(037) Is aggregate use of published case reports as a viable option for augmenting pharmacovigilance? An applied example of pembrolizumab-related adverse events in non-small cell lung cancer patients receiving first-line treatment
Background: Pharmacovigilance suffers from considerable underreporting, and causality is rarely confirmed. Case reports provide a rich source of adverse events, particularly rare and unexpected events. We explore the setting of first line pembrolizumab treatment for non-small cell lung cancer (NSCLC) as safety signals in this cohort are relatively well documented.
Objectives: To compare adverse event rates between a cohort of published case reports and three clinical trials of first line pembrolizumab treatment for NSCLC. To examine the validity of the aggregate use of published case reports as a viable option for augmenting pharmacovigilance.
Methods: We compared immune-related adverse event rates (Any and Severe) associated with pembrolizumab from three select first line NSCLC KEYNOTE clinical trials (024, 042, and 189) to clinician confirmed pembrolizumab-related adverse event rates from a synthetic cohort published case report data from a matching population. The latter were retrieved from the OpenCase(TM) database
Results: The synthetic cohort had a sample size of n=126, and the pembrolizumab groups in the three trials had sample sizes of n=150, n=636 and n=410 for KEYNOTE-024, -042, and -189, respectively. In the synthetic cohort, 110 patients had received pembrolizumab monotherapy and 16 pembrolizumab in combination with a platinum-based chemotherapy regimen, and 116 patients had locally advanced or metastatic NSCLC. Baseline characteristics with respect to age, sex, smoking history, non-drug treatment history, PD-L1 tumor proportion score, and key genetic markers (e.g. EGFR, ALK) were acceptably similar (results will be presented in a table). In the synthetic cohort of case reports, life threatening and severe treatment-related AEs were 7.4% and 65% respectively. These were substantially higher than reported in clinical trials 8.0%-9.7% experienced severe immune-related events. The majority of the most frequently occurring immune-related AEs in the KEYNOTE trials had similar reported proportions to the case reports. For example, hypothyroidism (any grade) ranged from 6.7% to 9.1% in the KEYNOTE trials compared to 4.9% among the case reports, and pneumonitis (any grade) ranged from 4.4% to 12.0% in the trials compared to 5.5% in the case reports. Other events such as myositis, type 1 diabetes, liver dysfunction, kidney injury, and colitis were either substantially higher or solely reported across case reports.
Conclusions: Overall, adverse event rates in the synthetic cohort of case reports were either similar or higher than observed in clinical trials of matching populations. This observation provides rationale for utilizing case reports in pharmacovigilance practice, specifically for rarer events.
