FAERS analysis flags checkpoint‑inhibitor signals

The new paper is not a clinical trial and it is not a warning from the FDA. It is a map. Researchers mined the FDA Adverse Event Reporting System, or FAERS, to see which side effects show up most often in reports tied to four major checkpoint classes: PD-1, PD-L1, CTLA-4, and LAG-3 inhibitors. They published the analysis in July 2024, after pulling the 25 most common adverse events for each class with the AERSMine pharmacovigilance tool. That matters because checkpoint blockade now spans a long list of cancers, while its toxicities still arrive in patterns that can look obvious only after thousands of patients have already been treated (mdpi.com, fda.gov). FAERS is built for signal detection, not proof. The database collects reports from drugmakers, clinicians, and patients, and the FDA itself says those reports cannot establish that a drug caused an event. Reporting is incomplete. Cases can be duplicated. Sicker patients are more likely to generate reports. But that is exactly why this kind of paper is useful to drug developers and safety teams. It shows where to look harder. It helps define what is expected for a class and what might be unusual for a specific agent or combination (fda.gov, fda.gov). What the authors found is mostly a story of class fingerprints. PD-1 inhibitors were dominated by diarrhea, fatigue, and fever, with hypothyroidism and neutropenia standing out in some drugs. PD-L1 inhibitors also clustered around fever, diarrhea, and fatigue, but interstitial lung disease and hypothyroidism looked more like class-wide signals there. CTLA-4 inhibitors were the blunt instruments of the group. Diarrhea and colitis rose to the top, especially with ipilimumab. That fits the older clinical picture of CTLA-4 blockade as the checkpoint class most tightly linked to inflammatory gut toxicity (mdpi.com, mdpi.com). The LAG-3 result is the one that makes the paper feel current. There is only one FDA-approved LAG-3 regimen in the United States, Opdualag, the fixed-dose combination of relatlimab and nivolumab, approved in March 2022 for unresectable or metastatic melanoma. In the FAERS analysis, relatlimab showed fewer reported events overall, with fever and pneumonia among the recurring signals. That does not mean LAG-3 blockade is inherently safer. It means the evidence base is still thin, because the class is new and the approved exposure is still concentrated in one combination product, in one disease setting, over a much shorter commercial history than PD-1 or CTLA-4 drugs have had (fda.gov, mdpi.com). That thinness is exactly why the paper is more useful as an operations document than as a bedside guide. If you are building an immuno-oncology portfolio, especially combinations that stack checkpoints, you need an “expectedness” framework before the safety database starts to fill up. A sponsor already knows to watch for diarrhea with CTLA-4 and thyroid dysfunction with PD-1 or PD-L1. What this analysis adds is a ranked, class-by-class sketch of which events are common enough to deserve front-row monitoring and which rare events are serious enough to stay on the list anyway. The authors specifically call out myocarditis and myasthenia gravis as uncommon but important across checkpoint inhibitors, the kind of events that may be numerically small and clinically huge (mdpi.com). The most concrete detail in the paper is also the simplest one: after more than a decade of PD-1 and CTLA-4 use, the gut still lights up for CTLA-4, the thyroid still lights up for PD-1 and PD-L1, and the newest class, LAG-3, is still too young for anyone to pretend its safety profile is finished. In pharmacovigilance, that is not a flaw in the analysis. That is the signal. (mdpi.com, fda.gov)