Network meta-analysis of phase 2-3 therapies for non-cirrhotic metabolic dysfunction-associated steatohepatitisNew therapies show promise for liver disease without cirrhosis

This is a network meta-analysis of phase 2-3 randomized trials. The setting was phase 2-3 randomized trials, and the population was 12,787 adults with non-cirrhotic metabolic dysfunction-associated steatohepatitis. The intervention was phase 2-3 therapies for metabolic-dysfunction-associated steatohepatitis, including metabolic dysfunction and insulin sensitivity targets, anti-inflammatory agents, anti-fibrotic agents, fibroblast growth factor 21 analogs, and incretin-based multi-agonists. The comparator was placebo.

The primary outcomes were resolution of steatohepatitis without worsening of fibrosis (biopsy based), at least one-stage improvement in fibrosis without worsening of steatohepatitis (biopsy based), and at least a 30% relative reduction in liver fat measured by magnetic resonance imaging-derived proton density fat fraction. For therapies targeting metabolic dysfunction and insulin sensitivity, the main results showed odds ratios of 2.5-7.1 versus placebo for both biopsy-based endpoints. The absolute numbers indicated that 35%-70% of treated participants did not meet prespecified biopsy endpoints, while biopsy placebo response rates were 11%-18%. For the imaging endpoint, these therapies showed the most consistent benefits versus placebo and the lowest failure rates across biopsy and imaging endpoints. No confidence intervals or p-values were reported for the pooled effects.

No key secondary outcomes were reported in the input. Safety and tolerability findings were not reported; adverse events, serious adverse events, and discontinuations were not reported.

These results can be compared to prior landmark studies in this therapeutic area, but the input does not specify prior studies. The network meta-analysis is based on phase 2-3 randomized trials, but the source does not report confidence intervals or p-values for the pooled effects. Methodological limitations include the lack of reported confidence intervals and p-values, which reduces certainty about effect precision. Potential biases were not reported, and the network meta-analysis reports associations and comparative efficacy, not causation.

For clinical practice, these findings support metabolic therapies as a rational foundation for combination regimens spanning complementary mechanisms and highlight the need for more durable approaches to achieve clinically meaningful biopsy outcomes. Clinicians should consider that many participants did not meet prespecified biopsy endpoints, and placebo response rates were 11%-18%.

Key questions remain unanswered, including the durability of treatment effects, long-term safety profiles, and head-to-head comparisons among specific therapy classes. The input does not report follow-up duration or specific drug names beyond class descriptions.