It is estimated that 40% of drug candidates have failed to make it to market because of toxicity.¹ One of the most common adverse drug reactions is drug-induced liver injury (DILI), whereby patients experience acute illness, often with symptoms similar to hepatitis and cholestasis. Today, DILI is the leading cause of drug candidate failure and post-market withdrawals.²
The high incidence of DILI in clinical trials, and even post-market events, is partly due to the use of conventional animal models at preclinical stages. Conventional models are notoriously poor predictors of efficacy and toxicity in the liver, due to species-to-species variation.
In contrast, the PXB-mouse® model, with its humanized liver, provides a highly predictive model of human physiology and human-specific hepatotoxicity, allowing for more accurate prediction of human outcomes, and, therefore, aiding the smooth progression of new therapeutics into the clinic.
Effective and early toxicology studies are a crucial part of a well-designed preclinical program. These studies are critical for de-risking drug compound selection and enabling better prediction of human outcomes when the therapies reach clinical stages.
However, to predict hepatoxicity accurately, key data is needed:
Despite the importance of these data requirements, conventional preclinical testing can fail to accurately predict human liver toxicity, mainly due to the inter-species variation between animal and human hepatocytes.
When traditional animal models are used, some drug candidates will go on to show hepatotoxicity for the first time in clinical trials, whereas others might be dropped at the preclinical phase, based on false positive toxicity results that won’t be replicated in the human environment.
Unlike traditional animal models, the PXB-mouse® model has a humanized liver with a high proportion of human hepatocytes, making it a more accurate predictor of human physiology due to its stable expression of human enzymes and transporters. Histopathology, serum biomarker analysis and toxicogenomic tests all demonstrate accuracy in reflecting human outcomes, as demonstrated across a broad range of small molecule and biologics candidates. Two case studies show the strength of PXB-mice: KMTR2, and troglitazone (TGZ).
A recent study tested a therapeutic antibody (KMTR2) that targets the tumor necrosis factor (TNF)- related apoptosis-inducing ligand receptor 2 (TRAIL-R2) in PXB-mice.³ TRAIL-R2 is known to induce apoptosis in tumor cells, leaving human cells unaffected. However, monoclonal antibodies that have been developed to target and activate TRAIL-R2 can induce severe unexpected hepatotoxicity in clinical applications, which had not been predicted by preclinical studies.
KMTR2 has not reached clinical trials but has shown promising results in animal studies. The anti-human monoclonal antibody was tested in the PXB-mouse® model and demonstrated increased alanine transferase (ALT) levels and human ALT1 levels in the blood, both of which are biomarkers of hepatotoxicity. In addition, increased levels of TdT-mediated dUTP nick end labeling (TUNEL)-positive human hepatocytes were measured, as well as elevated serum concentration of cleaved cytokeratin 18 — both markers for human-specific apoptosis. Cell death and degeneration were also observed in human hepatocytes but were not seen in mouse hepatocytes, further demonstrating the inter-species variation in drug toxicity. RNA sequence analysis confirmed the up-regulation of TRAIL-R2-related genes in human hepatocytes only, further substantiating the human-specific nature of the hepatotoxicity.
Troglitazone (TGZ) was approved as a treatment for type 2 diabetes by the FDA in 1997. However, the FDA withdrew approval in 2000 owing to serious idiosyncratic liver injury. During preclinical toxicology studies, traditional mouse models did not suggest any severe hepatotoxicity.
Researchers investigated whether they could retroactively observe this TGZ induced human-specific hepatotoxicity in PXB-mice. Teams examined metabolite profiling,⁴ histopathological and immunohistochemical analysis,⁵ and biochemical analysis⁶ compared to control mice (SCID) in three separate studies:
These data, and many more studies besides, show that PXB-mice are far more accurate predictors of drug toxicity in the human environment than conventional animal models. The high proportion of human hepatocytes in the humanized liver, the histologically normal liver constitution, and the human-specific metabolism and excretion pathways allow the model to accurately simulate human physiology and the expression of key biochemical pathways. In this way, the PXB-mice allow researchers to better predict how efficacious or toxic a therapeutic candidate will be in the clinic, allowing for both promising candidates to progress, while failing those that could cause an adverse reaction.
DiMasi JA (1995) Success rates for new drugs entering clinical testing in the United States. Clin Pharmacol Ther 58:1–14. https://doi.org/10.1016/0009-9236(95)90066-7