Open questions/metabolic, liver, type 2 diabetes

Open

Why does fat build up in the liver in type 2 diabetes?

Excess triglyceride accumulates in liver cells (steatosis), driving inflammation and scarring. It affects a large share of people with type 2 diabetes.

Adults with type 2 diabetesLiving model, version 1.029 papers and 7 dataset records, each verifiedCurators: open, volunteer

15Unknowns open
7Experiments proposed
0Sent to a lab
0Results returned
0Unknowns closed

Prototype. An illustrative model and lab network, built to show how the loop works.

The loop

Each run goes all the way round: the result changes the model, the model re-checks what is unknown, and the experiments are re-ranked. Press run and watch every step fill in.

1 · UnknownPrediction fails for pioglitazone
A trial measured 54% less liver fat; the model predicts 10%.
2 · ExperimentHow much does pioglitazone make fat cells hold on to fatty acids?
3 · Lab result
The lab runs the protocol and returns data.
4 · Model updated
Predictions, link confidence and the trial check are recomputed.
5 · What to run next
Unknowns are re-checked and every experiment re-ranked.

What's still unknown

15 open unknowns, ranked within each group by how much they hold back the model's predictions. Each names the experiment that would fill it.

2Failed predictions 5Missing links 5Numbers poorly pinned down 3Studies that conflict 0Closed so far

Where the model is wrong

Trials the model never saw disagree with its prediction. Something is unknown or misestimated.

FailedPrediction fails for pioglitazone
A trial measured 54% less liver fat; the model predicts 10%.
Filled by experiment 1
Details
A held-out trial measured a 54% reduction in liver fat; the model predicts 10% (90% interval -2% to 25%). Either a parameter is misestimated, most likely drop in fatty-acid supply on pioglitazone or share of liver fat delivered from fat tissue, or a mechanism is missing from the model.
FailedPrediction fails for SGLT2 inhibitors
A trial measured 25% less liver fat; the model predicts 10%.
Filled by experiment 5
Details
A held-out trial measured a 25% reduction in liver fat; the model predicts 10% (90% interval 4% to 18%). Either a parameter is misestimated, most likely liver-fat reduction per unit of weight lost or direct effect of SGLT2 inhibitors on liver fat production, or a mechanism is missing from the model.

Missing links 5

Steps in the mechanism that no study has measured, or not yet in humans or at the doses patients take.

UntargetedUnknown whether any current drug acts on fat export and droplet handling
Genetics says it matters; no one has checked whether current drugs act on it.
Filled by experiment 7
Details
Human genetics (PNPLA3, TM6SF2) shows that impaired lipid export and droplet handling can cause fatty liver, yet no treatment modeled here targets it.
No studyMetformin → fatty-acid supply to the liver
Metformin might lower fatty-acid release from fat tissue. No study in the evidence base measures it.
Filled by experiment 2
Details
Metformin might lower fatty-acid release from fat tissue. No study in the evidence base measures it. The model relies on a prior alone for this link.
Not at patient dosesACC phosphorylation → de novo lipogenesis
Every study used doses above what patients reach.
Filled by experiment 3, experiment 4
Details
Every study behind this link used concentrations above those reached in patients.
Not at patient dosesSREBP-1c activity → de novo lipogenesis
Every study used doses above what patients reach.
Filled by experiment 3, experiment 4
Details
Every study behind this link used concentrations above those reached in patients.
Not in humansMetformin → energy stress in liver cells
Shown in animals, not yet in human cells or people.
Filled by experiment 4
Details
Tested at clinically relevant concentrations, but not yet in human cells or people.

Numbers poorly pinned down 5

Quantities the model has to guess within a wide range, or that rest on a single study.

Wide rangeLiver-fat reduction per unit of weight lost
Accounts for 93% of the uncertainty in the GLP-1 receptor agonists prediction.
Only a study in people can measure this. This program's labs run cell and worm experiments, so it stays open until such a study is published.
Details
Accounts for 93% of the predictive uncertainty for GLP-1 receptor agonists. 90% credible interval: 0.9 to 5.0. Wide prior (roughly 1–6), then fitted to the Look AHEAD randomized comparison (see calibration).
Wide rangeDrop in fatty-acid supply on pioglitazone
Accounts for 70% of the uncertainty in the pioglitazone prediction.
Filled by experiment 1
Details
Accounts for 70% of the predictive uncertainty for pioglitazone. 90% credible interval: 8% to 46%. A randomized trial shows free fatty acids fall and lipolysis becomes insulin-sensitive, but reports no percentage for delivery to the liver. Centered near 25%, wide.
One studyShare of liver fat made by the liver itself
3 predictions lean on it, and only one study has measured it (Donnelly 2005, J Clin Invest).
Only a study in people can measure this. This program's labs run cell and worm experiments, so it stays open until such a study is published.
Details
3 treatment predictions depend on this number, and only one study has measured it (Donnelly 2005, J Clin Invest). Current estimate 15% to 37%. Stable-isotope labeling in nine fatty-liver patients.
One studyShare of liver fat delivered from fat tissue
2 predictions lean on it, and only one study has measured it (Donnelly 2005, J Clin Invest).
Only a study in people can measure this. This program's labs run cell and worm experiments, so it stays open until such a study is published.
Details
2 treatment predictions depend on this number, and only one study has measured it (Donnelly 2005, J Clin Invest). Current estimate 43% to 76%. Same nine patients.
Wide rangeDrop in fatty-acid supply on metformin
Accounts for 32% of the uncertainty in the metformin prediction.
Filled by experiment 2
Details
Accounts for 32% of the predictive uncertainty for metformin. 90% credible interval: 1% to 18%. Not measured by any study in the evidence base. Centered on a small effect and left wide on purpose.

Studies that conflict 3

Places where published studies point in different directions.

ConflictEnergy stress in liver cells → AMPK activation
Zhou 2001, J Clin Invest and Foretz 2010, J Clin Invest point different ways.
Filled by experiment 6
Details
Zhou 2001, J Clin Invest: Metformin activates AMPK in hepatocytes; AMPK is required for its effect on glucose output. But Foretz 2010, J Clin Invest: In liver lacking AMPK, metformin still lowers glucose output through the energy state.
ConflictSREBP-1c activity → de novo lipogenesis
Li 2011, Cell Metab and He & Wondisford 2015, Cell Metab point different ways.
Filled by experiment 3, experiment 4
Details
Li 2011, Cell Metab: Blocking SREBP-1c processing reduces lipogenesis and lipid accumulation in hepatocytes. But He & Wondisford 2015, Cell Metab: Most mechanism studies used 10–100 times the concentrations patients reach.
ConflictPPARγ in fat cells → insulin resistance of fat tissue
He 2015, Metabolism and Pravenec 2008, Int J Obes point different ways.
Filled by experiment 1
Details
He 2015, Metabolism: Thiazolidinediones inhibited lipolysis in rat fat cells, dose- and time-dependently. But Pravenec 2008, Int J Obes: Long-term pioglitazone increased fat-tissue lipolysis in rats, yet circulating fatty acids still fell through better insulin action and muscle oxidation.

Closed so far 0

Unknowns a returned result settled. They stay here so progress accumulates.

Nothing closed yet. Approve an experiment above and run it, or press “Watch one cycle”.

Experiments that would fill the gaps

Ranked by how much uncertainty each removes per dollar. The top 3 are shown in full. Each names the unknowns it fills and why; each is reviewed by an advocate, a skeptic, a methods reviewer and an editor; none runs without your approval.

Pioglitazone · Fatty-acid supply from fat tissue

How much does pioglitazone make fat cells hold on to fatty acids?

At plasma-level concentrations, pioglitazone cuts net fatty-acid release from insulin-resistant human fat cells by well over 25%, enough to explain the liver trials.

Recommend with revisions

Fills 3 unknowns

FailedPrediction fails for pioglitazone
The model's prediction for pioglitazone fails. This measures the most likely cause (drop in fatty-acid supply on pioglitazone). If the value differs from the model's guess, the prediction moves; if not, a mechanism is missing.
Wide rangeDrop in fatty-acid supply on pioglitazone
The model has to guess drop in fatty-acid supply on pioglitazone within a wide range. This measures it directly, so the range, and every prediction that uses it, narrows.
ConflictPPARγ in fat cells → insulin resistance of fat tissue
Studies disagree about PPARγ in fat cells → insulin resistance of fat tissue. A controlled test in differentiated human preadipocytes made insulin-resistant with TNF-α, day 12 can show which result holds.

Ranked here because it removes 32% of the uncertainty for $7,000 · 12 days

Skeptic: Pravenec 2008 found long-term pioglitazone increased lipolysis in rat adipose tissue while circulating fatty acids still fell; a 7-day culture may not capture that. The discrepancy may also reflect a mechanism the model lacks, not this parameter.

Metformin · Fatty-acid supply from fat tissue

Does metformin lower fatty-acid release from human fat cells?

At 10–100 µM, metformin reduces stimulated lipolysis in human fat cells by under 10%.

Recommend with revisions

Fills 2 unknowns

No studyMetformin → fatty-acid supply to the liver
No study has measured metformin → fatty-acid supply to the liver. This would be the first direct measurement, in differentiated human preadipocytes, day 12.
Wide rangeDrop in fatty-acid supply on metformin
The model has to guess drop in fatty-acid supply on metformin within a wide range. This measures it directly, so the range, and every prediction that uses it, narrows.

Ranked here because it removes 17% of the uncertainty for $4,000 · 5 days

Skeptic: Interpretable only if the clinically relevant doses carry the conclusion: 4 of 5 concentrations fall within the range patients reach.

Metformin · Fat production in the liver

The same dose-response in HepG2 cells

HepG2 cells show the same concentration dependence, at a fraction of the cost.

Not now

Fills 3 unknowns

Not at patient dosesACC phosphorylation → de novo lipogenesis
Every study of ACC phosphorylation → de novo lipogenesis used doses patients never reach. This runs a dose range that starts at patient-level concentrations (10, 25, 50 µM).
Not at patient dosesSREBP-1c activity → de novo lipogenesis
Every study of SREBP-1c activity → de novo lipogenesis used doses patients never reach. This runs a dose range that starts at patient-level concentrations (10, 25, 50 µM).
ConflictSREBP-1c activity → de novo lipogenesis
Studies disagree about SREBP-1c activity → de novo lipogenesis. A controlled test in hepG2 hepatoma cells (STR-authenticated) can show which result holds.

Ranked here because it removes 10% of the uncertainty for $2,500 · 4 days

Skeptic: HepG2 is a hepatoma line with altered lipid metabolism, and the model already discounts it. Inexpensive, but not decisive on its own.

More experiments (4)

Metformin · Fat production in the liver

Metformin dose-response on fat production in primary human liver cells

At concentrations patients reach (10–100 µM), metformin cuts fat production in human liver cells by less than 10%.

Recommend with revisions

Fills 4 unknowns

Not at patient dosesACC phosphorylation → de novo lipogenesis
Every study of ACC phosphorylation → de novo lipogenesis used doses patients never reach. This runs a dose range that starts at patient-level concentrations (10, 25, 50 µM).
Not at patient dosesSREBP-1c activity → de novo lipogenesis
Every study of SREBP-1c activity → de novo lipogenesis used doses patients never reach. This runs a dose range that starts at patient-level concentrations (10, 25, 50 µM).
Not in humansMetformin → energy stress in liver cells
So far metformin → energy stress in liver cells has been shown only in animals. This tests it in cryopreserved primary human hepatocytes, 3 donors, sandwich culture.
ConflictSREBP-1c activity → de novo lipogenesis
Studies disagree about SREBP-1c activity → de novo lipogenesis. A controlled test in cryopreserved primary human hepatocytes, 3 donors, sandwich culture can show which result holds.

Ranked here because it removes 20% of the uncertainty for $9,000 · 7 days

Skeptic: Interpretable only if the clinically relevant doses carry the conclusion: 4 of 8 concentrations fall within the range patients reach.

SGLT2 inhibitors · Fat production in the liver

Do SGLT2 inhibitors act on human liver cells directly?

Empagliflozin has no direct effect on fat production in liver cells; any liver benefit comes through weight and insulin.

Recommend with revisions

Fills 1 unknown

FailedPrediction fails for SGLT2 inhibitors
The model's prediction for SGLT2 inhibitors fails. This measures the most likely cause (direct effect of SGLT2 inhibitors on liver fat production). If the value differs from the model's guess, the prediction moves; if not, a mechanism is missing.

Ranked here because it removes 14% of the uncertainty for $6,000 · 6 days

Skeptic: A null result is the expected outcome, so the information gain is small unless it surprises. Justified mainly as confirmation that the benefit is indirect.

Metformin · Fat production in the liver

Metformin and fat storage in C. elegans, controlling for bacteria

Metformin lowers worm fat through AMPK only when the bacteria are alive.

Not now

Fills 1 unknown

ConflictEnergy stress in liver cells → AMPK activation
Studies disagree about energy stress in liver cells → AMPK activation. A controlled test in C. elegans N2 and aak-2(ok524), synchronized L4 can show which result holds.

Ranked here because it removes 1% of the uncertainty for $1,500 · 6 days

Skeptic: Worms are dosed at 25–50 mM, and metformin acts partly through their bacteria. Even with killed-bacteria controls, translational value is limited.

All treatments · Fat export and droplet handling

Does any current drug touch fat export?

None of the five patches changes VLDL-triglyceride secretion from human liver cells, in either PNPLA3 genotype, so the export module is genuinely unpatched.

Recommend

Fills 1 unknown

UntargetedUnknown whether any current drug acts on fat export and droplet handling
Genetics says this mechanism matters, but no one has checked whether current drugs act on it. This screens them in human liver cells.

Tests whether this mechanism is an untapped target · $11,000 · 8 days

Skeptic: 48 hours in culture may be too short to detect changes in export, and the genetic effect is lifelong. A null result would narrow the question, not close it.

Also considered: 1 experiment that fill no open unknown

Do GLP-1 drugs act on human liver cells directly? Semaglutide has no effect on fat production in primary human liver cells, confirming the benefit is indirect. It would confirm an assumption the model already makes, so it is not proposed.

Experiment record

Each approved experiment, what the lab returned, which unknowns it closed, and how the model's predictions changed. Saved in this browser only.

No experiments run yet. The top-ranked experiment addresses the model's failed prediction for pioglitazone.

Against the ARPA-H IGoR program metrics

	Program target	Now
Knowledge gaps detected algorithmically	≥ 3 (Phase I)	15	Met
Mechanistic sub-models (treatments)	≥ 3 (Phase I), ≥ 10 (Phase II)	5	Met
Out-of-sample prediction	improves with experiments (Phase II)	2/3	Met
Lab data returns ingested	≥ 1 (Phase I), ≥ 10 (Phase II)	0	Not yet
Model update latency	≤ 24 h, then ≤ 4 h	no data yet	–
Protocols generated automatically	≥ 3 (Phase I)	0	Not yet
Experimental modalities	≥ 3 (Phase II)	6	Met
Inter-lab concordance	≥ 85%	run one test at two labs	–
Exceptions handled automatically	≥ 30% (Phase II)	none yet	–

How well the model predicts trials

Each treatment's effect is predicted from mechanism alone, then compared with liver-fat trials the model has never seen (out-of-sample validation). An effect counts if liver fat falls by at least 10%.

Clinically meaningful reduction

2 of 3 trial outcomes predicted correctly from mechanism alone · Brier score 0.16

	Mechanism targeted			Held-out trials		Adverse effects
Lifestyle weight lossDiet and exercise		23% (8% to 43%)	90%	Calibration data, not a test	–	–
GLP-1 receptor agonistsIncretin mimetics		18% (6% to 33%)	83%	Effective · Tang 2015, Diabetes Metab Res Rev Armstrong 2016 (LEAN), Lancet	Correct	Gastrointestinal adverse events in 63.5% vs 34.3% on placebo (semaglutide 2.4 mg) Davies 2021 (STEP 2), LancetBoxed warning on the drug label DailyMed: Ozempic (semaglutide) label
PioglitazoneThiazolidinedione (PPARγ agonist)		10% (-2% to 25%) · measured 54%	47%	Effective · Belfort 2006, N Engl J Med Tang 2015, Diabetes Metab Res Rev	Failed	Weight gain of almost 3 kg Clar 2009, PLoS OneBoxed warning on the drug label DailyMed: pioglitazone tablet label
SGLT2 inhibitorsGliflozins	,	10% (4% to 18%) · measured 25%	45%	Inconsistent · Kuchay 2018 (E-LIFT), Diabetes Care Tang 2015, Diabetes Metab Res Rev	Trials disagree	Genital tract infections, odds ratio 4.2 vs placebo Liu 2015, J Diabetes Complications
MetforminBiguanide		10% (2% to 20%)	41%	No effect · Tang 2015, Diabetes Metab Res Rev Said 2017, Ann Hepatol Lavine 2011 (TONIC), JAMA	Correct	–

Bar: 90% credible interval with the posterior mean. Dots: reductions measured in held-out trials (filled when inside the interval).

What the model knows so far

Fatty liver means more fat enters the liver than leaves it. The bar shows where incoming fat comes from; the cards show each mechanism and the treatments that act on it. A treatment can only help as much as the source it targets.

59% delivered from fat tissue (43%–76%)26% made in the liver (15%–37%)16% from dietary fat (the remainder) (0%–35%)

Where incoming liver fat comes from, measured by tagging each source with isotopes. Most arrives from fat tissue, so a drug aimed only at fat-making in the liver can do little. These shares rest on one study of nine patients, so they are listed under numbers poorly pinned down.

Adds liver fat

Fatty-acid supply from fat tissue

Fat tissue that resists insulin keeps releasing fatty acids, which the liver takes up and stores.

Pioglitazone predicted 10% less liver fat

6 knowledge gaps

Adds liver fat

Fat production in the liver

Liver cells turn sugar into new fat (de novo lipogenesis), driven by insulin and SREBP-1c.

Metformin predicted 10% less liver fat
SGLT2 inhibitors predicted 10% less liver fat

9 knowledge gaps

Removes liver fat

Fat export and droplet handling

Liver cells package fat into VLDL for export and remodel their lipid droplets.

No current treatment acts on this mechanism. Human genetics (PNPLA3, TM6SF2) shows that impaired lipid export and droplet handling can cause fatty liver, yet no treatment modeled here targets it.

1 knowledge gap

Adds liver fat

Body weight and energy balance

Weight loss lowers both the fatty-acid supply and fat production.

GLP-1 receptor agonists predicted 18% less liver fat
SGLT2 inhibitors predicted 10% less liver fat
Lifestyle weight loss predicted 23% less liver fat

3 knowledge gaps

Hepatocyte triglycerideLiver fatFatty liver disease in type 2 diabetes

Causal model

The full model behind every number above. Mechanisms in rows, treatments on the left, the clinical outcome on the right. Darker links carry stronger evidence in humans; select a link to see the studies behind it.

Darker = stronger evidence in humans · thicker = more studies · teal = part of a ranked knowledge gap · red dashed = unmeasured · ↓ = inhibits · • = includes your experiments

The 17 model parameters and their priors

		Basis
Direct effect of GLP-1 drugs on fat production in liver cells	0% – 9%	No receptor found in mouse hepatocytes or in human liver expression data: expected near zero. Panjwani 2013, Endocrinology GTEx (platform data): GLP1R
Direct effect of SGLT2 inhibitors on liver fat production	1% – 23%	SGLT2 is not expressed in liver; lower insulin could still reduce fat production. Unmeasured. GTEx (platform data): SLC5A2 (SGLT2)
Drop in fatty-acid supply on metformin	1% – 18%	Not measured by any study in the evidence base. Centered on a small effect and left wide on purpose.
Drop in fatty-acid supply on pioglitazone	8% – 46%	A randomized trial shows free fatty acids fall and lipolysis becomes insulin-sensitive, but reports no percentage for delivery to the liver. Centered near 25%, wide. Gastaldelli 2009, Am J Physiol Endocrinol Metab
Extra liver fat cleared by fat burning on pioglitazone	4% – 43%	Part of a proposed mechanism (adiponectin raises fat burning in liver cells). Unmeasured; centered near 20% and wide. Bajaj 2004, J Clin Endocrinol Metab Yamauchi 2002, Nat Med
Liver-fat reduction per unit of weight lost	0.9 – 5.0	Wide prior (roughly 1–6), then fitted to the Look AHEAD randomized comparison (see calibration). Lazo 2010 (Look AHEAD), Diabetes Care
Metformin concentration reaching the liver	19 µM – 88 µM	Therapeutic levels sit 10–100× below most experimental doses; portal levels run above plasma. He & Wondisford 2015, Cell Metab Owen 2000, Biochem J
Metformin concentration that halves fat production (IC50)	43 µM – 3,638 µM	Cell and rodent studies show suppression at hundreds of µM to low mM; nothing measures it at patient concentrations, so the prior spans about 25 µM to 6 mM. Zhou 2001, J Clin Invest Li 2011, Cell Metab He & Wondisford 2015, Cell Metab
Most fat production metformin can switch off	41% – 88%	Strong suppression at high doses in rodent cells; the ceiling in human cells is unmeasured. Zhou 2001, J Clin Invest Fullerton 2013, Nat Med
Share of liver fat delivered from fat tissue	43% – 76%	Same nine patients. Donnelly 2005, J Clin Invest
Share of liver fat made by the liver itself	15% – 37%	Stable-isotope labeling in nine fatty-liver patients. Donnelly 2005, J Clin Invest
Share of pioglitazone's weight gain that acts on the liver like ordinary weight gain	13% – 86%	Unmeasured here. Part of the gain is fluid and fat under the skin, which should not load the liver.
Weight change on metformin	-1% – 3%	Not measured by any study in the evidence base; assumed small.
Weight gained on pioglitazone	2% – 5%	Almost 3 kg across trials; converted assuming a typical 90 kg trial participant. Clar 2009, PLoS One
Weight lost on GLP-1 receptor agonists	5% – 7%	Semaglutide 2.4 mg vs placebo in type 2 diabetes. Liver trials used other agents and doses; the model assumes a class effect. Davies 2021 (STEP 2), Lancet
Weight lost on SGLT2 inhibitors	2% – 3%	2.5 kg vs placebo at 1 year; converted assuming a typical 90 kg trial participant. Liu 2015, J Diabetes Complications
Weight lost with intensive lifestyle change	7% – 10%	Look AHEAD: −8.5% vs −0.05%. Lazo 2010 (Look AHEAD), Diabetes Care

Evidence

Every source, its study design, and how it is used: as model evidence, a prior, a calibration point, or an out-of-sample test. PubMed records are re-verified before every change is accepted.

Show all 36 sources and the version history

	Study design
Armstrong 2016 (LEAN), LancetLiraglutide safety and efficacy in patients with non-alcoholic steatohepatitis (LEAN): a multicentre, double-blind, randomised, placebo-controlled phase 2 study.	Study	out-of-sample test
Belfort 2006, N Engl J MedA placebo-controlled trial of pioglitazone in subjects with nonalcoholic steatohepatitis.	Study	out-of-sample test
Kuchay 2018 (E-LIFT), Diabetes CareEffect of Empagliflozin on Liver Fat in Patients With Type 2 Diabetes and Nonalcoholic Fatty Liver Disease: A Randomized Controlled Trial (E-LIFT Trial).	Study	out-of-sample test
Lavine 2011 (TONIC), JAMAEffect of vitamin E or metformin for treatment of nonalcoholic fatty liver disease in children and adolescents: the TONIC randomized controlled trial.	Study	out-of-sample test
Said 2017, Ann HepatolMeta-Analysis of Randomized Controlled Trials of Pharmacologic Agents in Non-alcoholic Steatohepatitis.	Study	out-of-sample test
Tang 2015, Diabetes Metab Res RevComparative efficacy of anti-diabetic agents on nonalcoholic fatty liver disease in patients with type 2 diabetes mellitus: a systematic review and meta-analysis of randomized and non-randomized studies.	Study	out-of-sample test
Bajaj 2004, J Clin Endocrinol MetabDecreased plasma adiponectin concentrations are closely related to hepatic fat content and hepatic insulin resistance in pioglitazone-treated type 2 diabetic patients.	Study	prior
Cabreiro 2013, CellMetformin retards aging in C. elegans by altering microbial folate and methionine metabolism.	Genetic perturbation (knockout / knockdown)	context
Clar 2009, PLoS OneAdding pioglitazone to insulin containing regimens in type 2 diabetes: systematic review and meta-analysis.	Randomized controlled trial	model evidence, prior, adverse effect
DailyMed: Ozempic (semaglutide) labelLabel carries a boxed warning	Dataset	adverse effect
DailyMed: pioglitazone tablet labelLabel carries a boxed warning	Dataset	adverse effect
Davies 2021 (STEP 2), LancetSemaglutide 2·4 mg once a week in adults with overweight or obesity, and type 2 diabetes (STEP 2): a randomised, double-blind, double-dummy, placebo-controlled, phase 3 trial.	Randomized controlled trial	model evidence, prior, adverse effect
Donnelly 2005, J Clin InvestSources of fatty acids stored in liver and secreted via lipoproteins in patients with nonalcoholic fatty liver disease.	Isotope tracer study (human)	model evidence, prior
Foretz 2010, J Clin InvestMetformin inhibits hepatic gluconeogenesis in mice independently of the LKB1/AMPK pathway via a decrease in hepatic energy state.	Genetic perturbation (knockout / knockdown)	model evidence
Fullerton 2013, Nat MedSingle phosphorylation sites in Acc1 and Acc2 regulate lipid homeostasis and the insulin-sensitizing effects of metformin.	Genetic perturbation (knockout / knockdown)	model evidence, prior
Gastaldelli 2009, Am J Physiol Endocrinol MetabDecreased whole body lipolysis as a mechanism of the lipid-lowering effect of pioglitazone in type 2 diabetic patients.	Randomized controlled trial	model evidence, prior
GTEx (platform data): GLP1ROnly tissue recorded: pancreatic islets, median 29.4 TPM	Expression atlas	model evidence, prior
GTEx (platform data): PNPLA3Liver 20.2 TPM, hepatocytes 18.5 TPM	Expression atlas	model evidence
GTEx (platform data): PPARGSubcutaneous fat 63.8 TPM, visceral fat 58.7 TPM	Expression atlas	model evidence
GTEx (platform data): SLC5A2 (SGLT2)Kidney cortex 33.8 TPM, testis 15.2 TPM; liver not recorded	Expression atlas	model evidence, prior
GTEx (platform data): TM6SF2Small intestine 76.3 TPM, liver 34.4 TPM	Expression atlas	model evidence
He & Wondisford 2015, Cell MetabMetformin action: concentrations matter.	Review	model evidence, prior
He 2015, MetabolismThiazolidinediones attenuate lipolysis and ameliorate dexamethasone-induced insulin resistance.	Pharmacological exposure	model evidence
Kozlitina 2014, Nat GenetExome-wide association study identifies a TM6SF2 variant that confers susceptibility to nonalcoholic fatty liver disease.	Human genetic association, Genetic perturbation (knockout / knockdown)	model evidence
Lazo 2010 (Look AHEAD), Diabetes CareEffect of a 12-month intensive lifestyle intervention on hepatic steatosis in adults with type 2 diabetes.	Randomized controlled trial	model evidence, prior, calibration
Li 2011, Cell MetabAMPK phosphorylates and inhibits SREBP activity to attenuate hepatic steatosis and atherosclerosis in diet-induced insulin-resistant mice.	Genetic perturbation (knockout / knockdown)	model evidence, prior
Liu 2015, J Diabetes ComplicationsEfficacy and safety of sodium-glucose cotransporter 2 inhibitors in type 2 diabetes: a meta-analysis of randomized controlled trials for 1 to 2years.	Randomized controlled trial	model evidence, prior, adverse effect
Madiraju 2014, NatureMetformin suppresses gluconeogenesis by inhibiting mitochondrial glycerophosphate dehydrogenase.	Genetic perturbation (knockout / knockdown)	model evidence
Onken & Driscoll 2010, PLoS OneMetformin induces a dietary restriction-like state and the oxidative stress response to extend C. elegans Healthspan via AMPK, LKB1, and SKN-1.	Genetic perturbation (knockout / knockdown)	context
Owen 2000, Biochem JEvidence that metformin exerts its anti-diabetic effects through inhibition of complex 1 of the mitochondrial respiratory chain.	Pharmacological exposure	model evidence, prior
Panjwani 2013, EndocrinologyGLP-1 receptor activation indirectly reduces hepatic lipid accumulation but does not attenuate development of atherosclerosis in diabetic male ApoE(-/-) mice.	Genetic perturbation (knockout / knockdown)	model evidence, prior
Pravenec 2008, Int J ObesLong-term pioglitazone treatment enhances lipolysis in rat adipose tissue.	Pharmacological exposure	model evidence
Romeo 2008, Nat GenetGenetic variation in PNPLA3 confers susceptibility to nonalcoholic fatty liver disease.	Human genetic association	model evidence
Vilar-Gomez 2015, GastroenterologyWeight Loss Through Lifestyle Modification Significantly Reduces Features of Nonalcoholic Steatohepatitis.	Observational cohort	model evidence
Yamauchi 2002, Nat MedAdiponectin stimulates glucose utilization and fatty-acid oxidation by activating AMP-activated protein kinase.	Study	prior
Zhou 2001, J Clin InvestRole of AMP-activated protein kinase in mechanism of metformin action.	Pharmacological exposure	model evidence, prior

Version history

Version 1, 2026-09-28: First model: five treatments over four mechanisms, 27 PubMed-verified papers plus GTEx and DailyMed records; liver-fat trials held out for testing.

Think a link is wrong or a study is missing? Suggest a correction. Every change is re-validated against the held-out trials before it is accepted.