Skip to content

Pathogenesis

TL;DR — Hepatic fat accumulates because supply exceeds disposal, and stable-isotope tracing in NAFLD patients apportions the excess: 59.0% ± 9.9% of hepatic triacylglycerol fatty acids came from serum non-esterified fatty acids, 26.1% ± 6.7% from de novo lipogenesis and 14.9% ± 7.0% directly from diet (Donnelly 2005, PMID 15864352). That ~26% DNL contribution is the abnormality — in health it is a few per cent — and it is the reason adipose-tissue insulin resistance and dietary carbohydrate both matter. Triglyceride itself is not the toxin: specific lipid species (saturated free fatty acids, free cholesterol, lysophosphatidylcholine, ceramides) drive ER stress, mitochondrial dysfunction and cell death (Marra 2018, PMID 29154964). Mitochondria adapt then fail: obese people with or without steatosis had 4.3–5.0-fold higher maximal respiration than lean controls, whereas NASH patients had higher mitochondrial mass but 31–40% lower maximal respiration alongside greater oxidative stress and DNA damage (Koliaki 2015, PMID 25955209). The "two-hit" model is obsolete; the multiple-parallel-hits framework — insulin resistance, adipokines, nutrition, microbiota, genetics, epigenetics acting together on a predisposed host — replaced it (Buzzetti 2016, PMID 26823198). Fibrosis, the outcome-relevant endpoint, emerges from hepatocyte–macrophage–hepatic stellate cell crosstalk with re-activated developmental pathways (TAZ, Notch, hedgehog) (Schwabe 2020, PMID 32044315). The unresolved question is why only a minority progress: no mechanism yet explains the ~25% who develop steatohepatitis with anything approaching the precision of the genetic effects described in genetics.

Where the fat comes from

Source Contribution to hepatic TAG in NAFLD Determinants
Serum non-esterified fatty acids (adipose lipolysis) 59.0% ± 9.9% adipose insulin resistance; failure of WAT to store lipid
De novo lipogenesis 26.1% ± 6.7% carbohydrate load, hyperinsulinaemia, SREBP-1c/ChREBP
Dietary fatty acids (chylomicron spillover) 14.9% ± 7.0% meal composition

Source: Donnelly 2005, PMID 15864352 — nine patients (5 male, 4 female, 44 ± 10 years) infused and orally fed stable isotopes over four days with liver biopsy. The pattern of labelling in VLDL mirrored the liver, and DNL was elevated in the fasted state with no diurnal variation, meaning it is constitutively upregulated rather than simply postprandial.

The adipose contribution is not passive overflow but tissue failure. White adipose tissue can expand and store triglyceride up to a "personalised adiposity threshold"; beyond it, storage becomes dysfunctional, producing metabolic inflexibility, inflammation and aberrant adipokine secretion, and lipid spills systemically — the liver receiving the largest share (Lee 2023, PMID 36740049).

Diet. Reviewing human stable-isotope and randomised evidence, glucose and fructose raise intrahepatic triglyceride comparably in the context of energy excess — the widely assumed unique lipogenic hazard of fructose is not supported when energy is matched — whereas saturated fat raises IHTG more than polyunsaturated or monounsaturated fat, with adverse effects on insulin sensitivity probably mediated in part by ceramide synthesis (Yki-Järvinen 2021, PMID 34257427). Fructose does have a distinct route: fructose-stimulated DNL is promoted by inflammation, with inflammatory signalling amplifying the lipogenic response (Todoric 2020, PMID 32839596). See lifestyle and weight loss.

From fat to injury: lipotoxicity

Total hepatocyte triglyceride is a poor predictor of damage. Specific lipid classes act as the damaging agents — palmitic and other saturated free fatty acids, free cholesterol, lysophosphatidylcholine and ceramides — acting through signalling cascades and death receptors, ER stress, mitochondrial dysfunction and oxidative stress (Marra 2018, PMID 29154964). Toxic intermediates accumulate from three converging sources: increased hepatic fatty acid uptake, elevated DNL and impaired mitochondrial oxidation; the resulting metabolites (saturated fatty acids, free cholesterol, ceramides, lactate, succinate) act as paracrine signals reinforcing inflammation and fibrogenesis across multiple liver cell types (Steinberg 2025, PMID 40955659).

Mitochondrial flexibility and its loss is one of the few places where human tissue gives a clean quantitative answer. High-resolution respirometry on liver biopsies from lean controls (n=12), obese insulin-resistant people without steatosis (n=18), with NAFL (n=16) and with NASH (n=7):

Group Maximal respiration vs lean Mitochondrial mass Other
Obese without NAFL / with NAFL 4.3–5.0× higher similar content adaptive hypermetabolism
NASH 31–40% lower higher mass greater hepatic insulin resistance, uncoupling, leak; raised H₂O₂ and lipid peroxides; raised 8-OH-deoxyguanosine; reduced antioxidant capacity; increased inflammatory response

Source: Koliaki 2015, PMID 25955209. The interpretation offered — hepatic mitochondrial flexibility that compensates early in obesity-related insulin resistance and is lost at the steatohepatitis transition — is one of the better-supported candidate explanations for why steatosis becomes steatohepatitis, though causal direction is not established by a cross-sectional design.

Why steatosis makes the liver insulin resistant

The lipotoxicity account above says which lipids are toxic; it does not say by what molecular step fat causes hepatic insulin resistance, which is the link between the liver lesion and the systemic disease. The best-supported mechanism is diacylglycerol-mediated: accumulation of hepatic diacylglycerol activates PKC-ε, which impairs insulin-receptor kinase activation and therefore insulin-stimulated glycogen synthesis. Peripheral (adipose and muscle) insulin resistance then acts on the liver indirectly, by increasing the flux of substrates that drive lipogenesis (glucose, fatty acids) and gluconeogenesis (glycerol, and fatty-acid-derived acetyl-CoA, an allosteric activator of pyruvate carboxylase) (Samuel 2018, PMID 28867301). This is a substrate-flux model rather than a signalling-defect model, and it explains why weight loss reverses hepatic insulin resistance so reliably (lifestyle and weight loss) and why interventions that raise hepatic energy expenditure — the liver-targeted mitochondrial uncouplers — reverse the phenotype in animals without acting on insulin signalling at all.

The competing candidate is ceramide, and the cleanest test of it is a double-bond experiment. Deleting dihydroceramide desaturase 1 (DES1) — the enzyme inserting the conserved double bond into the ceramide backbone — whole-body, or tissue-specifically in liver and/or adipose, resolved hepatic steatosis and insulin resistance in leptin-deficient and diet-induced obese mice; the ceramide actions that promoted lipid uptake and storage and impaired glucose utilisation could not be recapitulated by dihydroceramides lacking that double bond (Chaurasia 2019, PMID 31273070). This causal mouse experiment identifies a specific druggable enzyme. The two mechanisms are not exclusive — diacylglycerol and ceramide both derive from the same excess fatty-acid flux — and their relative contribution in people remains unresolved.

The multiple-parallel-hits framework

The two-hit hypothesis (steatosis then a second oxidative insult) cannot account for the molecular and metabolic changes observed, and has been superseded by a multiple-hit model in which insulin resistance, adipose-derived hormones, nutritional factors, gut microbiota, and genetic and epigenetic factors act in parallel on a predisposed host (Buzzetti 2016, PMID 26823198). Comprehensive mechanistic syntheses now organise pathogenesis around metabolic origins, hepatic glucose and lipid handling, bile-acid toxicity, macrophage dysfunction and stellate cell activation, with genetic, epigenetic and environmental modifiers of fibrosis progression and HCC risk (Loomba 2021, PMID 33989548).

The gut–adipose–liver axis is the systems-level version: gut-derived metabolites including acetate and ethanol, and non-esterified fatty acids from white adipose tissue, feed hepatic processes that depend on mitochondrial function; obesity, insulin resistance and T2D disrupt these fluxes (Steinberg 2025, PMID 40032040).

Inflammation and immune drivers

Cell type / axis Finding Source
Kupffer cells vs recruited macrophages Kupffer cells are reduced in fatty liver and replaced by bone-marrow-derived macrophages; recruited cells split into KC-like and lipid-associated macrophage (LAM) subsets. Hepatic LAMs express osteopontin, a NASH biomarker linked to fibrosis, and localise to regions with fewer Kupffer cells and increased desmin Remmerie 2020, PMID 32888418
TREM2⁺ macrophages TREM2 sustains macrophage–hepatocyte metabolic coordination; Trem2-deficient macrophages release exosomes rich in miR-106b-5p that block Mitofusin 2, impairing hepatocyte mitochondrial structure. NAFLD was an independent risk factor for sepsis in a large clinical cohort Hou 2021, PMID 33586673
Type 1 conventional dendritic cells cDC1 more abundant and activated in patients across the NAFLD/NASH spectrum; cDCs promote inflammatory T-cell reprogramming; cDC1 depletion or anti-XCL1 blockade attenuated pathology in mouse NASH Deczkowska 2021, PMID 34017133
Human single-cell atlas across stages Enriched hepatic regulatory T cells, monocytic MDSCs, TREM2⁺S100A9⁺ macrophages and S100ʰⁱHLAˡᵒ cDC2 with MASH progression; cytotoxic T-cell function rose with inflammation but fell with fibrosis while acquiring an exhausted signature; NK-driven toxicity intensified Martin 2025, PMID 40883571
Metabolism–microbes–immunity interplay Lipotoxicity instigates injury, inflammation and insulin resistance; dysbiosis fuels hepatic and systemic inflammation by activating innate and adaptive responses Tilg 2021, PMID 34931080

The exhaustion signature in cytotoxic T cells at the fibrotic stage is a notable finding because it offers a mechanistic link between MASH and the two clinical observations that MASLD raises susceptibility to infection (PMID 33586673) and to hepatocellular carcinoma — see MASLD-related hepatocellular carcinoma.

Fibrogenesis

Hepatic stellate cells are quiescent vitamin-A-storing pericytes in healthy liver. In MASH they are activated by metabolic stress, lipotoxicity and chronic inflammation, with input from injured hepatocytes, recruited macrophages, capillarised sinusoidal endothelial cells and a permeable intestinal epithelium. Activated HSCs shift metabolism toward glycolysis, glutaminolysis and lactate generation, and transform into myofibroblast-like cells producing excess extracellular matrix; they also occupy inflammatory and intermediate activation states rather than a single "activated" phenotype (Kisseleva 2025, PMID 40120772).

The upstream triggering is best described as hepatocyte–macrophage–HSC crosstalk: stressed and dying hepatocytes instigate profibrogenic signalling, in part by reactivating developmental pathways including TAZ, Notch and hedgehog; the efficiency of dead-cell clearance (efferocytosis) modulates the inflammatory and fibrogenic response; and single-cell sequencing has revealed substantial HSC and macrophage heterogeneity within this network (Schwabe 2020, PMID 32044315). This network structure — rather than a single dominant pathway — is why antifibrotic monotherapy has repeatedly failed in trials; see other pharmacotherapy.

Systemic read-out: the circulating proteome

Multicentre proteomics in 306 histologically characterised NAFLD patients measured 4,730 circulating proteins with paired liver transcriptomics, deriving a 31-marker proteo-transcriptomic signature of steatohepatitis and fibrosis. Deconvolution by single-cell RNA sequencing attributed the proteomic changes to specific hepatic cell types, and a four-protein model (ADAMTSL2, AKR1B10, CFHR4, TREM2) with BMI and T2D status identified at-risk steatohepatitis (Govaere 2023, PMID 37037945). TREM2 appearing in both the mechanistic macrophage literature (PMIDs: 33586673, 40883571) and the circulating diagnostic signature is one of the clearer mechanism-to-biomarker bridges in this field. See noninvasive assessment.

The gut–liver axis

One mechanistic result deserves separate statement because it supplies a candidate pathway for an epidemiological association. In the studied mouse model, intestinal nicotine activated AMPKα, which phosphorylated and stabilised sphingomyelin phosphodiesterase 3 (SMPD3), increased intestinal ceramide formation and worsened progression from steatosis to steatohepatitis. The gut bacterium Bacteroides xylanisolvens degraded nicotine; colonising nicotine-exposed mice with it lowered intestinal nicotine and improved NAFLD progression (Chen 2022, PMID 36261549). This links the ceramide mechanism, the gut–liver axis and smoking, but does not establish that the pathway mediates smoking-associated disease in humans.

Bacterial products and gut-derived metabolites reach the liver through the portal circulation, and barrier integrity is part of the mechanism rather than a bystander: microbiota-driven disruption of the gut vascular barrier has been reported as a prerequisite for NASH development in mouse models (Mouries 2019, PMID 31419514). A universal gut-microbiome-derived signature has been reported to predict cirrhosis (Oh 2020, PMID 32610095), and a metagenomic signature has been proposed for non-invasive detection of advanced fibrosis in human NAFLD (Loomba 2017, PMID 28467925). Disentangling microbial signatures from the metabolic disorders they accompany remains the central methodological problem in this literature (Aron-Wisnewsky 2020, PMID 32152478), and the fungal component has been much less studied than the bacterial (Demir 2022, PMID 34896404).

Open questions

  • Why do only a minority progress to steatohepatitis and fibrosis? Roughly 25% of NAFLD progresses to NASH (Schwabe 2020, PMID 32044315) and 3–5% to cirrhosis (natural history). The mitochondrial-flexibility hypothesis (PMID 25955209), lipid-species composition (PMID 29154964) and genotype (genetics) each explain part; no integrated model predicts progression at the individual level.
  • Is the loss of mitochondrial respiratory capacity in MASH cause or consequence? The human evidence is cross-sectional (n=7 with NASH) (PMID 25955209). No interventional human study has restored hepatic mitochondrial respiration and measured histological change.
  • Which immune population is the causal driver rather than a marker? cDC1 depletion attenuated murine NASH (PMID 34017133) and TREM2 overexpression improved outcomes in a murine model (PMID 33586673), but the human single-cell atlas describes a coordinated shift across regulatory T cells, MDSCs, macrophage and dendritic subsets and NK cells simultaneously (PMID 40883571). No human intervention has tested a single immune target.
  • Does microbiome modification change liver histology in humans? The mechanistic literature is largely murine and the human literature largely associative. Query run 2026-09-02: (NAFLD OR NASH OR MASLD OR MASH) AND (microbiome OR microbiota OR "gut-liver") — 4,335 records; the human interventional evidence remains small randomised faecal-transplant and diet studies rather than histological endpoint trials.
  • Diacylglycerol or ceramide? Both have strong mechanistic support — DAG–PKC-ε from human and rodent flux studies (PMID 28867301), ceramide from a DES1 double-bond deletion experiment that isolates the causal moiety (PMID 31273070) — and both derive from the same fatty-acid excess. No human study has measured hepatic DAG, DAG-PKC-ε activation and ceramide species in the same tissue with a functional insulin-sensitivity readout, so their relative weight in people is unresolved and each supports a different drug class.
  • Does the intestinal nicotine–ceramide pathway operate in human smokers? The mouse mechanism is complete (intestinal nicotine → AMPKα → SMPD3 → ceramide → NASH) and identifies a nicotine-degrading commensal (PMID 36261549). Whether Bacteroides xylanisolvens carriage modifies the smoking–MASLD association in humans has not been tested, and it is a directly answerable question in existing cohorts with paired microbiome and smoking data.
  • Why does saturated fat raise liver fat more than unsaturated fat, when both are energy-dense? The ceramide-mediated explanation is proposed but not proven in humans (PMID 34257427), and no trial has tested ceramide-lowering as a mechanism-specific intervention in MASLD.

References

  1. Donnelly KL, Smith CI, Schwarzenberg SJ, et al. Sources of fatty acids stored in liver and secreted via lipoproteins in patients with nonalcoholic fatty liver disease. J Clin Invest. 2005;115(5):1343-51. PMID 15864352
  2. Buzzetti E, Pinzani M, Tsochatzis EA. The multiple-hit pathogenesis of non-alcoholic fatty liver disease (NAFLD). Metabolism. 2016;65(8):1038-48. PMID 26823198
  3. Loomba R, Friedman SL, Shulman GI. Mechanisms and disease consequences of nonalcoholic fatty liver disease. Cell. 2021;184(10):2537-2564. PMID 33989548
  4. Marra F, Svegliati-Baroni G. Lipotoxicity and the gut-liver axis in NASH pathogenesis. J Hepatol. 2018;68(2):280-295. PMID 29154964
  5. Koliaki C, Szendroedi J, Kaul K, et al. Adaptation of hepatic mitochondrial function in humans with non-alcoholic fatty liver is lost in steatohepatitis. Cell Metab. 2015;21(5):739-46. PMID 25955209
  6. Yki-Järvinen H, Luukkonen PK, Hodson L, et al. Dietary carbohydrates and fats in nonalcoholic fatty liver disease. Nat Rev Gastroenterol Hepatol. 2021;18(11):770-786. PMID 34257427
  7. Todoric J, Di Caro G, Reibe S, et al. Fructose stimulated de novo lipogenesis is promoted by inflammation. Nat Metab. 2020;2(10):1034-1045. PMID 32839596
  8. Lee E, Korf H, Vidal-Puig A. An adipocentric perspective on the development and progression of non-alcoholic fatty liver disease. J Hepatol. 2023;78(5):1048-1062. PMID 36740049
  9. Steinberg GR, Carpentier AC, Wang D. MASH: the nexus of metabolism, inflammation, and fibrosis. J Clin Invest. 2025;135(18). PMID 40955659
  10. Steinberg GR, Valvano CM, De Nardo W, et al. Integrative metabolism in MASLD and MASH: Pathophysiology and emerging mechanisms. J Hepatol. 2025;83(2):584-595. PMID 40032040
  11. Schwabe RF, Tabas I, Pajvani UB. Mechanisms of Fibrosis Development in Nonalcoholic Steatohepatitis. Gastroenterology. 2020;158(7):1913-1928. PMID 32044315
  12. Kisseleva T, Ganguly S, Murad R, et al. Regulation of Hepatic Stellate Cell Phenotypes in Metabolic Dysfunction-Associated Steatohepatitis. Gastroenterology. 2025;169(5):797-812. PMID 40120772
  13. Remmerie A, Martens L, Thoné T, et al. Osteopontin Expression Identifies a Subset of Recruited Macrophages Distinct from Kupffer Cells in the Fatty Liver. Immunity. 2020;53(3):641-657.e14. PMID 32888418
  14. Hou J, Zhang J, Cui P, et al. TREM2 sustains macrophage-hepatocyte metabolic coordination in nonalcoholic fatty liver disease and sepsis. J Clin Invest. 2021;131(4). PMID 33586673
  15. Deczkowska A, David E, Ramadori P, et al. XCR1+ type 1 conventional dendritic cells drive liver pathology in non-alcoholic steatohepatitis. Nat Med. 2021;27(6):1043-1054. PMID 34017133
  16. Martin OP, Wallace MS, Oetheimer C, et al. Single-cell atlas of human liver and blood immune cells across fatty liver disease stages reveals distinct signatures linked to liver dysfunction and fibrogenesis. Nat Immunol. 2025;26(9):1596-1611. PMID 40883571
  17. Tilg H, Adolph TE, Dudek M, et al. Non-alcoholic fatty liver disease: the interplay between metabolism, microbes and immunity. Nat Metab. 2021;3(12):1596-1607. PMID 34931080
  18. Govaere O, Hasoon M, Alexander L, et al. A proteo-transcriptomic map of non-alcoholic fatty liver disease signatures. Nat Metab. 2023;5(4):572-578. PMID 37037945
  19. Mouries J, Brescia P, Silvestri A, et al. Microbiota-driven gut vascular barrier disruption is a prerequisite for non-alcoholic steatohepatitis development. J Hepatol. 2019;71(6):1216-1228. PMID 31419514
  20. Oh TG, Kim SM, Caussy C, et al. A Universal Gut-Microbiome-Derived Signature Predicts Cirrhosis. Cell Metab. 2020;32(5):878-888.e6. PMID 32610095
  21. Loomba R, Seguritan V, Li W, et al. Gut Microbiome-Based Metagenomic Signature for Non-invasive Detection of Advanced Fibrosis in Human Nonalcoholic Fatty Liver Disease. Cell Metab. 2017;25(5):1054-1062.e5. PMID 28467925
  22. Aron-Wisnewsky J, Vigliotti C, Witjes J, et al. Gut microbiota and human NAFLD: disentangling microbial signatures from metabolic disorders. Nat Rev Gastroenterol Hepatol. 2020;17(5):279-297. PMID 32152478
  23. Demir M, Lang S, Hartmann P, et al. The fecal mycobiome in non-alcoholic fatty liver disease. J Hepatol. 2022;76(4):788-799. PMID 34896404
  24. Samuel VT, Shulman GI. Nonalcoholic Fatty Liver Disease as a Nexus of Metabolic and Hepatic Diseases. Cell Metab. 2018;27(1):22-41. PMID 28867301
  25. Chaurasia B, Tippetts TS, Mayoral Monibas R, et al. Targeting a ceramide double bond improves insulin resistance and hepatic steatosis. Science. 2019;365(6451):386-392. PMID 31273070
  26. Chen B, Sun L, Zeng G, et al. Gut bacteria alleviate smoking-related NASH by degrading gut nicotine. Nature. 2022;610(7932):562-568. PMID 36261549