Canonical text (the exact bytes fingerprinted)

TRUVACE RECORD VERSION
record: TRV-2026-1295
version: 1
kind: certified
reason: Certified into the record
timestamp: 2026-10-06T06:56:13.694306Z
status: published
lens: g_space
sector: health
headline: A Multiparametric in vitro Strategy for Small Molecule Drug-Induced Liver Injury Risk Mitigation in Early Drug Development
dek: Drug-induced liver injury (DILI) remains a major cause of preclinical and clinical attrition, reflecting persistent gaps in the predictive translation of preclinical safety data to human outcomes. To address these gaps, we established a refined multiparametric framework for early DILI risk prediction using a balanced reference set of 170 drugs with established clinical outcomes. Thirty physicochemical, mechanistic endpoints and exposure-related features were systematically evaluated to identify high-specificity…
gain_title: Random Forest and flag-based models integrating exposure, lipophilicity, mitotoxicity and cytotoxicity improved early DILI risk prediction with high specificity, supporting safer molecule triage to reduce hepatotoxicity-related attrition.
problem_title: (none)
trace_subject: (none)
gain_reading: Random Forest and flag-based models integrating exposure, lipophilicity, mitotoxicity and cytotoxicity improved early DILI risk prediction with high specificity, supporting safer molecule triage to reduce hepatotoxicity-related attrition.
gain_evidence: Random Forest models further enhanced predictive performance | achieved 98% specificity and a positive likelihood ratio of 18 | reduce hepatotoxicity-related attrition in clinical development
problem_reading: (none)
problem_evidence: (none)
quick_read: By the publication date of 2026-10-03, researchers reported a refined multiparametric in vitro strategy for early DILI risk mitigation built on 170 reference drugs with known clinical outcomes. They combined cytotoxicity, mitotoxicity and transporter inhibition assays with exposure data and machine learning, showing that a >=2 flag system and Random Forest models improved specificity and balanced accuracy for early compound triage.

The work matters because DILI remains a major cause of preclinical and clinical attrition, and a high-specificity, interpretable platform could reduce late-stage failures and guide safer design. Uncertainty remains about how well the 98% specificity and 0.72 balanced accuracy observed in this reference set and Genentech portfolio will translate to diverse chemistries, patient populations, and in vivo outcomes.
limitation: Performance estimates are tied to a balanced reference set of 170 drugs and demonstration on contemporary Genentech portfolio molecules, limiting generalizability beyond those chemical spaces and in vitro systems.
tag: Evidence-backed gain
key_points: Study used a balanced reference set of 170 drugs with established clinical outcomes and evaluated 30 physicochemical, mechanistic and exposure-related features. | Top predictive features were cytotoxicity in primary human hepatocytes and human liver microtissues, mitochondrial toxicity and bile salt export pump inhibition. | Incorporation of clinical exposure data to derive margins of safety substantially improved assay performance. | Exposure-informed Random Forest models achieved a balanced accuracy of 0.72 with exposure, cLogD, mitotoxicity, cytotoxicity and hepatic clearance as major contributors.
rundown: Researchers systematically evaluated 30 endpoints including cytotoxicity in primary human hepatocytes and human liver microtissues, mitochondrial toxicity and bile salt export pump inhibition, and found that exposure normalization via margins of safety substantially improved prediction.

A flag-based hazard framework integrating >=2 hazard flags reached 98% specificity and positive likelihood ratio of 18, while Random Forest models identified exposure, predicted lipophilicity (e.g., cLogD), mitotoxicity, cytotoxicity and hepatic clearance as key drivers, with translational testing on Genentech portfolio molecules to refine borderline 0-1 flag compounds.
sources:
- peer_reviewed | Toxicological Sciences | https://doi.org/10.1093/toxsci/kfag124 | 2026-10-03
prev: 0000000000000000000000000000000000000000000000000000000000000000

sha256

a512c07e44703c7ea746745c5b712dc005bc5c3645ca32d307df000ecba3507f

previous

0000000000000000000000000000000000000000000000000000000000000000