Predicting the likelihood of elevated transaminases in lung cancer patients undergoing immune checkpoint blockade

Abstract

ABSTRACTBackgroundThe recent implementation of immune-checkpoint inhibitors (ICIs) has markedly changed the management and clinical outcomes for patients with advanced non-small cell lung cancer (NSCLC). Despite higher efficacy, ICIs are associated with a range of immune-related adverse events (irAEs). This retrospective study aimed to investigate the incidence of ICI-related transaminitis/hepatitis in NSCLC patients as well as to establish if any pre-treatment clinical parameters can predict the onset of liver toxicity.MethodsWe examined medical records of N=420 NSCLC patients treated across two health districts in Sydney, Australia between 2016 to 2020. R packages (corrplot and ggcorrplot) were used to construct correlation matrices and to calculate the correlation p-value using Spearman method. Logistic regression models were used to determine association of clinical parameters with elevated LFTs.ResultsN=185 patients were considered eligible. n=37 (20%) had elevation of liver transaminases at any stage post-ICI commencement, although only n=10 were deemed as those having ICI-related hepatitis. Most of these patients (n=29 [78%]) developed elevated LFTs within 3 months post-therapy initiation. Regression model established that pre-treatment ratio of serum protein (SP) to total bilirubin (TB) showed significant association with the elevated transaminases. Moreover, most of the patients (n=34 [94%]) with elevated LFTs had SP/TB<4. Using a second cohort of melanoma patients, the linear regression model did not establish a significant association between the SP/TB ratio and elevated LFTs. Nonetheless, n=31 (70.5%) patients with immune-mediated transaminitis in the melanoma cohort had SP/TB<4.ConclusionOur study has established a clinical risk factor associated with the elevation of LFTs in NSCLC patients, thus potentially enabling their prediction at the pre-treatment stage. However, its predictive accuracy was not confirmed in melanoma patients, stressing that ‘one size does not fit it all’ when developing predictive scores for irAEs in different patient cohorts.