases treated surgically be¬ Tumors classified as categories A to tween 1994 and 1996. Nine of these men C were considered potentially curable Among 36 men with nonpalpable pros¬ were excluded because they did not have since more than 75% of men with tu¬ tate cancers who had pretreatment PSA a PSA measurement at least 4 weeks mors in these categories will be pro¬ levels of 4.0 ng/mL or less, 34 (94%) had after DRE and prostate biopsy and com¬ gression free at 10 years using a detect¬ curable cancers (organ-confined or cap¬ plete step-section analysis of prostate able PSA level as the indicator of sular penetration with Gleason score <7 specimens for determination of tumor progression.10 Category D tumors were and negative margins), and 25 (69%) considered potentially noncurable since were small cancers (confined tumor 0.5 The mean and median ages for the 389 less than 50% of men with tumors in this cm3 with no Gleason pattern 4 or 5) men with nonpalpable cancer treated category will be progression free at 10 (Table 1). When the pretreatment PSA surgically was 58.7 years and 60 years level was greater than 4.0 ng/mL but (range, 41-70 years), respectively. Se¬ less than 5.0 ng/mL (n=36), 32 (89%) rum PSA levels for the 389 men who had curable cancers and 12 (33%) were underwent surgical treatment were de¬ The unit of observation for the PSA small cancers. When the pretreatment termined from fresh serum samples conversion analysis was pairs of PSA PSA level was greater than 5.0 ng/mL, drawn at the time of initial consultation.
tests that were collected IV2 to 2V2 years 96 (30%) of 317 cancers were noncur- The mean and median pretreatment PSA apart or 3Vfe to 4Vz years apart. Pairs of able. Logistic regression analysis dem¬ levels were 9.85 ng/mL and 7.5 ng/mL PSA measurements were analyzed de¬ onstrated that the probability of non- (range, 0.5-114 ng/mL), respectively.
pending on the PSA level at the initial curable cancer increased with PSA level visit and on whether the subsequent and that the probability of noncurable Radical Prostatectomy Specimens PSA test had increased to 4.1 to 5.0 cancer was higher in men aged 65 years After resection, prostate specimens ng/mL or exceeded 5.0 ng/mL. The crude compared with men aged 45 to 55 years were coated with india ink, completely proportion of pairs of PSA measure¬ at PSA levels less than 8.0 ng/mL embedded, and processed as previously ments that increased to these levels was described.7 Tumors were graded using compared statistically by means of 2 system. Capsular penetra¬ tests. A power analysis11 indicated that tion was defined as tumor extending out the 2 analysis was capable of detecting The probability of a PSA conversion of the prostate into periprostatic soft small to moderate effect sizes (effect after 2 years (Table 2) to above 4.0 ng/ tissue, and subclassified as focal and es¬ size index >0.125) with a power of more mL was statistically significantly dif¬ tablished (nonfocal).7 Capsular margins than 0.80 at an level of .05 for the test ferent (Mantel-Haenszel and 2 tests) of resection were designated as nega¬ when the initial PSA level was less than between the cancer and noncancer tive or positive, as previously described.8 2.0 ng/mL and had a power of more than groups when the baseline PSA level was Seminal vesicle invasion was diagnosed 0.80 to detect moderate effect sizes (ef¬ 2.1 to 3.0 ng/mL (P=.01), but not when when a tumor penetrated the muscular fect size index 2:0.3) when the initial the baseline PSA level was 0.0 to 1.0 ng/mL (P=.71), 1.1 to 2.0 ng/mL (P=.09), tively, at baseline PSA levels of 3.1 to acceptable range for maintaining the or 3.1 to 4.0 ng/mL (P=.06). The prob¬ detection of potentially curable cancer.
ability of a PSA conversion to 4.1 to 5.0 The probability of a PSA conversion ng/mL was only 4% at 2 years for cancer after 4 years (Table 3) to above 4.0 cases when the baseline PSA level was ng/mL was statistically significantly testing increases the lead time to 2.0 ng/mL. The probability of a (Mantel-Haenszel and 2 PSA conversion to 4.1 to 5.0 ng/mL and tests) between the prostate cancer diagnosis13 and re¬ cancer and noncan- sults in increased detection of to greater than 5.0 ng/mL at 2 cer cases when the baseline PSA level confined cancers.14 However, it is cur¬ cancer cases) was 27% and 0%, respec¬ was 1.1 to 2.0 ng/mL (P=.001) or 2.1 to rently unknown whether PSA screening at baseline PSA levels of 2.1 to 3.0 ng/mL (P=.005); but there was no 3.0 ng/mL; and 36% and 32%, prostate cancer mortality.
statistically significant difference be¬ Regardless, PSA testing is being used tween the groups when the baseline Table 1 .—Pathologic Classification of 389 Men With routinely today for the early detection was 0.0 to 1.0 ng/mL (P=.74) Nonpalpable Prostate Cancer Treated Surgically of prostate cancer, and annual testing is or 3.1 to 4.0 ng/mL (P=.15). When the the common approach.2 We previously Pathologic Categories baseline PSA level was 1.1 to 2.0 ng/mL, demonstrated that _by PSA Level*_No. (%) the probability of changes in PSA, on a PSA conversion (for Pretreatment PSA <2.5 ng/mL average, occur slowly at lower PSA lev¬ cancer cases) to 4.1 to 5.0 ng/mL and to els even among men who are destined to greater than 5.0 ng/mL was 30% and develop prostate cancer later in life.4 0%, respectively. The probability of a Thus, annual testing to 4.1 to 5.0 ng/mL Pretreatment PSA a2.5 to £4.0 ng/mL sary for those men with lower baseline among cancer cases was 18% at a base¬ PSA values; and PSA testing intervals line PSA level of 2.1 to 3.0 ng/mL and year may be consistent with a baseline PSA level of 3.1 to early detection of curable cancers.
Pretreatment PSA >4.0 to £5.0 ng/mL We have demonstrated for the first time in a longitudinal study the prob¬ Testing Guidelines ability of PSA conversions among men Logistic regression analysis revealed who were destined for later develop¬ Pretreatment PSA >5.0 to £6.0 ng/mL that the probability of a PSA conversion ment of prostate cancer and men with¬ among cancer cases after 4 years in¬ out prostate cancer. Men who later de¬ higher initial PSA levels veloped prostate cancer were more likely Pretreatment PSA >6.0 to £10.0 ng/mL (P<.02). While age was a statistically to experience a PSA conversion than significant factor related to the prob¬ men without prostate cancer at a given ability of a PSA conversion to 4.1 to 5.0 level, and this probability ng/mL (P=.04), age was not a statisti¬ was directly related to the initial serum Pretreatment PSA >10.0 ng/mL cally significant factor for PSA conver¬ PSA value. We did not observe a rela¬ sion to 2.1 to 3.0 ng/mL (P=.08), to 3.1 tionship between PSA conversions and to 4.0 ng/mL (P=.17), or to greater than age in cancer cases.
5.0 ng/mL (P=.93). When the initial With respect to the use of PSA testing PSA level is less than indicates prostate-specific antigen. A indicates early detection of prostate cancer, an organ-confined tumor smaller than 0.2 cm3 without conversion to greater than 5.0 ng/mL the major investigative focus has been to seminal vesicle or lymph node involvement and without determine the PSA cutoff level that will presence of Gleason pattern 4 years is a rare event. How¬ or 5. indicates an organ-confined tumor between 0.2 and 0.5 cm3 without ever, the risk of a PSA conversion produce both high sensitivity (detection seminal vesicle or lymph node involvement and without increases as baseline PSA level rises of the largest number of prostate can¬ the presence of Gleason pattern 4 or 5. C indicates confined tumors 0.5 cm3 or larger or a tumor with cap¬ ng/mL. Guidelines for appro¬ cers) and high specificity (exclusion of sular penetration of low grade (Gleason score <7) with priate PSA testing intervals are shown men without prostate cancer). However, negative margins and no involvement of the seminal vesicles and lymph nodes. D indicates a tumor with Figure based on the probability of goal of PSA testing should be the sular penetration of high grade (Gleason score £7) or a PSA conversion by initial PSA level detection of curable tumors that pose a established (nonfocal) capsular penetration with positive and the surgical data suggesting that a expectancy or quality of life or involvement of seminal vesicles or lymph nodes. Category D is considered noncurable.
PSA level of 4.1 to 5.0 ng/mL is an and not the detection of all cancers. The Table 2.—Two-Year Prostate-Specific Antigen (PSA) Interval Analysis Initial PSA Level PSA pairs, No.
Median initial PSA level, ng/mL (range) Median age at initial PSA test, y (range) Median time before diagnosis* at initial PSA test, y (range) 4.0 (0.0-19.9) 13.6(1.3-24) 2.1(0.0-17.5) 7.9(0.1-19.6) 2.0(0.0-16.1) 3.7(0.0-14.9) 2.0(0.0-15.0) 3.9(0.0-28.7) Median interval between PSA tests, y (range) Second PSA levels £2.0 ng/mL, No. (%) Second PSA levels 2.1-3.0 ng/mL, No. (%) Second PSA levels 3.1-4.0 ng/mL, No. (%) Second PSA levels 4.1-5.0 ng/mL, No. (%) Second PSA levels >5.0 ng/mL, No. (%) Time of diagnosis for noncancers is time of last PSA measurement.
Table 3.—Four-Year Prostate-Specific Antigen (PSA) Interval Analysis Initial PSA Level PSA pairs, No.
Median initial PSA level, ng/mL (range) Median age at initial PSA test, y (range) Median time before diagnosis* at initial PSA test, y (range) 2.7 (0.0-22.0) 7.5(1.4-20.3) 2.0(0.0-19.2) 6.1(0.2-20.6) 2.0(0.0-15.0) 2.2(0.0-20.5) 2.0(0.0-13.5) 3.8(1.9-6.2) Median interval between PSA tests, y (range) Second PSA levels £2.0 ng/mL, No. (%) Second PSA levels 2.1-3.0 ng/mL, No. (%) Second PSA levels 3.1-4.0 ng/mL, No. (%) Second PSA levels 4.1-5.0 ng/mL, No. (%) Second PSA levels >5.0 ng/mL, No. (%) *Time of diagnosis for noncancers is time of last PSA measurement.
current lack of knowledge regarding pre¬ there is no suspicion of cancer on DRE.
dictors of tumor behavior prevents an In contrast, annual testing may be an accurate determination ofwhich early can¬ appropriate interval when the PSA level cers are likely to progress and what PSA is above 2.0 ng/mL because ofthe higher cutoff is most likely to detect the impor¬ risk of a PSA conversion to 4.1 to 5.0 tant cancer. Recognizing these limitations, ng/mL and to greater than 5.0 ng/mL.
there is some rationale for classifying tu¬ In attempting to identify those men mors that are smaller than 0.2 to 0.5 cm3 that may not need yearly PSA testing, as potentially unimportant,6'15 especially we have taken into consideration the among older men, and a strong rationale probability of a PSA conversion given for using tumor grade as a predictor of the initial PSA level and the PSA range aggressive behavior.10 at which the detection of curable cancer In this case series, potentially curable remains high. We have recommended, cancer was almost always found at PSA as a guideline, a 2-year sampling inter¬ levels less than 4.0 ng/mL (95% of cases) val for men with PSA levels below 2.0 in men without suspicion of cancer on ng/mL and an annual sampling interval DRE, but it was found at the expense of for men with PSA levels of 2.0 ng/mL detecting a high percentage (69%) of and above. This conservative approach smaller, potentially unimportant tumors takes into consideration the individual (Table 1). In comparison, when pretreat¬ physiologic variability between consecu¬ Probability of PSA conversion to 4.1 to 5.0 ng/mL ment PSA levels were between 4.0 and tive PSA measurements that may be as after 2 years (open circles), to higher than 5.0 ng/ 5.0 ng/mL in the setting of a nonsuspi¬ high as 20% to 30%.16·17 This substantial mL after 2 years (open triangles) and 4 years (solid cious DRE, the majority of men still had variability could lead to false low PSA squares). A 2-year testing interval appears to be potentially curable appropriate when the initial PSA level is less than cancers (89%) and readings that could delay repeat PSA 2.0 ng/mL because there is a negligible risk of con¬ the risk of detecting a smaller, poten¬ testing and delay prostate biopsy. Thus, version to 5.0 ng/mL even after 4 years. The test¬ tially unimportant cancer was reduced a more conservative approach would at¬ ing interval guidelines may be adjusted depending (33%). These data suggest that a PSA tempt to reduce the possibility of miss¬ on individual patient risk factors for development of level of 4.0 to 5.0 ng/mL represents prostate cancer or based on factors that affect the ing a PSA conversion to above 5.0 ng/ potential benefits and disadvantages of therapy.
range at which the detection of poten¬ mL—a level at which detection of curable tially curable cancer is maintained and cancer becomes less likely. Although the detection of potentially unimportant there is an extremely low probability of 2.0 ng/mL, may be appropriate for men cancers is reduced.
PSA conversion when the baseline PSA at higher risk of prostate cancer devel¬ The 4-year PSA interval analysis for is 0.0 to 1.0 ng/mL, we feel that further opment in whom PSA thresholds below cancer cases revealed no PSA conver¬ studies will be necessary before a rec¬ 4.0 ng/mL would raise suspicion of can¬ sions to greater than 5.0 ng/mL when ommendation can be made for testing cer. In contrast, a 4-year PSA testing the initial PSA level was less than or intervals that exceed 2 years among in¬ interval when the baseline PSA level is equal to 2.0 ng/mL. The 2-year interval dividuals with very low PSA levels.
less than 2.0 ng/mL may be adequate for analysis for cancer cases revealed a low Interpretation of the significance of a an older individual with comorbidities risk of a PSA conversion to 4.1 to 5.0 given serum PSA level requires consid¬ who has less to gain from a prostate can¬ ng/mL (4%) when the baseline PSA level eration of risk factors such as age, family cer diagnosis and in whom a PSA thresh¬ was less than or equal to 2.0 ng/mL; history, and race and also factors such as old above 4.0 ng/mL might be used to whereas PSA conversions to 4.1 to 5.0 patient health and level of risk aversion signal a higher risk of cancer. Tables 2 ng/mL occurred in 27% and 36% of can¬ which affect the potential downstream and 3 provide the clinician with informa¬ cer cases when baseline PSA levels were benefits and disadvantages of therapeu¬ tion regarding the chance of a PSA con¬ 2.1 to 3.0 ng/mL and 3.1 to 4.0 ng/mL, tic options. Thus, a decision regarding version to various PSA levels (depend¬ respectively. Based on these data it the timing of repeat PSA testing should ing on the initial PSA level) that can be would appear that annual PSA testing also take these factors into consideration.
used to make recommendations for ap¬ ns not beneficial for most men with ini¬ An annual PSA testing interval, even propriate PSA testing intervals for the tial PSA levels less than 2.0 ng/mL if when the baseline PSA level is less than individual patient.
In the current study, we were unable an age effect in cancer cases may reflect referral population to evaluate the patho¬ to evaluate PSA testing intervals of 1 the fact that PSA conversion is closely logic characteristics ofnonpalpable pros¬ year and 3 years, which could have af¬ related to cancer progression and that tate cancers, and a referral population fected the results. Given the individual tumors arise at different ages in differ¬ may not be comparable to screened popu¬ biologic variability between PSA mea¬ ent patients. Thus, the current study lations. However, comparisons of the surements, more frequent PSA testing represents a unique longitudinal evalu¬ pathologic features of nonpalpable intervals could have resulted in the de¬ ation of the natural history of PSA screened cancers20 and nonpalpable re¬ tection of PSA conversions that were changes among men with prostate can¬ ferral cancers6 demonstrate minimal dif¬ not detected in this study. Smith et al18 cers that were detected without PSA ferences. Therefore, we believe that found that after 4 years of PSA screen¬ screening later in the natural history of these data are representative ofthe find¬ ing at 6-month intervals, 4% of the men the disease.
ings in a screened population.
screened had PSA conversions to greater Several limitations of the current In summary, we have presented evi¬ than 4.0 ng/mL when the baseline PSA study deserve discussion. First, the use dence that men with baseline PSA lev¬ level was less than 2.5 ng/mL; whereas of frozen serum samples to determine els less than 2.0 ng/mL are unlikely af¬ 48% had PSA conversions to greater serum PSA levels could have affected ter even 4 years to reach a PSA level than 4.0 ng/mL when the baseline PSA the results. However, the stability of that would be inconsistent with cure of level was 2.6 to 4.0 ng/mL. These data total PSA in frozen serum samples at cancer. Furthermore, even if the initial include both men with and without pros¬ -70°C has been shown in the BLSA "true" PSA level was 2.0 to 3.0 ng/mL tate cancer and are similar to our 4-year population and in other studies.4,19 Sec¬ and the initial serum measurement was interval analysis for men with baseline ond, we evaluated PSA conversions less than 2.0 ng/mL, it is unlikely that PSA values less than and greater than among men who were diagnosed in the noncurable cancer (PSA conversion to 2.0 ng/mL, despite the differences in pre-PSA testing era (BLSA partici¬ >5.0 ng/mL) would be detected 2 years study design.
pants) and may have had tumors with later. Thus, it would appear safe to alter Smith et al18 also reported that pa¬ greater biologic potential compared to the current recommendations regard¬ tient age, in addition to baseline PSA, men diagnosed with use of PSA testing ing PSA testing. For men between the was a significant predictor of PSA con¬ (Tic surgical series). Thus, we may have ages of 50 years and 70 years who have versions to greater than 4.0 ng/mL in a overestimated the PSA conversion rates no suspicion of prostate cancer on DRE, proportional hazards model; whereas we by using the BLSA cancer cases since and in whom a PSA threshold of 4.0 found that age was a significant factor these were tumors that were destined ng/mL would be considered abnormal, a related only to PSA conversions to 4.1 to progress and were diagnosed without 2-year PSA testing interval is appro¬ to 5.0 ng/mL after 4 years in cancer the use of PSA testing. In comparison, priate when the baseline PSA level is cases and for PSA conversions among cancers detected using PSA testing could less than 2.0 ng/mL. When the baseline non-cancer cases. The age effect in non- have slower PSA conversion rates, which PSA level is 2.0 ng/mL and higher, an cancer cases is presumably due to the favors the safety of lowering the fre¬ annual PSA testing interval is appro¬ age-associated increase in the prevalence quency of PSA testing in men with no priate. This approach would result in ofbenign prostatic hyperplasia that leads suspicion of cancer on rectal examina¬ substantial cost savings for the health to elevations of PSA levels. The lack of tion. Third, the present study used a care system.
1. US Preventive Services Task Force. Screening following radical prostatectomy: a study of 196 cases Detection of organ-confined prostate cancer is in- for prostate cancer. In: Guide to Clinical Preven- with long-term follow-up. J Urol. 1993;150:135-141.
creased through prostate-specific antigen-based tive Services. 2nd ed. Baltimore, Md: Williams & 8. Epstein JI, Pizov G, Walsh PC. Correlation of screening. JAMA. 1993;270:948-954.
pathologic findings with progression following radi- 15. Stamey TA, Freiha FS, McNeal JE, Redwine 2. Mettlin C, Jones G, Averette H, Gusberg SB, cal retropubic prostatectomy. Cancer. 1993;71:3582 x=req- EA, Wittemore AS, Schmid H-P. Relationship of Murphy GP. Defining and updating the American- tumor volume to clinical significance for treatment Cancer Society guidelines for the cancer-related 9. Partin AW, Epstein JI, Cho KR, Gittelsohn MA, of prostate cancer. Cancer. 1993;71:933-938.
checkup: prostate and endometrial cancers. CA Can- Walsh PC. Morphometric measurement of tumor 16. Carter HB, Pearson JD, Waclawiw Z, et al.
cer J Clin 1993;43:42-46.
volume and percent of gland involvement as pre- Prostate-specific antigen variability in men with- 3. Shock NW, Greulich RC, Andres R, et al. Nor- dictors of pathological stage in clinical stage B pros- out prostate cancer: effect of sampling interval in mal Human Aging: The Baltimore Longitudinal tate cancer. J Urol. 1989;141:341-345.
prostate-specific antigen velocity. Urology. 1995; Study ofAging. Washington, DC: US Government 10. Epstein JI, Partin AW, Sauvageot J, Walsh PC.
Printing Office; November 1984. NIH publication Prediction of progression following radical prosta- 17. Prestigiacomo AF, Stamey TA. Physiological tectomy: a multivariate analysis of variation of serum prostate specific antigen in the 4. Carter HB, Pearson JD, Metter JE, et al. Lon- long-term follow-up. Am J Surg Pathol. 1996;20: 4.0 to 10.0 ng/mL range in male volunteers. J Urol.
gitudinal evaluation ofprostate-specific antigen lev- men with and without prostate disease. JAMA.
J. Statistical Power Analysis for the 18. Smith DS, Catalona WJ, Herschman JD. Lon- Behavioral Sciences. Hillsdale, NJ: Lawrence gitudinal screening for prostate cancer with pros- 5. Oesterling JE, Jacobsen SJ, Chute CG, et al. Se- Erlbaum Associates; 1988.
tate-specific antigen. JAMA. 1996;276:1309-1315.
rum prostate-specific antigen in a community-based CY. Additive splines in statis- 19. Jacobsen SJ, Klee GG, Lilja H, Wright GL, population of healthy men: establishment of age-spe- tics. In: Proceedings of the Statistical Computing Oesterling JE. Stability of serum prostate-specific cific reference ranges. JAMA. 1993;270:860-864.
Section of the American Statistical Association.
antigen determination across laboratory, assay, and 6. Epstein JI, Walsh PC, Carmichael M, Brendler Washington, DC: 1985:45-48.
storage time. Urology. 1995;45:447-453.
CB. Pathologic and clinical findings to predict tu- 13. Gann PH, Hennekens CH, Stampter MJ. A pro- 20. Humphrey PA, Keetch DW, Smith DS, Shep- nonpalpable (stage T1c) prostate can- spective evaluation of plasma prostate-specific an- herd DL, Catalona WJ. Prospective characteriza- cer. JAMA. 1994;271:368-374.
tigen for detection ofprostatic cancer. JAMA. 1995; tion of pathological features of prostatic carcinomas 7. Epstein JI, Carmichael M, Pizov G, Walsh PC.
detected via serum prostate specific antigen based Influence of capsular penetration on progression 14. Catalona WJ, Smith DS, Ratliff TL, Basler JW.
screening. J Urol. 1996;155:816-820.

Source: http://www.prostata-shg.de/Literaturrecherche/Carter1997.pdf