Is Orforglipron Safe for the Liver? The Pooled Phase 3 Data

So far, the phase 3 data include no confirmed case of orforglipron-caused liver injury, though a few liver measures ran numerically higher than on comparators. In a Lilly-funded pooled analysis of seven trials and 11,220 participants, 6,920 of them on orforglipron, no one had confirmed drug-induced liver injury or Hy’s Law, and hepatic adverse events were balanced at 4.6% versus 4.4% on comparators^[1]. The trials excluded people with significant liver disease, and FDA asked Lilly for enhanced liver surveillance after approval^[1][3].

Why this question hangs over the drug

The worry comes from earlier small-molecule pills. Pfizer ended lotiglipron after a phase 2 study in which transaminase elevations occurred in 6.6% of the type 2 diabetes cohort and 6.0% of the obesity cohort, against 1.6% on placebo in the obesity cohort; attempts to identify who was at risk failed^[5]. Development of danuglipron was later halted as well, with suspected drug-induced liver injury cited^[6].

Orforglipron belongs to the same broad category, a non-peptide agonist taken by mouth, and it is cleared through the liver. Its label says it “is metabolized primarily via hepatic CYP3A4”^[2]. In laboratory tests it also inhibits the liver transporters OATP1B, BSEP and MRP2, though the pooled paper’s authors argue that concentrations at therapeutic doses are too low for this to matter clinically^[1]. A 2026 narrative review proposes that the liver problems are specific to each molecule and do not come from GLP-1 receptor activation, while acknowledging that definitive toxicology has not been fully disclosed^[7]. That remains a hypothesis.

What the pooled analysis did

Wharton and colleagues combined ATTAIN-1 and ATTAIN-2 in weight management with ACHIEVE-1 through ACHIEVE-5 in type 2 diabetes, comparing orforglipron with placebo, dapagliflozin, oral semaglutide and insulin glargine for up to 104 weeks including safety follow-up^[1]. Doses are reported in tablet equivalents up to 17.2 mg; the trials used capsules up to 36 mg, later shown to be bioequivalent^[1]. Results cover everyone who received at least one dose of study drug^[1].

Liver tests were drawn on a schedule, every 4 to 12 weeks depending on the program and phase^[1]. Cases crossing set thresholds, such as ALT or AST at 5 times the upper limit of normal (ULN) or higher, went to two blinded safety physicians employed by the sponsor, with a hepatologist as tiebreaker^[1]. Hy’s Law is the classic warning pattern: ALT or AST at 3 times ULN or more, bilirubin at 2 times ULN or more, no cholestasis, and no other explanation^[1].

The main numbers

All figures in the table come from the pooled analysis, across all seven trials^[1]. Exposure was unequal, at 8,889.1 patient-years on orforglipron and 5,922.6 on comparators, and the comparator pool mixes placebo with active drugs^[1].

The six orforglipron participants who hit the combined enzyme-and-bilirubin threshold were each assigned another cause: gallstone disease in two, acute hepatitis A in one, and three others attributed to possible alcohol-associated liver disease, suspected Gilbert’s syndrome with alcohol use and steatosis as confounders, and herbal medicine with Gilbert’s syndrome^[1]. That is the basis for the zero in the headline. An adjudicated zero depends on the judgment of the adjudicators, who here were sponsor employees using a modified US Drug-Induced Liver Injury Network classification^[1].

Where orforglipron ran numerically higher

• Very high ALT. ALT at 10 times ULN or more occurred in 16 versus 6 participants. The paper says the majority of orforglipron cases at that level had other causes reported, which leaves open that some did not^[1].
• Cholestasis quadrant. On the liver-injury screening plot, 21 (0.31%) versus 7 (0.16%) participants fell in the cholestasis quadrant. Gilbert’s syndrome, hyperbilirubinemia history or elevated baseline bilirubin was reported in 12 of the 21 and 2 of the 7. The authors attribute the imbalance mainly to greater weight loss (−15.0% vs −3.9% in those participants)^[1]. That is their interpretation, and the analysis cannot prove it.
• Gallbladder events. Gallbladder and biliary events occurred in 105 (1.5%) versus 49 (1.1%), driven by gallstones at 1.0% versus 0.6%^[1]. The authors note that gallstones were not systematically screened and that more abdominal imaging on orforglipron may have inflated the count^[1]. The label pool of the two weight-reduction trials shows cholelithiasis in 1% versus 0.7% and acute cholecystitis in 0.4% versus 0.3%, and acute gallbladder disease is a labeled warning^[2]. Our Foundayo side effects review covers it alongside the other labeled risks.

Liver enzymes fell on average

In the placebo-controlled weight-management trials at week 72, model-based estimates in the safety population show ALT falling 15.7% on 5.5 mg, 18.4% on 9 mg and 19.8% on 17.2 mg, versus 8.0% on placebo; the 17.2 mg difference was −12.9% (95% CI −16.1 to −9.6)^[1]. In open-label ACHIEVE-3 at week 52, ALT fell 28.7% on orforglipron 17.2 mg and 22.0% on oral semaglutide 14 mg (difference −8.6%; 95% CI −14.4 to −2.4)^[1]. These treatment comparisons were not adjusted for multiplicity, and the model assumes data are missing at random^[1].

A lower ALT is not the same as a healthier liver. The authors say dedicated studies assessing inflammation and fibrosis are needed to learn whether these drops reflect improved disease activity, because steatosis and fibrosis were not systematically evaluated^[1]. Falling enzymes are a familiar pattern with weight loss, as our review of GLP-1 drugs and ALT, AST and GGT describes. Orforglipron has not been shown to treat fatty liver disease in these data.

Who these results may not cover

Entry rules shaped the population. Participants needed ALT and AST under 3 times ULN in the weight-management trials and no more than 5 times ULN in the diabetes trials, and people with acute or chronic hepatitis or liver disease other than fatty liver disease were excluded^[1]. Results looked consistent in those with elevated baseline enzymes, defined as ALT or AST at 1.5 times ULN or higher, but that subgroup assessment was post hoc, and their fatty liver disease was presumed from their metabolic profile and lab values^[1].

For people with established liver disease, the guidance comes from a pharmacokinetic study, not from outcome data. After a single 0.8 mg dose, exposure (AUC) was similar with mild hepatic impairment and rose 1.7-fold with moderate and 4.6-fold with severe impairment^[2]. The label’s wording:

FOUNDAYO is not recommended for use in patients with severe hepatic impairment (Child-Pugh Class C). No dosage modification of FOUNDAYO is recommended in patients with mild or moderate hepatic impairment (Child-Pugh Class A and B, respectively).

— Foundayo prescribing information, section 8.6, July 2026 label version

Sponsor involvement and other limits

Eli Lilly funded the work and designed and oversaw the trials, provided the data analysis and medical writing support, and five of the six authors are Lilly employees and shareholders; the sixth, Sean Wharton, reports personal fees from Lilly and other drugmakers^[1]. Three of the seven trials were open-label, which could bias how hepatic events were identified and reported, and the authors caution that differences in design and duration limit cross-trial inference^[1].

Two outside sources add context. The NIH LiverTox entry states that aminotransferase elevation rates are “no higher than with placebo” and that orforglipron has not been linked to clinically apparent liver injury with symptoms or jaundice^[4]. In a lab study using human liver spheroids, danuglipron caused dose-dependent cell toxicity from 15 times its therapeutic peak concentration, while orforglipron did so only above 30 times^[6]. That is a preclinical finding, its senior author is CEO and a shareholder of a liver-spheroid company^[6], and a lab model cannot stand in for clinical data.

What FDA still wants to see

Clinical trials of this size can miss rare reactions, and FDA treated the question as open at approval. Its April 2026 approval letter required Lilly to complete ACHIEVE-4 and submit an assessment of potential liver injury, including Hy’s Law plots and case reports, on a timetable with the final report due in July 2026^[3]. That date has passed, and ACHIEVE-4’s liver data appear in the pooled analysis above^[1]. The letter also requests 5 years of enhanced pharmacovigilance: serious unexpected liver-injury cases resulting in jaundice, liver failure, transplant or death go to FDA as 15-day alert reports, with periodic analyses of potential Hy’s Law cases and transaminase elevations above 10 times ULN^[3]. Reporting of that kind is what can show whether rare injury appears in wider use.

The Foundayo label’s nine warnings and precautions do not include a hepatotoxicity warning, and it does not list routine liver enzyme monitoring in its dosing or warning instructions^[2].

What to raise with a prescriber

• Any known liver disease, cirrhosis or prior abnormal liver tests, since the trial evidence stops at ALT or AST of 5 times ULN and the label addresses severe impairment separately^[1][2].
• Whether baseline liver tests make sense for you, particularly with fatty liver disease or heavy alcohol use. Alcohol featured in two of the six adjudicated cases^[1].
• Every supplement and herbal product you take. One of the six adjudicated cases was attributed most likely to herbal medicine^[1].
• Medications affecting CYP3A4, because inhibitors raise orforglipron exposure and the label sets limits for some of them^[2].
• Upper abdominal pain, fever or yellowing of the skin or eyes, which the Medication Guide lists as possible signs of gallbladder problems and tells patients to report right away^[2].