Androgen-deprivation therapy, given for either 36 or 28 months, has been shown in two independent randomized clinical trials to prolong survival, relative to six or four months, respectively, in men with localized high-risk prostate cancer. However, the benefit comes with substantial side effects that can diminish quality of life. Therefore, investigators at Universitaire de Sherbrooke in Canada and colleagues designed a randomized controlled trial to test whether a shorter duration of therapy, 18 months, would have effects similar to those for 36 months in terms of overall survival.
The results of the trial, though, as reported at the Genitourinary Cancers Symposium ( Abstract 3 ), are insufficient to guide practice, according to experts.
“The bottom line is that the data are not ready to change management because the trial design was superiority; it needs to be non-inferiority, because 36 months is the standard of care and 18 months is less,” explained Anthony D'Amico, MD, PhD, Chair of the Division of Genitourinary Radiation Oncology at Dana-Farber Cancer Institute and Professor of Radiation Oncology at Harvard Medical School, the Discussant for the abstract.
“Based on this trial, we cannot tell with certainty whether or not 18 months is as good as 36,” he told OT. “What we do know is that if you give 18 months today, [your patients] run up to a 59 percent increased risk of dying; the average increased risk is 15 percent. If people are willing to take that risk, and doctors are willing to inform them of that increased risk, that's fine. Otherwise people should stay with the longer course.”
The researchers, led by Abdenour Nabid, MD, Associate Professor at Centre Hospitalier Universitaire de Sherbrooke, enrolled 630 men with high-risk localized prostate cancer between October 2000 and January 2008. To be eligible for the trial, patients had to have at least one of the following risk factors: PSA score of 20 ng/ml or higher, Gleason score above 7, or stage T3-T4 disease.
All patients received standard radiotherapy, which began four months after starting hormone therapy. The investigators randomly assigned patients to receive 50 mg of bicalutamide for one month plus 10.8 mg of goserelin every three months for 18 or 36 months.
With a median follow-up of 77 months, he reported that there was no significant difference in overall survival at 76.2 months in the 18-month group and 77.1 months in the 36-month group. Landmark analyses also appeared similar, at five years (86.8% vs. 92.1%; P = 0.052) and at 10 years (63.2% vs. 62.6%; P = 0.429). Disease-specific survival also appeared similar at five years (97.4% vs. 97.6%; P = 0.473) and at 10 years (87.2% vs. 87.2%; P = 0.838).
More Events Required to Reach Reasonable Risk
While the data look similar, the problem is that the trial analysis is not sufficient to show that 18 months is not inferior to 36 months, D'Amico said. To prove that one regimen is not inferior to another, one needs substantially more events than are required to show that one regimen is superior to the another. Usually that criteria is met by increasing the number of patients enrolled in the trial.
However, given that this trial is already ongoing, D'Amico and Wen Ling Xie, a biostatistician at Dana-Farber, calculated that the Canadian team could use longer follow-up to reach an adequate number of events for an acceptable non-inferiority boundary. If they do that they would have enough statistical power to prove that the shorter regimen was not significantly worse than the standard.
Thus far, 147 patients in the trial have died and that gives them a 95 percent confidence interval with an upper limit of 1.59, meaning there is as much as a 59 percent increased risk of death with the shorter regimen. If the researchers wait until there are 275 events—about five to seven years according to the Boston group's estimate—the upper limit would be under 1.35. That was the non-inferiority boundary planned in the EORTC trial that compared 36 versus six months of hormone therapy, and thus should work here too, D'Amico said.
He noted that although such ad hoc analyses do not strictly fit into evidence-based medicine, similar methods have been used and accepted by the community in the past. For example, EORTC investigators designed the 36-versus-six-month trial as a non-inferiority trial, but when the data did not meet that standard, the team reanalyzed the data as a superiority trial. The community—and the New England Journal of Medicine—accepted that approach.
“I think [Nabid et al] can do the same thing here,” D'Amico said. “They can do a post-hoc non-inferiority assessment once they get the adequate number of deaths to make the assessments. What they shouldn't do is publish it now, since it is speculative, not conclusive. No one will accept the non-inferiority hypothesis with an upper limit of 1.59—that's just too high a risk to take.”
Two-Year Regimen at Least a Start
Nabid initially presented his data as conclusive, but after D'Amico's discussion showing the critical requirement to show non-inferiority, he said his team will continue following the patients to get an adequate number of events.
In the meantime, however, he remains convinced that the shorter duration is better, given the substantial side effects of therapy. “I think all the numbers we brought really favor the reduction of duration of androgen blockade,” he told OT. “Really, there is no difference at all. You can look at it from any direction; there is no difference.”
He emphasized that these are very high-risk patients, with nearly 60 percent having a Gleason score above 7, and 20 percent having a score of 9 or 10. He also emphasized the timing of the trial, noting that the researchers started enrolling patients 13 years ago and have a median follow-up beyond six years.
“We are going to follow these patients, but in my mind the results represented will be validated—more validated—with time,” he said.
“So my recommendations for doctors is, if you look at guidelines, NCCN for example, patients should have 24 to 36 months [of hormone therapy]. As a first step, my recommendation would be to consider only 24 months for androgen blockade. Then, when we present the data, they can decide if they want to reduce it to 18 months.”
D'Amico, though, is substantially more cautious. First, he notes that the 24 month number actually comes from the RTOG trial that compared 28 with four months, so 24 is already shorter than trial data support. “It may very well be that 18 months can replace 36, but we are not ready to tell that with the data in its present form,” he said.
Moreover, without those additional events the trial outcome can't be known today. “We don't know that it will turn out that way [in favor of shorter duration]. It could turn out that the risk of death is much higher than the tradeoff [in reduced side effects].”
He also points out that even a non-inferiority analysis will not show that the two regimens are actually equal. “What ‘not inferior’ means is that it is not as good, but it is not bad enough that we would say, given the increased toxicity of longer treatment, that the tradeoff is not worth it. Right now it is too much risk to take for dropping the amount of treatment and reducing the amount of toxicity. Fifty-nine percent is just too much risk.”
The meeting is co-sponsored by the American Society of Clinical Oncology, the American Society for Radiation Oncology, and the Society of Urologic Oncology.
How the General Media Got It Wrong
Numerous major news outlets, as well as medical news sites, reported the study as Nabid interpreted it, as showing that 18 months is equivalent to 36 months.
How did that happen? The communications committee for the Symposium selected the abstract for presentation in a pre-meeting news conference. However, none of those committee members appear to have caught the problem with the trial design noted by D'Amico, who was scheduled to present the formal discussion of the paper during the meeting itself.
Given the widespread media coverage, many patients likely heard the incorrect—or at least incomplete—interpretation of the data. Some of them will likely bring questions to their oncologists about shortening their course of therapy.
“I'm sorry there is confusion out there. It is a shame,” D'Amico said, when asked about the media reports. “I'll be talking to a lot of media today about it.”
For physicians confronted with such questions, he said, well, they need to tell the truth, which is that there was a discussion that followed the talk that put it into perspective.
“People want to give shorter courses, they don't want to give that much hormonal therapy, but that is the way the data stand right now.”