© The Author 2006. Published by Oxford University Press.
EDITORIAL |
Prostate-Specific Antigen and Prostate Cancer Prognosis
Affiliation of author: Division of Environmental Health Sciences, University of Minnesota School of Public Health, Minneapolis, MN.
Correspondence to: Timothy R. Church, PhD, Division of Environmental Health Sciences, University of Minnesota School of Public Health, 200 Oak St. SE, Ste. 350, Minneapolis, MN 55455-2008 (e-mail: trc{at}cccs.umn.edu).
The impulse in clinical medicine is to reduce diagnostic testing to dichotomous decisions: positive for increased risk of the disease and negative for decreased risk (1). Some of this tendency stems from discomfort with or distrust of mathematical manipulation and some from the need for simplicity and clarity. However, ultimately a diagnosis is a complex judgment by the clinician, incorporating a variety of data, including test results, symptoms, appearance of the patient, and intuition. Combining these disparate sources into a single "posterior probability," in the language of Bayesian statistics, is an uncertain and undoubtedly biased process when done informally or intuitively (1). To counteract these problems, the sophisticated mathematics of Bayesian inference have been proposed. Far from unique, sophisticated mathematics has been used to improve diagnosis in many ways. For example, it used to suffice to capture the intensity pattern of X-rays on a two-dimensional photographic plate to examine the interior of the body. With the advent of computed tomography, clinicians have accustomed themselves to looking at the results of complicated mathematical calculations based on the variations in transmission of a beam of X-rays through the body as the transmitter rotates through space. That these computational results are represented as a three-dimensional "picture" of the interior of the body has certainly facilitated the acceptance of these methods and has led to the ready acceptance of magnetic resonance imaging, an even more complex physical and mathematical abstraction. Nonetheless, many radiologists are so comfortable with the mathematics that they give no thought to the difference between these mathematically generated three-dimensional images and the two-dimensional film recording.
How do these complex diagnostic methods relate to a relatively simple assay for a circulating protein? The impulse to divide prostate-specific antigen (PSA) results into two categories to determine whether to biopsy a man for the presence of cancer has dominated PSA use since its introduction. The paper by Carter et al. (2) in this issue of the Journal is another step on the road to more sophisticated use of quantitative methods that can improve the understanding and accuracy of screening and prognosis in prostate cancer. What is most compelling is the implication that applying a mathematical method to a modest screening test might actually help to distinguish the cancer that requires aggressive intervention to save the life of the patient from one that will remain indolent and can be safely ignored. The investigators go beyond the simplistic question of whether the current PSA value is more than or less than 4 (or 2.6) ng/dL to recognizing the potential of changes in PSA over time to predict not only the presence of cancer but also its lethal potential. Similarly to previous researchers, Carter and colleagues use a simple formula that estimates or calculates a PSA gradient over time, using two or three measurements to yield what urologists have come to call PSA velocity. The authors' analyses of subsequent mortality suggest that these PSA velocities are highly associated with the likelihood that a patient will die of prostate cancer, especially if he is measured 10–15 years before diagnosis.
PSA screening for prostate cancer has been a hope ever since introduction of the test in the mid-1980s. To date, no solid evidence exists to confirm the realization of this hope, but at least two large studies to address the efficacy of PSA screening are in progress. One, the National Cancer Institute–funded Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial, is expected to produce an answer within the next 5–7 years, and the other, the collaborative European Randomized study of Screening for Prostate Cancer, is expected to provide results in a similar time frame. Although many physicians believe in the efficacy of PSA screening today, the more skeptical practitioners await more reliable results to support the decision to use it.
The paper by Carter et al. is by no means definitive regarding PSA velocity, nor is it without weaknesses, some of which they address in their discussion. The validity of the design and methods is difficult to evaluate in description and application, and the implication of the results is not straightforward. For example, the authors do not explicitly define what substituted for the date of diagnosis among noncancer subjects in the cohort. Presumably it was the last PSA test, but if so, they neither justify the validity of using that definition nor investigate the potential impact of defining the beginning point for follow-up relative to this self-selected event. Further, the implication of, say, Cox regression results from such data is unclear. Unfortunately as well, the improvement in prognostic value of the PSA velocity relative to the PSA is never explicitly estimated; only qualitative comparisons are presented. The paper would benefit greatly from a more careful analysis and discussion of these points.
Raising the focus of PSA testing from individual test results to evaluating patterns of test results in an individual over time is a valuable endeavor and is a positive example of how clinical medicine should always strive to use mathematics to make the most of information gathered regarding disease occurrence and outcome. As it is, this study encourages optimism about the value of PSA velocity in predicting outcomes for prostate cancer, although some of the optimism may be unwarranted because of the study insufficiencies cited above. However, whatever false optimism for PSA velocity that this study engenders will ultimately be tempered by the results of the large randomized trials currently under way (3,4). With their generous sample sizes, the prospective nature of their follow-up (which does not depend on disparate definitions based on case status), and their rigorous application of screening, these trials can more definitively answer the question of the prognostic nature of PSA velocity. In the meantime, screening for prostate cancer by using PSA is not a practice grounded in evidence in the literature and should not be undertaken on the basis of the results of this or previous studies not based on randomized comparison of mortality and morbidity.
REFERENCES
(1) Edwards W. N = 1: diagnosis in unique cases. In Jacquez JA, editor. Computer diagnosis and diagnostic methods. Springfield (IL): Charles C. Thomas; 1972. p. 139–51.
(2) Carter HB, Ferrucci L, Kettermann A, Landis P, Wright EJ, Epstein JI, et al. Detection of life-threatening prostate cancer with prostate-specific antigen velocity during a window of curability. J Natl Cancer Inst 2006;98:1521–7.
(3) Gohagan JK, Prorok PC, Hayes RB, Kramer BS. The Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial of the National Cancer Institute: history, organization, and status. Control Clin Trials 2000;21:251S–72S.[CrossRef][Web of Science][Medline]
(4) Schroder FH, van der Maas P, Beemsterboer P, Kruger AB, Hoedemaeker R, Rietbergen J, et al. Evaluation of the digital rectal examination as a screening test for prostate cancer. Rotterdam section of the European Randomized Study of Screening for Prostate Cancer. J Natl Cancer Inst 1998;90:1817–23.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  U. Fedeli, N. Alba, G. Ciccone, C. Galassi, and P. Spolaore Re: Trends in Radical Prostatectomy Rates J Natl Cancer Inst, July 4, 2007; 99(13): 1052 - 1053. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||