At the heart of ICI function lies the blockade of key immune checkpoints—specifically, CTLA-4 and PD-1/PD-L1—that normally serve to regulate immune responses and maintain self-tolerance. By inhibiting these checkpoints, ICIs restore T cell activity against cancer cells but simultaneously perturb immune homeostasis, leading to off-target immune activation against normal tissues such as the liver. The immunopathology of IMH involves multiple intertwined pathways: activated cytotoxic CD8⁺ T cells directly target hepatocytes, while helper T cell subsets including Th1 and Th17 expand, and regulatory T cells (Tregs) are impaired. Concurrently, B cell activation promotes autoantibody production, exacerbating the hepatic immune assault. The local milieu becomes saturated with inflammatory cytokines such as interferon-γ, tumor necrosis factor-α, and interleukins (IL-1β, IL-6), which, alongside inflammasome activation, recruit monocytes and macrophages to the liver. This constellation of immune dysfunction culminates in the characteristic liver injury seen in IMH.

Clinically, IMH spans a broad spectrum of severity, from asymptomatic elevations of liver enzymes—ALT, AST, and alkaline phosphatase—to fulminant liver failure. The incidence varies widely in clinical reports, documented between 1 and 15 percent, yet the morbidity is profound, as IMH was implicated in over 20 percent of fatalities related to ICI therapies. Current management is largely reactive, relying on cessation of ICIs and administration of systemic glucocorticoids once liver injury manifests, underscoring a critical need for preemptive strategies based on reliable predictive markers.

Emerging evidence has identified multiple clinical risk factors that heighten the propensity for IMH. Female patients appear predisposed, with studies indicating more than double the odds compared to males, potentially due to inherent differences in immune regulation or hormonal milieu. Younger age cohorts display increased vulnerability, as does Asian ethnicity, particularly in populations from China, Japan, and Korea. Comorbid chronic liver conditions such as hepatitis B virus (HBV) infection and nonalcoholic fatty liver disease (NAFLD) significantly amplify risk, with NAFLD-associated hazard ratios soaring upward of 29 in some analyses. Baseline liver function abnormalities—elevated AST/ALT and diminished ALP—also herald impending hepatotoxic complications.

Tumor biology further modulates risk, with certain cancers exhibiting stronger associations. Melanoma, for instance, carries an odds ratio exceeding 11 for IMH development, while hepatocellular carcinoma (HCC), biliary tract, and gastric cancers also contribute to elevated incidence rates. Notably, the role of liver metastases remains contentious, with conflicting data on whether their presence exacerbates hepatotoxic risk or confers protective effects through immunosuppressive tumor environments.

Pharmacologic factors prove equally consequential. Dual ICI regimens, combining agents like anti-CTLA-4 with anti-PD-1/PD-L1 antibodies, dramatically escalate hepatotoxicity risk with odds ratios approaching 11. Prior exposure to ICIs augments susceptibility threefold, and higher doses of ipilimumab—common in dual-therapy protocols—worsen outcomes. Furthermore, concurrent administration of hepatotoxic medications including acetaminophen and statins compounds liver injury risk, highlighting the intricate interplay between therapeutic interventions.

The quest for predictive biomarkers has yielded a spectrum of candidates spanning immunologic, genetic, and microbial domains, yet none currently achieves clinical reliability. Baseline peripheral blood eosinophil count exceeding 130 cells per microliter correlates with a threefold increase in IMH hazard, while dynamic shifts in lymphocyte populations and the neutrophil-to-lymphocyte ratio (NLR) have shown variable predictive value contingent on tumor type and treatment stage. Reduction of CD4⁺/CD8⁺ T and B cells during therapy, alongside expansion of proliferative Ki67⁺ CD8⁺ T cells, are features noted in severe IMH, suggesting immune cell kinetics hold key insights into hepatotoxic pathogenesis.

Serum protein markers add another dimension, with elevated C-reactive protein (CRP) levels at hepatotoxic onset—specifically above 8.2 mg/L—forecasting grade 3 or higher liver injury. Composite scores such as CRAFTY, which combines CRP and alpha-fetoprotein (AFP), show promise particularly in HCC patient subsets. Conversely, hypoalbuminemia portends worse hepatic outcomes, reflecting systemic inflammatory burden and nutritional status.

Autoantibodies provide a window into autoimmune activation during IMH. Antinuclear antibody (ANA) positivity, particularly among pembrolizumab-treated patients, increases IMH odds nearly eightfold. Thyroid peroxidase antibodies (TPOAb) also surface as relevant markers, though liver-specific autoantibodies—anti-smooth muscle, antimitochondrial, and liver kidney microsomal antibodies—lack predictive consistency.

Cytokine profiling reveals an inflammatory signature centered on IL-1β, IL-6, CXCL9/10/11, IL-18, and TNF-α, all correlating with hepatotoxic severity. The innovative CYTOX score, an 11-cytokine panel, exhibits potential in identifying patients at risk for severe immune-related adverse events, predicating the necessity for longitudinal cytokine tracking during therapy.

Genomic studies have uncovered single nucleotide polymorphisms (SNPs) within genes such as EDIL3, SEMA5A, GABRP, SLCO1B1, SMAD3, and PACRG, which confer two- to ninefold increased risk of IMH. Copy number variations in immune-regulatory loci including CD274 (encoding PD-L1) further implicate a genetic predisposition. Additionally, human leukocyte antigen (HLA) typing has revealed associations between alleles like HLA-A*26:01, HLA-DR4, and HLA-B27:04 with hepatotoxicity risk, though findings remain inconsistent across cohorts.

Beyond genetic and immunological markers, the gut microbiome emerges as a frontier for IMH prediction. The abundance of Veillonella species correlates with hepatotoxicity, highlighting the gut-liver axis as a potential biomarker reservoir and therapeutic target. Microbial metabolites may modulate systemic and hepatic immunity, influencing ICI response and toxicity.

Despite these advances, the landscape lacks a definitive biomarker or predictive algorithm robust enough to inform personalized clinical decisions preemptively. Most existing studies suffer from retrospective design, heterogeneous patient populations, and lack longitudinal multi-omic integration, hindering validation and broad application. The way forward necessitates well-powered, prospective, multicenter cohorts employing integrative platforms combining clinical parameters, immunophenotyping, cytokine arrays, genetic screening, and metagenomic profiling. Harnessing artificial intelligence and machine learning approaches will be crucial for synthesizing these complex datasets into actionable risk stratification models.

In conclusion, immune checkpoint inhibitor-mediated hepatotoxicity is a multifactorial immune adverse event influenced by patient demographics, underlying liver disease, tumor characteristics, and therapeutic regimens. While promising biomarkers span circulating immune cells, serum proteins, autoantibodies, cytokines, genetics, and microbiota, no single indicator currently suffices for clinical prediction. Future research must embrace comprehensive, longitudinal, and integrative strategies to transcend current knowledge gaps, ultimately enabling precision medicine approaches that safeguard patient safety while harnessing the transformative power of cancer immunotherapy.

Subject of Research: Immune checkpoint inhibitor-mediated hepatotoxicity, risk factors, and predictive biomarkers
Article Title: Risk Factors and Biomarkers for Immune Checkpoint Inhibitor-mediated Hepatotoxicity: Emerging Insights and Future Perspectives
News Publication Date: 23-Jan-2026
Web References: https://doi.org/10.14218/JCTH.2025.00622
References: Published in the Journal of Clinical and Translational Hepatology, DOI: 10.14218/JCTH.2025.00622
Image Credits: Chengliang Zhang, Yingjie Hu