ALT, AST, ALP, total and direct bilirubin. You need the starting point, because the interpretive question later is always "is this new?"

The Phase 3 programmes excluded ALT/AST ≥3× ULN in weight management and >5× ULN in T2D, plus ALP ≥1.5× ULN or bilirubin ≥1.5× ULN (Gilbert's excepted). Above those thresholds you're outside the evidence base, not necessarily contraindicated.

Mild-to-moderate elevation consistent with MASLD is not a reason to withhold therapy — that subgroup was specifically analysed and did fine.

With MASLD in 65–81% of T2D and ~70% of overweight/obesity, most candidates for these drugs have some hepatic steatosis. The panel tells you about inflammation; it says nothing about fibrosis.

Stage it if the answer would change management — i.e., if F2–F3 would push you toward semaglutide 2.4 mg for the on-label MASH indication. FIB-4 first, then elastography or imaging, then hepatology.

A normal ALT does not exclude significant fibrosis.

In the pooled orforglipron data, ALT dropped 15.7–19.8% at 72 weeks in weight management and 9.3–16.1% at 40 weeks in T2D, dose-dependently. The same direction of travel holds across the semaglutide and tirzepatide programmes.

Reductions are largest in patients who started highest — 36.6% to 74.8% mean reduction in the elevated-baseline subgroup.

Frame this for the patient as a metabolic improvement marker, with the honest caveat that it doesn't prove fibrosis regression.

All six orforglipron-treated participants who hit ALT/AST ≥3× ULN plus bilirubin ≥2× ULN had a competing explanation. Biliary disease was the most common; the remainder were acute viral hepatitis, Gilbert's syndrome, alcohol-associated liver disease, and herbal/supplement intake:

Biliary disease (cholelithiasis ~1.0% vs 0.6% — a rapid-weight-loss effect, though possibly amplified by ascertainment bias) · acute viral hepatitis · Gilbert's syndrome (fractionate the bilirubin — predominantly indirect) · alcohol-associated liver disease · herbal and supplement intake.

Right-upper-quadrant ultrasound early. Gallstones are the single highest-yield finding.

Hy's Law = ALT or AST ≥3× ULN concurrent with total bilirubin ≥2× ULN, with ALP <2× ULN (no cholestasis), after alternative etiologies are excluded. That last clause does most of the work.

Temple's corollary — significant transaminase elevation without bilirubin elevation — signals possible hepatocellular injury but is not Hy's Law. It was balanced between orforglipron (2.3%) and comparators (2.0%).

Isolated bilirubin elevation with a normal direct fraction and normal ALP is Gilbert's until proven otherwise, not drug injury.

Permanent discontinuation for a hepatic event occurred in 0.17% of orforglipron-treated vs 0.12% of comparator-treated participants. Serious hepatic events: 0.07% vs 0.16%.

Stopping a drug that is improving the underlying metabolic driver, on the basis of a transaminase bump with an obvious alternative cause, is the more common error.

Escalate to hepatology for persistent ≥5× ULN elevations, any Hy's Law pattern surviving workup, or new synthetic dysfunction (INR, albumin).

Biopsy-confirmed MASH with F2–F3 (the population actually studied; non-invasive staging is a real-world extrapolation): semaglutide 2.4 mg is the only GLP-1 with a label for it, which also makes it the only one with a coverage pathway on that indication.

MASLD without staged fibrosis, plus obesity or T2D: treat the metabolic disease. Any effective agent producing substantial weight loss moves the liver in the right direction.

Compensated cirrhosis (F4): off-label everywhere. The July 2026 Phase 2 signal is encouraging but not practice-changing; co-manage with hepatology.

Hard clinical outcomes. Every liver approval to date rests on histologic surrogates. Cirrhosis-free survival — progression to cirrhosis, MELD ≥15, decompensation, transplant — arrives with ESSENCE Part 2.

Combination therapy. Semaglutide + resmetirom is unstudied. The one incretin-plus combination trial that did read out — zalfermin (FGF21 analogue) + semaglutide — failed to beat placebo. Other FGF21 analogues with different designs remain in development.

Durability after stopping. No trial has followed histology after discontinuation. Given weight regain patterns, this is a live question.

Underrepresented groups. Black patients were 0.6% of ESSENCE enrolment. Lean MASH was ~3%.

The orforglipron hepatic safety pooled analysis was funded by Eli Lilly; the sponsor designed the trials, performed the statistical analysis, and provided medical writing support. Five of six authors are Lilly employees and shareholders.

ESSENCE and the 2026 cirrhosis trial were funded by Novo Nordisk. SYNERGY-NASH was funded by Eli Lilly.

None of this invalidates the data — the analyses are prespecified and the datasets are large. It does mean independent replication carries extra weight here.