Eric M Mukherjee, Amir Asiaee, Dodi Park, Matthew S Krantz, Cosby A Stone, Michelle Martin-Pozo, Elizabeth Jane Phillips
Checkpoint blockade was associated not simply with additional toxicity, but with a change in treatment context: drug-phenotype associations shifted in both directions, adverse events formed cross-organ structure and documented onset varied by checkpoint pathway. These findings support ICIs as modifiers of drug toxicity and identify specific signals for longitudinal and mechanistic validation.
BACKGROUND: Immune checkpoint inhibitors (ICIs) are usually treated as direct culprits in immune-related adverse events, but checkpoint blockade may also reset tolerance to other medications. We tested whether ICI exposure reshapes the organization, drug specificity and timing of reported treatment toxicity.
METHODS: We analyzed 13,701,106 deduplicated FDA Adverse Event Reporting System reports from 2016 through 2025. Cancer-restricted reporting associations and cross-organ community detection characterized the ICI-associated toxicity landscape. Adjusted logistic models tested primary-suspect drug x ICI interactions for Stevens-Johnson syndrome/toxic epidermal necrolysis (SJS/TEN), drug reaction with eosinophilia and systemic symptoms (DRESS), acute generalized exanthematous pustulosis, interstitial nephritis, drug-induced liver injury, anaphylaxis and vomiting. Accelerated failure-time models evaluated documented onset according to ICI exposure and checkpoint pathway.
RESULTS: Among 2,365,278 cancer-associated reports, 256,940 contained an ICI. Of 3001 eligible Preferred Terms, 2091 differed between ICI-containing and non-ICI reports at false discovery rate (FDR) <0.05, and four cross-organ toxicity communities emerged. Drug-phenotype associations were reweighted: 71 of 147 eligible pairs had FDR-significant interactions, including 56 amplifications and 15 attenuations. Signals included marked moxifloxacin-SJS/TEN amplification (interaction OR 100.51, 95% CI 38.58 to 261.84), a submultiplicative enfortumab vedotin-SJS/TEN interaction (0.17, 0.13 to 0.23) and omeprazole-interstitial nephritis amplification (10.33, 7.60 to 14.04). Among 64,593 reports with documented onset of 1-365 days, ICI exposure was associated with longer adjusted onset for five of seven phenotypes (time ratios 1.37-1.58); 10 of 19 estimable checkpoint-phenotype coefficients remained FDR significant.
CONCLUSIONS: Checkpoint blockade was associated not simply with additional toxicity, but with a change in treatment context: drug-phenotype associations shifted in both directions, adverse events formed cross-organ structure and documented onset varied by checkpoint pathway. These findings support ICIs as modifiers of drug toxicity and identify specific signals for longitudinal and mechanistic validation.