(160) Optimizing Clinicaltrials.gov as a Tool to Evaluate the Potential of Real World (RW) Data Sources for Use as External Comparator Arms

Publication Number: 491

Background: Randomized clinical trials are the gold standard in evaluating clinical evidence for efficacy and safety of drug treatments at time of marketing approval. Where randomization is not feasible or unethical, the use of an external data source as a comparator is increasingly used. One challenge with an external comparator is identifying a data source that adequately reflects the target trial population. Clinicaltrials.gov is a data source that can be readily used to evaluate the ability of the RW data source to approximate a potential target trial.

Objectives: To evaluate the generalizability of potential external comparator data sources to the characteristics of patients enrolled in clinical trials evaluating chimeric antigen receptor T-cell (CART) therapy for Diffuse Large B-Cell Lymphoma (DLBCL) using data aggregated through clinicaltrials.gov.

Methods: We compared baseline characteristics of DLBCL patients enrolled in trials with CART interventions reported through clinicaltrials.gov to RW data sources: the nationwide (US-based) de-identified electronic health record (EHR)-derived Flatiron Health database and Optum. Trials matching the criteria were extracted from clinicaltrials.gov, along with additional metadata to enable further analysis, combination, and summarization of the subject baseline characteristics. Characteristics using standards-based, categorical taxonomies were combined and non-standard taxonomies were included where possible to accurately represent the data.

DLBCL patients treated with available CART treatments were identified in both Flatiron and Optum using ICD-10 codes and a record of a CART treatment in the patient’s clinical record. Baseline characteristics were described using summary statistics and evaluated for clinically meaningful differences.

Results: Six clinical trials of CART treatments for DLBCL were identified in the clinicaltrials.gov data base totaling 454 patients. 91 CART treated DLBCL patients were identified in the Flatiron and 109 in Optum. More RW patients were age 65 and older (43 - 44%%) than trials (0% - 30%). Most patients in both RW and trials were white and male. One trial reported both lymphoma type and line of therapy. One other study reported disease histology. Similar to RW patients, most trial patients had DLBCL subtypes and had 2 or more lines of therapy.

Conclusions: Clinicaltrials.gov provides an opportunity to evaluate potential RW data sources to be used as external comparators. Results showed differences and similarities in patient characteristics between trials and RW DLBCL patients. Variation in clinical characteristics reported across trials were observed. This study was limited to DLBCL patients receiving CART therapy. Further assessment in other disease areas is needed.