Epidemiology and characteristics of clinical trials supporting US FDA approval of novel cancer drugs

The US Food and Drug Administration (FDA) approves novel drugs that appear to be effective for their intended uses and whose benefits outweigh their risks. The legal standards for drug approval are widely understood to require evidence of efficacy from two or more clinical trials that independently demonstrate statistically significant effects in favor of the experimental drug and on endpoints that reflect a clinical benefit to patients. However, the FDA has the authority to be flexible in applying the approval standards, particularly in the case of drugs that are intended to treat serious medical conditions. This regulatory flexibility, however, is frequently and repeatedly criticized for putting patients at risk. The two overall aims of this thesis were a) to develop a guide on how to access, retrieve, and use of clinical trial data that supported FDA approval of novel treatments published by the agency itself; and b) to describe characteristics and extent of clinical trial evidence that supported FDA approval of novel drugs for cancer indications between 2000 and 2016. These aims were addressed in three manuscripts, and we describe the methods used to retrieve and manage the clinical trial information in a fourth manuscript. \n \nManuscript 1 “How to use FDA drug approval documents for evidence syntheses”: The FDA publishes information about the clinical trial evidence that supported approval of novel drugs and therapeutic biologics in the drugs@FDA database in form of “drug approval packages”. Information in the main document extends over hundreds of pages and is structured in a way that may be unintuitive for researchers. Although the value of this source of potentially unpublished clinical trial information is undisputed, its use in evidence syntheses of drug interventions remains limited. Based on our experience in using the drugs@FDA database and drug approval packages, we provide step-by-step instructions on how to navigate through the database as well as how to access, efficiently find and retrieve, and use the clinical trial information. Our guide may promote better use of this information, which may improve the completeness and validity of future evidence syntheses of drug interventions. \n \nManuscript 2 “The Comparative Effectiveness of Innovative Treatments for Cancer (CEIT-Cancer) project: rationale and design of the database and the collection of evidence available at approval of novel drugs”: We describe the rationale and efforts made to identify all clinical trials that supported FDA drug approval between 2000 and 2016 for the treatment of cancer and to retrieve pertinent information about trial design and treatment effects on overall survival, progression-free survival, and objective tumor response. Most data retrieval steps were conducted by two data reviewers (who worked independently and who were guided by an instruction manual) to reduce random errors that would affect the quality of the collected data. The study design can be applied in the future for projects with similar scopes, and the collected data will be used in the future for numerous research projects. \n \nManuscript 3 “Clinical trial evidence supporting US FDA approval of novel cancer therapies between 2000 and 2016”: Using the data collected in the CEIT-Cancer project, we analyzed the 127 clinical trials that supported FDA approval of 92 novel drugs for the treatment of 100 cancer indications. The median number of enrolled patients was 193 (interquartile range [IQR]: 106, 448). Fifty-one percent (51%) were randomized controlled, and 75% were open-label. The hazard ratio (HR) for the pooled average treatment effect on overall survival was 0.77 (95% confidence interval [CI]: 0.73, 0.81; I-squared [I2] = 47%), and HR = 0.52 (95% CI: 0.47, 0.57; I2 = 88%) for progression-free survival. The odds ratio for objective tumor response was 3.58 (95% CI: 2.77, 4.62; I2 = 87%). The median absolute survival gain was 2.40 months (IQR: 1.25, 3.89). These findings indicate that novel cancer treatments are supported by trials with design features that have the potential to threaten the validity of the findings and that the overall absolute survival difference is small. \n \nManuscript 4 “Corroborating characteristics of single pivotal trial evidence supporting FDA approval of novel cancer therapies”: For experimental new drugs intended to treat serious conditions, the FDA may grant marketing approvals based on evidence from a single trial alone (instead of two or more) if certain trial characteristics are met. The presence of one or more of these trial characteristics defined by the FDA may increase the FDA’s confidence in the validity of a single clinical trial and may therefore provide corroborating evidence of efficacy. Our results show that 36 out of 100 approvals of novel cancer treatments were based on evidence from a single trial alone. Sixty-four percent (64%) were large and multicentric trials; 64% had consistent effects across different subgroups; 42% had consistent effects across endpoints; and 33% had very low p-values for the primary endpoint. Overall, 92% of clinical trials that supported FDA approval alone fulfilled at least one of the corroborating characteristics. Whether the presence of one or more of these corroborating characteristics indeed provides a safeguard against threats to the validity of trials remains to be answered. \n \nThe background information provided in the manuscripts and this thesis improve the understanding of the regulatory considerations that are made to bring novel cancer treatments to market. The results of the analyses provide insight into the characteristics of the clinical trial evidence and the number of clinical trials that supported drug approval in the fields of oncology and malignant hematology. \n