After 13 years of follow-up, men who underwent annual screening for prostate cancer were no less likely to die of prostate cancer than were men who received usual care and opportunistic screening.
Cumulative mortality rates from prostate cancer differed by a statistically insignificant rate of 0.3 deaths per 10,000 person years in the intervention and usual-care arms of the Prostate, Lung, Colorectal, and Ovarian (PLCO) screening trial, published online Jan. 6 in the Journal of the National Cancer Institute. Further, age, comorbidity status, and pretrial prostate-specific antigen (PSA) testing did not influence the results, wrote Dr. Gerald L. Andriole of Washington University, St. Louis, and his colleagues.
The PLCO trial randomly assigned 76,693 men, aged 55-74 years, to either 6 years of annual PSA screening in combination with 4 years of annual digital rectal examination (38,343) or to usual care (38,350), which included screening tests as recommended by physicians.
The goal was to evaluate the effect of adding annual screening and compare outcomes to the opportunistic screening already in place, the researchers said. It was expected that the impacts of earlier diagnosis and a persistent excess of cases because of annual screening in the intervention arm would exceed the impacts of opportunistic screening.
Prior to the study, 44% of all participants had undergone PSA screening. During the trial, 52% of the usual-care group, compared with the entire intervention group, underwent PSA testing.
The researchers had previously reported 7- and 10-year follow-up results. At 7 years of follow-up, yearly screening was associated with an increased incidence of prostate cancer diagnosis as compared with usual care. The rates of prostate cancer mortality and of all-cause mortality, however, were the same for both groups. Similarly, after 10 years of follow-up, no mortality benefit was observed for the intervention, the investigators reported previously (N. Engl. J. Med. 2009;360:1320-1328).
For the current study, Dr. Andriole and his associates ascertained all incident prostate cancer diagnoses and deaths through 13 years of follow-up or through December 31, 2009, and estimated relative risks as the ratio of observed rates of diagnoses and deaths in the intervention and control arms. They examined the interactions between prostate cancer mortality by trial arm and age, comorbidities, and pretrial PSA testing using Poisson regression modeling (J. Natl. Cancer Inst. 6 Jan. 2012 [doi: 10.1093/jnci/djr500]).
At 13 years, 4,250 of the 38,340 participants in the intervention arm had been diagnosed with prostate cancer, compared with 3,815 of the 38,345 control participants. "The cumulative incidence rates for prostate cancer in the intervention and control arms were 108.4 and 97.1 per 10,000 person-years, respectively, resulting in a statistically significant 12% relative increase in the intervention arm," the authors wrote. Of the prostate cancer diagnoses, 401 in the intervention arm and 454 in the usual care arm were high-grade prostate cancers with Gleason scores of 8-10.
At 13 years of follow-up, there were 158 deaths in the intervention arm and 145 deaths in the usual-care arm. "The cumulative mortality rates from prostate cancer were 3.7 and 3.4 deaths per 10,000 person-years, respectively, resulting in a non-statistically significant difference between the two arms," according to the authors. The examination of mortality rates per 10,000 person-years and relative risks of prostate cancer mortality by age, comorbidity, and pretrial PSA testing produced no statistically significant interactions.
Deaths from all causes other than prostate, lung, and colorectal cancers differed by a factor of "borderline" statistical significance, with 5,783 deaths in the intervention arm and 5,982 in the usual-care arm, Dr. Andriole and his associates wrote. "Intervention and control arms showed 23% and 22% deaths, respectively, from non-PLCO cancers, and 21% and 19% deaths, respectively, from ischemic heart disease."
The authors acknowledged limitations of the study, including the possibility that a reduction in prostate cancer mortality in the analysis has somehow been masked by "the sticking diagnosis effect." In other words, more deaths were attributed to prostate cancer in the intervention arm. This possibility is supported by the statistically significant lower all-cause mortality in the intervention group as compared with the usual-care group. The percentage of deaths from other causes was higher in the intervention arm, so an error in cause of death attribution likely does not account for the excess prostate cancer deaths in the intervention group, they concluded.
The investigators plan to update the mortality findings through 15 years of follow-up, when those data become available.
Some of the study authors report relationships with Amgen, Augmenix, Bayer, Cambridge Endo, Caris, Envisioneering Medical, France Foundation, GenProbe, GlaxoSmithKline, Human Genome Sciences, Myriad Genetics, Steba Biotech, Ortho Clinical Diagnostics, and Viking Medical.