PSA Screening: An Expert Interview With David F. Penson
Editor's Note:
The January 2006 issue of the Archives of Internal Medicine carried a report of a study on the effectiveness of prostate cancer screening in extending survival among screened patients. The conclusion of that report ("These results do not suggest that screening with PSA or DRE is effective in reducing mortality.") drew headlines in the national press. Medscape Urology's David McNeel discussed the results of that study with one of its coauthors, David F. Penson, Associate Professor of Urology at the Keck School of Medicine in Los Angeles, California (and member of the Advisory Board for Medscape Urology).

Medscape: Can you briefly describe the study, particularly how it differs from other studies that have tried to address the usefulness of prostate-specific antigen (PSA) in reducing mortality?

Dr. Penson: This study differs from others in that it uses an unusual design for clinical research; it's a case-control study. Effectively, the study looks backwards. In other words, most clinical studies define 2 groups: one that gets intervention A and the other that gets intervention B. Follow-up is then done with the patients to see how many have the outcome of interest in each group. In this study, we take a group of men in Veterans Affairs (VA) medical centers who were diagnosed with prostate cancer from 1991-95 and who died from 1991-98 and define them as the cases. We define the controls as a group of men age-matched to the cases who were seen in the VA medical centers from 1991-95 and who were alive at the time of death of the cases. We then look to see if there are differences in whether the 2 groups received prostate cancer screening. In other words, the patients were identified on the basis of the outcome and were then analyzed to see if they had received the intervention as opposed to the other way around. There is one other study that used this methodology, but that this study looked at prostate-cancer specific death as opposed to overall mortality.

While the case-control study design approach is reasonable, it is not as convincing as a randomized clinical trial. All the same, in a short window (5-7 years) in a selected population (veterans), it looks like prostate cancer screening did not affect overall mortality. The findings may have been different had we looked at prostate cancer-specific mortality, used a more generalizable population, or followed the patients for a longer period of time. I certainly wouldn't make policy decisions regarding screening on the basis of the current study.

Medscape: I didn't entirely understand how the control group was selected. Can you explain that?

Dr. Penson: The cases were originally identified by the date of diagnosis of prostate cancer and the fact that they died in the 8 years following diagnosis. Once a case was identified, a control was randomly chosen by finding a veteran who had been a patient in a VA medical center, and who matched the case for age at the time of diagnosis. For example, Mr. Smith, a 65-year-old man, was diagnosed with prostate cancer at the Bill Clinton VA Medical Center on February 1, 1991 and died on April 1, 1995. We then looked at the center's ambulatory care clinic roles for February 1, 1991 and found all men age 65 who were seen in the clinic that day who were alive on April 1, 1995, when Mr. Smith died. We then randomly selected one of these men to serve as the matched control for Mr. Smith.

Medscape: Did this study simply confirm what you suspected or was there anything that surprised you about the results?

Dr. Penson: No surprises here. The study doesn't have adequate follow-up, nor does it examine prostate cancer-specific mortality. The men most likely to die in the 5- to 7-year period following a diagnosis of prostate cancer are the ones with the most advanced disease at presentation or the ones with the most associated comorbid disease. Therefore, it is unlikely that you would see any overall mortality benefit of prostate cancer screening in this short window.

Medscape: It seems that the more we know about PSA the less accurate it appears to be. And yet isn't it true that mortality from prostate cancer has significantly decreased since the advent of PSA testing? What is PSA good for?

Dr. Penson: Mortality has decreased since the advent of PSA testing. I would like to think that prostate cancer screening with the PSA test played a part in this, but I doubt that this is the only factor influencing mortality rates. There has been increased use of early hormone therapy, which some authors have postulated has a greater impact on mortality than screening. I think that people expect PSA to be a perfect test, and clearly it is not. However, it has definitely resulted in a downstaging of prostate cancer and will ultimately be shown to positively influence mortality in one of the randomized clinical trials. The real issue with PSA is that there are no logical thresholds or cut-offs, and this makes its clinical use more of a challenge. Providers really need to be up to date on how this test works in younger men, men with a family history of prostate cancer, and men with prior biopsies. It's clear that the 4-ng/mL threshold commonly used in primary care is not adequate.

Medscape: Did this study tell us anything new about the value of digital rectal exam (DRE)?

Dr. Penson: Not really. You have to remember that the DRE info was taken from VA medical records, and was often obtained by residents. I don't have a lot of faith in the DRE data in this study.

Medscape: It almost seems as if one of the most difficult jobs of doctors treating prostate cancer is going to be explaining all of the various test results to patients and the choices that face them. Do you have any advice to help doctors advise their patients?

Dr. Penson: There are a lot of resources available to providers regarding prostate cancer screening. Certainly Medscape is a good place to start, but there are many online patient-centered resources available. One of the best is the Center for Disease Control CDC Web site. They've put together a portfolio of both physician-centered and patient-centered information. Other places to look include the National Cancer Institute NCI Web site and some of the professional society Web sites.

Medscape: Should physicians do anything different from now on based on the results of this study?

Dr. Penson: No. Physicians should be explaining to patients that we really don't know if screening is a good thing or not, and that it is their choice if they want to get the test. Most men do choose to be screened, and that's very reasonable. The one thing providers must do, particularly if the patient states that he doesn't want to be screened, is document that the test was offered. Beyond that, we'll have to wait for the results from the randomized clinical trials.