Click here to read more Quest articles. print | Print this page
From the Summer/Fall 2016 Quest
At a critical time in the debate over PSA screening, we can look to the future. © Dan Oldfield

In the 1960s and 70s, several researchers independently identified proteins later shown to have the same amino acid sequence as PSA. T. Ming Chu’s laboratory at the Roswell Park Memorial Institute developed PSA as a biomarker and patented it in 1984. In 1987, the FDA approved PSA for monitoring patients already diagnosed with prostate cancer, but not for screening.

My research group first reported in the New England Journal of Medicine that PSA is useful as a screening test for prostate cancer in 1991, and in 1994, based upon a multi-institutional pivotal study that I led, the FDA approved PSA as an aid to the early detection of prostate cancer.

Improvements in PSA testing

Various enhancements of PSA testing have been developed to increase its specificity. Particularly important is the baseline PSA of a man in his 40s, which is the strongest risk factor for the development of metastatic and lethal prostate cancer decades later. There is a dramatic increase in the metastasis rates in men in their 40s with baseline PSAs ranging from the lowest 25th percentile to the highest 10th percentile. See page 3 in this issue for information on two recent studies on mid-life PSA testing.

Trends in the prostate cancer mortality rate

The widespread use of PSA testing has resulted in marked changes in the epidemiology of prostate cancer. During the PSA era in the US, there has been an 80% decrease in the percentage of patients who have advanced disease at diagnosis and more than a 50% decrease in the prostate cancer mortality rate. Much of this favorable trend has been attributed to PSA testing.

The USPSTF, CMS, and PSA Testing

The US Preventive Services Task Force (USPSTF) is currently updating its recommendations for prostate cancer screening. In addition, the Centers for Medicare and Medicaid Services (CMS) recently temporarily suspended the development of a measure that would discourage PSA screening in all men.

Dr. Catalona and some other experts in the prostate cancer field are encouraging these organizations to reconsider their use of the PLCO trial to recommend against the use of PSA testing. See page 2 for information on flaws in the trial.

Randomized trials on PSA screening

Randomized trials were launched to evaluate PSA screening. In my opinion, the best of these is the Goteborg trial that showed a 44% decrease in prostate cancer mortality. The largest trial from Europe with an older population and less frequent screening, mostly with PSA and not digital rectal exam, reported a 21% decrease in prostate cancer mortality.

In contrast, the US PLCO (Prostate, Lung, Colorectal, Ovarian) Cancer Screening Trial reported no significant difference in prostate cancer mortality and, paradoxically, the men in the control group, the non-screening arm of the trial, actually fared better than men in the screening arm.

The impact of the PLCO trial

The results reported from PLCO in the New England Journal of Medicine in 2009 have impacted the estimated benefits of PSA screening. In 2012, the US Preventive Services Task Force (USPSTF), relying on the reported 2009 published PLCO results, issued a grade D recommendation against PSA screening, concluding that the harms outweigh the benefits. Following the Task Force’s recommendation, there has been less PSA testing, fewer localized cancers detected and more men diagnosed with advanced disease at the time of diagnosis.

"Now that it is clear that the reporting of the PLCO study was inaccurate, professional organizations, such as the USPSTF, should bear this fact in mind and recommend prostate cancer screening."

– Dr. William Catalona


PLCO trial flaws: PSA testing in the control group

Importantly, the major criticism of the PLCO trial relates to the degree of PSA testing in the control arm as reported in the first 2009 PLCO publication. In developing recommendations for PSA screening, the Task Force and other panels relied on the reported 52% rate of PSA testing of controls cited in the PLCO 2009 report.

However, in a recent issue of the New England Journal of Medicine, Drs. Shoag, Mittal, and Hu reported on their reevaluation of raw PLCO data obtained from the National Cancer Institute. The investigators found that the original 2009 PLCO report inaccurately stated that approximately 52% of the controls had a PSA test during the study. In reviewing the raw data, they found that nearly 90% of controls had at least one PSA test before or during the trial.

My opinion

Thus, during the PSA era, the prostate cancer death rate fell. Then PLCO came out. There followed a flurry of papers stating the harms of prostate cancer screening outweighed the benefits. The reason was that people were looking at the benefits of screening through a prism distorted by inaccurate reporting.

This article was adapted from Dr. Catalona’s plenary presentation at the American Urological Association (AUA) annual meeting held in May.

The PLCO Trial: Reanalysis Reveals Flaws

The 2009 Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial randomly assigned men to annual prostate cancer screening (the screening arm) or no screening for prostate cancer (the control group). The study authors administered a follow-up questionnaire to a subset of participants in the study and used the responses to calculate rates of PSA testing in the control group. The questionnaire asked if men had ever undergone a PSA blood test for prostate cancer, as well as asking when and why they had the test. The original study counted men as being “tested” only if they had been screened within 1 year as part of a routine physical examination. Men who had PSA testing within 2 or 3 years, for a specific prostate problem or follow-up to a health problem were not counted as being tested in the original PLCO report.

The reevaluation of the data, published in May in the New England Journal of Medicine, revealed that PSA testing rates were actually higher in the control arm than in the screened arm.

Close this window