Correspondence

Verification Bias in Screening for Prostate Cancer

N Engl J Med 2003; 349:1672-1673October 23, 2003

Article

To the Editor:

Punglia and colleagues (July 24 issue)1 assert that prostate biopsy should be the gold standard for evaluating the sensitivity and specificity of the prostate-specific antigen (PSA) test. This suggestion is not appropriate; clinically significant prostate cancer should be the gold standard. The more biopsies performed, the more incidental, clinically irrelevant cancers, probably unrelated causally to slight PSA elevations, would be detected, leading to unnecessary surgeries with subsequent long-term morbidity or early death. Until we are better able to determine which cancers will truly affect men's lives, we should not lower the PSA threshold for recommending biopsy.

Steven Leiner, N.P.
Mission Neighborhood Health Center, San Francisco, CA 94110
sleiner@itsa.ucsf.edu

1 References

1. 1

   Punglia RS, D'Amico AV, Catalona WJ, Roehl KA, Kuntz KM. Effect of verification bias on screening for prostate cancer by measurement of prostate-specific antigen. N Engl J Med 2003;349:335-342
   Full Text | Web of Science | Medline

To the Editor:

Regarding the article on prostate-cancer screening by Punglia et al., in what percentage of the men found to have prostate cancer on biopsy would the disease have been missed (i.e., would biopsy have been skipped) if the decision to perform a biopsy had been based on PSA-test results only? What was the number needed to screen by digital rectal examination in order to detect a prostate cancer that would not have been detected by PSA screening only?

Howard T. Chatterton, M.D., Ph.D.
Marshfield Clinic, Ladysmith, WI 54848
chatterton.howard@marshfieldclinic.org

Author/Editor Response

We agree with Mr. Leiner that clinically significant prostate cancer is the most relevant clinical outcome. However, definitions vary as to what constitutes clinically insignificant cancer, and they also change with a patient's age and coexisting conditions. In the absence of a consensus definition, we included all diagnoses of prostate cancer in our analysis to study the performance characteristics of PSA testing. Prior analyses of these data have shown similar proportions of “clinically unimportant”1 and “insignificant”2 tumors among patients with a PSA level of 2.6 to 4.0 ng per milliliter and those with PSA levels higher than 4.0 ng per milliliter.3,4 Those studies also showed that significantly more patients in the former group than in the latter group have organ-confined tumors.

The answer to Dr. Chatterton's first question depends on the PSA threshold used to recommend biopsy. If we define an elevated PSA level as a value of 2.6 ng per milliliter or greater, then only 5 percent of the men in whom prostate cancer was detected had a suspicious finding on digital rectal examination without an elevated PSA level. The second question is more difficult to answer for a data set with verification bias. Prostate cancer was diagnosed in 24 percent of the men who had suspicious findings on digital rectal examination and a PSA value of less than 2.6 ng per milliliter and who underwent biopsy. In the entire screened population, there were 84 men with low PSA values and suspicious findings on digital rectal examination, which means that there were approximately 20 such cancers among 6691 men, or 335 examinations to detect 1 prostate cancer that would have been missed by PSA screening alone (without adjustment for age, race, and family history). If the definition of an elevated PSA level is a level higher than 4.0 ng per milliliter, then the relative importance of the digital rectal examination increases. With use of the higher threshold, the probability of finding cancer was 31 percent among men who underwent biopsy and who had suspicious findings on digital rectal examination without an elevated PSA level. Our unadjusted analysis would then predict approximately 33 cancers, and the number needed to screen by digital rectal examination in order to detect a prostate cancer that would not have been detected by PSA screening alone would be 203.

Rinaa S. Punglia, M.D., M.P.H.
Brigham and Women's Hospital, Boston, MA 02115

William J. Catalona, M.D.
Northwestern University, Fineberg School of Medicine, Chicago, IL 60611

Karen M. Kuntz, Sc.D.
Harvard School of Public Health, Boston, MA 02115

4 References

1. 1

   Ohori M, Wheeler TM, Dunn JK, Stamey TA, Scardino PT. The pathological features and prognosis of prostate cancer detectable with current diagnostic tests. J Urol 1994;152:1714-1720
   Web of Science | Medline

2. 2

   Epstein JI, Walsh PC, Carmichael M, Brendler CB. Pathologic and clinical findings to predict tumor extent of nonpalpable (stage T1c) prostate cancer. JAMA 1994;271:368-374
   CrossRef | Web of Science | Medline

3. 3

   Krumholtz JS, Carvalhal GF, Ramos CG, et al. Prostate-specific antigen cutoff of 2.6 ng/mL for prostate cancer screening is associated with favorable pathologic tumor features. Urology 2002;60:469-474
   CrossRef | Web of Science | Medline

4. 4

   Roehl KA, Antenor JA, Catalona WJ. Serial biopsy results in prostate cancer screening study. J Urol 2002;167:2435-2439
   CrossRef | Web of Science | Medline