Fatty Liver (MASLD) and Weight Loss
Opens the masld cluster. MASLD (renamed from NAFLD in 2023) is «steatotic liver disease (SLD) in the
presence of one or more cardiometabolic risk factor(s) and the absence of harmful alcohol intake» —
so a person with obesity + metabolic syndrome effectively has it by definition. The evidence was
«transferred» from NAFLD given «a nearly complete overlap» (99.8%).
(European Association for the Study of the Liver, 2024)
The cardiometabolic criteria (Table 3, adults; any one qualifies): waist «>=94 cm in men and >=80 cm in women (Europeans)» or BMI >=25; prediabetes/T2D (fasting glucose 5.6-6.9 / >=7.0 mmol/L); triglycerides «>=1.7 mmol/L»; low HDL; BP «>=130/85 mmHg». These are classification thresholds (they mark whether a factor counts), not outcome targets. (European Association for the Study of the Liver, 2024)
The lever, and its dose-response — the actionable finding
Weight loss is the base-layer treatment, and the histological benefit is dose-dependent and outcome-specific. EASL’s ladder (LoE 2, strong recommendation):
| Sustained weight loss | What it achieves |
|---|---|
| >=5% | reduces liver fat (steatosis) |
| 7-10% | improves liver inflammation (steatohepatitis / MASH) |
| >=10% | improves fibrosis |
The recommendation: weight loss «should aim at a sustained reduction of [>=5%] to reduce liver fat, 7-10% to improve liver inflammation, and [>=10%] to improve fibrosis»; «There is a dose-dependent association between the amount of weight loss and the extent of improvement in biomarkers of liver damage.» (European Association for the Study of the Liver, 2024)
Three caveats the guideline attaches, all decision-relevant:
- The ladder rests on a single histology-endpoint trial («A stringent interventional trial with histological end-points suggested…» — its ref 280). This is a guideline threshold traceable to one RCT, not a pooled effect with intervals — read it as name-the-curve guidance, not a precise dose.
- Few reach it, and it regains. «a limited proportion of individuals achieve a weight reduction of [>=5%]»; follow-up shows «maximal weight loss at 6 months, followed by a gradual weight regain to a net weight loss of about 5% at 12-24 months and partial regain of liver lipid content and stiffness.»
- No proven effect on advanced fibrosis/cirrhosis — «evidence for an effect of weight reduction by lifestyle modification on advanced fibrosis or cirrhosis is insufficient». So the lever works on the reversible stages; once bridging fibrosis/cirrhosis is present, lifestyle is not shown to reverse it. (European Association for the Study of the Liver, 2024)
The second guidance family — AASLD 2023 agrees on the ladder’s shape, and it is un-GRADEd [AASLD 2023]
A second national guidance body (AASLD, American) states the same weight-loss ladder, which raises the guidance-null strength — but with three decision-relevant qualifications: it is not GRADEd, it rests partly on the same single trial as EASL, and it draws the middle rung differently. AASLD: «Weight loss of 3%–5% improves steatosis, but greater weight loss (> 10%) is generally required to improve NASH and fibrosis.» (Rinella et al., 2023)
Parameter comparison (built before the cross-source claim — same quantity in each row = weight-loss % → histological outcome):
| Rung | EASL 2024 (GRADEd, LoE 2 strong) | AASLD 2023 (un-GRADEd Guidance) |
|---|---|---|
| Steatosis (liver fat) | sustained >=5% | 3-5% (slightly lower floor) |
| Steatohepatitis (NASH) | 7-10% (its own rung) | folded into the >10% tier — no distinct NASH rung |
| Fibrosis | >=10% | >10% (≈equal) |
The two agree on the shape (more loss for worse disease) and on the fibrosis rung; they differ in the middle — EASL gives NASH a 7-10% rung, AASLD requires >10% for both NASH and fibrosis. This is not a tension to file: it is the same curve read at different granularity, and both trace partly to the same histology trial (Vilar-Gomez 2015 — EASL’s ref 280, among AASLD’s refs 262/367-370), so it is shared-evidence refinement, not two independent readings.
Two caveats that bound how much this second body adds:
- Un-GRADEd Guidance, not a Guideline. AASLD states: «A “Guidance” differs from a “Guideline” in that it is not bound by the Grading of Recommendations, Assessment Development and Evaluation system. Thus, actionable statements rather than formal recommendations are provided herein.» So concordance with EASL is expert-consensus agreement, not a second structured GRADE appraisal — it strengthens the guidance-null less than two GRADEd bodies would. (Rinella et al., 2023)
- Not independent-E. Both bodies rest on the same single histology-endpoint trial, so the agreement is shared-source (a second guidance family, one evidence base), never independent-backing robustness.
AASLD refines three of the lever arms EASL already names:
- Exercise gets an intensity ladder mirroring the weight-loss ladder. «Exercise, independent of weight loss, has hepatic and cardiometabolic benefit … more vigorous exercise is needed to improve NASH histology, with even higher intensity exercise needed to reduce fibrosis.» This refines EASL’s arm (EASL: exercise reduces steatosis even without weight loss, HIIT ≈ moderate; histology LoE 5): AASLD adds that the intensity required rises with disease severity — more vigorous for NASH, higher still for fibrosis. Both bodies keep the histology/fibrosis benefit weakly evidenced. (Rinella et al., 2023)
- Alcohol’s modest-protective signal is superseded, and F2+ means abstinence. Where Peng’s umbrella (below) carries a modest-alcohol protective association (NASH OR 0.50), AASLD reports the same reversal the U/J-artifact caveat predicts: «Earlier epidemiological studies suggested a protective effect of mild alcohol consumption … but in a subsequent study, moderate alcohol use (defined broadly as > 20 g/d) was associated with less improvement in steatosis and aspartate aminotransferase (AST) and lower odds of NASH resolution.» And a clinical bright line: «Patients with clinically significant hepatic fibrosis (≥ F2) should abstain from alcohol use completely.» -> Alcohol and Mortality and Vascular Disease. (Rinella et al., 2023)
- Fructose — the guideline side asserts the molecule-specific harm Chung’s isocaloric trials did not find. AASLD: «Excessive fructose consumption in particular increases the risk of NAFLD, NASH, and advanced fibrosis independent of calorie intake» (observational, refs 380-382). This sits opposite Chung’s isocaloric controlled-feeding null (below) — but they are not cleanly joined: AASLD’s claim is observational and about long-term disease incidence, whereas Chung’s null is short-term isocaloric RCT on liver fat. AASLD thus populates exactly the chronic/observational regime the fructose [REVIEW] flag says Chung’s short-term isocaloric trials cannot cover — a horizon/design difference, not a settled contradiction. The higher-quality design for the net-of-energy question remains Chung’s isocaloric RCT; the decision-relevant lever stays cutting excess liquid energy. (Rinella et al., 2023)
The surrogate gap, one level deeper — from AASLD’s own mouth. The Peng section already shows the lifestyle treatments move surrogates (liver fat, enzymes), not histology. AASLD extends the ladder one rung further out: even histology is a surrogate for clinical outcomes. «Additional studies are needed to better understand the long-term association among changes in liver fat, histological response, and clinical outcomes.» So the evidence chain is liver-fat → histology → (unproven) → cirrhosis / liver mortality / all-cause mortality, and the guidance body itself flags the last link as unestablished. The patient-important endpoints (liver-related death, all-cause mortality, decompensation) are the under-measured dark region -> Surrogate Outcomes; the histological “response” the trials report is two surrogate-steps short of them. (Rinella et al., 2023)
The drug comparator — drugs hit the same surrogate ceiling; only bariatric surgery is shown on mortality [AASLD 2023]
Sizing the weight-loss rock at the margin (Layer 1) means asking what a mature drug already captures. For MASLD the answer is the surrogate but not the outcome: the pharmacological levers improve NASH histology but, like the lifestyle levers, are not shown to reduce fibrosis — the stage that carries the liver risk — so they do not shrink the weight-loss rock on the outcome that matters.
- Semaglutide (phase 2b RCT, 320 NASH F1-3, 72 wk): «NASH resolution was dose dependent and occurred in 59% in the treatment group versus 17% in the placebo group (p < 0.001). Despite evidence of fibrosis improvement in the treatment groups, there was no statistically significant reduction in fibrosis compared with placebo; however, a dose-dependent decrease in progression was observed.» So the fibrosis stage was not significantly reduced (the antifibrotic column stays empty below), though progression slowed dose-dependently — a slowing, not a reversal. (Rinella et al., 2023)
- The whole antifibrotic column is empty: «Available data on semaglutide, pioglitazone, and vitamin E do not demonstrate an antifibrotic benefit, and none has been carefully studied in patients with cirrhosis.» And «Metformin, UDCA, DPP-4, statins, and silymarin … should not be used as a treatment for NASH as they do not offer a meaningful histological benefit.» (Rinella et al., 2023)
- Bariatric surgery is the exception, and the only lever here that reaches a patient-important endpoint. «Resolution of NASH without worsening of fibrosis occurred in 80% of patients 1 year following bariatric surgery, which was maintained at 5 years»; Table 7 adds it «reduces mortality from CVD and malignancy.» Surgery is the aggressive form of the same weight-loss lever — so the mortality signal, where it exists, tracks large sustained weight loss, not a distinct drug mechanism. (Rinella et al., 2023)
So the drug landscape does not shrink the weight-loss rock on the decision that matters (fibrosis / liver outcomes): GLP-1 RAs offer a NASH-resolution alternative on the surrogate and carry their own cardiovascular benefit -> Semaglutide for Cardiovascular Risk in Obesity, but no agent is shown to reverse fibrosis, and the one lever with a mortality signal (bariatric surgery) is itself large sustained weight loss. The weight-loss lever stays non-substitutable for the fibrosis/outcome question. (inferred from Rinella et al., 2023)
The other lifestyle levers, with their grades
- Diet quality (LoE 2, strong): a Mediterranean pattern, limiting ultra-processed food and avoiding sugar-sweetened beverages. It «has repeatedly been shown to provide hepatic… benefits… even without weight loss»; low-carb and low-fat are «similarly effective» for liver fat, with the Mediterranean diet the preferred, more-maintainable option. Very-low-carbohydrate/ketogenic diets have «insufficient evidence on the efficacy or safety» (with «potential cardiovascular, kidney and other side effects»); time-restricted eating has «very little evidence for a beneficial effect of time-restricted eating over regular caloric restriction» on liver fat — the two were previously conflated under one quote (corrected 2026-07-30). But: «little evidence that improving diet quality beneficially impacts clinical liver-related outcomes» (LoE 3).
- Exercise (LoE 1 for steatosis; only LoE 5 for histology): «>150 min/week of moderate or 75 min/week of vigorous» — and «exercise alone, without… significant weight loss, reduces liver steatosis». HIIT and moderate-intensity are «equally effective»; «combination of aerobic plus resistance training is preferred». The histological (MASH/fibrosis) benefit of exercise is far less established than the steatosis one.
- Alcohol: discouraged in all SLD, and the SLD definition itself sets the alcohol ceiling — above it the disease is reclassified as MetALD then ALD -> Alcohol and Mortality and Vascular Disease.
- Coffee (LoE 4, observational only): «>=3 cups of coffee per day» associated with lower fibrosis markers, «stronger and more consistent for fibrosis than for steatosis» — but «RCTs… are scarce and negative or inconclusive, no firm conclusions can be drawn». (European Association for the Study of the Liver, 2024)
Macronutrient composition at fixed calories — the fat type and the carb->protein swap move liver fat; the total fat<->carb swap does not [Winters-van Eekelen 2020]
EASL calls low-carb and low-fat «similarly effective» for liver fat, and Peng (below) grades every diet arm only suggestive. A gold SR-MA of isocaloric RCTs refines which composition change actually moves liver fat, holding energy equal between the two arms of each trial — so its effects are the composition signal net of calories, the dimension that NAFL management otherwise ignores by going calorie-restriction-first: «interventions on NAFL mainly focus on decreasing total body fat by recommending calorie-restricted diets in overweight or obese patients». (Winters-van Eekelen et al., 2020)
26 RCTs, 32 comparisons; omega-3 and fructose trials excluded (covered by prior MAs) — so this is the other-macronutrient slice. Effects are pooled as standardized mean difference (SMD), not a percent, because liver fat was «measured with different measuring instruments of liver fat on different scales» (MRS/MRI in %, CT in Hounsfield units). SMD is a unitless effect: «an SMD of 0.2 can be considered small, 0.5 as medium and 0.8 as high», and «a negative standardized mean difference can be interpreted as a decrease in liver fat in the intervention arm». Read these as SMDs with that anchor — never as a percentage of liver fat removed. (Winters-van Eekelen et al., 2020)
| Swap (intervention vs control) | SMD (95% CI); n comparisons | Read |
|---|---|---|
| Total fat <-> carbohydrate | 0.01 (-0.36; 0.37); 12 | null — the fat/carb ratio is not the lever |
| Unsaturated vs saturated fat | -0.80 (-1.09; -0.51); 4 | large reduction (UFA favoured) |
| Carbohydrate -> protein | -0.33 (-0.54; -0.12); 5 | moderate reduction (protein favoured) |
(Winters-van Eekelen et al., 2020)
- The fat<->carb null is opposing effects cancelling, not fat being inert. «The fat type that is replacing the carbohydrates is likely relevant» — three trials replacing carbs with unsaturated fat found the low-carb-high-fat arm reduced liver fat, while most others found the reverse, so the pooled total-fat estimate averages to ~0. The decision content is in the type, not the fraction. (Winters-van Eekelen et al., 2020)
- The signal is composition, not energy exchanged: «the effect sizes of the studies were not proportional to the amount of energy percentage that was exchanged», and on visual inspection caloric intake did not modify the fat-vs-carb result. Caveat: too few arms for stratified analysis, substantial heterogeneity, and substantial risk of performance/attrition/reporting bias across the included trials. (Winters-van Eekelen et al., 2020)
- Named gap — no fat-vs-protein evidence exists: «there were no studies comparing dietary protein with fat», so the three swaps cannot be assembled into a full ranking (no network meta-analysis possible). (Winters-van Eekelen et al., 2020)
- Minor internal discrepancy: the results-section text reports marginally different point estimates and fewer described studies than the abstract (SFA/UFA SMD -0.75 (-1.11; -0.39), 3 studies; carb->protein SMD -0.32 (-0.58; -0.05), 3 studies). The abstract figures (above) are the paper’s stated conclusions; the direction and magnitude class are unchanged either way. (Winters-van Eekelen et al., 2020)
What this refines, and its two ceilings. It sharpens EASL’s «similarly effective» from composition does not matter to the fat/carb ratio does not, but the fat type and the carb->protein swap do — a type-F refinement below the umbrella-level menu. But it does not lift the ceilings already on this page. (1) Surrogate, not outcome: every effect is on liver-fat content — NAFL is «a liver fat content of more than 5.6%», «associated with an increased risk of liver- and cardiovascular disease-related mortality», but this MA measures none of that transmission, so it is the same liver-fat -> histology -> (unproven) -> mortality chain AASLD flags. (2) Weight-neutral by design: these are isocaloric swaps, so they do not substitute for the weight-loss lever ranked first above — they answer what to eat at a given calorie level, a second-order question once the big weight-loss rock is being pulled. (inferred from Winters-van Eekelen et al., 2020)
The SFA->UFA result converges with the SFA-vs-CVD evidence — noted, not stamped independent. That swapping saturated for unsaturated fat lowers liver fat reaches the same «SFA is worse than UFA» direction that Saturated Fat Intake and Replacement reaches via CVD events and LDL — here through a different endpoint (imaged liver fat) and largely different trials, which is real corroborating breadth. It is deliberately not marked independent-backing (type-E): the endpoint is itself a surrogate, and the routes share antecedents — this MA leans on the same de-novo-lipogenesis mechanism and cites the Imamura glucose-insulin feeding-trial MA the SFA page also holds — so it is shared-lineage refinement (F), not an independent second route. (inferred from Winters-van Eekelen et al., 2020)
The graded lifestyle-lever menu — only sugary soda is high-grade, and the treatments move surrogates not histology [Peng 2022 umbrella]
An umbrella review (37 meta-analyses of observational studies and RCTs) quantifies the same lever menu EASL asserts qualitatively, and — decisively — grades each arm’s evidence strength. Its two load-bearing findings are both decision-relevant, and both cut against over-confidence in the diet menu:
1. Of every lifestyle risk factor, only sugar-sweetened soda reaches high-quality evidence. Peng: «the quality of evidence was only high for the association of sugar-sweetened soda with increased NAFLD risk». The quantified risk menu (all observational, so associational):
| Direction | Factor (effect size; 95% CI) — GRADE unless noted |
|---|---|
| Increases risk | sugar-sweetened soda RR 1.53 (1.34, 1.75) — HIGH; SSB (broad) OR 1.40 (1.07, 1.82); soft drinks OR 1.33 (1.18, 1.49); obesity RR 3.53 (2.48, 5.03); WHR OR 4.10 (1.53, 10.79); highest serum uric acid OR 1.92 (1.66, 2.23); red meat OR 1.26 (1.08, 1.47); short sleep RR 1.19 (1.04, 1.36); smoking OR ~1.32-1.43 |
| Decreases risk | weight loss -> NASH OR 0.14 (0.04, 0.49); modest alcohol (<40 g/day) -> NASH OR 0.50 (0.34, 0.74) / NAFLD OR 0.68 (0.58, 0.81); green tea RR 0.65 (0.44, 0.98); coffee -> fibrosis RR 0.70 (0.60, 0.82); nut intake OR 0.94 (0.90, 0.97) |
Do not read the modest-alcohol-protective row as a lever. It is an observational U/J-shaped association carrying the standard artifact hazards (sick-quitter, confounding by frailty), and Peng itself cautions that «as alcohol causes adverse health effects such as liver cirrhosis, and increased risk for cancers, translation of these results into recommendations have to be considered carefully» — consistent with EASL discouraging alcohol in all SLD -> Alcohol and Mortality and Vascular Disease. (Peng et al., 2022)
2. The lifestyle treatments with the firmest evidence move biochemistry, not the liver. Peng: «green tea, omega-3 PUFAs, and exercise (RT) effectively improve liver enzymes, blood lipids and blood glucose rather than histological changes of the liver, with the higher strength of epidemiologic evidence»; whereas the arms that «can improve liver fat content or hepatic histopathology» (caffeine, low-carb, omega-3, other exercise, weight loss) all «had a large heterogeneity and/or small study effects» — i.e. lower strength. Only 7 treatment interventions (4 green tea, 2 omega-3, 1 resistance exercise) cleared the higher-strength bar, and every one of them hit only surrogates. (Peng et al., 2022)
This is the same surrogate-vs-histology gap EASL states from the guideline side (exercise LoE 1 for steatosis but LoE 5 for histology; weight-loss ladder with «no proven effect on advanced fibrosis/cirrhosis») — reached here from a pooled-MA base. Not independent corroboration: an umbrella pools meta-analyses that pool the same primary literature EASL rests on, so the agreement is a quantification/grading of the shared evidence, not a second independent route (volume != independence). What it adds is the ranking — sugary soda as the one firmly-evidenced dietary target, everything else suggestive — and an explicit warning that the RCT-backed diet treatments have not been shown to change what matters (fibrosis). (inferred from Peng et al., 2022)
And a guard against over-attributing prevention to a generic “healthy diet”: Peng found 15 factors with no significant NAFLD association — «current smoking, light smoking, heavy smoking, whole grains, refined grains, fish, fruits, vegetables, eggs, dairy, or legumes» (smoking appearing on both the increased-risk and the null list reflects conflicting constituent MAs, an unresolved heterogeneity Peng carries). So the actionable dietary signal narrows to cutting liquid sugar, not adding produce. (Peng et al., 2022)
Is fructose the specific hepatic culprit? Mostly no — it is an energy story [2026-08-06, Chung 2014]
EASL’s diet lever names «avoiding sugar-sweetened beverages» and finds low-carb and low-fat «similarly effective» for liver fat — both consistent with the driver being energy, not the fructose molecule. A gold SR-MA of controlled-feeding trials pins this down and defuses the “fructose is uniquely hepatotoxic” claim rather than confirming it:
- At equal calories, fructose is not shown to raise liver fat more than glucose. «The 2 isocaloric monosaccharide diets did not alter IHCLs (+0.11% ± 2.1%)»; fructose and glucose «did not differ in any hepatic outcome measure». (Chung et al., 2014)
- The liver-fat rise is the ADDED-ENERGY arm. Hypercaloric fructose raised IHCL 54% (95% CI 29-79%) vs weight-maintenance — but in healthy young men, at supra-physiological doses, from a single research group, with baseline liver fat far below the NAFLD threshold; and at equal excess energy fructose ≈ glucose. (Chung et al., 2014)
- Chung’s own verdict: the liver-health associations «appear to be confounded by excessive energy intake», and evidence is «not sufficiently robust to draw conclusions regarding effects of fructose, HFCS, or sucrose consumption on NAFLD». (Chung et al., 2014)
So the decision-relevant lever is cutting the excess (liquid) energy — SSB, fruit juice, HFCS — not avoiding the fructose molecule per se, and certainly not whole fruit at normal intakes (a modest fructose dose in a fibre matrix, not a hepatic bolus) -> Free Sugars Intake. The lever is the same energy/weight lever EASL already ranks first; fructose-restriction adds no distinct mechanism beyond the calories it removes. This is the hepatic-tissue instance of the energy-balance-vs-nutrient-source finding -> What Drives Fat Gain - Energy Balance vs the Carbohydrate-Insulin Model. Guard: Chung’s isocaloric leg is a single low-ROB study, so this is insufficient evidence for a fructose-specific effect, not proof of no effect; and the fructose->de-novo-lipogenesis mechanism is real even though the whole-organism liver-fat outcome at equal energy is null (a mechanism-vs-outcome gap). (inferred from Chung et al., 2014)
Bariatric surgery — the large lever for class II/III obesity
For a person at BMI > 35, the guideline names surgery (LoE 3, strong): the BRAVES RCT showed «histological resolution of MASH… in 55% of those assigned to Roux-en-Y gastric bypass or sleeve gastrectomy at 1-year follow-up vs. 15% in the lifestyle modification group», with >=1-stage fibrosis improvement in «37% and 39%… vs. 23% after lifestyle modification», and «about 6% of participants had severe adverse events». A separate post-surgery cohort (not BRAVES) found «advanced fibrosis… persisted in 47%» years later despite weight loss — so surgery does not reliably reverse the advanced stage either. BRAVES is the one held head-to-head of surgery vs lifestyle on liver histology — and its surgery arm is much larger than the lifestyle arm. (European Association for the Study of the Liver, 2024)
The reframe that matters for a metabolic patient — the risk is mostly NOT the liver
For most people with MASLD, the dominant risk it carries is cardiovascular and metabolic, not hepatic. Two guideline facts drive this:
- Steatosis per se is nearly benign for the liver; fibrosis stage is what matters. «the presence of steatosis in the general population is not associated with a clinically meaningful increase in the risk of liver-related outcomes» — liver risk tracks fibrosis, not fat.
- The extrahepatic risk is the real load: MASLD carries «higher risk of non-fatal cardiovascular disease (HR 1.40)», coronary heart disease, heart failure, chronic kidney disease, and «a more than two-fold increased risk of an incident diagnosis of T2D». (European Association for the Study of the Liver, 2024)
So for the metabolic-syndrome stratum, the NAFLD lever is the SAME weight-loss lever already ranked 1-2 for cardiometabolic reasons — MASLD does not add a separate intervention, it adds a reason and a target (>=7-10% for the inflammatory stage) to the weight loss already indicated, plus a monitoring handle (fibrosis, not fat). The dominant risk to act on is still the cardiovascular one -> Baseline Risk and the Relative-Absolute Split. Insulin resistance is the shared pathogenic hub, so the glycemia levers apply too -> Carbohydrate Restriction and Type 2 Diabetes Remission.
Why the levers are shared: liver fat IS the ectopic depot. MASLD is the clinically-visible edge of the intra-organ fat that the personal-fat-threshold frame puts at the centre of cardiometabolic risk — Taylor calls liver-fat accumulation «pivotal» to hepatic insulin resistance, and demotes visceral fat: «Extent of visceral fat accumulation is a surrogate marker for intra-organ fat ex- cess, but is not pathophysiologically related to adverse metabolic consequences». So the >5% / 7-10% / >=10% weight-loss ladder here is depot-drawdown made measurable, and the >2-fold incident-T2D risk MASLD carries is the same hepatic-fat -> IR pathway seen from the liver side -> Ectopic Fat and Depot-Specific Risk. (Taylor & Holman, 2014) (inferred from Taylor & Holman, 2014)
Limits
- The core dose-response is one guidance family (EASL), which transfers NAFLD evidence to MASLD by the 99.8%-overlap argument — a reasonable but stated assumption, not a re-derivation. The Peng umbrella adds a second source type (pooled MAs) that quantifies and grades the lever menu, but it pools the same primary literature, so it is a refinement of the shared evidence base, not independent confirmation.
- Umbrella-review caveat: Peng’s breadth (37 MAs) is not 37 independent tests — constituent MAs overlap in primary studies, and Peng did not re-appraise individual studies or run subgroup/sensitivity analyses. Its GRADE-per-arm is the usable output; read the effect sizes as ranked-associational, not as effects on hard liver outcomes (only surrogates were moved by the RCT-backed treatment arms).
- The weight-loss ladder is a single-RCT threshold, and the histological evidence for exercise and diet-quality on hard liver outcomes is explicitly weak (LoE 3-5). The firm parts are steatosis reduction and the >=10% -> fibrosis claim; the clinical-liver-outcome parts are not.
- Observational confounding on diet and coffee, flagged by the guideline itself.
- Coherence, not validity (R1): these are graded recommendations, not proof that acting changes a given person’s liver or life.