ASCO--A Canadian study is the first to show that prostate cancer screening, using a PSA of 3.0 ng/mL as the upper limit of normal, could significantly decrease deaths from the disease. Fernand Labrie, MD, in his presentation at the ASCO plenary session, said that 27,000 lives a year could be saved in the United States alone if PSA testing were performed on all men age 50 and older.
Dr. Labrie and his colleagues at Laval University, Quebec, began their study in 1988, using the names of 46,193 men age 45 to 80 from the 1985 electoral rolls of Quebec City. These men were randomized to receive no intervention or an invitation by letter to be screened. Prostate cancer deaths were identified in the official Province of Quebec death registry.
The men who actually underwent screening had a 69% reduction in prostate cancer mortality (137 deaths in the 38,056 unscreened men vs 5 deaths in the 8,137 screened men, for an odds ratio of 3.25). Of the five deaths in the screened group, four were diagnosed during the first visit, three with metastatic disease. With screening, Dr. Labrie said, "diagnosis of metastatic prostate cancer should practically disappear." But others sounded a note of caution, saying that screening guidelines should not be based on a single study and citing concerns about the study’s patient selection process.
Rather than using intention-to-screen methodology, the researchers based their analysis on men who were actually screened vs those who were not screened. Men who were invited for screening but were not screened were not counted in the screened group. However, Dr. Labrie said that the two groups could be pooled, since there was no difference in death rate between the unscreened men who were not invited (true controls) and the unscreened men who were invited.
Further Analysis Needed
Dr. Peter Boyle, director of epidemiology, European Institute of Oncology, Milan, congratulated Dr. Labrie for having "the insight and foresight" to undertake the first randomized trial of PSA screening, but noted in his discussion that further analysis of the data set is needed before final conclusions can be drawn.
Dr. Boyle’s analysis of the data on an intention-to-screen basis showed a relative risk of prostate cancer death in the invited group of about 1.0, suggesting a similar prostate cancer death rate in both groups. This points to a possible role for selection and/or ascertainment bias in Dr. Labrie’s analysis of compliers vs noncompliers. Dr. Labrie, however, said that the men who were screened (the compliers) did not have any special low risk for prostate cancer death and may have been at higher risk, since they were apparently more concerned about prostate cancer for various reasons such as prostate symptoms or a family history.
Dr. Boyle also cited as problems the lack of information on prostate cancer incidence and the lack of a committee to review causes of death. He suggested that these issues could be resolved with the application of advanced statistical epidemiology methods. "These analyses would give us a better understanding of the impact of PSA screening on reducing prostate cancer mortality," he said.
