MASLD — quick-reference statistics
Every figure carries its source, year, population and method. Conflicting estimates are shown side by side and never averaged. All PMIDs were verified against live PubMed queries on 2026-09-02; ClinicalTrials.gov figures came from the v2 API on the same date.
Read the "known conflicts and caveats" section at the end before quoting anything here in isolation.
1. Prevalence — general adult population
| Estimate (95% CI) |
Definition |
Method |
Study period |
n |
Source |
| 30.05% (27.88–32.32) |
NAFLD |
92 population-based studies, random effects |
1990–2019 |
9,361,716 |
Younossi 2023, PMID 36626630 |
| 30.69% (28.40–33.09) |
NAFLD, ultrasound-defined |
subset of the above |
1990–2019 |
— |
PMID 36626630 |
| 32.4% (29.9–34.9) |
NAFLD, imaging-diagnosed |
72 studies, 17 countries |
to May 2021 |
1,030,160 |
Riazi 2022, PMID 35798021 |
| 29.8% (28.6–31.1) |
NAFLD |
245 studies |
to Mar 2020 |
5,399,254 |
Le 2022, PMID 34890795 |
| 30.2% (28.7–31.7) |
NAFLD |
479 studies, 38 countries |
to May 2023 |
78,001,755 |
Amini-Salehi 2024, PMID 39094335 |
| ~38% |
NAFLD |
narrative review |
to 2023 |
— |
Wong 2023, PMID 37169151 |
| 30–40% |
MASLD |
narrative review |
to 2025 |
— |
Tilg 2026, PMID 41212550 |
United States, elastography-based
| Metric |
Estimate (95% CI) |
Survey |
Source |
| All steatotic liver disease |
34.2% (31.9–36.5) |
NHANES 2017–Mar 2020, n=7,367 |
Lee 2024, PMID 37732946 |
| MASLD |
31.3% (29.2–33.4) |
same |
PMID 37732946 |
| MetALD |
2.0% (1.6–2.9) |
same |
PMID 37732946 |
| ALD |
0.7% (0.5–0.9) |
same |
PMID 37732946 |
| Etiology-specific/cryptogenic |
0.03% (0.01–0.08) |
same |
PMID 37732946 |
| All SLD, age-adjusted (CAP ≥285 dB/m) |
35.0% (33.4–36.7) |
NHANES 2017–2023, n=12,199 |
Kim 2025, PMID 39610192 |
| MASLD, age-adjusted |
31.9% (30.4–33.4) |
same |
PMID 39610192 |
| MetALD, age-adjusted |
2.2% (1.8–2.6) |
same |
PMID 39610192 |
| ALD, age-adjusted |
0.8% (0.6–1.1) |
same |
PMID 39610192 |
| NAFLD / MAFLD / MASLD, same sample |
18.5% / 19.3% / 20.8% |
NHANES III, n=7,519 (ultrasound) |
Song 2024, PMID 38293788 |
| MASLD / MetALD / ALD |
30% / 2.3% / 1.0% |
NHANES III, n=9,939 (ultrasound) |
Sripongpun 2024, PMID 39290401 |
Regional (NAFLD unless stated)
By subgroup
| Population |
Prevalence (95% CI) |
Source |
| Type 2 diabetes |
65.33% (62.35–68.18), 123 studies, N=2,224,144 |
Younossi 2024, PMID 38521116 |
| Type 2 diabetes |
65.04% (61.79–68.15), 156 studies, N=1,832,125 |
Cho 2023, PMID 37491159 |
| Overweight |
69.99% (65.40–74.21) |
Quek 2023, PMID 36400097 |
| Obesity |
75.27% (70.90–79.18) |
PMID 36400097 |
| Men vs women |
39.7% (36.6–42.8) vs 25.6% (22.3–28.8), p<0.0001 |
Riazi 2022, PMID 35798021 |
| Men vs women |
36.6% (34.7–38.4) vs 25.5% (23.9–27.1) |
PMID 39094335 |
| Children, general population |
13% (9–18) |
Lee 2024, PMID 38771552 |
| Children with obesity |
47% (41–53) |
PMID 38771552 |
| Children, general population |
14.3% (10.3–18.8) |
PMID 39094335 |
| Children with obesity |
38.0% (31.5–44.7) |
PMID 39094335 |
| Lean, as fraction of the NAFLD population |
19.2% (15.9–23.0) |
Ye 2020, PMID 32413340 |
| Non-obese, as fraction of NAFLD |
40.8% (36.6–45.1) |
PMID 32413340 |
| Lean NAFLD in the general population |
5.1% (3.7–7.0) |
PMID 32413340 |
| Non-obese NAFLD in the general population |
12.1% (9.3–15.6) |
PMID 32413340 |
| Lean-MAFLD in the general population |
1.94% (1.10–3.39), I²=98.7% |
Cheah 2025, PMID 40087205 |
| T2D among people with NAFLD |
28.3% (25.2–31.6), 395 studies |
Cao 2024, PMID 38448943 |
| T2D among people with MAFLD |
26.2% (23.9–28.6) |
PMID 38448943 |
2. Incidence
| Estimate (95% CI) |
Population |
Source |
| 46.9 per 1,000 person-years (36.4–57.5) |
16 studies, 5 countries, N=381,765 |
Riazi 2022, PMID 35798021 |
| 46.13 per 1,000 PY (39.3–52.9) |
63 studies, N=1,201,807 (26 China/HK, 22 Korea, 14 Japan) |
Le 2023, PMID 37040843 |
| 50.9 per 1,000 PY (44.8–57.4) |
Asia, 237 studies |
Li 2019, PMID 30902670 |
| 24.6 per 1,000 PY (13.4–39.2) |
non-obese population without NAFLD at baseline |
Ye 2020, PMID 32413340 |
| Men 59.4 vs women 36.7 per 1,000 PY (p=0.0013) |
same as Le 2023 |
PMID 37040843 |
| Obese 86.7 vs non-obese 29.6 per 1,000 PY (p<0.0001) |
same |
PMID 37040843 |
| Smokers 80.4 vs non-smokers 46.9 per 1,000 PY (p=0.046) |
same |
PMID 37040843 |
| Incident T2D in NAFLD: 24.6 per 1,000 PY (20.7–29.2) |
395 studies |
Cao 2024, PMID 38448943 |
3. Severity distribution
| Metric |
Estimate (95% CI) |
Population |
Source |
| Advanced fibrosis, general population |
3.3% (2.4–4.2) |
46 studies, 21 countries, ~8 million |
Zamani 2025, PMID 39209202 |
| Cirrhosis, general population |
1.3% (0.9–1.7) |
same |
PMID 39209202 |
| Advanced fibrosis among people with SLD, pre-COVID |
7.4% (CAP ≥285) |
NHANES |
Kim 2025, PMID 39610192 |
| Advanced fibrosis among people with SLD, COVID era |
9.8% (p=0.039 vs pre-COVID) |
NHANES |
PMID 39610192 |
| High-risk NASH (NAS≥4, F≥2), FAST ≥0.35 |
5.8% age-adjusted |
NHANES 2017–2018, n=4,218 |
Vilar-Gomez 2023, PMID 34958922 |
| High-risk NASH, FAST ≥0.67 |
1.2% age-adjusted |
same |
PMID 34958922 |
| High-risk NASH in T2D |
8.7%–22.5% (cut-off dependent) |
same |
PMID 34958922 |
| NASH in overweight |
33.50% (28.38–39.04) |
Quek 2023 |
PMID 36400097 |
| NASH in obesity |
33.67% (28.45–39.31) |
PMID 36400097 |
|
| F2–4 in overweight with NAFLD |
20.27% (11.32–33.62) |
PMID 36400097 |
|
| F3–4 in overweight with NAFLD |
6.65% (4.35–10.01) |
PMID 36400097 |
|
| F3–4 in obese with NAFLD |
6.85% (3.85–11.90) |
PMID 36400097 |
|
| NASH in T2D (biopsy subset) |
66.44% (56.61–75.02) |
12 studies, N=2,733 |
Younossi 2024, PMID 38521116 |
| F2–4 in T2D with NAFLD |
40.78% (24.24–59.70) / 35.54% (19.56–55.56) |
PMID 38521116 / PMID 37491159 |
|
| F3–4 in T2D with NAFLD |
15.49% (6.99–30.99) / 14.95% (11.03–19.95) |
PMID 38521116 / PMID 37491159 |
|
| Advanced fibrosis, prospectively assessed, age ≥50 with T2D |
14% |
n=501, MRI-PDFF + MRE + VCTE |
Ajmera 2023, PMID 36410554 |
| Cirrhosis, same cohort |
6% |
same |
PMID 36410554 |
| Advanced fibrosis (≥F3) in biopsy-confirmed MASLD |
35% |
Global-MASLD, n=17,792, 41 countries |
Younossi 2026, PMID 41231627 |
| NASH in lean/non-obese NAFLD |
39.0% (24.1–56.3) |
Ye 2020 |
PMID 32413340 |
| Significant fibrosis in lean/non-obese |
29.2% (21.9–37.9) |
PMID 32413340 |
|
| Cirrhosis in lean/non-obese |
3.2% (1.5–5.7) |
PMID 32413340 |
|
4. Progression and regression
| Metric |
Estimate (95% CI) |
Source |
| Fibrosis progression from F0, MASL |
0.07 stages/yr (0.02–0.11) = 1 stage per 14.3 y (9.1–50.0) |
Singh 2015, PMID 24768810 |
| Fibrosis progression from F0, MASH |
0.14 stages/yr (0.07–0.21) = 1 stage per 7.1 y (4.8–14.3) |
PMID 24768810 |
| Paired-biopsy trajectory |
33.6% progressed, 43.1% stable, 22.3% regressed |
PMID 24768810 |
| Progression to cirrhosis |
3–5% of patients, usually >20 years |
Hagström 2024, PMID 39243773 |
| Fibrosis progression events |
49.0 per 1,000 PY |
Le 2024, PMID 38281814 |
| Progression to LSM ≥10 kPa, over mean 4.4 y |
29% of those at risk |
Gawrieh 2024, PMID 38762169 |
| Regression to LSM <10 kPa |
44% of those at risk |
PMID 38762169 |
| First decompensation in cACLD, 5 y |
3.5% (3.0–4.1) |
Pennisi 2025, PMID 40550340 |
| Further decompensation after the first, 5 y |
43.9% (37.2–50.2) |
PMID 40550340 |
| Decompensation, CSPH present, 5 y |
30.7% |
Paternostro 2024, PMID 38823501 |
| Decompensation, no CSPH, 5 y |
9.4% |
PMID 38823501 |
5. Mortality and events
Rates per 1,000 person-years
| Cause |
Unselected NAFLD |
T2D with NAFLD |
Referral-enriched cohorts |
Asian biopsy registry |
| All-cause |
12.60 (6.68–23.67) |
16.79 (10.64–26.40) |
14.6 |
5.34 (4.02–7.08) |
| Cardiac |
4.20 (1.34–7.05) |
4.19 (1.34–7.05) |
4.53 |
— |
| Extrahepatic cancer |
2.83 (0.78–4.88) |
6.10* |
4.53 |
— |
| Liver-specific |
0.92 (0.00–2.21) |
2.15* |
3.10 |
2.34 (1.52–3.58) |
| Source |
Younossi 2023, PMID 36626630 |
Younossi 2024, PMID 38521116 |
Le 2024, PMID 38281814 |
Fujii 2023, PMID 35051649 |
*The published confidence intervals for these two figures do not contain their point estimates as reported in the source abstract; the point estimates are quoted as published and the intervals are omitted.
Fibrosis-stage-specific mortality
| Stage |
All-cause MRR (Dulai 2017) |
Liver-related MRR (Dulai 2017) |
All-cause HR (Ng 2023) |
Liver-related HR (Ng 2023) |
Death/transplant HR (Angulo 2015) |
| F1 |
1.58 (1.19–2.11) |
1.41 (0.17–11.95) |
— |
— |
1.88 (1.28–2.77) |
| F2 |
2.52 (1.85–3.42) |
9.57 (1.67–54.93) |
1.46 (1.08–1.98) |
4.07 (1.44–11.5) |
2.89 (1.93–4.33) |
| F3 |
3.48 (2.51–4.83) |
16.69 (2.92–95.36) |
1.96 (1.41–2.72) |
7.59 (2.80–20.5) |
3.76 (2.40–5.89) |
| F4 |
6.40 (4.11–9.95) |
42.30 (3.51–510.34) |
3.66 (2.65–5.05) |
15.1 (5.27–43.4) |
10.9 (6.06–19.62) |
| Sources |
PMID 28130788 |
PMID 28130788 |
PMID 35513235 |
PMID 35513235 |
PMID 25935633 |
Taylor 2020 (PMID 32027911) gives unadjusted F0-vs-F4 risk ratios of 3.42 (2.63–4.46) all-cause, 11.13 (4.15–29.84) liver-related, 5.42 (1.05–27.89) transplant and 12.78 (6.85–23.85) liver-related events.
Absolute mortality
| Metric |
Value |
Source |
| All-cause mortality at 1/5/10 y, F0–F2 |
0.1% / 3.3% / 7.7% |
Ng 2023, PMID 35513235 |
| All-cause mortality at 1/5/10 y, F4 |
0.3% / 20.6% / 41.5% |
PMID 35513235 |
| 20-y cumulative major adverse liver outcomes: population / F0 / F3 |
2% / 3% / 35% |
Akbari 2024, PMID 38293684 |
| 5-y mortality, biopsy-confirmed MASLD overall |
2.1% |
Younossi 2026, PMID 41231627 |
| 5-y mortality, cirrhosis |
8.3% |
PMID 41231627 |
| Median survival from first decompensation to death or transplant |
2.0 years |
Noureddin 2024, PMID 38571305 |
| All-cause mortality, biopsy NAFLD vs matched controls |
28.6 vs 16.9 per 1,000 PY; aHR 1.93 (1.86–2.00) |
Simon 2021, PMID 33037056 |
| …by stage (aHR vs controls) |
steatosis 1.71; non-fibrotic NASH 2.14; non-cirrhotic fibrosis 2.44; cirrhosis 3.79 |
PMID 33037056 |
| Excess mortality contributors (aHR) |
extrahepatic cancer 2.16; cirrhosis 18.15; CVD 1.35; HCC 11.12 |
PMID 33037056 |
Adverse-event incidence per 1,000 person-years (79 studies, N=1,377,466)
| Event |
Rate |
| Any liver-related event |
24.3 |
| Cirrhosis |
10.9 |
| Liver transplant |
12.0 |
| Hepatocellular carcinoma |
3.39 |
| Incident CVD |
24.77 |
| Renal impairment |
30.3 |
| Depression/anxiety |
29.1 |
| Non-liver cancer |
10.5 |
| Incident diabetes |
19.0 |
| Incident hypertension |
25.8 |
| Incident dyslipidaemia |
26.4 |
| Incident metabolic syndrome |
25.4 |
Source: Le 2024, PMID 38281814. Baseline composition: 21.97% cirrhosis, 58.85% biopsy-proven NASH — a referral-enriched population.
| Outcome |
Effect (95% CI) |
Population |
Source |
| Fatal or non-fatal CVD |
HR 1.45 (1.31–1.61), I²=86.18% |
36 studies, 5,802,226 people, 99,668 events |
Mantovani 2021, PMID 34555346 |
| …with advanced fibrosis |
HR 2.50 (1.68–3.72) |
same |
PMID 34555346 |
| MACE, biopsy cohort |
24.3 vs 16.0 per 1,000 PY; aHR 1.63 (1.56–1.70) |
Sweden, n=10,422 vs 46,517 |
Simon 2022, PMID 34489307 |
| …IHD / CHF / stroke / CV death |
aHR 1.64 / 1.75 / 1.58 / 1.37 |
same |
PMID 34489307 |
| …cirrhosis stratum |
aHR 2.15 (1.77–2.61) |
same |
PMID 34489307 |
| CVD, national screening cohort |
HR 1.38 (1.37–1.39) for MASLD/related SLD |
Korea, 8,808,494 followed 12.3 y, 272,863 events |
Lee 2024, PMID 37907259 |
| …MASLD / MetALD / other combined |
1.39 (1.38–1.40) / 1.28 (1.26–1.30) / 1.30 (1.26–1.34) |
same |
PMID 37907259 |
| Incident CKD stage ≥3 |
HR 1.43 (1.33–1.54), I²=60.7% |
13 studies, 1,222,032 people, 33,840 events |
Mantovani 2022, PMID 33303564 |
| GI cancers (oesophagus, stomach, pancreas, colorectal) |
~1.5–2-fold |
10 studies, 182,202 people, 8,485 cancers |
Mantovani 2022, PMID 33685968 |
| Lung, breast, gynaecological, urinary cancers |
~1.2–1.5-fold |
same |
PMID 33685968 |
| Event |
F0–F2 |
F3 |
F4 |
| All-cause death |
0.32 |
0.89 |
1.76 |
| Variceal haemorrhage |
0.00 |
0.06 |
0.70 |
| Ascites |
0.04 |
0.52 |
1.20 |
| Encephalopathy |
0.02 |
0.75 |
2.39 |
| Hepatocellular carcinoma |
0.04 |
0.34 |
0.14 |
| Incident type 2 diabetes |
4.45 |
— |
7.53 |
| >40% eGFR decline |
0.97 |
— |
2.98 |
Source: Sanyal 2021, PMID 34670043. Cardiac events and non-hepatic cancers did not differ across stages.
7. Hepatocellular carcinoma
| Metric |
Value |
Source |
| MASH cirrhosis, annual HCC incidence |
0.5–2.6% per year |
Huang 2021, PMID 33349658 |
| Non-cirrhotic MASLD |
0.1–1.3 per 1,000 patient-years |
PMID 33349658 |
| NAFLD overall (pooled cohorts) |
3.39 per 1,000 PY |
Le 2024, PMID 38281814 |
| Asia (pooled) |
1.8 per 1,000 PY (0.8–3.1) |
Li 2019, PMID 30902670 |
| Proportion of all HCC attributable to NAFLD |
15.1% (11.9–18.9) |
Tan 2022, PMID 35255263 |
| Single-MAFLD as sole aetiology of HCC |
12.4% (8.3–17.3) |
Crane 2024, PMID 38623613 |
| Total MAFLD (sole or contributory) in HCC |
48.7% (34.5–63.0) |
PMID 38623613 |
| Non-cirrhotic at HCC diagnosis: NAFLD vs other |
38.5% (27.9–50.2) vs 14.6% (8.7–23.4), p<0.0001 |
Tan 2022, PMID 35255263 |
| Underwent surveillance: NAFLD vs other |
32.8% (12.0–63.7) vs 55.7% (24.0–83.3), p<0.0001 |
PMID 35255263 |
| Overall survival vs non-NAFLD HCC |
HR 1.05 (0.92–1.20), p=0.43 |
PMID 35255263 |
| Disease-free survival vs non-NAFLD HCC |
HR 0.79 (0.63–0.99), p=0.044 |
PMID 35255263 |
| Liver-cancer risk vs no SLD (UK Biobank, n=464,556) |
MetALD 1.70 (1.37–2.09); MASLD 1.91 (1.66–2.21); MAFLD 2.01 (1.76–2.29); ALD 3.16 (2.54–3.93); MASLD+viral 22.0 (10.8–44.4) |
Zeng 2025, PMID 39949980 |
8. Treatment effect sizes (histological endpoints)
| Intervention |
MASH resolution, no fibrosis worsening |
≥1-stage fibrosis improvement |
Comparator |
Source |
| Semaglutide 2.4 mg, 72 wk |
62.9% vs 34.3% (Δ28.7, 21.1–36.2) |
36.8% vs 22.4% (Δ14.4, 7.5–21.3) |
placebo |
PMID 40305708 |
| Tirzepatide 15 mg, 52 wk |
62% vs 10% (Δ53, 37–69) |
51% vs 30% (Δ21, 1–42) |
placebo |
PMID 38856224 |
| Survodutide 4.8 mg, 48 wk |
62% vs 14% |
36% vs 22% |
placebo |
PMID 38847460 |
| Lanifibranor 1200 mg, 24 wk |
49% vs 22% |
48% vs 29% |
placebo |
PMID 34670042 |
| Efruxifermin 50 mg, 96 wk (mITT) |
— |
49% vs 19% (Δ31, 12–49) |
placebo |
PMID 40818852 |
| Efruxifermin 50 mg, 96 wk (biopsied) |
— |
75% vs 24% (Δ52, 31–73) |
placebo |
PMID 40818852 |
| Pegozafermin 30 mg, 24 wk |
23% vs 2% |
26% vs 7% (Δ19, 5–32) |
placebo |
PMID 37356033 |
| Resmetirom 100 mg, 52 wk |
29.9% vs 9.7% |
25.9% vs 14.2% |
placebo |
PMID 38324483 |
| Denifanstat 50 mg, 52 wk |
26% vs 11% (Δ13.0, 0.7–25.3) |
— |
placebo |
PMID 39396529 |
| Obeticholic acid 25 mg, 18 mo |
12% vs 8% (not met) |
23% vs 12% (p=0.0002) |
placebo |
PMID 31813633 |
| Cenicriviroc 150 mg, 1 y |
8% vs 6% (p=0.49) |
20% vs 10% (p=0.02) |
placebo |
PMID 28833331 |
| Selonsertib, 48 wk (STELLAR-3/4) |
— |
10–14% vs 13% (both negative) |
placebo |
PMID 32147362 |
| Vitamin E 800 IU, 96 wk |
43% vs 19% NASH improvement (p=0.001) |
no improvement (p=0.24) |
placebo |
PMID 20427778 |
| Pioglitazone 45 mg, 18 mo (T2D/prediabetes) |
51% resolution (Δ32, 13–51) |
fibrosis score −0.5 (p=0.039) |
placebo |
PMID 27322798 |
| Lifestyle, ≥10% weight loss |
90% |
45% regression |
within-cohort subgroup |
PMID 25865049 |
| Lifestyle, pooled meta-analysis |
0.12 (0.08–0.16); 0.25 (0.20–0.30) if >3% weight loss |
0.17 (0.13–0.22) |
— |
PMID 41510965 |
| Bariatric surgery, RYGB / sleeve, 1 y |
56% / 57% vs 16% lifestyle |
— |
lifestyle + best medical care |
PMID 37088093 |
| Bariatric surgery, 5 y |
84% (73.1–92.2) |
70.2% decreased; 56% fibrosis-free |
single-arm |
PMID 32553765 |
Placebo-arm response, for calibration
| Trial |
Placebo MASH resolution |
Placebo fibrosis improvement |
| ENLIVEN (pegozafermin) |
2% |
7% |
| MAESTRO-NASH (resmetirom) |
9.7% |
14.2% |
| SYNERGY-NASH (tirzepatide) |
10% |
30% |
| Denifanstat phase 2b |
11% |
— |
| Survodutide phase 2 |
14% |
22% |
| HARMONY 96 wk (efruxifermin) |
— |
19% |
| NATIVE (lanifibranor) |
22% |
29% |
| ESSENCE (semaglutide) |
34.3% |
22.4% |
| Test |
Target |
AUROC |
Population |
Source |
| LSM-VCTE |
advanced fibrosis |
0.85 |
IPD meta-analysis, 37 studies, n=5,735 |
Mózes 2022, PMID 34001645 |
| FIB-4 |
advanced fibrosis |
0.76 |
same |
PMID 34001645 |
| NFS |
advanced fibrosis |
0.73 |
same |
PMID 34001645 |
| VCTE |
significant / advanced fibrosis / cirrhosis |
0.83 / 0.85 / 0.89 |
82 studies, n=14,609 |
Selvaraj 2021, PMID 33991635 |
| MRE |
significant / advanced fibrosis / cirrhosis / NASH |
0.91 / 0.92 / 0.90 / 0.83 |
same |
PMID 33991635 |
| Point SWE |
significant / advanced fibrosis / cirrhosis |
0.86 / 0.89 / 0.90 |
same |
PMID 33991635 |
| 2D-SWE |
significant / advanced fibrosis / cirrhosis |
0.75 / 0.72 / 0.88 |
same |
PMID 33991635 |
| CAP |
S≥1 / S≥2 / S3 |
0.87 / 0.77 / 0.70 |
n=404, prospective |
Eddowes 2019, PMID 30689971 |
| LSM |
F≥2 / F≥3 / F4 |
0.77 / 0.80 / 0.89 |
same |
PMID 30689971 |
| ELF |
advanced fibrosis |
0.85 (0.79–0.92) |
screening cohort, n=3,378 |
Kjaergaard 2023, PMID 37088311 |
| FIB-4 (same cohort) |
advanced fibrosis |
0.73 (0.64–0.81) |
same |
PMID 37088311 |
| NFS (same cohort) |
advanced fibrosis |
0.66 (0.57–0.76) |
same |
PMID 37088311 |
| ADAPT (PRO-C3) |
advanced fibrosis |
0.86 / 0.87 |
derivation n=150, validation n=281 |
Daniels 2019, PMID 30014517 |
| NIS4 |
at-risk NASH |
0.81 (0.78–0.84) |
NASH CRN DB2, n=1,073 |
Sanyal 2023, PMID 37679433 |
| FAST |
at-risk NASH |
0.80 (0.76–0.85) derivation; 0.85 (0.83–0.87) pooled external |
n=350 + 1,026 |
Newsome 2020, PMID 32027858 |
| MAF-5 |
LSM ≥12 kPa |
0.86 training / 0.85 validation |
n=21,797 |
van Kleef 2024, PMID 38513745 |
| LiverPRO |
TE ≥8 kPa |
0.80 (0.78–0.82) |
DECIDE, n=6,468 |
Lindvig 2025, PMID 39674225 |
| SAFE |
≥F2 vs F0/1 |
≥0.80 |
NASH CRN n=676 + testing sets |
Sripongpun 2023, PMID 35477908 |
| ANTICIPATE-NASH |
liver-related events in F3–F4 |
C statistic 0.93 (vs histology 0.67) |
n=699; validated n=1,396 |
Aceituno 2026, PMID 41212130 |
Global variation (G-MASLD, n=17,792, 41 countries; Younossi 2026, PMID 41100867)
| Test |
Pooled AUC |
Lowest region |
Highest region |
| FIB-4 (advanced fibrosis) |
0.80 (0.79–0.81) |
Latin America 0.75 (0.71–0.79) |
MENA 0.84 (0.82–0.87) |
| ELF (advanced fibrosis) |
0.77 (0.76–0.79) |
Europe 0.72 (0.69–0.76) |
North America 0.80 (0.78–0.82) |
| LSM (advanced fibrosis) |
0.84 (0.83–0.85) |
North America 0.78 (0.76–0.81) |
— |
| FAST |
0.75 (0.74–0.76) |
— |
— |
| Agile 3+ |
0.87 (0.86–0.88) |
— |
— |
| Agile 4 (cirrhosis) |
0.90 (0.89–0.91) |
North America 0.85 (0.83–0.87) |
MENA 0.96 (0.94–0.98) |
| Modality |
tAUC (95% CI) |
| LSM-VCTE |
0.76 (0.70–0.83) |
| FIB-4 |
0.74 (0.64–0.82) |
| Histology |
0.72 (0.62–0.81) |
| NFS |
0.70 (0.63–0.80) |
10. Histological scoring reproducibility
| Feature |
Inter-rater weighted κ (NASH CRN, adult cases) |
| Fibrosis |
0.84 |
| Steatosis |
0.79 |
| Injury (ballooning) |
0.56 |
| Lobular inflammation |
0.45 |
| Diagnostic category |
0.61 |
Source: Kleiner 2005, PMID 15915461. With the SAF score and FLIP algorithm applied, concordance with a reference reading rose from 77% to 97% (experienced readers, κ 0.54→0.66) and from 42% to 75% (less experienced, κ 0.35→0.61) (Bedossa 2014, PMID 24753132).
11. Burden projections
| Projection |
Horizon |
Result |
Source |
| Markov, USA |
2015→2030 |
NAFLD 83.1 M→100.9 M (+21%); NASH 16.52 M→27.00 M (+63%); decompensated cirrhosis +168% to 105,430/yr; HCC +137% to 12,240/yr; liver deaths +178% to 78,300/yr; ~800,000 excess liver deaths |
Estes 2018, PMID 28802062 |
| Markov, 8 countries |
2016→2030 |
NAFLD cases +0–30%; NASH prevalence +15–56%; liver mortality and advanced liver disease more than double |
Estes 2018, PMID 29886156 |
| Agent-based, US adults |
2020→2050 |
MASLD 33.7% (86.3 M)→41.4% (121.9 M); MASH 14.9 M→23.2 M; MASH with F≥2 6.7 M→11.7 M; HCC 11,483→22,440/yr; transplant 1,717→6,720/yr; liver deaths 30,500 (1.0% of adult deaths)→95,300 (2.4%) |
Le 2025, PMID 39821400 |
| Hierarchical Bayesian, global |
→2040 |
NAFLD prevalence 55.7% (from 38.9% in 2020); +2.16%/yr |
Le 2022, PMID 36117442 |
| ARIMA, Asia |
→~2041 |
MASLD incidence +34% vs 1990; nine CKM conditions = 50.9% of Asian disease prevalence and 34.4% of deaths in 2021 |
Duo 2026, PMID 41643809 |
GBD context, age-standardised prevalence per 100,000, 2021, and 2010–2021 change
| Condition |
Prevalence (95% UI) |
Change 2010–2021 |
| Cirrhosis and other chronic liver disease |
20,302.6 (18,845.2–21,791.9) |
+2.6% |
| MASLD |
15,018.1 (13,756.5–16,361.4) |
+11.2% |
| Hepatitis B |
3,583.6 (3,293.6–3,887.7) |
−20.4% |
| Hepatitis C |
1,717.8 (1,385.5–2,075.3) |
−5.1% |
Source: Feng 2025, PMID 40062742. Liver disease overall accounts for ~2 million deaths annually, 4% of all deaths, about two-thirds in men (Devarbhavi 2023, PMID 36990226).
12. Clinical trial landscape (ClinicalTrials.gov v2 API, 2026-09-02)
| Query |
Count |
query.cond=MASH OR NASH OR MASLD OR nonalcoholic steatohepatitis OR nonalcoholic fatty liver |
2,011 |
| …interventional and recruiting |
178 |
query.cond=MASH OR NASH OR MASLD OR nonalcoholic steatohepatitis + AREA[Phase](PHASE3) |
110 |
Largest active phase 3 programmes by enrolment: Eli Lilly master protocol (tirzepatide, retatrutide) NCT07165028 n=4,500; efruxifermin cirrhosis NCT06528314 n=2,150; efruxifermin non-cirrhotic NCT06215716 n=1,650; survodutide LIVERAGE NCT06632444 n=1,800; pemvidutide NCT07795164 n=1,800; MAESTRO-NASH NCT03900429 n=1,759; efimosfermin NCT07701993 n=1,740; survodutide LIVERAGE-Cirrhosis NCT06632457 n=1,590; pegozafermin NCT06318169 n=1,350; ESSENCE NCT04822181 n=1,205.
Terminated or withdrawn phase 3: REGENERATE (obeticholic acid) NCT02548351 n=2,477; RESOLVE-IT (elafibranor) NCT02704403 n=2,157; STELLAR-4 (selonsertib) NCT03053063 n=883; denifanstat NCT06594523 withdrawn at n=0.
13. Added in the 2026-09-02 depth pass
Every figure below carries source, population and method, as elsewhere in this file. All PMIDs were retrieved from live PubMed E-utilities queries on 2026-09-02.
13.1 Alcohol misclassification within "MASLD" (phosphatidylethanol vs self-report)
| Measure |
Value |
Population / method |
Source |
| Median PEth, alcohol-recruited group |
172 ng/mL (IQR 45–434) |
1,482 at-risk adults 30–75 y, Odense, prospective, LC-MS |
Torp 2025, PMID 40945520 |
| Median PEth, metabolically-recruited group |
11 ng/mL (5–37) |
1,442 adults, same cohort |
PMID 40945520 |
| Under-reported intake by PEth |
39.5% (alcohol group); 11.1% (metabolic group) |
same |
PMID 40945520 |
| Classified MASLD by self-report, but PEth ≥20 ng/mL (MetALD/ALD range) |
559 of 1,433 (39.0%) |
same |
PMID 40945520 |
| Over-reporting (high self-report, PEth <20 ng/mL) |
0.7% and 0.1% |
same |
PMID 40945520 |
| PEth testing diagnostically redundant |
812 of 2,042 with SLD (39.8%) |
decision tree on self-report + AUDIT-C |
PMID 40945520 |
13.2 Lean MASLD outcomes — the cardiovascular disagreement
| Study |
CVD events, lean vs non-lean |
CVD mortality |
All-cause mortality |
Design |
| Wakabayashi 2024, PMID 38570344 |
aHR 0.73 (0.64–0.84) |
— |
— |
2.9M Japanese health checkups |
| Huo 2026, PMID 41093635 |
HR 0.89 (0.83–0.95) |
HR 1.22 (1.05–1.41) |
HR 1.26 (1.14–1.39) |
UK Biobank + Kailuan + CKB, 5,030 lean, median 14.2 y |
| Souza 2024, PMID 39117942 |
adj HR 0.89 (0.77–1.02) ns |
HR 1.26 (0.89–1.78) ns |
— |
22 cohorts, >1M, 13.0% lean |
| Nso 2024, PMID 38599554 |
MACE OR 0.9 (0.7–1.2) ns |
OR 1.5 (1.2–1.8) |
OR 1.4, p=0.06 |
21 studies, 7,153 lean |
| Wongtrakul 2024, PMID 38278181 |
— |
— |
HR 1.61 (1.37–1.89), I²=77% |
14 studies, 94,181, median 8.4 y |
| Al Ta'ani 2026, PMID 41614694 |
composite HR 1.21 (1.13–1.30); HF 1.23 (1.16–1.31); cerebrovascular 1.33 (1.24–1.43) |
— |
HR 1.48 (1.38–1.59) |
TriNetX, 67,519 matched pairs, 7 y |
| Ghani 2025, PMID 40406910 |
higher after adjustment |
— |
higher |
75,921 US, 4.99% lean |
Liver-related mortality, by contrast, is consistent: OR 3.56 (3.45–3.67) (PMID 40087205); RR 2.22 (1.57–3.15) and HR 2.26 (1.14–4.51) (PMID 39117942); HR 2.31 (1.54–3.46) (PMID 41093635).
13.3 Sampling and reader variability of liver biopsy
| Quantity |
Value |
Source |
| Paired-core discordance, fibrosis stage ≥1 stage |
41% of 51 patients |
Ratziu 2005, PMID 15940625 |
| Bridging fibrosis on one core, mild/none on the other |
6 of 17 (35%) |
PMID 15940625 |
| Ballooning missed with a single core |
24% of patients |
PMID 15940625 |
| Negative predictive value of a single biopsy for NASH |
≤0.74 |
PMID 15940625 |
| Inter-reader κ, trial dataset: steatosis / fibrosis / ballooning / lobular inflammation |
0.609 / 0.484 / 0.517 / 0.328 |
Davison 2020, PMID 32610115 (678 biopsies, 3 hepatopathologists) |
| Inter-reader κ, NASH resolution without worsening fibrosis |
0.396 |
PMID 32610115 |
| Inter-reader κ, fibrosis improvement without worsening NASH |
0.366 |
PMID 32610115 |
| Enrolled patients failing ≥1 other reader's entry criteria |
46.3% |
PMID 32610115 |
| Simulated power loss from endpoint unreliability |
>90% → as low as 40% |
PMID 32610115 |
13.4 Non-invasive test failure modes
| Setting |
Finding |
Source |
| FIB-4 specificity for advanced fibrosis, age ≥65 |
35% (NFS 20%) |
McPherson 2017, PMID 27725647 (n=634, 5 age bands) |
| Re-derived thresholds, age ≥65 |
FIB-4 2.0 (sens 77%), NFS 0.12 (sens 80%), specificity 70% |
PMID 27725647 |
| FIB-4/NFS/AST-ALT AUROC, age ≤35 |
0.52 / 0.52 / 0.60 |
PMID 27725647 |
| Population screening: LSM ≥8 kPa with normal FIB-4 |
43% (NFS 31%) |
Graupera 2022, PMID 34971806 (5,129, 5 cohorts) |
| False-positive rate of elevated FIB-4/NFS |
28–29% |
PMID 34971806 |
| FIB-4 false-negative rate in diabetes |
11% (NFS 2.5%) |
PMID 34971806 |
| Ceiling of routine-variable models (superlearner over 23 variables) |
AUC 0.79 (0.73–0.84) FLINT; 0.74 (0.68–0.79) NHANES; SAFE equivalent |
Charu 2024, PMID 38687634 |
| Baveno VII LSM ≥25 kPa rule-in PPV |
92% overall; 0.67 in MASLD with obesity, rising to 0.83 with ANTICIPATE±NASH ≥75% |
Bañares 2026, PMID 41138818 (1,433 patients) |
| Baveno VII rule-out NPV |
99% all aetiologies |
PMID 41138818 |
| Spleen stiffness (100 Hz) added to Baveno VII: grey zone |
60% → 15–20% |
Odriozola 2023, PMID 36912787 |
| NICER (SSM+LSM+platelets+BMI) vs ANTICIPATE±NASH, AUC |
0.889 vs 0.849 (p=0.022); validation 0.906 vs 0.863 (p=0.012) |
Jachs 2024, PMID 39326431 |
| LITMUS: best AUC for at-risk NASH among 17 markers |
0.81 (SomaSignal) — none reached the prespecified 0.80 threshold significantly |
Vali 2023, PMID 36958367 (n=966) |
| LITMUS: AUC for advanced fibrosis |
SomaSignal 0.90; ADAPT 0.85; LSM 0.83 |
PMID 36958367 |
| LITMUS: screen-failure reduction with marker prescreening |
to 33%; NNT to find one true positive 4 (SomaSignal) to 9 (PRO-C3) |
PMID 36958367 |
13.5 Treatment: additions
| Intervention |
Endpoint |
Result |
Source |
| Dapagliflozin 10 mg, 48 wk, biopsy MASH, n=154 |
MASH improvement without fibrosis worsening |
53% vs 30%; RR 1.73 (1.16–2.58), p=0.006 |
Lin 2025, PMID 40467095 |
| same |
MASH resolution without fibrosis worsening |
23% vs 8%; RR 2.91 (1.22–6.97), p=0.01 |
PMID 40467095 |
| same |
Fibrosis improvement without MASH worsening |
45% vs 20%; RR 2.25 (1.35–3.75), p=0.001 |
PMID 40467095 |
| Empagliflozin 10 mg, 20 wk, T2D+NAFLD, n=50 |
MRI-PDFF, between-group |
−4.0% (p<0.0001) |
Kuchay 2018, PMID 29895557 |
| Empagliflozin 10 mg, 52 wk, non-diabetic MASLD, n=97 |
median MRI-PDFF change |
−2.49% vs −1.43% (p=0.025); resolution 44.9% vs 28.6% (p=0.094) |
Cheung 2024, PMID 38536017 |
| ATLAS, 48 wk, F3–F4, n=392 |
≥1-stage fibrosis improvement without worsening NASH |
placebo 11%; cilofexor+firsocostat 21% (p=0.17); all other arms 12–19%, all ns |
Loomba 2021, PMID 33169409 |
| Resmetirom, qFibrosis (AI digital pathology), MAESTRO-NASH n=966 |
categorical ≥1-stage improvement over placebo |
+24.4% (80 mg), +22.3% (100 mg); continuous qFC −0.95 (−1.22 to −0.69) and −1.08 (−1.34 to −0.81) |
Schattenberg 2026, PMID 41895606 |
| Metabolic surgery in compensated MASH cirrhosis (SPECCIAL), n=168, mean 10.0±4.5 y |
15-y major adverse liver outcomes |
20.9% (2.5–35.9) vs 46.4% (25.6–61.3); aHR 0.28 (0.12–0.64) |
Aminian 2025, PMID 39870816 |
| same |
15-y decompensation |
15.6% (0–31.3) vs 30.7% (12.9–44.8); aHR 0.20 (0.06–0.68) |
PMID 39870816 |
| Lanifibranor 800 mg, 24 wk, T2D+MASLD, n=38 |
IHTG (¹H-MRS) |
−44% vs −12%; LS mean difference −31% (−51 to −12); weight +2.7% |
Barb 2025, PMID 39824443 |
13.6 Incretin observational hepatic outcomes
| Comparison |
Outcome |
Effect |
Source |
| Tirzepatide vs DPP-4i, T2D, 2 y, n=10,165 matched pairs |
incident major adverse liver outcome |
HR 0.53 (0.40–0.71) |
Henney 2025, PMID 40980971 |
| Semaglutide vs DPP-4i, n=56,702 pairs |
same |
HR 0.81 (0.72–0.90) |
PMID 40980971 |
| Liraglutide vs DPP-4i, n=8,301 pairs |
same |
HR 1.04 (0.79–1.36) — null |
PMID 40980971 |
| GLP-1RA vs DPP-4i, MASLD+T2D, FIB-4 <2.67, 2,238 pairs |
progression to FIB-4 >2.67 |
HR 0.75 (0.65–0.87) |
Choi 2025, PMID 41250965 |
| same |
cirrhosis/decompensation/HCC/transplant composite |
HR 0.98 (0.72–1.34) — null |
PMID 41250965 |
| Semaglutide vs other GLP-1RA, MASLD, 20,384 pairs |
all-cause mortality |
aHR 0.68 (0.59–0.80) |
Kuo 2025, PMID 40536520 |
| same |
major adverse liver outcomes |
aHR 0.79 (0.66–0.94) |
PMID 40536520 |
| GLP-1RA class, 22 RCTs, n=2,258 |
lean mass change |
−0.86 kg (−1.30 to −0.42); ~25% of total weight lost |
Karakasis 2025, PMID 39719170 |
13.7 Alcohol use disorder after bariatric surgery
| Procedure vs usual obesity care |
AUD |
Alcohol-related mortality |
Source |
| Gastric bypass (n=266) |
aHR 5.07 (3.11–8.25) |
sub-HR 6.18 (2.48–15.40) |
Sjöholm 2025, PMID 41066138 (SOS, median 25.2 y) |
| Vertical banded gastroplasty (n=1,365) |
aHR 2.28 (1.56–3.34) |
sub-HR 3.56 (1.79–7.08) |
PMID 41066138 |
| Gastric banding (n=376) |
aHR 2.34 (1.37–4.01) |
sub-HR 2.52 (0.89–7.15), ns |
PMID 41066138 |
| Any surgery (95% bypass) vs obese controls, Denmark, 13,430 vs 21,021, median 6.9 y |
AUD, IPTW |
HR 7.29 (5.06–9.48); year 1 HR 2.77 (1.39–5.53) |
Bramming 2021, PMID 33085738 |
| Post-bariatric incretin vs non-incretin anti-obesity drug, 3,990 pairs |
new-onset AUD |
2.4 vs 5.2 per 1,000 PY; HR 0.45 (0.25–0.81) |
Fakhoury 2025, PMID 41632137 |
13.8 Health economics of resmetirom
| Analysis |
ICER vs standard of care |
QALY gain |
Price threshold |
Source |
| Pre-approval Markov (phase 2 inputs), US payer |
$53,929/QALY |
+1.24 |
cost-effective to $72.00/day at $100k/QALY |
Javanbakht 2023, PMID 36104546 |
| Post-approval agent-based microsimulation, 200,000 F2/F3 patients, 2023 USD |
$140,134/QALY |
+0.26 |
$10,914 / $15,406 / $19,879 per year at $50k / $100k / $150k per QALY |
Le 2025, PMID 40577015 |
| Same, assuming no discontinuation |
$318,740/QALY |
— |
$5,645–$10,619/year |
PMID 40577015 |
| Real-world access, 137 prescribed / 113 treated ≥3 months |
— |
— |
24.1% required insurance appeal; 93.4% eventually approved; mean 12.5-day dispensing delay; 8.8% discontinued |
Shuaibi 2025, PMID 40688390 |
13.9 Thyroid dysfunction and MASLD
| Exposure |
Outcome |
Effect |
Source |
| Primary hypothyroidism (24 cross-sectional studies, ~76.5M) |
prevalent MASLD |
OR 1.43 (1.23–1.66), I²=89% |
Mantovani 2024, PMID 38782564 |
| same (5 studies) |
MASH or advanced fibrosis |
OR 2.84 (2.07–3.90), I²=0% |
PMID 38782564 |
| same (4 longitudinal cohorts, median 4.5 y) |
incident MASLD |
HR 1.39 (0.98–1.97) — ns |
PMID 38782564 |
| Subclinical hypothyroidism (10 studies, n=71,332) |
prevalent MASLD |
OR 1.46 (1.23–1.73) |
Amdetsion 2025, PMID 41357781 |
| same (4 cohorts, n=31,518) |
incident MASLD |
HR 1.59 (1.05–2.40); prospective-only 1.90 (1.50–2.39) |
PMID 41357781 |
| Hypothyroidism, biopsy-confirmed MASLD (12,172 vs 56,831) |
MASLD |
OR 1.68 (1.36–2.06); prevalence 2.5% vs 1.4% |
Yuan 2025, PMID 40342634 |
| Hyperthyroidism, same |
MASLD |
OR 0.17 (0.05–0.56) |
PMID 40342634 |
13.10 Pregnancy
| Population |
Outcome |
Effect |
Source |
| 240 births to 162 women with biopsy-proven MASLD vs 1,140 matched (Sweden 1992–2017) |
preterm birth |
16.7% vs 4.7%; aOR 3.41 (1.98–5.88); vs overweight/obese comparators aOR 4.60 (2.00–10.60) |
Marxer 2025, PMID 40630617 |
| same |
medically indicated / spontaneous preterm |
aOR 11.90 (2.46–57.59) / 2.42 (1.16–5.04) |
PMID 40630617 |
| same |
caesarean section |
aOR 1.63 (1.17–2.27); vs overweight/obese 1.20 (0.77–1.86) |
PMID 40630617 |
| same |
Apgar, malformation, stillbirth, neonatal death |
no difference |
PMID 40630617 |
| 290,527 Korean women, pre-pregnancy health check, singleton delivery |
composite adverse pregnancy outcomes vs no SLD |
MASLD aOR 2.44 (2.33–2.55); MetALD 2.45 (2.26–2.66); ALD 2.28 (2.11–2.47) |
Jung 2026, PMID 41307877 |
| same |
low birthweight |
MASLD 1.15 (1.05–1.27); ALD 1.21 (1.02–1.43); MetALD 0.96 (0.79–1.18) |
PMID 41307877 |
13.11 Burden, re-examined
| Metric |
Value |
Source |
| MASLD DALYs, 2021 |
3.67 million (2.90–4.61) |
Zhang 2024, PMID 39151887 |
| Type 2 diabetes DALYs, 2021 |
75 million (63–90) |
PMID 39151887 |
| Hypertension DALYs, 2021 |
226 million (190–259) |
PMID 39151887 |
| MASLD DALY annual percentage change, 1990–2021 |
+0.05% (−0.06 to 0.17) — not significant |
PMID 39151887 |
| NAFLD prevalence, India, adults (62 datasets, n=23,581) |
38.6% (32.0–45.5) |
Shalimar 2022, PMID 35677499 |
| …average-risk / high-risk subgroups |
28.1% (20.8–36.0) / 52.8% (46.5–59.1) |
PMID 35677499 |
| …hospital-based vs community-based |
40.8% (32.6–49.3) vs 28.2% (16.9–41.0) |
PMID 35677499 |
| NAFLD prevalence, Indian children (8 datasets, n=2,903) |
35.4% (18.2–54.7); obese 63.4% (59.4–67.3); non-obese 12.4% (4.4–23.5) |
PMID 35677499 |
| NAFLD prevalence in type 1 diabetes (20 studies, n=3,901) |
19.3% (12.3–27.5) overall; ultrasound 27.1%, biopsy 19.3%, MRI 8.6%, elastography 2.3% |
de Vries 2020, PMID 32827432 |
13.12 ClinicalTrials.gov, re-queried 2026-09-02
| Query (v2 API) |
Count |
query.cond=(MASLD OR NAFLD OR MASH OR NASH OR "steatotic liver disease") |
2,056 |
…plus AREA[StudyType]INTERVENTIONAL AND AREA[Phase]PHASE3 |
112 |
…plus AREA[OverallStatus]RECRUITING AND AREA[StudyType]INTERVENTIONAL AND AREA[Phase]PHASE3 |
17 |
Known conflicts and caveats
-
Prevalence estimates 29.8%–38% are not in conflict; they answer different questions. Modality (ultrasound vs CAP), study period (prevalence is genuinely rising at ~0.7%/yr) and case definition (NAFLD 18.5% vs MAFLD 19.3% vs MASLD 20.8% in the same NHANES III sample) each move the number. Quote with all three attached, and never average them.
-
The leading cause of death depends on the population. In unselected NAFLD, cardiac deaths outnumber liver deaths ~4.5:1 (PMID 36626630). In the Swedish nationwide biopsy cohort, extrahepatic cancer (aHR 2.16) and cirrhosis (aHR 18.15) dominate excess mortality with CVD contributing modestly (aHR 1.35) (PMID 33037056). In the Asian CLIONE biopsy registry, liver-related death leads and fibrosis is not associated with overall mortality after adjustment (PMID 35051649). In the NASH CRN prospective cohort, cardiac events do not differ across fibrosis stages while liver events rise steeply (PMID 34670043). These are four different populations, not four contradictory facts about one population.
-
Fibrosis-stage mortality ratios disagree by two- to three-fold at F4 — MRR 6.40 (Dulai), HR 3.66 (Ng), RR 3.42 (Taylor), HR 10.9 (Angulo, death or transplant). Adjustment, follow-up duration and reference-group composition explain most of it. Dulai pooled simple steatosis and NASH without fibrosis into the reference stratum, which attenuates the ratios.
-
Liver-related mortality confidence intervals are enormous (e.g. MRR 42.30, 95% CI 3.51–510.34 at F4) because liver deaths are rare in the reference stratum. The direction is certain; the magnitude is not.
-
Placebo response in MASH trials ranges from 2% to 34.3% for steatohepatitis resolution and 7% to 30% for fibrosis improvement. This spread exceeds most drug effects, so cross-trial comparison of the two approved drugs — resmetirom's 9.7% placebo arm against semaglutide's 34.3% — is not valid, and neither are network-meta-analytic rankings built on top of it.
-
Non-invasive test accuracy varies by world region by up to 0.09 AUC for FIB-4 and 0.11 for Agile 4 within a single harmonised cohort (PMID 41100867). No explanation has been established, and no region-specific thresholds have been derived and validated.
-
Incidence estimates are dominated by East Asia, but the abstract's counts are internally inconsistent. It reports 63 studies overall and lists 26 China/Hong Kong, 22 South Korea, 14 Japan and two elsewhere, which sum to 64 (PMID 37040843). Do not quote “62 of 63”; the geographic concentration is clear, but that fraction is not arithmetically supportable from the abstract.
-
MASLD prevalence peaks at moderate socio-demographic index (PMID 40062742) and correlates inversely with human development index (coefficient −0.523, p=0.005) (PMID 39094335) — the opposite of the naive affluence model of metabolic disease.
-
Every projection assumes no effective therapy. All five models in §11 were built before or without incorporating resmetirom and incretin uptake; the largest (PMID 39821400) states this explicitly.
-
Two figures in §5 have published confidence intervals that do not contain their point estimates, both from the T2D global-epidemiology abstract (PMID 38521116). They are reproduced as published and flagged rather than silently corrected.
-
"MASLD-related HCC" means two different things. 12.4% of HCC has MAFLD as its sole aetiology; 48.7% has MAFLD present as sole or contributory factor (PMID 38623613). The NAFLD-defined figure is 15.1% (PMID 35255263). These are not interchangeable.
-
All ClinicalTrials.gov counts are a snapshot. The registry changes daily; §12 is valid only for 2026-09-02.
-
New 2026 audit findings alter prior absence claims. MASLD recompensation occurred in 17.7% of an MASLD subgroup (PMID 42467948); MELD 3.0 up-categorised 65.4% and down-categorised 11.2% of MASH transplant candidates without significantly increasing transplant access (PMID 41284513); 33% of statin-indicated people with MASLD were untreated versus 19% without MASLD (PMID 41861677); and sequential FIB-4 excluded 43% of biopsy-confirmed fibrotic MASLD from second-line testing in a 186-person prospective study (PMID 42566274).
-
Lean-MASLD cardiovascular risk is directionally contested, and the split maps onto study design. Population-based cohorts report lower CVD incidence in lean disease (aHR 0.73–0.89; PMIDs: 38570344, 41093635), pooled adjusted analysis reports no difference (HR 0.89, 0.77–1.02; PMID 39117942), and US health-system EHR cohorts report higher CVD, heart failure, cerebrovascular events and mortality (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), which implies higher case fatality and has no proposed mechanism. Liver-related mortality (2–3.5 fold higher) is not contested.
-
Biopsy sampling error exceeds reader error, and neither is small. Paired cores from the same session differ by ≥1 fibrosis stage in 41% of patients and miss bridging fibrosis in 35% of those who have it (PMID 15940625); inter-reader κ for the two licensing endpoints is 0.366–0.396 in a real trial dataset, 46.3% of enrolled patients fail at least one other reader's entry criteria, and simulated power falls from >90% to 40% (PMID 32610115). Applying the sampling-error benchmark retrospectively, no natural-history or trial study has shown a change in activity grade or ballooning exceeding sampling variability (PMID 17767466). Negative MASH trials cannot currently be distinguished from unmeasurable endpoints.
-
Two lifetime cost-effectiveness models of resmetirom differ ~5-fold on QALYs and land on opposite sides of $100,000/QALY — +1.24 QALYs and $53,929/QALY pre-approval (PMID 36104546) versus +0.26 QALYs and $140,134/QALY post-approval (PMID 40577015). Both extrapolate the same 52-week histological surrogate. In the later model, lower discontinuation worsens cost-effectiveness (ICER $318,740/QALY with none).
-
Two contemporaneous network meta-analyses of overlapping MASH trials produce different podiums. Souza ranks pegozafermin, cilofexor+firsocostat and cilofexor+selonsertib highest for fibrosis (PMID 39903735); Han ranks pegbelfermin first and obeticholic acid first at 1.5 years, and pegbelfermin does not appear in Souza's ranking at all (PMID 41811770). SUCRA rankings are unstable to inclusion criteria, endpoint definition and time window.
-
About 39% of people classified as MASLD by self-reported alcohol intake have a phosphatidylethanol concentration in the MetALD or ALD range (PMID 40945520). Because misclassification is essentially one-directional (under-reporting 39.5%/11.1%; over-reporting 0.7%/0.1%), every MASLD-versus-MetALD outcome contrast in this file is attenuated by an unmeasured amount, and the reported MetALD hepatic hazard ratios of 1.23–1.62 (PMIDs: 42120953, 40953570) are lower bounds.
-
MASLD's measured global disability burden is small relative to its prevalence and has not changed in three decades — 3.67 million DALYs against 75 million for type 2 diabetes and 226 million for hypertension, APC +0.05% (ns) (PMID 39151887) — while GBD's own age-standardised MASLD prevalence rose 11.2% over 2010–2021 (PMID 40062742). This is an attribution artefact (burden reassigned to cirrhosis and to cardiovascular disease), not a measurement of unimportance, and it has not been re-attributed under alternative rules.
-
Reported NAFLD prevalence in type 1 diabetes spans a twelve-fold range within one meta-analysis, by modality — 27.1% by ultrasound to 2.3% by transient elastography (PMID 32827432). No modern reference standard has been applied to an adequately sized type 1 cohort, so the size of this comorbidity is genuinely unknown.
-
Glycaemic control slows fibrosis progression without changing liver events. In 7,543 MASLD patients with serial elastography, T2D raised stiffness progression (HR 1.501) and liver-related events (HR 2.030), and poor long-term control raised progression within T2D (HR 1.524) — but control made no difference to liver-related events (p=0.625) or to regression (p=0.957) (PMID 41076043). Whether that is insufficient follow-up or a genuine dissociation is unresolved.