Background

Despite the success of immune checkpoint inhibitors (ICPI) in treating metastatic cancers, immune-related adverse events remain a major clinical problem. However, the characterisation of baseline risk factors associated with these complications remain scarce. This study interrogates a comprehensive set of accessible clinical parameters in relation to ICPI-induced hepatitis.

Method

We retrospectively identified 69 (10.9%) patients who experienced ICPI-induced hepatitis from 633 patients who received ICPI therapy at Cambridge University Hospitals NHS Foundation Trust between 2014 and 2020.  Of 69 patients, we were able to obtain comprehensive records of 61 from prior to ICPI commencement. We examined associations between age, sex, ICPI regime, site of primary tumour and metastases, BMI, previous oncology therapies, past medical conditions, alcohol and smoking history, in addition to known factors associated with impaired liver function, and routine bloods. We compared this data to randomly identified drug and primary-tumour matched controls (1:1) to identify clinically meaningful associations with ICPI-induced hepatitis.

Results

Age under 50, use of ipilimumab-nivolumab combination therapy and melanoma were associated (all p < 0.05) with ICPI-induced hepatitis. However, multi-variate analysis revealed that only combination therapy was independently associated with hepatitis (OR=3.50, 95%CI 1.92–6.25). In the matched cohort, higher (>31.2 IU/L) baseline serum ALT (OR=4.45, 95%CI 1.16-18.97; HR=2.08, 95%CI 1.06-4.09) and a history of high (>14 units/week) alcohol consumption (OR=5.71, 95%CI 1.30-33.49) were independently associated with ICPI-induced hepatitis, and severity of hepatitis by CTCAE grade. A logistic model of these parameters predicted occurrence of hepatitis in our cohort with a sensitivity and specificity of 0.35 and 0.91. Interestingly, Kaplan-Meier analysis suggested that pre-existing lung metastases is inversely correlated with drug-induced hepatitis, while liver metastases may be associated with increased risk, although these observations were not statistically significant (p = 0.12, p = 0.066 respectively). Other parameters, including measures of liver synthetic function, do not predict risk of hepatitis.

Conclusion

Given these findings, we recommend careful risk-benefit discussions when prescribing combination immunotherapy to patients with elevated baseline ALT and previous high alcohol consumption, with increased vigilance for ICPI-induced hepatitis.

Impact statement

Clinical evidence suggesting pre-existing liver injury may predict risk of drug-induced hepatitis in patients on immune-checkpoint inhibitors, with high specificity.