According to an article published this week in NEJM, not all men with prostate cancer require surgery to treat it.
The findings come from the Prostate Cancer Intervention Versus Observation Trial (PIVOT), a study that enrolled 731 patients over the span of seven years (November 1994 to January 2002) at more than 50 sites across the U.S. Trial participants – men with clinically localized prostate cancer detected by prostate specific antigen (PSA) screening – were randomized to either radical prostatectomy or observation. They were then followed for an average of 10 years.
Overall mortality was the primary outcome. Of the men who had been assigned to undergo surgery, 47% died (171 out of 364 patients) by the end of the study; of the men who had been assigned to observation, 50% died (183 out of 367 patients). This mortality difference between the two arms was not statistically significant (95% CI: -4.1 to 10.3; P=0.22), and absolute reductions in mortality were not statistically significant at any interim time interval.
A secondary outcome of the study was prostate cancer-specific mortality. Death attributed to prostate cancer occurred in 5.8% of men randomized to surgery, versus 8.4% of men randomized to observation. This, too, was not a statistically significant difference (95% CI: -1.1 to 6.5; P=0.09).
There were certain groups of patients, however, for whom surgery was associated with a statistically significant mortality reduction as compared to observation. For instance, men with a PSA level above 10 ng/mL who were randomized to surgery had a 13.2% lower overall mortality rate as compared to those randomized to observation (hazard ratio = 0.67; 95% CI: 0.48 to 0.94). The results also suggested that men with intermediate- or high-risk tumors might benefit from surgery, while those with low-risk tumors likely would not. (Tumor risk classification was based on PSA level, Gleason score, and clinical stage.)
So where does this leave us?
Taken at face value, these results suggest that for many men diagnosed with prostate cancer, prostatectomies should not be performed. What’s more, if such a conclusion were valid in the early days of PSA testing when this study was conducted, its relevance is likely to be all the greater today, when PSA testing is more common and rates of prostate cancer detection are higher.
On the other hand, drawing such a conclusion from this study might be premature. In an accompanying editorial, Dr. Ian M. Thompson of the University of Texas Health Science Center and Dr. Catherine M. Tangen of the Fred Hutchinson Cancer Research Center point out that the study was originally designed to enroll 2,000 patients but actually enrolled fewer than 740 patients. They caution that due to the reduced sample size, the study may have been underpowered to detect statistically significant mortality differences between the two study arms. “A 41% increased risk of death on the observation arm cannot be excluded with 95% confidence,” they write, although they acknowledge that an increase of such magnitude would be unlikely given a measured difference of only 3%.
Another concern they raise is the considerable rate of contamination in the study. More than 20% of the patients assigned to the observation arm still received some form of curative therapy, including 10% for whom surgery was attempted.
Seen in this light, Thompson and Tangen imply, what is most remarkable about the study findings isn’t the absence of a statistically significant mortality difference for many patients. It’s the fact that “[d]espite these effects that ought to blunt our ability to discern a treatment effect, in high-risk cancers, a trend toward a reduction in mortality was seen.”
The results of PIVOT represent an important contribution to the literature in that they show that low risk patients may not require treatment. However, if screening and treatment truly do not improve survival, as PIVOT and other studies have suggested, then why has there been a dramatic fall in prostate cancer mortality over the last two decades (a 44% reduction between 1993 and 2009)?
As health care costs continue to rise, and as the male population continues to age, the need to make sense of this paradox – and its implications for the diagnosis and management of prostate cancer – will only grow more urgent. In the meantime, the results of PIVOT have given patients and providers yet another variable to factor into the increasingly complex calculus of costs versus benefits and action versus deliberate inaction.