Apoptotic signatures allow early and rapid screening of drug-induced liver injury to accelerate drug discovery

TL;DR

AEGIS is a new tool that uses gene activity to detect early signs of liver damage from drugs, helping to speed up drug development and improve safety.

Contribution

AEGIS introduces a novel transcriptomics-based approach for early detection of drug-induced liver injury with high accuracy and cross-model applicability.

Findings

AEGIS achieved 86% specificity, 75% sensitivity, and 90% precision in predicting DILI in human liver cells.

AEGIS accurately predicted DILI across species, in vitro models, and therapeutic modalities.

Cells from patients with fatty liver disease showed increased vulnerability to drug-induced liver injury.

Abstract

Early detection of drug-induced liver injury (DILI) during drug development is crucial for reducing drug attrition and ensuring the safety of patients. A versatile, biologically interpretable, and dose-dependent screening approach is therefore needed to inform early stop/go decisions and therapeutic margins.

We have developed AEGIS (Apoptotic Effector Genes In Safety), a preclinical DILI risk screening and prioritization tool that quantifies dose dependent perturbation of apoptosis-regulating transcription factors from transcriptomics data. We profiled transcriptomic responses after short exposures across primary human hepatocytes (PHH), HepG2/C3A cells, RAW 264.7 cells, and an acute Balb/c mouse study. From these profiles, AEGIS provides quantitative risk scores to rank and prioritize compounds and exposures.

Here we show that AEGIS distinguishes compounds with different degree of DILI concern, achieving 86% specificity, 75% sensitivity and 90% precision in PHHs. We demonstrate versatility in data type usage and clinical translation of AEGIS with accurate predictions across species, in vitro and in vivo models, and therapeutic modalities. In addition, we apply AEGIS in a precision medicine context during drug-development within the pharmaceutical industry and investigate the contribution of underlying liver disease on DILI severity. Our findings indicate that cells from patients with metabolic dysfunction-associated steatotic liver disease (MASLD) develop more severe DILI from treatment with troglitazone, aligning with preclinical observations.

Using AEGIS early in drug discovery exemplifies a more efficient approach to identify and mitigate potential safety concerns. This can reduce the need for animal testing, and accelerates drug discovery, ultimately providing the right medicines to patients more quickly.

Some medicines can harm the liver, leading to side effects in people taking the drug, failure of drugs being trialed for use in patients, and delayed new treatments. We developed AEGIS, a screening tool that looks at how cells respond to drugs to spot early warning signs that a drug may injure liver cells. We tested it across human liver cells and animals. It consistently identified drugs that might damage the liver at realistic doses. It also showed that cells from people with fatty liver disease may be more vulnerable to liver injury. Using AEGIS early in development could help remove unsafe medicines sooner, reduce animal use, and bring safer treatments to patients faster.

Hellgren et al. develop AEGIS, a transcriptomics-based tool that quantifies apoptosis-linked gene signatures to assess drug-induced liver injury risk across doses, in vitro models, species, and modalities. AEGIS ranks compounds by liver risk, translates from in vitro to in vivo, and can accelerate safer drug development.

Figures