MASH Drug Discovery: How to Choose the Right Mouse Model for Your MoA

Introduction: "Which Model Should I Use?"

In MASH (metabolic dysfunction-associated steatohepatitis) drug discovery, selecting the right preclinical animal model for a given drug candidate is one of the first — and most consequential — decisions a project team must make.

More than ten distinct MASH mouse models are currently available, each recapitulating different pathological mechanisms, running on different timelines, and offering different levels of clinical translatability. A poor model choice at the preclinical stage can lead directly to the "Lost in Translation" problem — promising preclinical efficacy data that fail to replicate in Phase 2b clinical trials.

This article differs from general model overviews (→ MASH Model Overview) or head-to-head comparisons of specific models (→ AMLN Diet vs. GAN Diet). Instead, it provides a mechanism-of-action (MoA)-driven decision matrix designed to help drug discovery program leaders and project managers rapidly identify the optimal model strategy for their compound.

1. Quick Reference: MoA-Based Model Selection Decision Table

Start with the bottom line. Use the table below to identify recommended models based on your drug candidate's MoA.

Drug MoA	First-Choice Model	Complementary Model	Rationale
Metabolic improvement (GLP-1, GIP/GLP-1, THR-β, FXR)	GAN diet	AMLN diet	Reproduces obesity, IR, and dyslipidemia — ideal for detecting metabolic efficacy
Anti-fibrotic (TGF-β inhibitor, LOXL2 inhibitor, HSC-targeted)	CCl4	WD+CCl4	Clean fibrogenesis pathway; reaches F3 rapidly
Anti-inflammatory (CCR2/5 inhibitor, ASK1 inhibitor, NF-κB-targeted)	CDA-HFD	GAN diet	Generates a robust inflammatory and fibrotic microenvironment in a short timeframe
Lipid metabolism (ACC inhibitor, SCD1 inhibitor, DGAT2 inhibitor)	GAN diet	CDA-HFD	Recapitulates systemic lipid metabolism dysfunction
HCC prevention / late-stage complications (immune checkpoint, etc.)	STAM	GAN diet (long-term)	Reliable MASH-to-HCC progression
Dual pharmacology / combination MoA	GAN diet + CCl4	WD+CCl4	Evaluates metabolic background and severe fibrosis simultaneously

Key principle: No single model recapitulates the full spectrum of human MASH. Both FDA and EMA guidance documents recommend the complementary use of multiple models.

2. Major MASH Models at a Glance

Diet-Induced Models

GAN Diet (Gubra-Amylin NASH Diet) — High fat (40 kcal% palm oil) + high fructose + high cholesterol (2%). Reaches F1–F2 in 16–20 weeks. Produces pronounced obesity, insulin resistance (IR), and dyslipidemia, giving it the highest overall clinical translatability^[1]. Inter-individual variability is substantial, making biopsy-confirmed enrollment advisable. Note that its predecessor, the AMLN diet, was discontinued following a 2018 FDA regulatory action; most programs have since transitioned to the GAN diet (→ AMLN vs. GAN comparison).

CDA-HFD (Choline-Deficient, High-Fat Diet) — Reaches F2–F3 in as few as 6–12 weeks, making it one of the fastest models available. However, it induces weight loss and does not produce IR, limiting clinical translatability to fibrosis and inflammation endpoints only. Best suited for high-throughput screening of anti-fibrotic and anti-inflammatory agents^[2].

Western Diet (WD) — Reaches F1–F2 in 16–24 weeks. Moderate obesity and IR. Fibrosis can be accelerated by combining with CCl4 (WD+CCl4).

Chemically Induced Models

CCl4 (Carbon Tetrachloride) — Intraperitoneal injection (twice weekly) reaches F2–F3 in 4–6 weeks and F4 in 8–12 weeks — the fastest fibrosis induction available. No metabolic phenotype, but cleanly recapitulates the core fibrogenesis cascade: HSC activation → myofibroblast differentiation → collagen deposition^[3].

TAA (Thioacetamide) — Administered in drinking water; reaches F2–F3 in 6–8 weeks. Produces prominent periductular fibrosis. Oral administration simplifies handling.

Combination Models

WD+CCl4 — Western diet feeding combined with CCl4 injection reaches F3–F4 in 12–16 weeks. Combines a metabolic background with severe fibrosis^[4].

STAM (STZ + High-Fat Diet) — Neonatal STZ injection followed by high-fat diet. Reaches F1–F2 by week 9 and HCC by weeks 16–20. Produces hyperglycemia but only mild obesity. Particularly suitable for evaluating "lean MASH" (prevalent in Asian populations) and MASH-to-HCC progression^[5].

3. Recommended Model Strategies by Drug Target (MoA)

3-1. Metabolic Modulators (GLP-1 Receptor Agonists, THR-β Agonists, FXR Agonists)

Drugs that improve liver pathology through systemic metabolic correction require a model that faithfully reproduces obesity, insulin resistance, and dyslipidemia.

Recommended: GAN diet model (16–24 weeks of feeding)
Rationale: The GAN diet model was used in preclinical evaluation of semaglutide (Novo Nordisk), and the directionality of body weight reduction, NAS improvement, and hepatic fat reduction was consistent with human clinical outcomes^[6]
See also: GLP-1 Receptor Agonists and Fibrosis: Beyond Semaglutide

For THR-β agonists (e.g., resmetirom), evaluation requires capturing thyroid hormone pathway-mediated hepatic fat reduction, so a GAN diet model with sufficient steatosis development (≥20 weeks) is recommended.

3-2. Anti-Fibrotic Agents (TGF-β Inhibitors, LOXL2 Inhibitors, Integrin Inhibitors)

Drugs directly targeting the fibrosis cascade require a model that isolates fibrogenesis from metabolic confounders, enabling high-fidelity assessment of anti-fibrotic activity.

Recommended: CCl4 model (4–8 weeks of dosing)
Rationale: The CCl4 model cleanly recapitulates the core fibrosis pathway — hepatic stellate cell (HSC) activation → myofibroblast differentiation → collagen production
Fibrosis assessment: Combined use of Sirius Red staining and hydroxyproline quantification is recommended
Complementary: Add WD+CCl4 to confirm efficacy in a metabolic background

3-3. Anti-Inflammatory Agents (CCR2/5 Inhibitors, ASK1 Inhibitors, Caspase Inhibitors)

Drugs targeting the hepatic inflammatory microenvironment require a model that generates robust inflammation within a short timeframe.

Recommended: CDA-HFD model (6–12 weeks of feeding)
Rationale: Choline deficiency induces endoplasmic reticulum stress and oxidative stress, driving potent macrophage infiltration and cytokine production
Complementary: Add a GAN diet arm to confirm anti-inflammatory efficacy in a metabolic context

Practical note: In preclinical studies, cenicriviroc (a CCR2/5 inhibitor) showed pronounced anti-inflammatory activity in a CDA-HFD model. However, the compound failed to meet its primary endpoint in the Phase 3 AURORA trial. This case underscores the risk of relying exclusively on positive data from inflammation-dominant models without metabolic validation.

4. Comprehensive Model Comparison Matrix

Parameter	GAN Diet	CDA-HFD	CCl4	TAA	WD+CCl4	STAM
Time to F2 fibrosis	16–20 wk	6–12 wk	4–6 wk	6–8 wk	8–12 wk	9 wk
Obesity	+++	✕ (weight loss)	✕	✕	+	✕
Insulin resistance	+++	✕	✕	✕	+	Hyperglycemia (IR−)
Steatosis	+++	+++	+	+	++	++
Lobular inflammation	++	+++	++	++	++	++
Hepatocyte ballooning	++	++	+	+	++	+
HCC progression	++ (long-term)	+	✕	+	✕	+++
Overall clinical translatability	+++	+	+	+	++	++
Study cost	High (long housing)	Medium	Low	Low	Medium–High	Medium
Inter-animal variability	High	Medium	Low	Low	Medium	Low
Best-fit MoA	Metabolic modulators	Anti-inflammatory / anti-fibrotic	Anti-fibrotic	Anti-fibrotic	Dual MoA	HCC prevention

+++ = Excellent / ++ = Good / + = Limited / ✕ = Not applicable

5. Biomarker Endpoint Design

Equally important to model selection is designing a biomarker strategy that aligns with clinical trial endpoints.

Essential Endpoints (Tier 1)

Biomarker	Method	Clinical Relevance
NAS (NAFLD Activity Score)	Histological scoring (H&E)	FDA-accepted endpoint (≥2-point NAS improvement)
Fibrosis stage	Sirius Red staining + quantitative image analysis	Clinical primary endpoint (≥1-stage fibrosis improvement)
Hepatic collagen content	Hydroxyproline quantification	Biochemical confirmation of fibrosis
Hepatic triglyceride	Biochemical quantification	Indicator of steatosis resolution

Recommended Endpoints (Tier 2)

Biomarker	Method	Significance
Serum ALT/AST	Clinical chemistry	Hepatocellular injury monitoring
αSMA immunostaining	IHC + quantification	Marker of HSC activation
Col1a1, Acta2, Tgfb1 gene expression	RT-qPCR	Fibrosis-related gene modulation
F4/80, CD68 immunostaining	IHC + quantification	Macrophage infiltration assessment

Clinical Translation Enhancement (Tier 3)

To bridge preclinical and clinical biomarker data, consider adding serum Pro-C3 (a fibrogenesis marker measurable in mice), serum CK-18 (M30/M65) (apoptosis marker), and hepatic transcriptomics (RNA-seq) to assess gene expression concordance with human MASH liver tissue.

6. Study Design Timeline Examples

Example 1: PoC Study for a Metabolic Modulator (GLP-1 Agonist)

GAN diet feeding for 24 weeks → liver biopsy (disease confirmation + stratified randomization) → 8-week drug treatment → terminal analysis. Groups: vehicle, low-dose, high-dose, and positive control (n=12–15 per group). At week 24, only animals with NAS ≥ 5 and F ≥ 1 on biopsy are enrolled, ensuring a uniform baseline.

Example 2: Anti-Fibrotic Screening Study

CCl4 dosing initiated (0.5 mL/kg, twice weekly i.p.) → drug treatment started at week 2 in parallel → terminal analysis at weeks 6–8. Groups: normal control, vehicle, and drug-treated (n=8–10 per group). A "therapeutic" design — where drug treatment overlaps with ongoing CCl4 — enhances clinical relevance.

Example 3: Comprehensive Dual-Pharmacology Evaluation

Run Study A (metabolic assessment: GAN diet 24 weeks + 8-week treatment) and Study B (fibrosis assessment: CDA-HFD 6 weeks + 6-week treatment) in parallel, building evidence for both metabolic improvement and anti-fibrotic activity.

7. Key Considerations When Engaging a CRO

When outsourcing MASH preclinical studies to a contract research organization, confirming the following points helps ensure study quality and data reliability.

Model Establishment Track Record

Availability of historical control data for the target model and reproducibility of fibrosis stages
Established standard protocols for mouse strain (C57BL/6J vs. C57BL/6N) and age at study initiation

Pathology Assessment Infrastructure

Blinded scoring (NAS, fibrosis stage) capability and workflow
Automated, standardized Sirius Red quantitative image analysis pipeline
Access to pathologists with hepatic pathology expertise

Liver Biopsy and Data Quality

Intra-operative liver biopsy technique and survival rate data for GAN diet models
Digital scanning and sharing of histopathology images; raw data delivery formats
Housing capacity for running multiple models concurrently

8. Summary: Three Principles for Model Selection

MoA first: Choose the model that is most sensitive to your compound's mechanism of action — not the most popular model in the field.
Complementary multi-model strategy: Combine a metabolic model (GAN diet) with a fibrosis-focused model (CCl4 or CDA-HFD) to maximize clinical translatability.
Align with clinical endpoints: Incorporate the same evaluation axes used in Phase 2b/3 primary endpoints — NAS improvement, fibrosis stage regression — into your preclinical study design.

The MASH therapeutic pipeline is expanding rapidly (→ MASH Therapeutic Landscape 2025), and as competition intensifies, appropriate preclinical model selection has never been more critical. We hope the decision matrix presented here serves as a practical tool for strategic decision-making in your drug discovery program.

References

1. Tølbøl KS, et al. Metabolic and hepatic effects of liraglutide, obeticholic acid and elafibranor in diet-induced obese mouse models of biopsy-confirmed nonalcoholic steatohepatitis. World J Gastroenterol. 2018;24(2):179-194. PubMed

2. Matsumoto M, et al. An improved mouse model that rapidly develops fibrosis in non-alcoholic steatohepatitis. Int J Exp Pathol. 2013;94(2):93-103. PubMed

3. Tsuchida T, et al. A simple diet- and chemical-induced murine NASH model with rapid progression of steatohepatitis, fibrosis and liver cancer. J Hepatol. 2018;69(2):385-395. PubMed

4. Teufel A, et al. Comparison of gene expression patterns between mouse models of nonalcoholic fatty liver disease and liver tissues from patients. Gastroenterology. 2016;151(3):513-525.e0. PubMed

5. Fujii M, et al. A murine model for non-alcoholic steatohepatitis showing evidence of association between diabetes and hepatocellular carcinoma. Med Mol Morphol. 2013;46(3):141-152. PubMed

6. Newsome PN, et al. A placebo-controlled trial of subcutaneous semaglutide in nonalcoholic steatohepatitis. N Engl J Med. 2021;384(12):1113-1124. PubMed