Correspondence
Prostate-Specific Antigen in Black Men
N Engl J Med 1997; 336:133-136January 9, 1997
- Article
To the Editor:
Morgan et al. (Aug. 1 issue)1 report on age-specific reference ranges for serum prostate-specific antigen (PSA) in black men. Their analysis is based on the distribution of serum PSA levels in a large population of blacks and whites both with and without prostate cancer. I believe there is a flaw in their data for men without prostate cancer (the controls).
Their control PSA data are derived from a group of 3764 men without known prostate cancer, 576 of whom had abnormal digital rectal examinations, serum PSA values higher than 4.0 ng per milliliter (most had elevated PSA values), or both. This group was then analyzed further: 127 men with biopsy-confirmed prostate cancer were appropriately excluded, and 287 men with negative biopsies, negative ultrasonographic studies, or both were appropriately included. Unfortunately, 162 men with abnormal digital rectal examinations, abnormal serum PSA levels, or both did not undergo further clinical evaluation and were excluded from the analysis. Thus, a substantial number of patients, many of whom had elevated PSA levels presumably without cancer (indicating false positive results), were not included in the analyses.
Although we cannot determine how many of these excluded patients had elevated PSA levels in the absence of cancer, if we assume that the rates of elevated PSA levels and prostate cancer in this group were the same as the rates in the group that could be evaluated, 105 additional men had PSA values higher than 4.0 ng per milliliter without prostate cancer.
In a study in which 287 of 3475 controls had elevated PSA levels, abnormal digital rectal examinations, or both, adding an estimated 105 patients with elevated PSA levels significantly reduces the specificity and alters the results and conclusions based on these numbers.
Because so many controls with elevated PSA levels were not evaluated with ultrasonography or biopsy, I believe that the data required to assess specificity accurately are not available from this study.
1 ReferencesMichael W. Jacobson, M.D., M.P.H.
Lenox Hill Hospital, New York, NY 100211
Morgan TO ,Jacobsen SJ ,McCarthy WF ,Jacobson DJ ,McLeod DG ,Moul JW . Age-specific reference ranges for serum prostate-specific antigen in black men. N Engl J Med 1996;335:304-310
Full Text | Web of Science | Medline
To the Editor:
Morgan et al. report high ranges of serum PSA levels in a black population, supporting our findings.1,2 In our study of approximately 3000 black men and 3000 white men with no detectable prostate cancer who were screened in the Maryland Veterans Affairs health care system, the 95th percentile of PSA values was higher in black men than in white men in every age group examined. The largest difference was in the group of men who were at least 70 years old; in this group, the ratio of the 95th percentile of PSA values in black men to the 95th percentile of values in white men was 1.6.
Black men may have higher PSA levels because of non-neoplastic conditions that increase the PSA level, or the presumably disease-free population of black men may include an undetermined number of men with undiagnosed cancers. Calculations of the sensitivity, specificity, and receiver-operating-characteristic curves require a determination of the total number of subjects with prostate cancer. Since we do not know the number of undiagnosed cancers in the population, even when the most thorough diagnostic procedures are used, the conclusions Morgan et al. draw from their data are questionable.
As for how to count multiple PSA assays performed over time in a single subject, Morgan et al. addressed the problem by ignoring all PSA tests but the first, thus excluding about 30,000 PSA assays from their data. This exclusion seems arbitrary and a potential source of bias. Why is the first PSA test more relevant than a later test? Our study included every PSA assay in every patient, and we assigned a weighted score for each assay. If a patient with 10 PSA tests was less than 50 years old when 4 of the 10 tests were performed, we counted 4 test scores, each with a weight of 0.1, for a total weight of 0.4 in the 40-to-49-year-old age group, and 6 test scores, each with a weight of 0.1, in the 50-to-59-year-old age group. In this manner, all tests were counted, and each subject had one composite PSA test score.
2 ReferencesRobert Sawyer, M.D.
Jules J. Berman, Ph.D., M.D.
G. William Moore, M.D., Ph.D.
Veterans Affairs Medical Center, Baltimore, MD 212011
Sawyer R ,Berman JJ ,Borkowski A ,Moore GW . Prostate-specific antigen in black men. Lancet 1996;347:1329-1329
CrossRef | Web of Science | Medline2
Sawyer R ,Berman JJ ,Borkowski A ,Moore GW . Elevated prostate-specific antigen levels in black men and white men. Mod Pathol 1996;9:1029-1032
Web of Science | Medline
To the Editor:
I was dismayed to see that subspecialists are still reporting recommendations for screening that are based solely on the accuracy of a test. Morgan et al. state that because the receiver-operating-characteristic curves for the PSA assay are better than those for Papanicolaou smears, PSA is “clearly valuable as a screening test.” There is no discussion of other criteria for selecting a screening test (e.g., lead-time bias, treatment effectiveness, or potential adverse effects). Consideration of such additional criteria has led the U.S. Preventive Services Task Force to recommend against routine PSA screening.1 Similarly, in his editorial, Oesterling states that “every man 50 years of age or older . . . should receive an annual prostatic evaluation,” without discussing effectiveness or other important criteria.2 Unfortunately, the question is not only whether PSA screening can detect prostatic cancer more accurately or earlier but whether the use of PSA screening can make a difference in a patient's chances of having a better quality of life, a longer life, and so forth. (And simply detecting the causes earlier does not necessarily improve the outlook.)
I was also disappointed in the apparent lack of appreciation of the true effect of a reduction in specificity as proposed by Morgan and colleagues. Those of us in family medicine are reluctant to use assays of lower specificity, even in younger patients, given the potentially disastrous effects on the positive predictive value of such assays in a primary care setting.
Although PSA is thought to be more sensitive in detecting clinically important disease (i.e., aggressive prostatic cancer1), one study found that the incidence of aggressive prostatic cancer was only 1 percent over a period of 10 years. (The incidence of curable cancer was not even reported.1) For family physicians treating a patient population with a 1 percent prevalence of prostate cancer (with the incidence being only a fair marker of prevalence) and using the 95 percent sensitivity recommended by Morgan and colleagues, the drop in specificity from 95 percent to 93 percent (which Morgan et al. label “acceptable”) would result in a 25 percent reduction in the positive predictive value (from 16 percent to 12 percent). If the specificity were dropped to the recommended level of 78 percent, the positive predictive value would be reduced by 75 percent (from 16 percent to 4 percent). None of these are pleasant choices for a primary care physician. In fact, given the low positive predictive value and, even more important, the lack of demonstrated effectiveness of PSA testing, most family physicians do not recommend routine PSA screening to their patients. Neither the report by Morgan et al. nor the editorial by Oesterling should change their minds.
2 ReferencesDavid E. Swee, M.D.
University of Medicine and Dentistry of New Jersey–Robert Wood Johnson Medical School, New Brunswick, NJ 08903-00191
Preventive Services Task Force. Guide to clinical preventive services. 2nd ed. Baltimore: Williams & Wilkins, 1996:118-34.
2
Oesterling JE . Age-specific reference ranges for serum PSA. N Engl J Med 1996;335:345-346
Full Text | Web of Science | Medline
To the Editor:
In his editorial on the report by Morgan et al., Dr. Oesterling's statement that “the prostates of older men are more `leaky' with regard to PSA” is misleading. Mean PSA levels in groups of older men are higher because older men have higher prevalences of both benign and malignant disorders of the prostate gland. However, I am unaware of direct evidence that with aging there is a general breakdown of the physiologic barriers to the diffusion of PSA from the prostatic acinar and ductal system into the bloodstream. Moreover, the leaky-prostate hypothesis is not supported by observations from our PSA screening studies (including 30,000 men), in which the majority of older men without prostatic disease had low PSA levels: 75 percent of men 60 to 69 years old and 63 percent of those 70 to 79 years old had serum PSA values of 2.5 ng per milliliter or less. This point is important, because a 70-year-old man with a serum PSA level of 5.0 ng per milliliter and a normal-sized, benign-feeling prostate gland has a substantial chance of having a medically important prostate cancer. However, with the use of the age-specific reference ranges recommended by Dr. Oesterling (a PSA level higher than 6.5 ng per milliliter in a 70-year-old man), biopsy would not be recommended for this man.
It is true that age-specific reference ranges make the PSA test more sensitive in younger men and more specific in older men, but the trade-off is that age-specific ranges lead to more unnecessary biopsies in younger men and miss a substantial proportion of cancers in older men for whom curative treatment may be an option.1 Furthermore, the cancers missed are those most likely to be curable (because of the lower PSA level).
Studies by my colleagues and me suggest that the calculation of the percentage of free PSA in serum is a promising alternative method of improving both the sensitivity and the specificity of PSA testing in men whose total PSA levels are 2.6 to 10.0 ng per milliliter.2,3
3 ReferencesWilliam J. Catalona, M.D.
Washington University, St. Louis, MO 631101
Catalona WJ ,Hudson MA ,Scardino PT , et al. Selection of optimal prostate specific antigen cutoffs for early detection of prostate cancer: receiver operating characteristic curves. J Urol 1994;152:2037-2042
Web of Science | Medline2
Catalona WJ ,Colberg JW ,Smith DS ,Ornstein DK ,Shayka JJ . Measurement of percent-free PSA improves specificity for lower PSA cutoffs in prostate cancer screening. J Urol 1996;155:Suppl:422A-422A abstract.
CrossRef | Web of Science3
Catalona WJ ,Smith DS ,Wolfert RL , et al. Evaluation of percentage of free serum prostate-specific antigen to improve specificity of prostate cancer screening. JAMA 1995;274:1214-1220
CrossRef | Web of Science | Medline
Author/Editor Response
The authors reply:
To the Editor: Dr. Jacobson believes that the controls excluded from our study may have had higher PSA values and that their exclusion may have shifted the distribution of PSA levels for the control group to the right, thereby increasing our 95th percentiles for specificity and decreasing sensitivity. We thought that we could not correctly assign these men to the case or control group because of their incomplete evaluation. Furthermore, our study presents PSA reference ranges based on sensitivity, or maximal cancer detection in this high-risk group. These values are derived from the case group and are therefore not affected by the omission of controls. The effect on the estimated receiver-operating-characteristic curve would have been only slight.
Sawyer et al. ask about the chance of having missed cancers in some of the controls. In our protocol, the controls were subjected to a rigorous clinical examination for prostate cancer. We may have missed some cancers, but even among men with sextant biopsies (those in which six specimens are obtained), it is possible to miss a tumor. The only way to determine the presence or absence of prostate cancer with complete accuracy is to perform a prostatectomy and submit the specimen for a thorough pathological examination. The ethical problems with this approach are obvious. Given the discrepancy between the prevalence of prostate cancer at autopsy and the prevalence of clinically evident prostate cancer, it is not certain that it would be desirable to detect the tumors that may have been missed.
Sawyer et al. discuss their method of pooling measurements to estimate a man's “true” PSA level, which controls for within-person and laboratory variability. Our approach of considering only the first measurement does not do this but addresses the distribution on the basis of a single PSA value, allowing us to examine the trade-offs in sensitivity and specificity for any given cutoff point for a single value. Our method is the only use of PSA values approved by the Food and Drug Administration for the early diagnosis of prostate cancer. There has been interest in using changes in serum PSA levels over time (PSA velocity) as a marker for prostate cancer, but the approach used by Sawyer et al. ignores the possibility of true changes in serum PSA levels over time.
Dr. Swee points out that many factors must be taken into consideration in evaluating screening tests. Our data cannot be used in and of themselves to make a recommendation for or against screening. We have, in fact, not explicitly recommended screening for any population. However, for black American men, who have the world's highest incidence of and rate of death from prostate cancer, what are we to do? Even in the U.S. military, with equal access to care, black men with clinically localized, surgically treated disease have tumor volumes more than one and a half times as large as those of their white counterparts, extraprostatic extension of the cancer more than one and a third times as often, and significantly higher recurrence rates.1,2 Facing the fact that this disparity is primarily due to delayed diagnosis because of lack of awareness and late detection is not a pleasant prospect for primary care physicians or specialists. Dr. Swee's arguments against screening are based on the incidence of aggressive prostate cancers and prevalence rates that may not apply to the black population. To argue against screening for black men on the basis of data that may not be applicable could perpetuate the problem.
Regardless of any recommendations for or against screening by any organization, it is being performed in the community. In many settings, a single cutoff point of 4.0 ng per milliliter is used, without taking into consideration characteristics such as race and age. A recent study showed that the specificity of 4.0 ng per milliliter as the cutoff point in black men was only 39 percent.3 Despite Dr. Swee's concern about our specificities (93 to 78 percent), they are still superior to 39 percent and will improve the positive predictive value of screening in this high-risk group.
3 ReferencesJudd W. Moul, M.D.
Uniformed Services University of the Health Sciences, Bethesda, MD 20814Ted O. Morgan, M.D.
Walter Reed Army Medical Center, Washington, DC 20307Steven J. Jacobson, M.D., Ph.D.
Mayo Clinic, Rochester, MN 559051
Moul JW ,Mooneyhan R ,Lin TH , et al. Three dimensional (3D) computerized tumor volume determination in radical prostatectomy specimens from black and white patients. J Urol 1996;155:Suppl:509A-509A abstract.2
Moul JW ,Douglas TH ,McCarthy WF ,McLeod DG . Black race is an adverse prognostic factor for prostate cancer recurrence following radical prostatectomy in an equal access health care setting. J Urol 1996;155:1667-1673
CrossRef | Web of Science | Medline3
Smith DS ,Catalona WJ ,Bullock AD . Operating characteristics of prostate cancer screening tests in African-American and white men. J Urol 1996;155:Suppl:376A-376A abstract.
Author/Editor Response
The age-specific reference ranges for whites,1 Asians,2 and blacks3 are the only reference ranges for serum PSA levels that have been determined in a scientifically and statistically correct manner. From numerous independent studies, it is clear that the serum PSA concentration is dependent on both age and race. Taking these two factors into account will considerably improve the clinical utility of the PSA test.
A recent survey of practicing urologists indicates that 80 percent use these age-specific reference ranges instead of the traditional reference range of 0.0 to 4.0 ng per milliliter (unpublished data). This change in the interpretation of the PSA test will lead to increased detection of cancers in younger men and fewer unnecessary prostate biopsies in older men. Indeed, this has already been shown in a large screening study in Innsbruck, Austria.4
Calculation of the percentage of free PSA appears to enhance the ability of the PSA test to distinguish early, curable prostate cancer from benign prostatic hyperplasia. The results, however, are still preliminary. Nevertheless, like Catalona, my colleagues and I at the University of Michigan believe that determining the percentage of free PSA can improve the sensitivity and the specificity of the PSA test. Like Catalona, we believe that calculating the percentage of free PSA is most useful when the total PSA value is in the range of 3.0 to 10.0 ng per milliliter.5 When the total PSA value is at the upper limit of normal, the appropriate cutoff point for free PSA is 19 percent. When the total PSA value is elevated but less than 10.0 ng per milliliter, the appropriate cutoff point is 24 percent. A man with a free PSA value less than or equal to the cutoff point should undergo a prostate biopsy. Using the percentage of free PSA in this manner improves sensitivity (increases the number of cancers detected) when the total PSA value is normal and improves specificity (reduces the number of negative prostate biopsies) when the total PSA value is elevated. Both are clinically desirable outcomes when detecting early, curable prostate cancer in men with a high prevalence of benign prostatic hyperplasia.
Lastly, prostate cancer kills — a man dies from this cancer every 13 minutes in the United States. Every man has a right to know if he has prostate cancer when it is still at a curable stage (organ-confined). Until it has been proved that early detection of prostatic cancer does not save lives, men should not be denied this right. This means that men should undergo an annual prostatic evaluation, consisting of a serum PSA test and a digital rectal examination, starting at the age of 50 years in general and at the age of 40 for black men and those with a family history of prostate cancer. Since we do not have a systemic, curative treatment for prostate cancer, this is our only hope of preventing 40,000 deaths from prostate cancer annually.
5 ReferencesJoseph E. Oesterling, M.D.
University of Michigan, Ann Arbor, MI 481091
Oesterling JE ,Jacobsen SJ ,Chute CG , et al. Serum prostate-specific antigen in a community-based population of healthy men: establishment of age-specific reference ranges. JAMA 1993;270:860-864
CrossRef | Web of Science | Medline2
Oesterling JE ,Kumamoto Y ,Tsukamoto T , et al. Serum prostate-specific antigen in a community-based population of healthy Japanese men: lower values than for similarly aged white men. Br J Urol 1995;75:347-353
CrossRef | Medline3
Morgan TO ,Jacobsen SJ ,McCarthy WF ,Jacobson DJ ,McLeod DG ,Moul JW . Age-specific reference ranges for serum prostate-specific antigen in black men. N Engl J Med 1996;335:304-310
Full Text | Web of Science | Medline4
Reissigl A ,Pointner J ,Horninger W , et al. Comparison of different prostate-specific antigen cutpoints for early detection of prostate cancer: results of a large screening study. Urology 1995;46:662-665
CrossRef | Web of Science | Medline5
Oesterling JE . Age-specific reference ranges for serum PSA. N Engl J Med 1996;335:345-346
Full Text | Web of Science | Medline
