ASCO 2011—Clearly, the realities of current health care economics are bringing to the fore important questions in clinical trial design. Not only does the United States spend far more on health care than other developed countries, with significantly worse outcomes, but the costs of cancer care account for a large and steadily increasing chunk of that spending. One way in which this trend can be curbed—and the money spent on cancer be made to produce more “bang for the buck”—is through changes to our drug development processes.
Drs. Eileen Mary O’Reilly, Neal J. Meropol, and Donna Niedzwiedki reviewed recent phase III studies in upper gastrointestinal malignancies (gastric and gastro-esophageal junction cancer, pancreatic neuroendocrine tumors [PNETs], pancreatic adenocarcinoma, hepatocellular carcinoma)—with an eye to what can be learned about ways in which future clinical trials can be designed so as to yield a greater wealth of truly useful clinical information.
The chief trials looked at included:
•The ToGA trial of trastuzumab(Drug information on trastuzumab) plus chemotherapy vs chemotherapy alone for HER2-positive advanced gastric and gastro-esophageal cancer.
•The RADIANT-3 trial of everolimus vs placebo in PNETs.
•The SUN IIII trial of sunitinib vs placebo in PNETs.
•The NCIC PA.3 trial of gemcitabine(Drug information on gemcitabine) /erlotinib vs gemcitabine/placebo in pancreatic adenocarcinoma.
Dr. Niedzwiecki pointed out that phase III trials are supposed to be randomized, comparative—and traditionally definitive. The above trials all meet the first two of these criteria, but whether they were definitive is less clear. The trastuzumab trial was clearly practice-changing: there was a 2.8-month improvement in overall survival (OS), with similar levels of toxicity. The gemcitabine/erlotinib trial produced statistically significant results (there was a 0.33-month improvement in OS that slightly favored the experimental arm), but the clinical significance of the results has been hotly debated. However, with the everolimus and sunitinib trials, the definitiveness of the results is not yet known.
Since these agents are all palliative, all expensive, and all have added toxicity, it is important to ask whether we can afford to continue to do trials such as these in which results fall short of definitive.
Dr. Meropol suggested several goals for future trial design that would help optimize increasingly scarce resources. A key goal, clearly, is to raise the bar for approval. This would lessen the likelihood of less-than-definitive results. He also proposed that cost-effectiveness data be discussed along with more traditional trial data—published at the same time, although perhaps in a separate article. He proposed using the incremental cost-effectiveness ratio (ICER) as a measure of cost-effectiveness. ICER data on the agents tested in the above trials, for example, definitely throw a new light on the trial results. According to a recent survey, 70% of oncologists think that a cost of $100,000 per year or less represents good value. Using this benchmark, trastuzumab for patients with gastric cancer (according to ICER data) would represent good value—although not quite as high a value as trastuzumab for patients with breast cancer. However, because everolimus costs twice as much as sunitinib, ICER data on these two drugs show a marked difference in their value.
Dr. Niedzwiecki pointed out that in a number of these trials, a large percentage of patients crossed over. Consequently, she stressed the importance, in future trials, of incorporating interim monitoring. This would have real potential to save money. She also proposed that researchers consider using estimation methods that allow for early stopping (although this would involve ethical considerations).
The bottom line, though, as both Dr. Meropol and Dr. Niedzwiecki stressed, is that we need to discuss how to better set the bar in clinical trials.