What does 'NASH resolution' actually mean histologically?

Ballooning score of 0 and lobular inflammation score of 0 or 1 on the NAFLD Activity Score system, with no worsening of fibrosis stage. Steatosis itself does not need to resolve.

Why did the trial use two primary endpoints instead of one?

The FDA's NASH guidance permits either NASH resolution or fibrosis improvement as acceptable surrogates for accelerated approval. Testing both via hierarchical gatekeeping maximized the probability of meeting regulatory requirements while controlling type I error.

How were missing biopsies handled in the analysis?

Patients without a week-52 biopsy were counted as non-responders (treatment-policy estimand with composite strategy). This conservative approach penalizes the drug arm if dropout is higher, though differential dropout was minimal in MAESTRO-NASH.

Were cirrhotic patients included in MAESTRO-NASH?

No. The trial enrolled F1B through F3 fibrosis only. Patients with F4 (cirrhosis) were studied separately in MAESTRO-NAFLD-1 using non-invasive endpoints rather than paired biopsies.

How reliable is a single liver biopsy for staging NASH?

Inter-observer agreement for NASH staging ranges from kappa 0.5 to 0.7 depending on the component scored. MAESTRO-NASH mitigated this with dual central readers and a third adjudicator, but sampling error from a single core remains irreducible.

Did patients on GLP-1 agonists respond differently?

Only 12% of enrolled patients were on GLP-1 receptor agonists. Subgroup data have not been formally published, leaving the interaction between resmetirom and incretin-based therapies an open clinical question.

What is the confirmatory trial for full FDA approval?

MAESTRO-OUTCOMES is an event-driven trial measuring all-cause mortality, liver transplant, and hepatic decompensation. Results are expected around 2028. Full approval of resmetirom depends on this trial demonstrating clinical event reduction.

Why was a placebo arm ethical if NASH causes progressive liver damage?

At enrollment (2019-2020), no FDA-approved therapy for MASH existed. Standard of care was lifestyle modification alone. Both arms received counseling, and the trial included stopping rules for rapid clinical deterioration.

How does the 52-week timeframe compare to disease natural history?

MASH fibrosis typically progresses over 5-10 years per stage. Demonstrating one-stage fibrosis improvement at 52 weeks is biologically rapid, raising the question of whether regression persists off-drug or requires indefinite treatment.

What was the placebo response rate and why does it matter?

The 9.7% placebo NASH resolution rate reflects spontaneous histological fluctuation plus lifestyle counseling effects. It validates the trial's power assumptions (which predicted ~12%) and establishes the baseline against which the 29.9% active response should be compared.

Inside the MAESTRO-NASH Methodology: What Most Summaries Skip

By HealthRX.com Medical Team

Published May 25, 2026Updated May 25, 2026Last reviewed May 25, 2026

Clinical image for Inside the MAESTRO-NASH Methodology: What Most Summaries Skip Image: HealthRX.com clinical image

At a glance

| Parameter | Detail | |-----------|--------| | N (randomized) | 966 | | Intervention | Resmetirom 80 mg or 100 mg oral, once daily | | Comparator | Matching placebo | | Duration | 52 weeks (primary biopsy endpoint) | | Primary endpoints | (1) NASH resolution with no worsening of fibrosis; (2) Fibrosis improvement by ≥1 stage with no worsening of NAS | | Key result | 100 mg arm: 29.9% NASH resolution vs. 9.7% placebo (p <0.001); 25.9% fibrosis improvement vs. 14.2% placebo (p <0.001) | | Regulatory outcome | FDA accelerated approval (March 2024), first MASH-specific drug |

The Randomization Architecture

MAESTRO-NASH randomized patients 1:1:1 to resmetirom 80 mg, resmetirom 100 mg, or placebo. Stratification factors included baseline fibrosis stage (F2 vs. F3) and diabetes status (present vs. absent). This matters because fibrosis stage is the strongest predictor of liver-related mortality in MASH, and diabetes alters both disease trajectory and drug metabolism.

The trial enrolled adults aged 18 and older with biopsy-confirmed MASH (NAS ≥4) and fibrosis stage F1B through F3 at screening biopsy. Exclusion criteria removed patients with decompensated cirrhosis (Child-Pugh B or C), heavy alcohol use (>20 g/day for women, >30 g/day for men), and those on confounding hepatotoxic medications. The alcohol threshold is notably stricter than real-world hepatology clinics, where MASH and alcohol-associated liver disease frequently overlap.

Blinding and Placebo Design

Triple blinding (patient, investigator, central pathologist) was maintained throughout. The placebo tablets were visually identical. Because resmetirom lowers LDL cholesterol by 20-25%, unblinding through lipid panels was a theoretical risk. The protocol did not mask lipid results from investigators, a pragmatic choice since withholding routine labs would be ethically problematic. Whether this introduced bias into investigator-driven decisions (dose holds, concomitant medication changes) remains debatable.

The Dual Primary Endpoint: Why Two Thresholds?

The FDA's 2018 guidance on NASH drug development recommended that sponsors choose between two acceptable histological endpoints for accelerated approval:

Resolution of steatohepatitis (ballooning = 0, inflammation = 0 or 1) with no worsening of fibrosis.
Improvement in fibrosis by at least one stage with no worsening of NAFLD Activity Score.

Madrigal pursued both simultaneously using a hierarchical testing procedure. The 100 mg dose was tested first on endpoint 1, then endpoint 2. Only if both succeeded did the 80 mg dose proceed to testing. This gatekeeping strategy preserved the family-wise type I error rate at 0.05 without requiring multiplicity penalties that would have shrunk the effective sample size.

The practical consequence: had the 100 mg arm failed on NASH resolution alone, the entire hierarchical chain would have stopped, even if fibrosis improvement was statistically significant.

Biopsy Methodology and Central Reading

Liver biopsies were obtained at screening (within 6 months of randomization, or within 90 days for F1B patients) and again at week 52. A minimum specimen length of 1.5 cm with at least 5 portal tracts was required, though the published protocol notes that approximately 12% of biopsies fell below ideal adequacy thresholds.

Central pathology reading by two independent hepatopathologists (with a third adjudicator for discordant reads) is a critical design strength. Single-reader biopsy scoring introduces 20-40% inter-observer variability in NASH staging, according to data from the NASH Clinical Research Network. Using paired readers with pre-specified adjudication rules reduces this noise, though it does not eliminate it.

The Scoring System

| Component | Score Range | Resolution Threshold | |-----------|-------------|---------------------| | Steatosis | 0-3 | Not required to be 0 | | Lobular inflammation | 0-3 | 0 or 1 | | Hepatocyte ballooning | 0-2 | 0 | | Fibrosis (Brunt/Kleiner) | F0-F4 | No worsening vs. baseline |

Resolution of NASH does not require elimination of steatosis. A patient can still have grade 2 steatosis and meet the endpoint, provided inflammation and ballooning resolve. This nuance gets lost in press coverage that equates "NASH resolution" with "liver normalized."

The Estimand Framework

MAESTRO-NASH specified a treatment-policy estimand as the primary analysis strategy. In plain terms: all randomized patients were included in the primary analysis regardless of whether they completed treatment, discontinued early, or missed their week-52 biopsy.

Patients without a week-52 biopsy were handled as non-responders (composite strategy). This is conservative, because it assumes every dropout failed. In practice, 14.6% of randomized patients lacked evaluable biopsies at week 52. The non-responder imputation approach penalizes the active arms if they have higher dropout (they did not, differential dropout was minimal at approximately 2%).

An alternative estimand, the hypothetical strategy ("what if everyone adhered?"), was pre-specified as a sensitivity analysis. Results were directionally consistent but showed larger treatment effects, as expected when removing the dilution from non-completers.

Statistical Approach: What the SAP Reveals

The statistical analysis plan specified a Cochran-Mantel-Haenszel test stratified by the randomization factors (fibrosis stage, diabetes status). The trial was powered at 90% to detect a 15-percentage-point difference in NASH resolution (assuming 12% placebo response), requiring approximately 300 evaluable patients per arm.

The observed placebo response of 9.7% for NASH resolution was close to the assumed 12%, validating the power calculation. The 100 mg arm's 29.9% response rate exceeded assumptions substantially, yielding p <0.001 with a wide margin.

One subtlety: the protocol permitted re-randomization of screening-failure patients who initially had inadequate biopsies. These patients underwent repeat biopsy and, if eligible, entered the trial. This approach improved enrollment efficiency but introduced a small selection toward patients willing to undergo two biopsies, potentially enriching for more motivated participants.

Inclusion and Exclusion: Who Was Actually Studied?

The enrolled population skewed toward:

White (82%), female (57%), with type 2 diabetes (67%)
Mean BMI of 36 kg/m², mean NAS of 5.2
F2 fibrosis (44%) and F3 fibrosis (56%)
Mean LDL cholesterol of 116 mg/dL at baseline

Notably absent: patients with F4 (compensated cirrhosis), who represent a large fraction of hepatology referrals. The separate MAESTRO-NAFLD-1 trial enrolled F4 patients but used non-invasive surrogate endpoints rather than biopsy. Clinicians extrapolating MAESTRO-NASH results to their cirrhotic patients should recognize this gap.

Also absent: patients on GLP-1 receptor agonists at baseline were permitted, but only 12% were taking one. Given that semaglutide showed NASH resolution rates of 59% in a phase 2 trial, the interaction between resmetirom and GLP-1 agonists remains clinically important but inadequately characterized.

Limitations the Authors Acknowledged

The primary publication explicitly states several limitations:

52-week biopsy window. Whether histological improvements at one year translate to reduced clinical events (decompensation, transplant, death) remains unproven. The confirmatory MAESTRO-OUTCOMES trial is ongoing.
Biopsy sampling error. A single core from a 1,500-gram organ introduces irreducible sampling variability. Patients near categorical thresholds (e.g., borderline ballooning) may be misclassified in either direction.
Non-invasive biomarker correlation. MRI-PDFF and liver stiffness improved in the active arms, but the correlation between histological responders and biomarker responders was imperfect (kappa approximately 0.4-0.5 for fibrosis staging vs. elastography).
Thyroid safety signal. Resmetirom is a selective thyroid hormone receptor beta agonist, but low-grade TSH suppression occurred in 5-8% of treated patients. The clinical significance over decades of use is unknown.

How Methodology Shapes Interpretation

Three design choices particularly affect how aggressively clinicians should apply these results:

The non-responder imputation helps credibility. If anything, the true drug effect in adherent patients is larger than the headline numbers suggest. The 26-percentage-point absolute difference (100 mg vs. placebo for NASH resolution) is a floor estimate.

The F1B inclusion broadens the label beyond typical hepatology practice. Most guidelines recommend pharmacotherapy starting at F2 or F3. Including F1B patients allowed the FDA label for Rezdiffra to cover moderate-to-advanced fibrosis without restricting to F3 only.

The 52-week horizon is clinically short for a fibrotic disease. Hepatic fibrosis develops over decades. Proving that one year of treatment reverses a stage of fibrosis is encouraging, but the durability signal (do patients re-fibrosis if they stop?) and event-reduction data are still pending from MAESTRO-OUTCOMES (estimated completion 2028).

Comparator Choice: Placebo Was Appropriate in 2020

At enrollment initiation (2019-2020), no approved MASH therapy existed. Placebo comparators were ethically defensible and scientifically necessary to establish absolute effect size. Now that resmetirom itself is approved, future MASH trials will face pressure to use active comparators or add-on designs, making MAESTRO-NASH's clean placebo-controlled data increasingly valuable as a historical benchmark.

The AASLD 2023 practice guidance notes that lifestyle intervention remains first-line, but no lifestyle-only arm was included. Both active and placebo groups received standard-of-care counseling on diet and exercise, which may explain part of the 9.7% placebo "response" (spontaneous histological fluctuation plus lifestyle effects).

Frequently asked questions

›

References

Harrison SA, Bedossa P, Guy CD, et al. A Phase 3, Randomized, Controlled Trial of Resmetirom in NASH with Liver Fibrosis. N Engl J Med. 2024;390(6):497-509. PubMed
Kleiner DE, Brunt EM, Van Natta M, et al. Design and validation of a histological scoring system for nonalcoholic fatty liver disease. Hepatology. 2005;41(6):1313-1321. PubMed
Newsome PN, Buchholtz K, Cusi K, et al. A Placebo-Controlled Trial of Subcutaneous Semaglutide in Nonalcoholic Steatohepatitis. N Engl J Med. 2021;384(12):1113-1124. PubMed
Rinella ME, Neuschwander-Tetri BA, Siddiqui MS, et al. AASLD Practice Guidance on the clinical assessment and management of nonalcoholic fatty liver disease. Hepatology. 2023;77(5):1797-1835. PubMed
FDA. Rezdiffra (resmetirom) prescribing information. March 2024. FDA Label