Open questions — MASLD¶
Last curated: 2026-09-02. This replaces the seed set written at scoping. Every question is grounded in the built wiki pages and their citations; every asserted absence carries the exact query that was run and the date.
The binding search rule. Every search behind these questions unions old and new vocabulary — (MASLD OR NAFLD OR MASH OR NASH OR "steatotic liver disease"). An absence found under one name is not an absence. This was not a theoretical concern: an absence claim about the patient-experience literature was asserted from one search formulation during the build and found to be false when re-run with a targeted formulation (see OQ-16 and literature/patient-voice/sources.md).
This file was extended in the 2026-09-02 depth pass: seven Tier-1 and six Tier-2 questions added (OQ-31 to OQ-43), seven new dots (D13–D19), and OQ-14 partly answered.
Tier 1 = answering it would change practice, and a study is designable today. Tier 2 = blocked on tools, numbers, or a Tier-1 answer.
Stable IDs. OQ-1 to OQ-5 carry forward from the seed set with their scope sharpened; OQ-6 onward are new.
Dots not yet connected¶
Cross-domain junctions where two bodies of evidence in this knowledge base both exist and no study has joined them.
| # | Dot A | Dot B | The missing junction | Powers |
|---|---|---|---|---|
| D1 | Fibrosis stage predicts outcomes; a histological diagnosis of steatohepatitis adds essentially nothing once fibrosis is known (PMID 25935633, 28803953, 27616339, 38293684) | Both approved drugs were licensed on co-primary endpoints, one of which is resolution of steatohepatitis (PMID 38324483, 40305708) | The endpoint that licenses the drugs is not the variable that predicts the outcome, and no trial has yet reported hard hepatic outcomes | OQ-1 |
| D2 | Cardiovascular disease is the leading cause of death in unselected MASLD; extrahepatic cancer is the largest contributor to excess mortality in biopsy cohorts (PMID 36626630, 33037056, 35418240) | Diagnosis, staging, treatment thresholds and follow-up are all organised within hepatology (PMID 41950980, 41201884) | The disease's leading causes of death sit outside the specialty that owns its care pathway, and no care model has been tested with cardiovascular risk reduction as the primary managed outcome | OQ-2 |
| D3 | PNPLA3 genotype reclassifies indeterminate-FIB-4 patients into a high-risk band and has become an antisense and siRNA target (PMID 36758837, 40334848, 39798707, 40581300) | Risk stratification worldwide is FIB-4 followed by elastography or ELF, containing no genetic information (PMID 41950980, 34001645) | Genotype predicts risk, directs two clinical programmes, and is absent from every stratification pathway | OQ-3 |
| D4 | MASLD requires a cardiometabolic criterion that NAFLD did not (PMID 37363821) | Essentially the entire evidence base was generated under NAFLD/NASH criteria; in one NHANES III analysis NAFLD was not associated with all-cause mortality while MASLD was (PMID 38293788) | Every effect estimate in the field was measured in a population that overlaps but does not coincide with the one now being diagnosed | OQ-4 |
| D5 | 38.5% of MASLD-related HCC arises in non-cirrhotic liver, and surveillance uptake is 32.8% versus 55.7% for other aetiologies (PMID 35255263) | Surveillance eligibility is anchored on cirrhosis in every guideline (PMID 40089151, 37542503) | A large minority of these tumours are outside the surveillance net by construction, and no surveillance strategy has been trialled in the pre-cirrhotic population | OQ-5 |
| D6 | Clinically significant portal hypertension separates 5-year liver-related mortality by a factor of ~26 (21.4% vs 0.8%), and ANTICIPATE-NASH outperforms histology for events (C 0.93 vs 0.67) (PMID 38823501, 41212130) | Trials enrol, and drugs are licensed, by histological fibrosis stage F2–F3 (PMID 38324483, 40305708, 41950980) | The best available risk stratifier is not the one that defines trial eligibility or treatment indication | OQ-6 |
| D7 | Non-invasive test accuracy varies by world region — FIB-4 AUC 0.75 in Latin America vs 0.84 in MENA; Agile 4 0.85 vs 0.96 — within one harmonised cohort (PMID 41100867) | A single global set of thresholds (FIB-4 1.3/2.67; LSM 8/10/15/20 kPa; ELF 9.2/11.3) is recommended everywhere (PMID 41950980, 34001645) | Thresholds derived in one population are exported worldwide despite measured region-dependent accuracy, and no region-specific thresholds have been derived | OQ-7 |
| D8 | Dietary effects on liver fat are up to 3.8-fold larger in PNPLA3 GG than CC individuals and 1.4–3.0-fold larger in the top PRS quartile (PMID 38582304) | Lifestyle advice is delivered uniformly, and the pooled MASH-resolution response to it is 0.12 (PMID 41510965) | Genotype predicts who benefits most from the cheapest intervention, and is used to target nobody | OQ-8 |
| D9 | Bariatric surgery is associated with 10-year major adverse liver outcomes of 2.3% vs 9.6% and MACE of 8.5% vs 15.7% (PMID 34762106) | Incretin therapy now delivers comparable weight loss pharmacologically, with histological endpoints only (PMID 40305708, 38856224) | The two interventions that produce the same weight loss have never been compared, and only one has outcome data | OQ-9 |
| D10 | Patients report three times more stigma from body weight than from the liver diagnosis, and 79.7% do not find "fatty" stigmatising while 85.1% find "alcohol" in a disease name stigmatising (PMID 37984709, 40337974) | The field's principal anti-stigma intervention was a liver-disease rename premised on panellist perceptions of "fatty" and "nonalcoholic" (PMID 37363821) | The stigma patients report and the stigma the rename addressed are different things, and no study spans the change | OQ-16 |
| D11 | Concept-elicitation interviews find fatigue in 18/23, right-upper-quadrant pain in 14/23 and cognitive complaints in 11–13/23 in non-cirrhotic NASH (PMID 33336323) | Guidance, scientific statements and public portals describe early MASLD as asymptomatic and silent (PMID 35418240) | Two incompatible accounts of the same disease stage coexist, and no registration trial carries a patient-reported endpoint that would arbitrate | OQ-17 |
| D12 | MASLD-cACLD listed for transplant at lower MELD despite more portal hypertension, transplanted less often (HR 0.867 at 1 year) and dying on the list more often (HR 1.25), driven by creatinine (PMID 37307997) | Incident CKD risk in MASLD is 1.43-fold, and >40% eGFR decline occurs at 2.98 vs 0.97 per 100 person-years at F4 vs F0–F2 (PMID 33303564, 34670043) | The renal component of MASLD is what eventually moves the allocation score, and no allocation model has been adapted for it | OQ-13 |
| D13 | 39.0% of self-report-classified MASLD carries a phosphatidylethanol concentration in the MetALD/ALD range, essentially all from under-reporting (PMID 40945520) | MetALD carries 1.23–1.62-fold the liver-event hazard of MASLD across 11.6M people and UK Biobank (PMIDs: 40953570, 42120953) | Nobody has reclassified an outcome cohort by biomarker; the measured MetALD excess is a lower bound of unknown size, and the two datasets have never been joined | OQ-14, OQ-31 |
| D14 | Inter-reader κ for the two licensing endpoints is 0.366–0.396 and simulated power falls >90% → 40% (PMID 32610115); paired cores disagree on fibrosis stage in 41% of patients (PMID 15940625) | A long list of MASH drugs is recorded as having failed on those endpoints (PMIDs: 32147362, 31813633, 28833331, 33169409) | No negative MASH trial has been re-analysed with reader and sampling unreliability modelled, so an ineffective drug and an unmeasurable endpoint are currently indistinguishable | OQ-34 |
| D15 | Unsupervised clustering separates a liver-specific from a cardiometabolic MASLD type with distinct transcriptomes, metabolomes and outcome trajectories (PMID 39653777) | Lean MASLD shows a liver-heavy, heart-light outcome profile in population cohorts (PMIDs: 38570344, 41093635) and lean cases carry a proportionally larger genetic burden (PMID 40944478) | Nobody has tested whether lean MASLD is the liver-specific cluster, which would convert an unexplained phenotype into a subtype with a mechanism | OQ-32, OQ-38 |
| D16 | Metabolic surgery reduces 15-year major adverse liver outcomes in compensated MASH cirrhosis (aHR 0.28) (PMID 39870816) | The same surgery raises alcohol use disorder 2.3–7.3-fold and alcohol-related mortality up to 6.2-fold, procedure-dependently (PMIDs: 41066138, 33085738) | No decision model has combined hepatic benefit and alcohol harm for a MASH population, and no surgical liver cohort has measured AUD as an outcome | OQ-37 |
| D17 | Herbal and dietary supplements cause 20% of US hepatotoxicity, with turmeric injury rising and strongly HLA-B*35:01-linked (PMIDs: 27677775, 36252717) | MASLD patients are told they have a liver condition for which, until 2024, no drug existed — the population most likely to self-treat | No MASLD cohort reports adjudicated DILI incidence or systematically collects supplement histories, so a rising transaminase is attributed to progression by default | — |
| D18 | Liver disease activity by cT1 predicts cardiovascular events, atrial fibrillation, heart failure and mortality, including in people without metabolic syndrome, while liver fat by PDFF predicts none of them (PMID 37348789) | MASLD and atherosclerotic cardiovascular disease share every major driver, and Mendelian randomisation supports only a mixed-hyperlipidaemia pathway (PMID 38147315) | The single strongest argument against pure shared-driver confounding rests on one proprietary measurement in one cohort with 2.5 years of follow-up, and has never been replicated with an independent measure of hepatic inflammation | OQ-2 |
| D19 | MASLD is associated with 3.4-fold odds of preterm birth independent of obesity and familial factors, with no gradient by liver severity (PMID 40630617) | Adverse pregnancy outcomes scale with the number of cardiometabolic risk factors across 290,527 women (PMID 41307877) | The two together imply the operative variable is the metabolic milieu rather than the liver lesion — which no antenatal risk model uses, and no guideline mentions | OQ-36 |
Tier 1 — answerable now, would change practice¶
OQ-1. Do the approved drugs change hepatic clinical outcomes, or only histology?¶
Both accelerated approvals rest on histological surrogates (PMID 38324483, 40305708), and Taylor's systematic review concluded that further studies were needed to establish that change in fibrosis stage is a valid trial endpoint (PMID 32027911) — a conclusion not since superseded. The trials that will answer it are running: MAESTRO-NASH to 2028 (NCT03900429), MAESTRO-NASH OUTCOMES in well-compensated cirrhosis to December 2026 (NCT05500222), ESSENCE to April 2029 (NCT04822181). Nothing else in this condition matters as much. → resmetirom-and-thyromimetics.md, glp1-and-incretin-therapy.md, clinical-trials-landscape.md
OQ-2. Who should own cardiovascular risk in MASLD, and does hepatology-led care deliver it?¶
Cardiovascular disease is the leading cause of death in unselected MASLD (PMID 36626630, 35418240) and the risk scales with fibrosis stage (HR 1.45 overall, 2.50 with advanced fibrosis; PMID 34555346) — yet Sanyal's prospective cohort found cardiac events flat across fibrosis stages (PMID 34670043), a contradiction nobody has reconciled. Absence checked 2026-09-02: (NASH OR MASH OR NAFLD OR MASLD) AND ("cardiovascular" OR "coronary" OR "atrial fibrillation" OR "heart failure") AND (risk OR cohort OR meta-analysis) — 5,908 records, none testing a service model with cardiovascular risk reduction as the primary managed outcome.
→ cardiovascular-and-extrahepatic-outcomes.md, guidelines.md
OQ-3. Should genotype enter risk stratification now that it is also a drug target?¶
PNPLA3 GG plus diabetes gives indeterminate-FIB-4 patients a cirrhosis incidence comparable to high-risk FIB-4 and 2.9–4.8× that of CC/CG carriers (PMID 36758837); genetic risk score predicts liver-stiffness trajectory with divergence from age 44 (PMID 40334848); polygenic scores predict HCC independently of cirrhosis (PMID 33248170). AGA states current evidence is inadequate for routine genetic testing (PMID 35842345). Absence checked 2026-09-02: (NAFLD OR NASH OR MASLD OR MASH OR "fatty liver") AND (TM6SF2 OR HSD17B13 OR MBOAT7 OR GCKR OR "GWAS" OR "polygenic risk score") — 1,023 records, all observational; no prospective test of a genotype-augmented pathway.
→ genetics.md, noninvasive-assessment.md
OQ-5. Does HCC surveillance work when the tumour can arise before cirrhosis?¶
38.5% of MASLD-related HCC is non-cirrhotic and only 32.8% of these patients had been surveilled (PMID 35255263); incidence in non-cirrhotic MASLD is 0.1–1.3 per 1,000 patient-years (PMID 33349658), too low for universal surveillance across a third of adults. Candidate stratifiers exist and are all retrospective: polygenic scores at ~90% specificity (PMID 33248170), a 133-gene prognostic liver signature with 15-year HCC incidence of 22.7% vs 0% by risk stratum (PMID 35731891), deep learning on H&E slides detecting risk at mild fibrosis (PMID 38768142). Absence checked 2026-09-02: (NASH OR MASH OR NAFLD OR MASLD) AND ("hepatocellular carcinoma") AND (cirrhosis OR "non-cirrhotic" OR surveillance OR incidence) — 5,216 records, no randomised surveillance trial in a non-cirrhotic MASLD population.
→ masld-related-hepatocellular-carcinoma.md, hepatocellular carcinoma
OQ-6. Should portal hypertension or a validated event model replace fibrosis stage for trial entry and treatment indication?¶
CSPH (HVPG ≥10 mmHg) separates 5-year liver-related mortality 21.4% vs 0.8%, and variceal bleeding did not occur without it (PMID 38823501). ANTICIPATE-NASH achieves C statistic 0.93 for liver-related events against 0.67 for histology, with histology adding nothing to the model, and identifies thresholds above which F3 patients have events and below which F4 patients do not (PMID 41212130). Trials still enrol by stage. → cirrhosis-and-decompensation.md, clinical-trials-landscape.md
OQ-7. Why does non-invasive test accuracy vary by world region, and should thresholds be region-specific?¶
Within one harmonised 41-country cohort of 17,792 biopsy-confirmed patients, FIB-4's AUC for advanced fibrosis ranged 0.75–0.84 and Agile 4's for cirrhosis 0.85–0.96 (PMID 41100867). Candidate explanations — biopsy practice, fibrosis-stage distribution, platelet reference ranges, genetic background, assay standardisation — have never been separated. A single global threshold set is recommended everywhere. → noninvasive-assessment.md, guidelines.md
OQ-9. Does metabolic surgery beat incretin therapy?¶
BRAVES showed surgery beat lifestyle plus best medical care (56–57% vs 16% MASH resolution; PMID 37088093), and SPLENDOR associates surgery with 10-year major adverse liver outcomes of 2.3% vs 9.6% (adjusted HR 0.12) and MACE 8.5% vs 15.7% (HR 0.30) (PMID 34762106). Semaglutide achieves 62.9% resolution with −10.5% weight (PMID 40305708). Absence checked 2026-09-02: (NASH OR MASH OR NAFLD OR MASLD) AND (bariatric OR "metabolic surgery" OR "sleeve gastrectomy" OR "gastric bypass") AND (histolog* OR fibrosis OR outcomes) — 1,210 records, no head-to-head trial. This is now the most consequential unanswered comparison in MASH treatment.
→ bariatric-and-metabolic-surgery.md, glp1-and-incretin-therapy.md
OQ-10. Does fibrosis screening in diabetes clinics improve outcomes?¶
The ADA calls for fibrosis screening in prediabetes and type 2 diabetes as a new standard of care (PMID 40434108), on the basis of prevalence (65% MASLD, 14% advanced fibrosis, 6% cirrhosis in prospectively assessed adults ≥50 with T2D; PMID 36410554) rather than a screening trial. Cost-effectiveness modelling exists (PMID 41196592). Absence checked 2026-09-02: (NASH OR MASH OR NAFLD OR MASLD) AND ("type 2 diabetes") AND (bidirectional OR incident OR screening OR prevalence OR risk) — 4,936 records, no randomised trial of a screening programme with clinical endpoints.
→ masld-and-type-2-diabetes.md, noninvasive-assessment.md
OQ-11. Should pioglitazone be first-line in MASH with type 2 diabetes?¶
Cusi's trial produced a treatment difference of 41 percentage points on its primary histological endpoint, 51% MASH resolution and fibrosis improvement of 0.5 stages (p=0.039), sustained over 36 months — arguably a larger effect than resmetirom's (PMID 27322798). It is a single-centre trial of 101 patients, weight gain is 2.5 kg, and discontinuation is associated with recurrence (PMID 36495442). Pioglitazone appears in Latin American and endocrinology guidance and is absent from the 2026 global hepatology consensus. → other-pharmacotherapy.md, guidelines.md
OQ-12. What replaces FIB-4 at the age extremes?¶
FIB-4's AUC for liver stiffness ≥8 kPa was 0.51 in young people and 0.55 in older adults, versus 0.86 and 0.75 for MAF-5 (PMID 38513745); age-adapted thresholds lowered both NPV and PPV in every algorithm tested (PMID 39887699); and in an unselected primary-care population FIB-4's rule-out sensitivity was 53.8% (PMID 39674225). MAF-5 and LiverPRO both outperform it and have not been compared head-to-head in a screening programme with biopsy or outcome adjudication. → noninvasive-assessment.md, red-flags-and-safety-concerns.md
OQ-13. Should MELD be adapted for MASLD?¶
MASLD-cACLD patients are listed at lower MELD despite greater portal hypertension, transplanted less often (HR 0.867 at 1 year, 0.672 at MELD >30) and die on the waiting list more often (HR 1.25 at 1 year), with creatinine rather than bilirubin the driver of score rise (PMID 37307997). No prospective evaluation of MELD 3.0's effect on this disadvantage has been retrieved. → cirrhosis-and-decompensation.md
OQ-14. How much alcohol is being misclassified, in both directions?¶
MetALD is defined by self-reported grams per week; the expert position statement recommends AUDIT-C and phosphatidylethanol confirmation and warns that a single metabolic criterion above the weekly alcohol thresholds may attribute to metabolism a disease driven by alcohol (PMID 39608457). Meanwhile mild-to-moderate alcohol use in metabolic syndrome is associated with worse liver outcomes (PMID 36063967). Every MetALD prevalence and outcome estimate in this knowledge base is exposure-misclassified by an unmeasured amount.
Partly answered, 2026-09-02 — and the magnitude is large. In 2,924 prospectively recruited at-risk adults, phosphatidylethanol showed under-reporting in 39.5% of an alcohol-recruited group and 11.1% of a metabolically-recruited group, with essentially no over-reporting (0.7% and 0.1%); of 1,433 participants classified as MASLD by self-report, 559 (39.0%) had PEth in the MetALD or ALD range (Torp 2025, PMID 40945520). Because the error is one-directional, MASLD cohorts are contaminated with alcohol-associated disease while ALD cohorts are not contaminated with MASLD, so the reported MetALD-versus-MASLD hepatic hazard ratios of 1.23–1.62 (PMIDs: 42120953, 40953570) are lower bounds. What remains open: no outcome cohort has been reclassified by biomarker; the 20/80/200 ng/mL cut-offs are conventions rather than outcome-anchored thresholds; PEth covers only the preceding 1–4 weeks, so lifetime exposure is still unmeasured; and no guideline requires the test. → nomenclature-and-definitions.md, red-flags-and-safety-concerns.md
OQ-15. Is statin underuse in MASLD measurable, and what does it cost?¶
The safety question is settled — 22 studies and 2,345 patients show ALT falling 35.4%, AST 31.8% and GGT 25.6% on statins (PMID 34133035) — and underuse is asserted in the field as "safety concerns, decreased awareness or both" (PMID 34741325). No study has quantified the prescribing gap in MASLD against comparable non-MASLD cardiovascular risk, or the excess events attributable to it. This is a cheap, designable study addressing the leading cause of death in the condition. → other-pharmacotherapy.md, cardiovascular-and-extrahepatic-outcomes.md
OQ-16. Did the rename change patient experience, and was it aimed at the right stigma?¶
The stigma rationale rested on panellist perception — 61% "nonalcoholic", 66% "fatty" (PMID 37363821). Patients report 8% liver-disease stigma against 26% weight stigma (PMID 37984709, n=1,976), and in a second survey 79.7% did not find "fatty" stigmatising while 85.1% found "alcohol" in a disease name stigmatising (PMID 40337974, n=222). No study spans the nomenclature change. Method note: an earlier absence claim here — that no interview-based literature existed — was asserted from one search formulation and was false; the corrected search is recorded in literature/patient-voice/sources.md. → patient-experience-and-advocacy.md, nomenclature-and-definitions.md
OQ-17. Is early MASLD actually asymptomatic?¶
Guidance and public statements describe it as silent until advanced (PMID 35418240). Concept-elicitation interviews in non-cirrhotic NASH elicit fatigue in 18/23, right-upper-quadrant pain in 14/23, impaired memory in 13/23, poor sleep in 12/23 and reduced focus in 11/23 (PMID 33336323), and both the NHS and Liver Canada patient pages list fatigue and abdominal discomfort. The two accounts have never been reconciled. The resolution determines whether patient-reported outcomes belong in registration trials, where they are currently absent as endpoints (PMID 36536958). → patient-experience-and-advocacy.md, literature/patient-voice/themes.md
OQ-18. Should the F2–F3 treatment window be redrawn?¶
It derives from trial entry criteria, not from a risk analysis. ANTICIPATE-NASH identifies F3 patients above a threshold who have events and F4 patients below it who do not (PMID 41212130); CSPH separates outcomes far more sharply than F3-vs-F4 (PMID 38823501); and the licensed population is defined histologically but must be identified non-invasively, with a large indeterminate zone (PMID 36375686, 41201884). → guidelines.md, cirrhosis-and-decompensation.md
OQ-31. Does MASLD add mortality risk beyond the cardiometabolic dysfunction that defines it?¶
Because cardiometabolic dysfunction is now a criterion rather than a confounder, comparing MASLD against a general population is confounded by construction. The one study that restricted the comparator to people who also carry a cardiometabolic risk factor found MASLD null for all-cause (aHR 1.04, 0.95–1.14), cardiovascular and cancer mortality over 26.7 years of NHANES III follow-up, while MetALD was not (all-cause 1.41, 1.05–1.89; liver 15.04, 2.96–76.35) (Kwak 2025, PMID 38739848). Every other mortality cohort in this knowledge base uses unmatched controls. The stage-specific hepatic gradients are unaffected; what is in question is whether unstaged MASLD carries mortality risk at all. Designable today: repeat the metabolically-matched design in a biopsy-staged cohort. → natural-history-and-fibrosis-progression.md, overview.md, epidemiology-and-burden.md
OQ-32. Is MASLD one disease, or two?¶
Unsupervised clustering on six routine clinical variables in 1,389 people with obesity, applied to three independent biopsy cohorts (n=1,099) and to UK Biobank, separates a genetically-linked liver-specific type (rapid hepatic progression, limited cardiovascular risk) from a cardiometabolic type (same baseline liver phenotype, higher cardiovascular and diabetes risk), with distinct liver transcriptomes (n=831) and plasma metabolomes (n=1,322) (Raverdy 2024, PMID 39653777). If it replicates, the population-average statement that "most people with MASLD die of cardiovascular disease" conceals two opposite risk profiles, and the lean-MASLD organ split becomes a special case rather than an anomaly. No prospective validation and no therapy allocated by cluster. → nomenclature-and-definitions.md, lean-masld.md, genetics.md
OQ-33. Should an SGLT2 inhibitor be treated as MASH therapy?¶
Dapagliflozin 10 mg for 48 weeks met all three histological endpoints in 154 biopsy-diagnosed MASH patients — MASH improvement without fibrosis worsening 53% vs 30% (RR 1.73, 1.16–2.58), resolution 23% vs 8% (RR 2.91, 1.22–6.97), fibrosis improvement 45% vs 20% (RR 2.25, 1.35–3.75) — with 1% discontinuation for adverse events (Lin 2025, PMID 40467095, NCT03723252). The effect sizes equal or exceed the approved drugs' from a generic-track agent already prescribed to much of this population. Absence checked 2026-09-02: no guideline in literature/guidelines/REGISTRY.md cites it, and no confirmatory trial outside China was retrieved. → other-pharmacotherapy.md, guidelines.md, masld-and-type-2-diabetes.md
OQ-34. Are negative MASH trials negative, or unmeasurable?¶
Inter-reader κ for the two licensing endpoints is 0.396 (MASH resolution without worsening fibrosis) and 0.366 (fibrosis improvement without worsening MASH) in a real trial dataset of 678 biopsies read by three hepatopathologists; 46.3% of enrolled patients failed at least one other reader's entry criteria; the observed treatment effect was smallest for the features with the worst reliability; and simulation puts the resulting power loss at >90% → 40% (Davison 2020, PMID 32610115). Sampling error is larger still: paired cores taken at the same session differ by ≥1 fibrosis stage in 41% of patients (Ratziu 2005, PMID 15940625). No published negative MASH trial has been re-analysed with reader and sampling unreliability modelled. This is designable today from existing trial datasets and would change how the whole failure literature is read. → histology-and-biopsy.md, clinical-trials-landscape.md, other-pharmacotherapy.md
OQ-35. Is resmetirom worth its price, and which model is right?¶
Two lifetime models of the same drug differ ~5-fold on QALYs gained (+1.24 vs +0.26) and land on opposite sides of $100,000/QALY ($53,929 vs $140,134) (PMIDs: 36104546, 40577015). In the later model, lower discontinuation makes the drug less cost-effective (ICER $318,740/QALY with none), supporting an annual price of $5,645–$10,619. Real-world access data show 24.1% requiring insurance appeals, 12.5-day dispensing delays and payers demanding invasive fibrosis testing (PMID 40688390). Only the MAESTRO-NASH outcome phase can adjudicate, because both models extrapolate the same 52-week surrogate. → resmetirom-and-thyromimetics.md, clinical-trials-landscape.md
OQ-36. Should MASLD change antenatal management?¶
Preterm birth occurred in 16.7% vs 4.7% of births to women with biopsy-proven MASLD (aOR 3.41, 1.98–5.88), independent of obesity (aOR 4.60, 2.00–10.60 against overweight/obese comparators) and confirmed in sibling analyses, with no excess stillbirth, malformation or neonatal death and no gradient with liver disease severity (Marxer 2025, PMID 40630617). In 290,527 Korean women, adverse pregnancy outcomes rose 2.44-fold with MASLD and scaled with the number of cardiometabolic risk factors (PMID 41307877). The absence of a severity gradient means liver staging would be the wrong triage variable. Absence checked 2026-09-02: no hepatology or obstetric guideline retrieved identifies MASLD as an antenatal risk factor. → red-flags-and-safety-concerns.md, guidelines.md
OQ-37. Can compensated MASH cirrhosis be treated after all?¶
Every randomised drug trial in compensated MASH cirrhosis has missed its primary endpoint, but metabolic surgery has not: in SPECCIAL, 62 surgical versus 106 non-surgical patients with histologically proven compensated MASH cirrhosis had 15-year major-adverse-liver-outcome incidence of 20.9% vs 46.4% (aHR 0.28, 0.12–0.64) and decompensation 15.6% vs 30.7% (aHR 0.20, 0.06–0.68) over a mean 10.0 ± 4.5 years (Aminian 2025, PMID 39870816). It is observational, single-system and small. The question is no longer whether cirrhosis is modifiable but whether the effect survives randomisation — and whether the alcohol-use-disorder hazard of surgery (aHR 2.28–7.29; PMIDs: 41066138, 33085738) offsets it in a liver population. → bariatric-and-metabolic-surgery.md, cirrhosis-and-decompensation.md
Tier 2 — blocked on tools, numbers, or a Tier-1 answer¶
OQ-4. How much does the NAFLD-to-MASLD reclassification move measured effects?¶
Overlap is ≥94% (κ 0.968; PMID 38286339), yet in the same NHANES III sample NAFLD was not associated with all-cause mortality while MASLD was (aHR 1.19, 1.06–1.34; PMID 38293788). Answerable by re-analysing existing cohorts under both definitions; largely not done at scale. Until it is, every figure in this condition carries a definitional caveat. → nomenclature-and-definitions.md, epidemiology-and-burden.md
OQ-8. Is a single cardiometabolic criterion enough?¶
One criterion suffices for diagnosis, and 95% of one tertiary cohort qualified on BMI alone (PMID 38286339). Whether requiring ≥2 criteria would improve prognostic discrimination has not been tested prospectively — and the MetALD position statement warns specifically against single-criterion diagnosis above the alcohol thresholds (PMID 39608457). → nomenclature-and-definitions.md
OQ-19. Is MASH cirrhosis modifiable at all?¶
Three cited randomised trials in compensated MASH cirrhosis have missed their primary endpoint — semaglutide (PMID 36934740), efruxifermin at 36 weeks (PMID 40341827), and selonsertib STELLAR-4 (PMID 32147362) — and a large observational analysis found no GLP-1 benefit once cirrhosis was established, versus benefit before it (PMID 39283612). Efruxifermin's week-96 signal (29% vs 11%) favours "too short" over "irreversible". MASLD-specific recompensation has now been observed, but not causally established, in an observational cohort (PMID 42467948). Blocked on an adequately powered, sufficiently long positive trial; phase 3 cirrhosis programmes are recruiting (NCT06528314, NCT06419374, NCT06632457). → cirrhosis-and-decompensation.md
OQ-20. Can a reproducible definition of steatohepatitis be constructed?¶
The histological label loses prognostic association after fibrosis adjustment (PMID 27616339, 38293684), and the field's own diagnosis is that this reflects subjectivity: inter-rater κ is 0.45 for lobular inflammation and 0.56 for ballooning against 0.84 for fibrosis (PMID 15915461), and ballooning is the least reproducible feature of all (PMID 37017559). Akbari's authors call explicitly for "more objective means by which to define NASH". Blocked on digital-pathology validation (PMID 37789057, 40262132). → histology-and-biopsy.md, clinical-trials-landscape.md
OQ-21. Why do only a minority progress?¶
About 25% of MASLD develops steatohepatitis and 3–5% reaches cirrhosis (PMID 32044315, 39243773). Partial explanations exist — loss of hepatic mitochondrial flexibility at the steatohepatitis transition (31–40% lower maximal respiration; PMID 25955209), lipid-species composition (PMID 29154964), genotype and its interactions with age, sex, BMI and diabetes (PMID 40998180). No integrated model predicts progression at the individual level. → pathogenesis.md, natural-history-and-fibrosis-progression.md
OQ-22. What are the seven NAFLD subtypes?¶
PheWAS across 17 GWAS loci suggested at least seven subtypes (PMID 37709864). None has been characterised clinically, and no study has tested differential treatment response. Blocked on cohort-scale phenotyping. → genetics.md
OQ-23. Why is TM6SF2 hepatotoxic and cardioprotective, and does that matter for drugging it?¶
The variant raises liver fat while lowering circulating lipids and cardiovascular risk, reaches genome-wide significance for advanced fibrosis cross-sectionally (PMID 32298765), and yet does not influence histological progression in serial biopsies (PMID 40998180). The discordance is unexplained and would matter for any therapy mimicking it. → genetics.md
OQ-24. Does combination therapy add?¶
The 2026 global consensus recommends against upfront combination (PMID 41950980) and AASLD states resmetirom plus semaglutide has not been studied (PMID 41201884), while network meta-analysis ranks cilofexor+firsocostat and cilofexor+selonsertib among the top three for fibrosis regression (PMID 39903735) — consistent with the multi-node structure of fibrogenesis (PMID 32044315). One registered trial combines mechanisms (NCT07527910, ALN-PNP ± tirzepatide). → other-pharmacotherapy.md, clinical-trials-landscape.md
OQ-25. What proportion of "cryptogenic" cirrhosis is burnt-out MASLD?¶
Steatosis is lost as MASLD reaches cirrhosis, making aetiology unassignable on morphology exactly where the burden is greatest (PMID 33111374, 40113099). Every estimate of MASLD-attributable cirrhosis is therefore biased downward by an unquantified amount. Blocked on a biomarker that recovers the lost attribution. → cirrhosis-and-decompensation.md, histology-and-biopsy.md
OQ-26. Why does MASLD prevalence peak at moderate socio-demographic index?¶
GBD shows MASLD prevalence peaking at moderate SDI (PMID 40062742) and a 479-study meta-analysis finds higher human development index associated with lower prevalence (coefficient −0.523, p=0.005; PMID 39094335) — the opposite of a simple affluence model. No study has separated nutrition-transition from surveillance-intensity explanations. → epidemiology-and-burden.md
OQ-27. What is the incidence outside East Asia?¶
The pooled global incidence estimate rests on 63 studies of which 62 are from China/Hong Kong, South Korea and Japan (PMID 37040843). No comparably powered incidence estimate exists for Africa, South Asia or Latin America — the last having the highest reported prevalence (44.37%; PMID 36626630). Blocked on cohort infrastructure. → epidemiology-and-burden.md
OQ-28. Should lean MASLD be treated differently?¶
Liver-related mortality is higher in lean than non-lean disease (1.33 vs 0.76 per 1,000 patient-years; OR 3.56) while cardiovascular risk is lower (aHR 0.73 vs non-lean, 0.99 vs normal liver) (PMID 40087205, 38570344) — the opposite of the profile BMI-anchored screening assumes. Weight loss as a lever is constrained in a lean patient, and bile-acid-axis modulation improved steatohepatitis in a lean murine model (PMID 31442319). Absence checked 2026-09-02: (NASH OR MASH OR NAFLD OR MASLD) AND (lean OR "non-obese" OR "normal weight" OR "normal BMI") — 2,043 records, no BMI-stratified interventional trial.
→ lean-masld.md
OQ-29. Does microbiome modification change liver histology in humans?¶
The mechanistic literature is largely murine and the human literature largely associative — gut vascular barrier disruption as a prerequisite in mice (PMID 31419514), microbiome signatures predicting cirrhosis and advanced fibrosis in humans (PMID 32610095, 28467925), with the central methodological problem of separating microbial signatures from the metabolic disorders they accompany (PMID 32152478). Absence checked 2026-09-02: (NAFLD OR NASH OR MASLD OR MASH) AND (microbiome OR microbiota OR "gut-liver") — 4,335 records; human interventional evidence remains small randomised faecal-transplant and diet studies without histological endpoints.
→ pathogenesis.md
OQ-30. Should the two vocabularies be reconciled?¶
APASL and Chinese guidance use MAFLD; AASLD, EASL, ALEH, NICE, Japan and the global consensus documents use MASLD; the populations differ measurably (18.5% NAFLD / 19.3% MAFLD / 20.8% MASLD in one sample, with divergent mortality associations; PMID 38293788). No body has proposed a reconciliation procedure, and the 2025 Delphi across 61 documents did not attempt one. → nomenclature-and-definitions.md, guidelines.md
OQ-38. Is cardiovascular risk in lean MASLD lower, equal, or higher?¶
Population-based cohorts say lower (aHR 0.73–0.89; PMIDs: 38570344, 41093635), pooled adjusted analysis says no difference (HR 0.89, 0.77–1.02; PMID 39117942), and two US health-system cohorts say higher (HR 1.21–1.48; PMIDs: 41614694, 40406910). Two independent syntheses simultaneously report fewer CVD events and more CVD deaths in lean disease (PMIDs: 41093635, 38599554), implying higher case fatality, for which no mechanism has been proposed. The split maps onto data source rather than onto population, and no study has compared the designs within one population. Blocked on a cohort with both symmetric ascertainment and clinical-grade phenotyping. → lean-masld.md, cardiovascular-and-extrahepatic-outcomes.md
OQ-39. Can at-risk MASH be identified without a biopsy at all?¶
Across 17 biomarkers and multimarker scores compared head-to-head against paired histology in 966 patients, none reached the prespecified AUC 0.80 for at-risk NASH (best: SomaSignal 0.81, 0.75–0.86), while three exceeded it for advanced fibrosis (SomaSignal 0.90, ADAPT 0.85, LSM 0.83) (Vali 2023, PMID 36958367). A superlearner ensemble over 23 routine clinical variables reached only 0.74–0.79 and did no better than SAFE, implying the remaining error is in the inputs rather than the models (Charu 2024, PMID 38687634). Blocked on new analytes: candidate mechanism-derived panels exist (IGFBP7/SSc5D/Sema4D; PMID 38811591) but are unreplicated in an independent biopsy cohort. Until then every registration trial must biopsy to enrol, at a 33% screen-failure rate even with optimal prescreening. → noninvasive-assessment.md, clinical-trials-landscape.md
OQ-40. Diacylglycerol or ceramide?¶
Both have strong mechanistic support for causing hepatic insulin resistance — DAG–PKC-ε from human and rodent flux studies (PMID 28867301), ceramide from a DES1 double-bond deletion that isolates the causal moiety and resolves both steatosis and insulin resistance (PMID 31273070) — and both derive from the same fatty-acid excess. No human study has measured hepatic DAG, PKC-ε activation and ceramide species in the same tissue with a functional insulin-sensitivity readout. Each supports a different drug class, so the answer is not academic. Blocked on tissue access and assay standardisation. → pathogenesis.md
OQ-41. Do the loss-of-function protective genes translate?¶
CIDEB rare-variant burden lowers cirrhosis odds to 0.50 (0.36–0.70) across aetiologies (PMID 35939579), and rare predicted-loss-of-function variants in MTARC1 and GPAM are protective, with GPAM corroborated in the largest cirrhosis GWAS (PMIDs: 36280732, 38632349). All three show the HSD17B13 pattern — protection by loss of function, across aetiologies and severities — which is the easy direction to drug. Absence checked 2026-09-02: no clinical-stage programme with human liver endpoints was retrieved for any of them. → genetics.md
OQ-42. Is clonal haematopoiesis a modifiable liver risk factor?¶
CHIP roughly doubles chronic liver disease risk (OR 2.01, 1.46–2.79) across 214,563 exome-sequenced individuals in four cohorts, with supporting Mendelian randomisation (OR 2.37, 1.57–3.60) and a Tet2/NLRP3 macrophage mechanism in a dietary NASH model (Wong 2023, PMID 37046084). This makes an age-acquired bone-marrow mutation a hepatic fibrosis risk factor operating through innate immunity, captured by no germline risk score. Blocked on: CHIP status is not measured in any MASLD cohort with paired histology, and NLRP3 inhibition has not been tested in carriers. → genetics.md, pathogenesis.md
OQ-43. How should cognitive dysfunction in MASLD be measured?¶
Patients report memory and concentration problems unprompted (13/23 and 11/23 in concept-elicitation interviews; PMID 33336323), and a dedicated liver–brain-axis review finds consistent associations between MASLD and cognitive dysfunction with inflammation, vascular disease and brain ageing as candidate contributors — while stating that there is no agreed test for diagnosing it, no established correlation with MASLD stage, and no evidence that any MASLD therapy improves it (Mikkelsen 2025, PMID 39701123). Blocked on instrument development. This is the clearest case in this condition of a patient-reported symptom with no measurement tool. → patient-experience-and-advocacy.md
Note to the next sweep¶
Re-run every asserted absence above with both vocabularies and with at least two search formulations before restating it. The one absence claim in this build that was checked with a second formulation turned out to be false (OQ-16), which is a reasonable base rate to assume for the rest.