Fred Hutch study finds prostate cancer screening more effective at reducing mortality than previously reported

Study provides new statistical analysis of data from two major clinical trials

SEATTLE — Sept. 4, 2017 — A study led by Fred Hutchinson Cancer Research Center in collaboration with the University of Michigan School of Public Health and Erasmus University Medical Center shows that prostate-specific antigen (PSA) screening for prostate cancer is more effective than previously reported. The research, published today in the Annals of Internal Medicine, employed new statistical techniques to re-analyze two major clinical trials and arrived at a consensus estimate of efficacy showing that screenings as conducted in the trials can lower prostate cancer mortality by 25 to 32 percent.

Dr. Ruth Etzioni
Dr. Ruth Etzioni and 21 researchers from 10 countries re-evaluated data from two clinical trials to calculate potential benefits of PSA screening for prostate cancer Fred Hutch

“What we’re trying to do here is go beyond the straightforward analyses to get the real message from these numbers,” said Dr. Ruth Etzioni (left), a biostatistician in the Public Health Sciences Division at Fred Hutch and senior author of the paper. “And what these numbers are telling us is that prostate cancer screening works.”

Current U.S. Preventive Services Task Force (USPSTF) guidelines, which are in the process of being updated, recommend against PSA screening based on the finding that there is a “very low” probability of screenings preventing long-term mortality.

However, these guidelines are largely based on the previous analyses of two clinical trials: the European Randomized Study of Screening for Prostate Cancer (ERSPC) and the Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial (PLCO). The ERSPC reported that men who were invited to screening had a 21 percent reduction in mortality compared with those who weren’t. By contrast, the PLCO showed no difference between the two groups.

A new look by the study authors shows that, under careful analysis, the two studies actually point to the same conclusion: screenings significantly lower the risk for prostate cancer mortality.

“Unfortunately, screening trials are complicated,” Etzioni said. “There are lots of issues in screening trials and if we want to learn from them, we have to go beyond the straightforward analyses.”

The researchers re-examined the two trials and identified key differences in PSA screening implementation, participant compliance, participant age and practice settings. They found that these factors clouded initial analyses of the data and could account for the conflicting results. The authors confirmed that, essentially, the U.S. study compared the effects of an organized versus opportunistic screening program while the European study provided a better comparison of screening versus no screening.

Using individual records from both trials, the authors conducted several computer-based statistical analyses to quantify these differences. They studied data from the trials —161,394 patients in the ERSPC and 76,683 in the PLCO — to formally test whether the effects of PSA screening on prostate cancer mortality differed in men who were screened versus those who were not.

The results showed that once the discrepancies were accounted for, the data told the same story. In fact, there was a strong benefit that increased with earlier detection due to screening. The earliness of detection was comparable in the screened and non-screened groups of the PLCO trial but different in the screened and non-screened groups of the ERSPC. Overall, the risk of prostate cancer mortality was lowered by 25 to 31 percent in the ERSPC screening arm and 27 to 32 percent in the PLCO screening arm compared to no screening.

The study shows how complicated screening trials can be, and, according to Etzioni, is evidence that good policy needs to be informed by in-depth interpretation of the data.

“That’s really the big message,” she said. “If we want to make sound policies, we have to acknowledge that we may need some mathematical or statistical modeling to help interpret our data. And we modelers have to communicate what we’re doing in a way people can understand. We have to come toward each other.”

The National Cancer Institute funded the study. Authors on the paper are Alex Tsodikov, Ph.D., of the University of Michigan School of Public Health; Roman Gulati, M.S. of Fred Hutch; Eveline A.M. Heijnsdijk, Ph.D., of Erasmus Medical Center; Paul F. Pinsky, Ph.D., of the National Cancer Institute; Sue M. Moss, Ph.D., of Queen Mary University of London; Sheng Qiu, M.S., of University of Michigan School of Public Health; Tiago M. de Carvalho, M.S., at Erasmus Medical Center; Jonas Hugosson, M.D., of Sahlgrenska University Hospital; Christine D. Berg, M.D., of Johns Hopkins Medicine; Anssi Auvinen, M.D., of University of Tampere; Gerald L. Andriole, M.D., of Washington University School of Medicine; Monique J. Roobol, Ph.D., of Erasmus Medical Center; E. David Crawford, M.D., of University of Colorado, Denver; Vera Nelen, M.D., of Provinciaal Instituut voor Hygie¨ne; Maciej Kwiatkowski, M.D., of Kantonsspital Aarau; Marco Zappa, Ph.D., of Institute for Cancer Prevention (Italy); Marcos Luja´n, M.D., of Universidad Complutense de Madrid; Arnauld Villers, M.D., of Universite´ de Lille; Eric J. Feuer, Ph.D., of the National Cancer Institute; Harry J. de Koning, M.D., of Erasmus Medical Center; Angela B. Mariotto, Ph.D., of the National Cancer Institute; and Ruth Etzioni, Ph.D., of Fred Hutch.


Claire Hudson
O: 206.667.7365
M: 206.919.8300