Impact of proton pump inhibitor (PPI) pharmacogenomics (PGx) on toxicities associated with immune checkpoint inhibitor (ICI) therapy.

12128 Background: There is evidence for the impact of PPIs on clinical outcomes when given concurrently with ICI therapies. Moreover, CYP2C19 genotype impacts PPI metabolism; poor and intermediate metabolizers having higher PPI effectiveness. Cancer patients routinely receive PPIs for acid suppression; we therefore undertook a PGx analysis to evaluate the impact of PPIs on adverse events (AEs) in patients receiving curative-intent ICI therapy. Methods: We performed a retrospective analysis on data from four Roche studies with atezolizumab (A) (NCT03197935, NCT02486718, NCT02450331, NCT03038100) and included patients who had available germline whole-genome sequencing information. PGx variants of the CYP2C19 gene were assessed using PharmCAT which predicted metabolizer phenotypes- normal, rapid/ultra-rapid, and poor/intermediate. Exposure to PPIs was categorized using the ConMedClassify R package. The primary outcome was time to the first Grade ≥ 3 AE. We used an extended multivariate Cox proportional hazards model that accounted for delayed entry into the PPI exposure class and stratified by study ID. An interaction term was included between PPI exposure, primary cancer treatment (A containing: yes/no), and CYP2C19 metabolizer phenotype. We adjusted for baseline covariates: race, sex, age, BMI, ECOG status, plasma albumin level, and neutrophil-to-lymphocyte ratio. Missing data were imputed with missRanger. Marginal effect estimates were pooled across the completed datasets using the mice R package. Results: This analysis included 816 patients who received A as adjuvant/neoadjuvant therapy in triple negative breast cancer, non-small cell lung cancer, and urothelial cancer, or for stage III/IV ovarian cancer indications; and 775 patients who received non-A based treatment as standard of care across corresponding tumor types . Patients receiving A-containing regimen and having a predicted intermediate/poor metabolizer phenotype [(N = 241 (30%)] were more likely to experience Grade ≥ 3 AEs if they were exposed to a PPI versus if they were exposed to H2 blocker (HR = 3.117, 1.319 - 7.368 95% CI, p-value = 0.01, FDR = 0.172, Ns in contrast groups: 46 vs 89). The analysis to evaluate the impact of PPI PGx on ICI efficacy outcomes is ongoing. Conclusions: Given the widespread use of PPIs in cancer patients receiving ICI therapies, consideration may be needed regarding CYP2C19 genotype data to guide acid suppression therapy. Where possible, prescribing a lower dose of PPI, considering PPIs with less CYP2C19 metabolism, or substituting with H2 blockers in patients receiving curative ICI therapy may minimize the risk of toxicity in patients with intermediate/poor metabolizer status (30% in our cohort from four A trials). This would be especially important in early disease settings where there may be less risk tolerance for toxicity compared with advanced disease.