The relative benefits of surgery or observation in men with prostate cancer detected by prostate-specific antigen (PSA) testing have not been defined. Randomized trials comparing radical prostatectomy with observation were conducted before widespread use of PSA testing and an observational study conducted during the PSA testing era suggested high 10-year survival rates among men treated conservatively.
Recent studies of the effects of PSA screening on prostate cancer mortality have reached opposite conclusions: The Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial on prostate cancer mortality found no benefit of PSA screening vs usual care; the 11-year follow-up of the European Randomized Study of Screening for Prostate Cancer (ERSPC) showed a significant benefit of PSA screening vs control on prostate cancer mortality.
In a recent article in The New England Journal of Medicine, Wilt and colleagues report findings of the Prostate Cancer Intervention Versus Observation Trial (PIVOT),1 which enrolled patients during the early part of the PSA testing era. The results indicate a lack of survival benefit or reduction in prostate cancer mortality of radical prostatectomy compared with observation in men with localized prostate cancer detected during the early PSA era.
In the PIVOT trial, 731 men aged ≤ 75 years with localized prostate were randomly assigned to radical prostatectomy (n = 364) or observation (n = 367) between November 1994 and January 2002 and followed through January 2010. The primary endpoint was all-cause mortality. Patients had a median age of 67 years and a median PSA value of 7.8 ng/mL. Nearly one-third were black, approximately 50% had stage T1c disease (nonpalpable, detected by PSA testing), and approximately 25% had histologic scores ≥ 7 on the Gleason scale. Tumors were low-risk in 40% of patients, intermediate-risk in 34%, and high-risk in 21%. On central pathology review, 48% of patients had Gleason scores ≥ 7 and 66% had intermediate- or high-risk tumors.
Overall, 79% of the patients randomized to radical prostatectomy underwent an attempted procedure (at a median time from randomization of 35 days) and 85% of those received definitive therapy. Among those randomized to observation, 10% underwent attempted radical prostatectomy (at a median of 61 days) and 20% received definitive therapy.
Over a median follow-up of 10.0 years, all-cause mortality rates were 47.0% in the prostatectomy group and 49.9% in the observation group (HR = 0.88, 95% CI = 0.71–1.08, P = .22), a nonsignificant absolute 2.9% difference. Overall, 354 patients (48%) died from any cause, and 52 (7%) died from causes related to prostate cancer or its treatment (prostate cancer mortality). Median overall survival was 13.0 years in the prostatectomy group and 12.4 years in the observation group. At 12 years, 40.9% of the prostatectomy group and 43.9% of the observation group had died. The absolute reduction in mortality in the prostatectomy group was not significant at any time point and decreased over time from 4.6% at 4 years to 2.9% at 12 years.
Subgroup analyses indicated no difference in mortality between the two groups according to age, Gleason histologic score, race, self-reported performance status, or Charlson comorbidity score, but showed a significant interaction between treatment group and baseline PSA level (P = .04) and a borderline interaction for tumor risk category (P = .07). Radical prostatectomy was associated with an absolute 13.7% reduction in all-cause mortality among men with baseline PSA level > 10 ng/mL (HR = 0.67, 95% CI = 0.48–0.94), an absolute 12.6% reduction among those with intermediate-risk tumors (HR = 0.69, 95% CI = 0.49–0.98), and a nonsignificant absolute 6.7% reduction among those with high-risk tumors. Among patients with low-risk tumors, the prostatectomy group had a nonsignificant relative 15% increase in risk for mortality.
Prostate Cancer Mortality
Rates of prostate cancer mortality (mortality due to prostate cancer or its treatment), a secondary endpoint of the study, were 5.8% in the prostatectomy group and 8.4% in the observation group (HR = 0.63, P = .09), a nonsignificant absolute 2.6% difference. Death considered to be definitely due to prostate cancer or its treatment occurred in 4.4% and 4.9% of patients, respectively. Rates of prostate cancer mortality in the two groups were identical at 4 years, with the prostatectomy group having a nonsignificant absolute 3.0% reduction at 12 years (4.4% vs 7.4%). Prostate cancer mortality rates were lower among patients in the prostatectomy group who had baseline PSA ≥ 10 ng/mL (5.6% vs 12.8%, P = .02) or high-risk tumors (9.1% vs 17.5%, P = .04).
Bone metastases occurred in 4.7% of patients in the prostatectomy group compared with 10.6% of the observation group (HR = 0.40, P < .0001). Complications within 30 days after surgery, including 1 death, occurred in 21% of men in the prostatectomy group who underwent the procedure. Complications occurring in > 2% of these patients included urinary tract infection, surgical repair, bleeding requiring transfusion, and urinary catheterization more than 30 days after surgery. At 2 years, urinary incontinence and erectile dysfunction, but not bowel dysfunction, were significantly more common in the prostatectomy group.
The investigators concluded, “[O]ur study showed that … radical prostatectomy did not significantly reduce all-cause or prostate cancer mortality through at least 12 years among men with clinically localized prostate cancer that had been diagnosed in the era of PSA testing.” They noted that their findings “support observation for men with localized prostate cancer, especially those who have a low PSA value and those who have low-risk disease,” while “[s]ubgroup analyses suggested that surgery might reduce mortality among men with higher PSA values and possibly among men with higher-risk tumors.”
Do Studies Miss Benefit of Screening?
In an accompanying editorial, Thompson and Tangen2 point out that prostate cancer mortality was reduced 44% between 1993 and 2009 and wonder how this can be if screening and treatment do not provide benefit. They suggest that part of the problem may be flaws in the studies that have indicated lack of benefit, including the fact that in the PLCO study, more than half of the control group underwent PSA testing.
With regard to the PIVOT trial, the commentators question whether statistical analysis of the trial was adequate to detect what might be meaningful differences in mortality. Due to slow enrollment, the target population of the trial had to be reduced from 2,000 to 740 patients. The statistical analysis was adjusted based on the estimate that an enrollment of 740 over a period of 7 years with 8 additional years of follow-up would provide 91% power to detect a 25% relative reduction in all-cause mortality on the assumption of a median survival of 10 years.
The commentators noted that given the actual 12-year median survival in the study’s observation group, the patient sample would need to be 1,200 in order to detect a 25% relative reduction in mortality with 90% power and a two-sided alpha level of 0.05. Given the actual enrollment of 731 patients, the study was underpowered to detect this relatively large clinical effect. The wide 95% confidence interval (0.71–1.08) for the hazard ratio for mortality in the treatment group illustrates this issue: A relative increase of up to 8% to a relative decrease of up to 29% in mortality in the prostatectomy group could not be excluded with 95% confidence.
The commentators also note that while all-cause mortality is an important endpoint, only 15% of endpoints in the study (52 of 354 deaths) were prostate cancer–related (informative endpoints). The treatment effect in prostate cancer would have to be enormous to show an overall 25% difference in all-cause mortality when 85% of events were not informative endpoints.
The commentators agree with the authors of the study that the finding of no benefit of prostatectomy in men with lower-risk disease argues for active surveillance in this population. They note that a rational approach to controlling prostate cancer is likely to include biopsy only for those with lethal cancer, treatment focused on such cancer, and individualized treatment approaches. Efforts under way to improve identification of patients with lethal cancer—including Early Detection Research Network studies of biomarker combinations and novel biomarkers—will help to achieve these objectives. ■
Disclosure: Dr. Wilt reported no potential conflicts of interest.
1. Wilt TJ, Brawer MK, Jones KM, et al: Radical prostatectomy versus observation for localized prostate cancer. N Engl J Med 367:203-213, 2012.
2. Thompson IM Jr, Tangen CM: Prostate cancer—uncertainty and a way forward (editorial). N Engl J Med 367:270-271, 2012.
PIVOT is the largest and longest randomized trial conducted in men with early-stage prostate cancer detected during the era of widespread PSA testing. The trial enrolled 731 men and followed them for up to 15 years after randomization. PIVOT results demonstrate that compared to observation, radical ...
I am happy to accept the invitation by The ASCO Post to comment on the recent, long-awaited publication of the PIVOT study (Wilt et al: N Engl J Med 367:203-213, 2012) and the accompanying editorial by Thompson and Tangen (N Engl J Med 367:270-271, 2012). The editorial both points out weaknesses of ...