Cutaneous reporting patterns of drugs relevant to hepatocellular carcinoma: a cross-indication FAERS analysis and proposed management framework

Background First-line therapy for unresectable hepatocellular carcinoma (HCC) includes combinations of an immune checkpoint inhibitor (ICI) with an anti-angiogenic agent or tyrosine kinase inhibitor. Cutaneous adverse events are frequent, but immune-mediated eruptions and drug-induced reactions such as hand–foot skin reaction are routinely counted together, distorting the apparent skin toxicity signal and risking inappropriate interruption of the checkpoint inhibitor. Methods In a disproportionality analysis using cross-indication reports from the US FDA Adverse Event Reporting System via openFDA, we computed reporting odds ratios (RORs; signal: ≥3 reports and lower 95% confidence limit >1) for 16 cutaneous reaction search phrases across 9 ICIs and 5 anti-angiogenic/multikinase agents relevant to HCC or used as class comparators. Proportional reporting ratios and information components were sensitivity measures. We examined co-reporting within selected combinations; reports were not restricted to HCC. An independent quarterly data sensitivity analysis compared primary-suspect, primary/secondary-suspect, and any-role coding with and without case version consolidation. Results In the original analysis, vitiligo (ROR up to 40.95 for ipilimumab) and lichen planus generated signals for the six more extensively represented ICIs but not for the five anti-angiogenic/multikinase agents. Hand–foot skin reaction signals emerged for all five anti-angiogenic/multikinase agents (ROR up to 135.40); durvalumab and tremelimumab did not generate ROR signals. Severe reactions (Stevens–Johnson syndrome, toxic epidermal necrolysis) generated signals for several ICIs and some multikinase agents. Of 91 ROR signals, 68 also met both sensitivity criteria. Hand-foot RORs decreased after excluding reports containing selected co-reported medicines (atezolizumab: from 4.19 to 1.82 after excluding bevacizumab-containing reports), suggesting possible confounding by concomitant therapy without establishing drug causality. Camrelizumab and sintilimab had zero reports in the original generic-name query output, and tislelizumab had limited coverage. Quarterly primary-suspect analyses retained all five anti-angiogenic/multikinase hand–foot signals, with agent/event-specific exceptions among ICIs. Conclusions These cross-indication reporting patterns focus attention on concomitant medicines but do not establish HCC-specific risks or individual drug causality. A proposed multidisciplinary monitoring framework with pharmacist involvement combines these observations with external clinical evidence on reactive cutaneous capillary endothelial proliferation, systemic organ assessment, and severe reactions. The framework and monitoring schedule have not been prospectively validated.