Well, yes; it's easy to avoid confoundment of OS. Don't have it as an endpoint. After all, AA based on a surrogate that is 'reasonably likely' to do something, is a pretty low bar! Especially if identifying toxicities prior to approval is optional (as has been the case in all ICI approvals). You can then have maybe a decade of unfettered access to the market before all toxicities become apparent, and it is discovered that that the 'reasonably likely' benefits don't exist!
So why go to all the bother of testing OS? You can save yourself a heap of cash and several years of trial duration, by demonstrating something that is reasonably likely to do something that may be of benefit to patients, but usually doesn't.. And forget about bothersome crossover confoundment.
As you say, big trials also help your cause. If you have 10,000 patients, you only need to demonstrate PFS advantage of 6.73 days to have stat sig! (OK, a slight bit of poetic licence employed here..)
Most contemporary approvals of new cancer drugs are made on the basis of a surrogate end point, such as response rate or progression-free survival (PFS).1 When the approval is based on a surrogate end point, subsequent studies are advised and often obligated to clarify the drug’s effect on overall survival. One such drug is bevacizumab, which received accelerated approval on the basis of PFS for patients with metastatic breast cancer. Later findings revealed no improvement in overall survival and significant toxicity, which required a removal of marketing authorization.2
A 2009 Government Accountability Office report criticized the US Food and Drug Administration (FDA) for failing to enforce postmarketing study commitments for surrogate approvals. Among the more than 400 postmarketing studies requested, approximately 30% were pending, ongoing, delayed, or terminated years later, yet the FDA never exercised its authority to remove a product from the market.3 For these reasons, we sought to investigate how often cancer drugs are approved based on a surrogate end point, whether subsequent studies for these drugs are reported, and whether the drugs improve overall survival. Methods
We examined all marketing approvals by the FDA from January 1, 2008, through December 31, 2012. We identified the pathway for approval (accelerated vs traditional) and the surrogate end point used, such as tumor response rate or PFS. This investigation of published reports was exempt from institutional review board approval.
For all drugs approved on the basis of a surrogate end point, we performed a systematic search of the published literature using Google Scholar as of August 22, 2015, and identified any subsequent reports of the drug’s effect on overall survival. We credited a drug for improving overall survival if that drug improved survival as the sole investigational agent in any combination or in any line of treatment (eg, if approved for second-line treatment of metastatic disease, but the drug improved survival in first-line treatment, we would credit the drug as improving survival). We identified whether crossover (from the control arm to the investigational agent) was used in the randomized clinical trial or via a postprotocol expansion study. We analyzed the study data from August 22 to September 1, 2015. Results
We identified 54 approvals made during our search period, with 36 drugs (67%) approved on the basis of a surrogate end point. Figure 1 shows all surrogate approvals, the efficacy end point at the time of approval, and the regulatory pathway. Approval was granted on the basis of a surrogate for all 15 accelerated approvals (100%) and 21 of 39 traditional approvals (54%). Rate of response, measured by a reduction in tumor size or volume, was the primary measure of efficacy for 19 of 36 surrogate-based approvals (53%), whereas PFS or disease-free survival was cited as the basis of 17 of 36 approvals (47%).
With a median follow-up of 4.4 years, 5 drugs were subsequently shown to improve overall survival in randomized studies (in 1 of 15 accelerated approvals and in 4 of 21 traditional approvals), 18 drugs failed to improve overall survival (in 6 of 15 accelerated approvals and in 12 of 21 traditional approvals) as primary or secondary outcomes, and 13 drugs continue to have unknown survival effects, meaning they remain untested or they have no reported survival results as primary or secondary outcome (in 8 of 15 accelerated approvals and in 5 of 21 traditional approvals). Figure 2 compares the percentage of approved drugs with known and unknown effects on overall survival based on our systematic review of subsequent literature. The use of crossover occurred in 11 of 36 trials (31%) and did not differ among trials that found a survival advantage vs those that did not (1 of 5 [20%] vs 10 of 18 [55%]; P?=?.16). Discussion
During our study period, 36 of 54 contemporary cancer drug approvals (67%) were made on the basis of a surrogate end point. With several years of follow-up, 31 (86%) of these approvals (57% of the 54 drugs approved) have unknown effects on overall survival or fail to show gains in survival. Our results show that most cancer drug approvals have not been shown to, or do not, improve clinically relevant end points.
Since 2008, the FDA has approved a higher percentage of drugs than previously,4 and cancer drugs are approved on the basis of surrogates that have poor correlations with overall survival.2 Our results suggest that the FDA may be approving many costly, toxic drugs that do not improve overall survival. Enforcement of postmarketing studies is therefore of critical importance.
AI
