Atallah, Edmond
(2024)
Biomarkers of drug-induced liver injury.
PhD thesis, University of Nottingham.
Abstract
Idiosyncratic drug-induced liver injury (DILI) is an unpredictable clinical event with no available biomarkers; specific phenotypes of DILI have been associated with certain drugs. The research presented in this doctoral thesis sought to evaluate the characteristics of diagnostic and prognostic non-genetic DILI biomarkers and set a direction for future studies to allow the clinical application of biomarkers in practice. First, a systematic review of DILI biomarkers that included 14 studies found that there is a clear need for research in this area and highlighted the necessary steps such as the inclusion of non-DILI control groups, needed to develop and validate diagnostic biomarkers and prognostic models. Two specific types of DILI were also investigated: chronic DILI due to methotrexate (MTX) and acute DILI due to checkpoint inhibitors (CPI).
MTX exposure has been previously associated with chronic DILI and liver fibrosis based on retrospective cohorts with small sample size; a prospective study was needed to inform clinical practice. Results from a multicentre longitudinal cohort study that included 999 patients with rheumatoid arthritis (RA) and psoriasis demonstrated that neither MTX cumulative dose (median 4.8 grams) nor duration of exposure (median of 6 years) was associated with liver fibrosis using two non-invasive markers: liver stiffness and ELF score. Type 2 diabetes showed the strongest independent association with liver stiffness ≥ 7.9 kPa in multivariable analysis (adjusted OR = 3.19, 95% CI 1.95-5.20, p<0.001).
Checkpoint inhibitor-induced liver injury (ChILI) is a new and clinically challenging phenotype of DILI due to the lack of evidence on its frequency, risk factors and underlying mechanism. To determine the frequency, probability and risk factors of ChILI, cancer patients receiving CPI were assessed: a retrospective cohort of all melanoma and renal cancer patients who received CPI over 10 years and a prospective cohort recruited over 30 months (ChILI Study: NCT04476563). Using DILI definitions and formal causality assessment, the risk of ChILI was 8.8% in the retrospective cohort (432 patients) and 9.3% in the prospective cohort (129 patients). The incidence rate of ChILI was 11.5 per 1,000 person-months and the highest probability of ChILI over 1 year was in patients receiving combination CPI (32.1%). Factors associated with higher risk of ChILI were combination therapy, female sex, higher baseline alanine transaminase and lower baseline alkaline phosphatase.
To investigate the mechanism of ChILI and develop a model that differentiates between types of liver injury, peripheral immune cells at the time of liver injury in prospectively recruited patients with acute liver injury adjudicated as ChILI, DILI and autoimmune hepatitis (AIH) in addition to cancer controls who received CPI for 12 weeks without developing ChILI, were analysed. Mass cytometry revealed a unique cluster of CD8+ effector memory T cells with high expression of CD38, HLA-DR and CXCR3 that was significantly increased in ChILI compared to cancer controls after CPI (p=3.26x10-11), DILI (p=1.14x10-5) and AIH (p=0.024). A linear discriminant analysis (LDA) model using 19 inflammatory cytokines classified patients into three types of liver injury. In the LDA model, IL-17A and IL-1 beta were key cytokines differentiating ChILI whereas CXCL8 had the highest coefficient in the LDA distinguishing AIH from the other types of liver injury.
In conclusion, the risk of chronic DILI and liver fibrosis due to MTX is likely to have been previously overestimated in patients with RA and psoriasis who are at higher risk of metabolic syndrome; therefore, current monitoring strategies need to be revised. In contrast, the risk of ChILI is higher than previously estimated in clinical trials. Immune characterisation of the liver tissue in ChILI, functional studies and transcriptomic analysis will inform the role of CD8+CD38+ effector memory T cells in ChILI pathogenesis. The LDA model using cytokine profiling showed promising results in distinguishing between ChILI, DILI and AIH; if the model is validated in an independent cohort, it might provide a useful tool to improve diagnostic certainty and guide management in clinical practice.
Actions (Archive Staff Only)
 |
Edit View |
Tools
Tools
![[thumbnail of Updated thesis with corrections-clean]](https://eprints.nottingham.ac.uk/style/images/fileicons/application_pdf.png "Updated thesis with corrections-clean")