The widespread use of serum prostate-specific antigen (PSA) testing for early prostate cancer detection has increased the proportion of early-stage cancers detected and is at least partially responsible for the recent decrease in prostate cancer mortality rates in the United States. 1
- Conclusions
- Acknowledgements
- References and Notes
A PSA cutoff of 4 ng/mL is generally used for recommending a biopsy, 2 and cancer is detected in initial or early follow-up biopsies in more than one third of men with PSA levels greater than 4 ng/mL. 2,3 2,3 However, about 20% of detectable cancers occur in men with PSA values below this cutoff, and the cancer has extended beyond the prostate in one third of men with PSA levels between 4 and 10 ng/mL and in more than half with PSA levels greater than 10 ng/mL. [2] [3] [4] [5] Therefore, the use of lower PSA cutoffs may detect prostate cancer more frequently in its curable stages. However, lower cutoffs require more biopsies, and there are concerns that they may detect more harmless cancers. 6,7 6,7 Thus, lower PSA cutoffs should be used in conjunction with other risk-assessment markers, such as free PSA.
In 1993, we recommended against using lower PSA cutoffs because we found that only 7% of men with PSA levels between 2.9 and 4 ng/mL had cancer detected in their initial biopsy. However, at that time, our biopsy protocol sampled only palpable or sonographic abnormalities. 8 During 4 years of serial screening, we found that 20% of this cohort was diagnosed with prostate cancer. 9 Also, our longitudinal PSA screening studies revealed that 48% of men whose initial PSA levels were between 2.6 and 4 ng/mL had an increase in their PSA level to greater than 4 ng/mL within 4 years, and 13% had cancer detected during this interval. 9 Similar PSA conversion rates were reported by Carter et al. 10
Several studies have suggested potential advantages of lower PSA cutoffs. [11] [12] [13] [14] Gann et al. 11 reported that, compared with men whose PSA levels were less than 1 ng/mL, those with PSA levels of 2.01 to 4.0 ng/mL were 5.5 to 8.6 times more likely to develop aggressive prostate cancer within 10 years.
In 1995, we lowered our PSA cutoff for biopsy recommendation to 2.6 ng/mL in our ongoing screening studies. 14 Because the proportion of our subjects with a serum PSA level of 2.6 to 4 ng/mL and benign digital rectal examination (DRE) findings was similar to that of men with PSA levels greater than 4 ng/mL and benign DRE findings (9% and 8%, respectively) (unpublished data), lowering the cutoff to 2.6 ng/mL increased the percentage of recommended biopsies from 8% to 17% (36% of men complied with the recommendation for biopsy using the 2.6 ng/mL cutoff). Similar findings were reported in the European Randomized Study of Screening for Prostate Cancer. 15
Since we lowered our PSA cutoff, men with a PSA of 2.6 to 4 ng/mL and benign DRE findings have accounted for nearly 20% of our screen-detected cancers. We reported a cancer detection rate of 22% in men with these findings using sextant biopsies in our screening population, with the cancer detection rate being higher in older men and black men. Of the cancers detected, 81% were organ confined and 17% were low-volume and low-grade or moderately low-grade tumors that some might consider potentially unimportant. 14 The positive predictive value of approximately 20% for a PSA of 2.6 to 4 ng/mL is lower than but still substantial compared with that of approximately 30% for a PSA greater than 4 ng/mL.
Our screening studies have suggested that a 2.6 ng/mL PSA cutoff detects more potentially curable cancers without overdetecting harmless ones. Few published reports have evaluated lower PSA cutoffs using routine biopsy protocols. In our study, including 676 consecutive patients treated with radical prostatectomy, the rates of organ-confined cancer were 81% for the PSA range of 2.6 to 4 ng/mL, 70% for the range of 4.1 to 5 ng/mL, 71% for the range of 5.1 to 10 ng/mL, and 53% for the range of 10 ng/mL or higher (P = 0.001) (unpublished data). In a subset of these patients, including 94 whose surgical specimens were totally embedded and serially sectioned using a research protocol to obtain more accurate data (the others were processed using our institution's standard clinical protocol), the organ-confined rate was 88% for the 2.6 to 4 ng/mL PSA range versus 63% for the 4 to 10 ng/mL range (P = 0.01). 16 Similar results were obtained in the European screening trials (ie, organ-confined disease in 84% with PSA 2 to 3.9 ng/mL, 62% with PSA 4 to 9.9 ng/mL, and 39% with PSA greater than 10 ng/mL). 15
The great majority of cancers detected by PSA screening with levels greater than 4 ng/mL have the characteristics of clinically significant cancers [1] [2] [3] [4] [5] [6] [7] [8] that are similar to pathologic Stage T2a tumors but less advanced than Stage T2b tumors. 18 In our cohort with step-sectioned prostatectomy specimens, a substantial overlap was found between the tumor characteristics of cancers detected with a PSA level of 2.6 to 4 ng/mL and those detected with a level between 4 and 10 ng/mL. 16 Approximately 14% of cancers in the 2.6 to 4 ng/mL range and 12% in the 4 to 10 ng/mL range had a Gleason grade of 7 or higher, with both groups having a similar percentage (11.9% versus 11.5%) of possibly harmless cancers, using the criteria of Epstein et al. 19 The European screening trials also found no significant difference in the percentage of Gleason grade 7 or higher tumors between cancers detected at a PSA range of 2 to 3.9 ng/mL (58%) versus those detected at a range of 4 to 9.9 ng/mL (54%), although the percentage of tumors with Gleason grade 7 or higher was unusually high in these trials. 15
In our series, 16 the mean tumor volume in the step-sectioned specimens was 1.1 mL for cancers detected with a PSA of 2.6 to 4 ng/mL and 1.8 mL for those with a PSA of 4 to 10 ng/mL (P = 0.02), both considerably larger than those reported in most autopsy series in which most tumors are smaller than 0.5 mL. In the European screening trials, a similar trend was observed, although the mean tumor volume was generally smaller: 0.48 mL (range 0.01 to 1.8) for the 2 to 2.9 ng/mL PSA range, 0.70 mL (range 0.02 to 3.1) for the 3 to 3.9 ng/mL PSA range, and 1.27 mL (range 0.01 to 13.5) for the 4 to 9.9 ng/mL PSA range. 15 Babaian et al. 20 reported that 15 (88%) of 17 prostate cancers detected in the 2.6 to 4 ng/mL range were larger than 0.5 mL and 10 (59%) were larger than 1.0 mL.
Our results differ from those reported from Johns Hopkins, 6,10 6,10 which suggested that lower cutoffs detect few additional curable cancers and more potentially unimportant ones. Several reasons are possible to explain these differences. Their study cohort consisted of two data bases: one consisting of patients in a study on aging for estimating PSA progression rates and another consisting of patients with nonpalpable prostate cancer referred for radical prostatectomy for assessing pathologic findings 10; our results are from a screening study. They used logistic regression analysis to predict the probability of curable cancer and estimated a 2% to 4% increase within the range of PSA values from 2.5 to 6 ng/mL. 6 We found an 11% (95% confidence interval 2% to 20%) difference in organ-confined cancer between the 2.6 to 4 ng/mL range and the 4 to 5 ng/mL range. They used a broader definition of curable cancer, including patients with low-grade cancer that had extended beyond the prostate. 6 Our definition included only men with organ-confined disease with clear surgical margins. Our use of a screening population (earlier cancer detection) and a more rigorous definition of curable cancer (better discrimination of early lesions) may explain why we found a larger increment of curable cancers with lower PSA cutoffs.
The Johns Hopkins group considered all organ-confined tumors smaller than 0.5 mL without a Gleason pattern of 4 or 5 to be potentially unimportant 10; we believe some may be dangerous. They reported that the percentage of potentially unimportant cancers was 69% in patients with a PSA of 2.5 to 4 ng/mL versus 33% in those with PSA levels between 4 and 5 ng/mL. 10 They recommended using a higher PSA cutoff that would reduce the detection of these cancers. Their more liberal definition of unimportant cancer may explain their higher estimate of unimportant cancer detected at the lower PSA cutoff.
The Johns Hopkins referral cohort was generally younger (mean age 58 years versus 63 years in our subjects), and their group with low PSA levels was further enriched with young men, because their biopsy criteria for men with PSA values lower than 4 ng/mL included age-specific cutoffs (men with PSA values less than 4.0 ng/mL underwent biopsies only if they were younger than 60 years old). In our screening population, the criterion for recommending biopsy was a PSA of 2.6 ng/mL or higher in all subjects. As a result, only 8% of their referred patients had PSA levels of 2.6 to 4 ng/mL 6 compared with 20% in our screening population. Thus, their study contained few data of men older than 60 years of age with PSA values between 2.6 and 4 ng/mL; in the general population, many men 60 years old and older have PSA levels within this range. Because older men have a higher prevalence of cancer, the underrepresentation of older men with a PSA level of 2.6 to 4 ng/mL may have lowered their estimate of additional cancers detected in the lower PSA range.
Undoubtedly, many cancers were present in the larger population from which the Johns Hopkins referral cohort was derived that were not included in their study because most men with PSA values of 2.6 to 4 ng/mL did not undergo biopsy. The absolute number whose cancer may have progressed during the interval required for the PSA to rise to greater than 4 ng/mL may be substantial. Accordingly, results from a referral population with a limited representation of men with lower PSA values must be interpreted cautiously and may not accurately reflect the performance of the 2.6 ng/mL PSA cutoff in a screening setting.
If PSA testing could detect all ''harmless autopsy cancers,'' the overall cancer detection rates in screening programs would approximate those of the autopsy series (22% to 53%). 7 However, in our screening program, using a PSA cutoff of 2.6ng/mL, the overall cancer detection rate for the initial screening was only 7%; in the European trial, using a cutoff of 2 ng/mL, it was 4%. 15
On the basis of the PSA conversion rates reported by us 9 and by Carter et al., 10 an average of 5 to 7 years would be required for the PSA level to rise to greater than 4 ng/mL if the baseline level was 2 to 4 ng/mL. Thus, even considering the slowest published prostate cancer growth rates, 17 the tumor volumes would increase substantially for most men during the time required for the PSA to rise to greater than 4 ng/mL.
Early progression-free survival benefits of lower PSA cutoffs were also demonstrated in men eligible for 4 years of follow-up in our screening study, with a higher 4-year probability of nonprogression of 94% (95% confidence interval 82% to 100%) (2% progressed) in 48 patients with preoperative PSA levels of 2.5 to 4 ng/mL and nonsuspicious DRE findings compared with 76% (95% confidence interval 52% to 100%) (7% progressed) in 151 men with levels of 4.1 to 6 ng/mL and nonsuspicious DRE findings (unpublished data).
If one accepts that detection of clinically important cancer can be enhanced by lowering the PSA cutoffs, the challenge becomes to reduce the false-positive results in the lower PSA ranges. Unnecessary biopsies can be reduced by restricting the biopsy recommendation to men at relatively higher risk in this generally low-risk population. Risk stratification can be achieved by combining risk factors (such as age, race, family history, biopsy history, and prostate volume), biochemical markers (such as free PSA and human kallikrein 2 [hK2]), and statistical methods (such as artificial neural networks and regression analysis) to provide better discrimination of men more likely to have prostate cancer.
The use of measurements of the percentage of free PSA improves the accuracy of cancer detection when PSA levels are mildly elevated and also helps identify more aggressive cancers. 14,21 14,21 Gann et al. 11 demonstrated in samples from the Physicians Health Study that the percentage of free PSA was superior to total PSA for discriminating between cancer and benign disease within the total PSA range of 3 to 10 ng/mL. In Gann's study, a percentage of free PSA cutoff of 20% for men with a total PSA of 3 to 10 ng/mL actually detected 10% more cancers with 12.5% fewer false-positive results than a simple total PSA cutoff of 4 ng/mL strategy (unpublished data).
Prostate cancer cells also exhibit increased expression of hK2, and new assays for hK2 show promise for increasing the specificity of cancer detection. [22] [23] [24] In a pilot study, we evaluated the hK2/free PSA ratio to reduce the number of prostate biopsies in men with PSA values of 2.6 to 4 ng/mL and benign DRE findings. 24 Using the hK2/free PSA ratio of greater than 0.25 as the cutoff for recommending biopsies, we could have identified 41% of the cancers, performing biopsies in only 14% of the subjects with a PSA level of 2.6 to 4 ng/mL. Unnecessary biopsies could have been avoided in 91% of the men who did not have detectable cancer. Of the cancers detected, 91% were organ confined, but only 7% fulfilled the criteria of Epstein et al. 19 for possibly harmless cancers. 24 Using this model, which favors specificity over sensitivity, the patients who did not undergo biopsy would need to be followed up carefully because some of the high Gleason grade and locally advanced cancers would be missed; however, none of these cancers would be detected immediately with the traditional 4 ng/mL cutoff (sensitivity 0%).
Lower PSA cutoffs may be especially beneficial for black men, in whom the incidence of prostate cancer is higher than in white men. 25 Black men are more likely to have advanced disease at the time of detection and are more likely to die of prostate cancer. 1 Early-stage prostate cancer in black men may be missed by the standard cutoffs determined from white populations. 25
Approximately 20% of potentially important prostate cancers, including those with the greatest prospect for a cure, can be detected in men with PSA levels of 2.6 to 4 ng/mL and benign DRE findings. The use of a 2.6 ng/mL cutoff increases the proportion of patients detected with early disease and therefore increases the chances for a cure. Our results suggest that the proportion of patients with possibly harmless cancers is similar in men with PSA levels between 2.6 and 4 ng/mL and those with levels between 4 and 10 ng/mL.
Lower PSA cutoffs require more biopsies; however, stratification of the population with PSA levels of 2.6 to 4 ng/mL into different risk groups with the use of additional markers allows better identification of men at higher risk. The final decision about a biopsy must rest with the patient who will actually experience the risks and benefits of early prostate cancer detection. For some patients, a 15% to 20% risk may be deemed sufficient to justify a biopsy. In this regard, it is noteworthy that the positive predictive value of mammography for detecting breast cancer is only 3% to 17%, depending on the patient's age. 26 Moreover, the positive predictive value for a suspicious DRE is 20% overall, and only 5% to 15% when the PSA is less than 4 ng/mL. 5 Yet, biopsies are routinely recommended in men with suspicious DRE findings.
In the future, statistical methods may be used to more accurately estimate a man's risk of prostate cancer. One may enter data concerning age, race, family history, history of prostatitis, previous biopsy history, total PSA, free PSA, hK2, and any other markers and risk factors identified, and the risks of having a positive biopsy will be computed. Then, the patient and doctor can decide whether the risk justifies a biopsy.
For the present, physicians and patients should be aware of the 15% to 20% risk of prostate cancer associated with a PSA level of 2.6 to 4 ng/mL. New screening models using lower PSA cutoffs have the potential to detect clinically important and potentially curable cancers that would not be detected with the current practice. To the extent that prostate cancer mortality and morbidity rates are due to late detection, diagnosing cancers in men with lower PSA levels may help reduce these rates.
To Kimberly Roehl and Joshua Hileman for expert statistical assistance and Drs. Paula C. Southwick, Peter H. Gann, and Deborah S. Smith for excellent critical review of the manuscript.
[1] Hankey B.F., Feuer E.J. and Clegg L.X. et al. Cancer surveillance series: interpreting trends in prostate cancer. Part I: Evidence of the effects of screening in recent prostate cancer incidence, mortality, and survival rates. J Natl Cancer Inst 1999, 91:1017-1024.[Medline]
[2] Catalona W.J., Smith D.S. and Ratliff T.L. et al. Measurement of prostate-specific antigen in serum as a screening test for prostate cancer. N Engl J Med 1991, 324:1156-1161.[Medline]
[3] Smith D.S., Humphrey P.A. and Catalona W.J. The early detection of prostate carcinoma with prostate specific antigen: the Washington University experience. Cancer 1997, 80:1852-1856.[Medline]
[4] Catalona W.J., Smith D.S. and Ratliff T.L. et al. Detection of organ-confined prostate cancer is increased through prostate-specific antigen-based screening. JAMA 1993, 270:948-954.[Medline]
[5] Catalona W.J., Richie J.P. and Ahmann F.R. et al. Comparison of digital rectal examination and serum prostate specific antigen in the early detection of prostate cancer: results of a multicenter clinical trial of 6,630 men. J Urol 1994, 151:1283-1290.[Medline]
[6] Carter H.B., Epstein J.I. and Partin A.W. Influence of age and prostate-specific antigen on the chance of curable prostate cancer among men with nonpalpable disease. Urology 1999, 53:126-130.[Medline]
[7] Woolf S.H. Screening for prostate cancer with prostate-specific antigen. N Engl J Med 1995, 333:1401-1405.[Medline]
[8] Colberg J.W., Smith D.S. and Catalona W.J. Prevalence and pathological extent of prostate cancer in men with prostate specific antigen levels of 2.9 to 4 ng/mL. J Urol 1993, 149:507-509.[Medline]
[9] Smith D.S., Catalona W.J. and Herschman J.C. Longitudinal screening for prostate cancer with prostate-specific antigen. JAMA 1996, 276:1309-1315.[Medline]
[10] Carter H.B., Epstein J.I. and Chan D.W. et al. Recommended prostate-specific antigen testing intervals for the detection of curable prostate cancer. JAMA 1997, 277:1456-1460.[Medline]
[11] Gann P.H., Hennekens C.H. and Stampfer M.J. A prospective evaluation of plasma prostate-specific antigen for detection of prostate cancer. JAMA 1995, 273:289-294.[Medline]
[12] Labrie F., Dupont A. and Suburu R. et al. Serum prostate specific antigen as pre-screening test for prostate cancer. J Urol 1992, 147:846-852.[Medline]
[13] Littrup P.J. Prostate cancer screening—appropriate choices? Cancer Suppl 1994, 74:2016-2022.
[14] Catalona W.J., Smith D.S. and Ornstein D.K. Prostate cancer detection in men with serum PSA concentrations of 2.6 to 4 ng/mL and benign prostate examination: enhancement of specificity with free PSA measurements. JAMA 1997, 277:1452-1455.[Medline]
[15] Schroder F.H., van der Cruijsen-Koeter I. and de Koning H.J. et al. Prostate cancer detection at low values of prostate specific antigen (PSA) J Urol 2000, 163:806-811.[Medline]
[16] Krumholtz J., Carvalhal G.F. and Ramos C.G. et al. Total PSA cutoff of 2.5 ng/mL correlates with favorable pathologic features of prostate cancer (abstract) J Urol 1999, 161:99.[Medline]
[17] Dugan J.A., Bostwick D.G. and Myers R.P. The definition and preoperative prediction of clinically insignificant prostate cancer. JAMA 1996, 275:288-294.[Medline]
[18] Ramos C.G., Carvalhal G.F. and Smith D.S. et al. Clinical and pathological characteristics and recurrence rates of stage T1c versus T2a or T2b prostate cancer. J Urol 1999, 161:1525-1529.[Medline]
[19] Epstein J.I., Walsh P.C. and Carmichael M. et al. Pathologic and clinical findings to predict tumor extent of nonpalpable (stage T1c) prostate cancer. JAMA 1994, 271:368-374.[Medline]
[20] Babaian R.J., Troncoso P. and Steelhammer L.C. et al. Tumor volume and prostate specific antigen: implications for early detection and defining a window of curability. J Urol 1995, 154:1808-1812.[Medline]
[21] Catalona W.J., Partin A.W. and Finlay J.A. et al. Use of percentage of free prostate- specific antigen to identify men at high risk of prostate cancer when PSA levels are 2: 51 to 4 ng/mL and digital rectal examination is not suspicious for prostate cancer: an alternative model. Urology 1999, 54:220-224.[Medline]
[22] Darson M.F., Pacelli A. and Roche P. et al. Human glandular kallikrein 2 expression in prostate adenocarcinoma and lymph node metastases. Urology 1999, 53:939-944.[Medline]
[23] Kwiatkowski M., Recker F. and Piironen T. et al. In prostatism patients the ratio of human glandular kallikrein to free PSA improves the discrimination between prostate cancer and benign hyperplasia within the diagnostic ''gray zone'' of total PSA 4 to 10 ng/mL. Urology 1998, 52:360-365.[Medline]
[24] Partin A.W., Catalona W.J. and Finlay J.A. et al. The use of human glandular kallikrein 2 (hK2) for the detection of prostate cancer: preliminary analysis. Urology 1999, 54:839-845.[Medline]
[25] Smith D.S., Carvalhal G.F. and Mager D.E. et al. Use of lower prostate specific antigen cutoffs for prostate cancer screening in black and white men. J Urol 1998, 160:1734-1738.[Medline]
[26] Kerlikowske K., Grady D. and Barclay J. et al. Positive predictive value of screening mammography by age and family history of breast cancer. JAMA 1993, 270:2444-2450.[Medline]
Article Footnote.Dr. Catalona's research is supported in part by a grant from Hybritech Incorporated, a subsidiary of Beckman Coulter, Inc., San Diego, California.