DGAT2 inhibition in MASH: histologic promise and lipid paradox

The therapeutic landscape of metabolic dysfunction-associated steatohepatitis (MASH) is evolving rapidly. Within only a few years, metabolic modulators targeting glucagon-like peptide-1 (GLP-1), thyroid hormone receptor-β, peroxisome proliferator-activated receptors, and fibroblast growth factor pathways have demonstrated histologic improvements in steatohepatitis and, in selected cases, fibrosis regression (1,2). As therapies move from weight reduction toward targeted metabolic reprogramming, the efficacy is not judged merely by whether it can improve MASH, but by whether it can improve without worsening the cardiovascular risk. Against this backdrop, the MIRNA phase 2, randomized, double-blind, double-dummy study by Wong et al., evaluating the diacylglycerol acyltransferase-2 (DGAT2) inhibitor ervogastat alone and in combination with the acetyl-CoA carboxylase (ACC) inhibitor clesacostat in biopsy-confirmed MASH with F2–F3 fibrosis, represents more than a readout of histologic response (3,4). It is a stress test of a therapeutic thesis: that de novo lipogenesis (DNL) inhibition can reshape the histologic course of MASH while preserving cardiometabolic equilibrium.

The biological rationale is compelling. ACC sits upstream, catalyzing malonyl-CoA formation and governing flux through DNL, whereas DGAT2 executes a terminal step in triglyceride synthesis (5-7). In principle, dual inhibition should reduce intrahepatic triglyceride accumulation more effectively than either agent alone and may attenuate lipotoxic signaling that sustains hepatocellular injury and inflammation (5,6). Yet DNL is not confined to the liver; it is interwoven with VLDL assembly, peripheral lipid handling, and atherogenic particle trafficking (4). The MIRNA trial therefore asks a clinically consequential question: can the liver be improved without exporting risk?

At week 48, ervogastat monotherapy did not meet the composite primary endpoint, with response rates ranging from 45% to 52% across doses compared with 38% for placebo (3). In contrast, both combination arms met the composite endpoint: 66% of patients receiving ervogastat 150 mg plus clesacostat 5 mg {difference vs. placebo 0.27 [90% confidence interval (CI): 0.07–0.43]} and 63% receiving ervogastat 300 mg plus clesacostat 10 mg [difference 0.25 (0.04–0.42)] achieved response (3).

This divergence is a central signal. It suggests that terminal triglyceride synthesis blockade alone may be insufficient to consistently shift biopsy-based endpoints in established fibrotic disease, and that meaningful histologic traction may require upstream flux reduction in tandem. DGAT2 inhibition may be better positioned as one component of a multi-node metabolic strategy than as a standalone lever in F2–F3 MASH.

A second signal is equally instructive: the benefit in the combination arms appears to be driven primarily by steatohepatitis resolution rather than fibrosis improvement beyond placebo on the prespecified endpoint. This is not a minor nuance; it speaks to the biology of response. Steatosis and inflammatory activity are dynamic and can improve over months when lipid flux is reduced (8). Fibrosis remodeling, by contrast, reflects extracellular matrix architecture that may lag behind improvements in injury signals, particularly over a 48-week horizon (9).

For a clinician who performs EUS-guided liver biopsy and is accustomed to the variability inherent in liver biopsy interpretation, the magnitude of placebo-associated improvement in fibrosis underscores how fragile a single-time-point histologic endpoint remains in metabolic liver disease. The MIRNA study therefore reinforces an increasingly familiar dissociation in MASH drug development: large shifts in liver fat and activity can occur without commensurate separation in fibrosis endpoints.

The trial’s imaging and biomarker findings make that dissociation difficult to dismiss as noise. The combination regimens produced substantial reductions in liver fat, with the greatest relative reduction approaching 70% from baseline at week 48, compared with minimal change in the placebo group, as measured by MRI-proton density fat fraction (3). Improvements were also observed in aminotransferases and several non-invasive markers (3). A plausible inference is that the MIRNA trial achieved substantial metabolic “de-loading” of the liver—sufficient to improve steatohepatitis and related biomarkers—but that either the duration was inadequate for consistent matrix remodeling, or that DNL inhibition preferentially attenuates lipotoxic injury without directly modulating established fibrogenic pathways. Both possibilities carry practical implications. If time is the dominant constraint, longer exposure may permit fibrosis separation to emerge. Contemporary metabolic agents, including GLP-1-based therapies, thyroid hormone receptor-β agonists, and pan-PPAR agonists, have demonstrated that fibrosis regression may require extended treatment horizons and careful patient selection (1,2,9). The absence of clear fibrosis superiority in the MIRNA trial therefore does not negate the relevance of lipogenesis inhibition; rather, it reinforces that fibrosis endpoints are intrinsically more resistant to short-term metabolic modulation. If mechanism is the limiting factor, future regimens may need to couple metabolic unloading with therapies that more directly engage matrix remodeling biology in patients with established F2–F3 disease.

Methodological context further tempers interpretation. The trial enrolled fewer participants than planned and was conducted during a period in which trial participation itself can intensify lifestyle optimization (3). The placebo response, including for fibrosis improvement, was substantial. High placebo response is not unique to the MIRNA study but remains a recurring feature of biopsy-based MASH trials, reflecting sampling variability, regression to the mean, and behavioral co-interventions during participation (10-12). Even with central pathology adjudication the data still underscore a broader lesson: when placebo response is high, the most informative questions often shift from “did the endpoint move?” to “what moved, how coherently, and at what systemic cost?”

It is that cost—metabolic rather than hepatic—that may define MIRNA’s long-term significance. Ervogastat monotherapy was not associated with clear differences from placebo in fasting triglycerides, direct VLDL-C, or LDL-related apolipoproteins (3). In contrast, combination therapy was associated with sustained elevations in serum triglycerides and direct VLDL-C, greater reductions in HDL-C and apolipoprotein A1, and sustained increases in apolipoprotein C3 and apolipoprotein E relative to placebo (3). Notably, apolipoprotein B and Apo B100 confidence intervals overlapped with placebo (3). These changes reversed after drug discontinuation, consistent with a pharmacologic effect rather than irreversible injury (3). Nevertheless, reversibility does not neutralize relevance. In MASH population where the where cardiovascular morbidity and mortality exceeds the liver related morbidity and mortality, and therapies that worsen atherogenic lipid profiles demand careful scrutiny (13).

The lipid signal in the MIRNA trial should not be interpreted as an incidental laboratory footnote; it is mechanistically coherent with ACC inhibition and clinically salient in a population already enriched for cardiometabolic risk (13). The trial thus illuminates a paradox that future development must confront directly: the same upstream blockade that may be needed to amplify hepatic histologic response can shift lipoprotein biology in an unfavorable direction. This is not an argument against DNL inhibition. It is an argument that DNL inhibition, particularly when ACC is involved, cannot be advanced as a liver-only strategy. Cardiovascular adjudication, lipid-mitigation protocols, and longer follow-up are not optional accessories; they are prerequisites for determining net clinical value.

The field-wide implication is that “efficacy” in MASH is becoming multidimensional. A therapy that delivers compelling MASH resolution but introduces a sustained atherogenic lipid shift occupies a complicated space in clinical decision-making. It may remain attractive for a subset of patients, particularly if lipid perturbations can be mitigated with background therapy or dosing strategies, but broad adoption would require confidence that hepatic gains are not purchased with cardiovascular harm. Conversely, a therapy that is cardiometabolically favorable but modest on histology may be more acceptable as chronic treatment (14). The MIRNA trial’s data therefore contributes to a necessary recalibration: in MASH, the target organ is the liver, but the outcome space is systemic. Its findings underscore the need to evaluate therapeutic efficacy in MASH within an integrated cardiometabolic framework, including potential effects on cardiovascular risk and long-term cardiovascular outcomes.

Where does this leave DGAT2-directed therapy? The MIRNA trial supports several disciplined inferences. First, DGAT2 inhibition appears biologically active, with improvements in liver fat and biomarkers, but may not be sufficient alone to meet composite biopsy endpoints in F2–F3 disease over 48 weeks. Second, upstream–downstream DNL blockade can increase histologic response rates, particularly for steatohepatitis resolution. However, combination strategies may introduce additive safety considerations, and their real-world applicability will depend on achieving a balance between enhanced efficacy and tolerability (15). Third, the addition of ACC inhibition introduces a lipid liability that will likely shape patient selection, monitoring intensity, and future regimen design.

These inferences naturally point toward next-step questions that are more strategic than descriptive. Can trial duration be extended to test whether fibrosis separation emerges after sustained metabolic unloading? Can dose selection or exposure patterns preserve hepatic gains while reducing lipid perturbations? Should lipid-lowering co-therapy be standardized within trial design to reflect real-world practice and to de-risk cardiovascular exposure? Are there patient subgroups—by baseline triglycerides, diabetes status, or metabolic inflammatory signatures—in whom the benefit–risk balance is more favorable? The MIRNA phase 2 trial does not answer these questions, but it clarifies that they are the correct questions.

In sum, MIRNA is best viewed as a transitional trial. It validates DNL inhibition as a credible histology-moving axis in MASH, demonstrates that dual metabolic blockade can improve endpoint attainment, and simultaneously exposes the central challenge of metabolic precision: the liver rarely changes in isolation within a metabolically integrated system. The path forward is not to retreat from mechanism-based therapy, but to broaden the development frame to include cardiometabolic equilibrium as a co-primary consideration. If MASH therapeutics are to mature into durable chronic regimens, the next generation of trials will need to prove not only that the liver improves, but that systemic risk is not inadvertently amplified in the process. In metabolic disease, a single organ is never the whole story.

Supplementary

The article’s supplementary files as

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.