Assessing the Benefit-Risk for New Drugs

The U.S. Food and Drug Administration (FDA) regulates which drugs can bemarketed in the U.S. as well as their label requirements. A congressionally authorized advisory committee process allows the FDA to receive input from outside experts on important regulatory decisions (1). Further, the advisory committee process enhances transparency by allowing the public, including health care professionals, to understand the issues being considered. The Endocrinologic and Metabolic Drugs Advisory Committee (EMDAC) “reviews and evaluates data concerning the safety and effectiveness of marketed and investigational human drug products for use in the treatment of endocrine and metabolic disorders, andmakes appropriate recommendations to the Commissioner of Food and Drugs” (2). Operationally, the FDA’s Division of Metabolism and Endocrinology Products (DMEP) is responsible for the review of most drugs used for indications relevant to endocrinologists and coordinates the EMDAC’s meetings. In recent years, the EMDAC has been asked to review a number of controversial drugs, particularly those for the treatment of diabetes and obesity. Often the assessment of these drugs requires integration of preclinical and clinical data to formulate a benefit-risk assessment. The assessment of benefit-risk for drugs is increasingly recognized to be a challenging task, and formal tools have been proposed to make this process more effective (3–7). In contrast with the DMEP and other FDA staff, most EMDAC members will have limited experience in these types of integrative, complex evaluations. Further, EMDAC members will spend considerably less time evaluating the available data in comparison with DMEP staff. Thus, it seems important to compare the recommendations made by EMDAC with the subsequent marketing and label decisions to assess concordance between the evaluative approaches used by EMDAC and that of DMEP (DMEP will be used as shorthand for the full FDA’s involvement in regulatory decisions). At most of the EMDAC meetings, the committee is asked to vote on a benefitrisk recommendation or other specific questions. These votes reflect the best overall EMDAC position on the issue and are widely cited in the press as the meeting’s primary outcome. This interpretation is overly simplistic as the committee’s discussions and rationales for its votes may be highly informative to DMEP deliberations and have greater importance than the votes (8). Nonetheless, the vote represents the consensus of the committee members’ deliberations and opinions, as well as their integration of the totality of the information into an overall benefitrisk calculus. To assess the concordance of EMDAC votes and subsequent actions, all EMDAC meetings from 1 January 2009 through 31 December 2012 were reviewed. Only the meetings that involved the review of a specific drug were considered. The recommendations of the committee were then compared with the actual actions taken with respect to marketing or labeling. EMDAC recommendations were considered concordant with subsequent actions when a majority of the EMDAC members’ votes reflected the ultimate action. EMDAC met to discuss specific drugs 20 times during the period of interest (Table 1). In only 12 (60%) of these cases was the ultimate action concordant with the EMDAC vote. In two cases, concordance could not be assessed because of the complexity of the questions or actions considered. In two cases where lack of concordance was identified, a change in one vote would have resulted in a majority agreement with the action. In most cases of EMDAC-DMEP discordance, EMDAC members seemed to view the benefit-risk of the candidate more favorably. In the examples for miglustat, naltrexone/bupropion, and insulin degludec, clear majorities of EMDAC members favored approval while the FDA did not agree. In the case of sibutramine, a majority of EMDAC members did not support withdrawal from the market, but the drug was withdrawn soon after the meeting. In contrast, in the case of liraglutide, a majority of the EMDAC members did not support approval, but the drug was approved 9months after the meeting. Importantly, lack of concordance between the EMDAC members and subsequent actions should not be interpreted as either EMDAC or FDA action being correct or incorrect. In fact, the action taken may be in full agreement with key points elicited during the committee’s discussions but not in the polling of the individual members. Alternatively, the FDA’s decision may have reflected considerations or data not available to the committee at the time of the vote. For example, if committeemembers indicated that certain additional information would change their votes, and this information is available to the FDA, the FDA’s action may be concordant with the committee’s recommendations but not their dichotomous vote. Rather, the relatively low degree of concordance should be viewed as reinforcing the challenges faced by the EMDAC. The discordance also raises the question as to whether EMDAC members are prepared to make a “yes” or “no” vote on issues as complex as benefit-risk assessment based on only a few hours of readings and deliberations. Discordance also suggests that the weighting of benefits and risks by committee members may differ from those of the DMEP and/or that they are being incompletely communicated during the meeting. The challenges the EMDAC faces in its deliberations are also illustrated in