Dear Editor,
As the therapeutic landscape for metabolic dysfunction- associated steatohepatitis (MASH) begins to shift from lifestyle modification toward pharmacologic intervention, the recent phase II trial by Wong and colleagues [
1] investigating HTD1801 provides timely and encouraging data. However, several methodological considerations warrant closer scrutiny to enhance the translational relevance of its findings.
First, the trial did not adequately account for baseline metabolic heterogeneity, a key determinant of treatment response in MASH. While patients with T2DM were included, no subgroup analyses were reported based on levels of insulin resistance such as HOMA-IR, baseline liver fat content measured by MRI-PDFF, or specific obesity phenotypes, including the distribution of visceral compared with subcutaneous adiposity. These factors substantially influence responsiveness to metabolic interventions such as AMPK-activating agents. For instance, each 1-unit increase in HOMA-IR corresponds to a significant increase in the likelihood of advanced fibrosis, with an odds ratio significantly above 1 for patients with higher insulin resistance. Additionally, patients with higher visceral adiposity, as measured by imaging, are 2–3 times more likely to progress to NASH-related fibrosis [
2]. Without stratified analyses, important differential responses across subgroups may be obscured, limiting the study’s value for clinical decisionmaking and precision trial design.
Second, the study’s imaging-based response definition—≥ 30% reduction in PDFF and ≥88 ms reduction in cT1—may overestimate therapeutic effect in the absence of adjustments for natural variability. MRI-PDFF and cT1 are useful non-invasive biomarkers, but both are susceptible to physiological and technical variation. Notably, MRIPDFF can vary by ±10–15% without any intervention, and cT1 is influenced by hepatic iron deposition and transient inflammatory changes if not corrected for T2 relaxation [
3,
4]. The 24% placebo response rate observed suggests that a portion of the responses could be artefactual. Incorporating a minimal detectable change (MDC) framework and, ideally, a test–retest control subgroup could help clarify the threshold for clinically meaningful change and reduce misclassification bias.
Third, the trial did not report dietary or physical activity parameters, both of which are potent modifiers of hepatic steatosis and aminotransferase levels. The observation that 22% of patients in the high-dose group experienced ≥5% weight loss raises the possibility that lifestyle changes contributed to the observed improvements. Yet, without concurrent data on caloric intake, macronutrient balance, or activity levels—such as those derived from structured dietary assessments or wearable accelerometers—it is difficult to isolate drug-specific effects from behavioural co-interventions. Moreover, given that berberine may modulate gastrointestinal motility and appetite regulation, lifestyle-related confounding may not be random. Rigorous behavioural monitoring should therefore be integrated into future metabolic liver disease trials.
In addition, the rapid advancement of artificial intelligence, particularly large language models (LLMs), offers a novel pathway to address clinical heterogeneity and refine therapeutic targeting. By integrating high-dimensional clinical, biochemical, and imaging datasets, AI-driven frameworks can uncover latent phenotypic clusters and predict subgroup-specific responsiveness to agents like HTD1801. For example, AI models could analyze patient data to predict subgroup-specific responses based on factors such as insulin resistance (HOMA-IR), liver fat content (measured by PDFF), and visceral adiposity, rather than relying on more generalized markers like ALT alone. Cross-disciplinary approaches combining hepatology, bioinformatics, and machine learning could unlock dynamic, individualised response modelling that transcends conventional stratification by ALT or PDFF alone. Embedding AI-based enrichment and predictive modelling within trial design may therefore improve both internal validity and clinical translatability of findings.
These considerations do not diminish the therapeutic promise of HTD1801 but rather highlight essential areas for methodological strengthening. As the treatment landscape for MASH evolves rapidly, rigorous trial design—including attention to metabolic stratification, robust imaging endpoints, behavioural confounders, and AI-enabled modelling— will be critical to ensure that emerging therapies reach the patients most likely to benefit.
FOOTNOTES
-
Conflicts of Interest
The author has no conflicts to disclose.
Abbreviations
metabolic dysfunction-associated steatohepatitis
minimal detectable change
REFERENCES
- 1. Wong VW, Neff GW, Di Bisceglie AM, Bai R, Cheng J, Yu M, et al. HTD1801 demonstrates promising potential for histologic improvements in metabolic dysfunction-associated steatohepatitis in both a preclinical and phase 2 study. Clin Mol Hepatol 2025;31:1071-1083.
- 2. Friedman SL, Neuschwander-Tetri BA, Rinella M, Sanyal AJ. Mechanisms of NAFLD development and therapeutic strategies. Nat Med 2018;24:908-922.
- 3. Pavlides M, Banerjee R, Sellwood J, Kelly CJ, Robson MD, Booth JC, et al. Multiparametric magnetic resonance imaging predicts clinical outcomes in patients with chronic liver disease. J Hepatol 2016;64:308-315.
- 4. Kim HJ, Cho HJ, Kim B, You MW, Lee JH, Huh J, et al. Accuracy and precision of proton density fat fraction measurement across field strengths and scan intervals: A phantom and human study. J Magn Reson Imaging 2019;50:305-314.
Citations
Citations to this article as recorded by
