Shortages in funding, manpower, and willing patients have created the proverbial perfect storm in the current clinical trial system. The fact that traditional clinical trial endpoints in assessing novel agents are being reconsidered only puts more pressure on an already strained system. As the cancer research community navigates the troubled waters of clinical trials, one question is: Should traditional phase II trials be phased out? Richard S. Kaplan, MD, spoke for modifying traditional phase II clinical trials while Alberto Sobrero, MD, defended them during a debate at ECCO/ESMO 2009 in Berlin. Dr. Kaplan, an associate director of the National Cancer Research Network in the UK, argued that traditional phase II clinical trials suffer from unreliable outcomes among other problems. But Dr. Sobrero, director of medical oncology at Ospedale San Martino in Genoa, Italy, countered that phase II trials are worth keeping as long as certain caveats are taken into consideration.
Limited benefits with current phase II set-up
“Traditional phase II trial design is a legacy of the formative years of oncology, where needs and assumptions were different,” Dr. Kaplan said. “Traditional phase II studies need to change to address the fundamental challenges for oncology trials, including more rapid development and testing of new agents, shortening time to patient access, reliable assessment of drug activity, and identification of responding cohorts.”
Dr. Kaplan pointed out that by analyzing the “track record” of phase III trials, the problem with traditional phase II trials becomes clear. Conventional phase II trials are adept at ruling out inactive or toxic monotherapies, he explained. But traditional phase II trials are not good at screening out unpromising combinations and identifying treatments that are more likely to prove effective in a phase III trial.
“Only about 30% of oncology drugs in phase III trials are brought to market, less than half the rate for anti-infectives or cardiovascular agents. And we carry out at least three major phase III trials for every one that is positive,” he said. Other weaknesses of traditional phase II trials: They are not useful for eliminating unpromising therapeutic combinations or estimating the frequency and impact of serious toxicities. Conventional phase II trials also cannot “determine the characteristics of the cohort most likely to respond, especially in the absence of an identified biomarker,” he said.
Traditional phase II trials are inadequate in several ways, Dr. Kaplan explained. Their outcomes are unreliable when they use endpoints other than response rates (RRs), such as progression-free survival (PFS), which is now common in trials of targeted drugs. They often use historical controls that are out of date, and they suffer from selection bias, especially nonrandomized and single-center trials. Finally, conventional phase II trials often avoid intent-to-treat analyses in favor of evaluating “adequately treated patients,” although “patients who drop out early reflect clinical reality,” he said.
Dr. Kaplan called attention to the wide confidence intervals around the observed outcomes in studies that are based on limited numbers of patients recruited from a few centers. “These outcomes are not very easy to generalize; no wonder they do not accurately predict outcomes in conventional phase III trials. They also do not help us prioritize the most promising agents to go forward with in phase III,” he said.
Finally, equivocal or even negative outcomes in traditional phase II trials are often disregarded, and compounds move to phase III regardless, he observed. “Principal investigators and sponsors are sometimes anxious to continue development even when planned outcomes have not been reached,” Dr. Kaplan said.
Dr. Kaplan suggested that phase II trials could be worth keeping if they were modified to speed up the drug development process. In particular, he advocated larger cohorts enrolled from community centers and academic sites, randomization schemes, target validation and inclusion, and testing of biomarkers.
“While this will cost more, the data will be a better predictor of phase III outcomes,” he said.
But it is critical to design these randomized phase II trials so they are not just underpowered phase III trials, he emphasized. They should contain an intermediate endpoint to use as a prospective criterion for expanding to a phase III trial; clear study objectives for the phase II element; and specifications, such as a regulatory endpoint for definitive comparison, for any subsequent phase III trial.
This could be accomplished with a multi-arm, multi-stage design with the principal endpoint of survival, Dr. Kaplan said. An interim analysis of PFS would be performed, allowing for the weakest players to be discarded and the promising compounds to continue to the “mega-trial” with one or two arms, he said.
This multi-arm, multi-stage trial would require fewer patients and take less time to complete because it would be randomized from the start and would evaluate data concurrently and not sequentially (no need to wait for one result before proceeding). Also, there would be no delay between phase II and III. Finally, this type of trial would necessitate fewer grant applications.
“This is an efficient way to test multiple agents at the same time, to use fewer patients, and take less time, but it will take work to organize,” he conceded.