鶹ý

Landmark Trial Not a Measure of PSA Value

— High rate of 'contamination' in control arm of PLCO trial

MedpageToday

SAN DIEGO -- The U.S. Preventive Services Task Force (USPSTF) relied on flawed data to make its controversial 2012 recommendation against routine PSA testing to screen for prostate cancer, investigators in a new analysis asserted.

The USPSTF cited results of the (PLCO) trial in concluding that routine PSA testing could not be recommended for men of any age. In reporting results from the landmark randomized trial, PLCO investigators estimated that 52% of the men in the control (unscreened) group had undergone PSA testing by the sixth year, the period of active screening in the intervention arm.

In fact, the "contamination" rate from screening in the control arm was closer to 90%, making a comparison between the intervention (screened) and control arms impossible, , said here at the (AUA) annual meeting.

"There was obviously a much higher rate of PSA testing in the control arm than previously cited -- so much so that comparing the control and intervention arms may not be informative," said Shoag. "In light of this, recommendations against screening, particularly those based on misinterpretations of the PLCO trial, should be reconsidered."

The results were published simultaneously in the .

PLCO investigators have acknowledged the high rate of PSA testing in the control, beginning with a "contamination analysis" published a year after the primary results were reported, said principal investigator , of Washington University in St. Louis.

"The data ... are not new nor substantially different from the published by the PLCO statisticians in 2010," Andriole told 鶹ý by email. "It is very important to realize that during the screening phase of PLCO, there were many more PSA tests in the screened men than in the control group, as is corroborated by the higher prostate cancer incidence in the screened men. Later, over the many years after screening stopped in PLCO, progressively more men in the control arm had PSA tests."

In a , PLCO investigators concluded that the trial showed no mortality benefit from organized annual PSA testing versus "opportunistic screening," Andriole continued. The USPSTF was aware of limitations of the PLCO when recommending against PSA testing.

The USPSTF also had access to results of the , which showed one prostate cancer death averted per 1,000 men screened for a decade. In all likelihood, the task force had not seen the modest benefit as outweighing the "human and economic costs of PSA testing," said Andriole.

According to Shoag, the inaccuracy resulted from PLCO investigators' narrow definition of screening contamination. Baseline contamination in the control group (defined as two or more PSA tests within the past 3 years) was about 10%. During the trial, screening was self-reported by means of the Health Status Questionnaire (HSQ).

The HSQ included multiple PSA testing intervals (<1 year, 1 to 2 years, 2 to 3 years, >3 years, and "don't know") and reasons for screening (follow-up for a prostate problem, follow-up for some other health reason, or part of a routine physical). During the trial, PLCO investigators defined screening contamination in the control group as a PSA test within the previous year for a routine physical. Other responses to the HSQ were not counted as contamination, but should have been, said Shoag.

The 2009 and 2012 PLCO publications attracted considerable attention, whereas the 2010 contamination analysis "suffered from poor visibility and a complex presentation of the data in the body of the manuscript only," Shoag continued. As evidence, he noted that USPSTF, the American Academy of Family Physicians, and the Cochrane Library have all cited the 2009 report and the estimated cumulative contamination rate of 52%. The AUA and the National Comprehensive Cancer Network have estimated the contamination rate to be 75%.

"This number appears to be critical, as those who have interpreted the [contamination] rate to be lower have advocated against PSA screening, while those who cite a higher rate have advocated for PSA screening," said Shoag.

Taking another look at PSA screening in the control arm, investigators found that 56% of men without baseline contamination had undergone a PSA test within the previous year by the sixth (last) year of the PLCO. By including men who provided other answers to the HSQ, the screening rate increased to 87% by year 6 when the 10% of men with baseline contamination were added.

Reviewing the testing rate in the control arm from year 1 to year 13 (the last year of follow-up) showed that 90% of men in the control arm had undergone one or more PSA tests during the last 4 years of the follow-up. Total contamination (including baseline contamination) resulted in testing rates of 67% in year 1 to 92% in year 13.

Cumulative testing rates in the intervention arm of the PLCO were actually lower than in the control arm during years 7 through 13. From 74% to 86% of men in the intervention arm had one or more PSA tests by the various HSQ criteria.

Using an expanded definition of screening to evaluate the PLCO control arm is not unreasonable given that different indications for a PSA test might be ordered, said , of NYU Langone Medical Center in New York City.

"[The high contamination rate] is unfortunate because the PLCO represented a really noble effort to study an important issue, but you really use the data to understand whether screening is useful or not," said Markarov. "The populations are essentially the same. If this were a placebo-controlled drug trial and everyone in both groups got the drug, it wouldn't be a comparison anymore. It's the same situation with the PLCO."

  • author['full_name']

    Charles Bankhead is senior editor for oncology and also covers urology, dermatology, and ophthalmology. He joined 鶹ý in 2007.

Disclosures

Shoag disclosed no relevant relationships with industry.

Primary Source

American Urological Association

Shoag J, et al "Reevaluating PSA testing rates in the PLCO Trial" AUA 2016; Abstract PL-LBA02.

Secondary Source

New England Journal of Medicine

Shoag J, et al "Reevaluating PSA testing rates in the PLCO Trial" N Engl J Med 2016; 374: 1795-1796.