Roche Bets $190 Million Upfront on Hanmi's Muscle-Sparing Obesity Peptide
The $2.3 billion licensing deal for Phase 1 asset HM17321 targets a core drawback of current GLP-1 therapies: the loss of lean skeletal muscle alongside fat.
Genentech has signed an exclusive global license for Hanmi Pharmaceutical's HM17321, a long-acting analog of urocortin-2. The cash check is $190 million up front. Development, regulatory, and commercial milestones could lift the package to about $2.3 billion, with tiered royalties on net sales. Hanmi keeps South Korea. Genentech takes the rest of the world, plus manufacturing and commercialization once early human testing is done.
The asset is still in Phase 1. That is an unusual amount of money for a peptide whose muscle-sparing claim has been shown in mice, rats, and monkeys, not yet in patients. Roche is late to the incretin obesity market that Eli Lilly and Novo Nordisk already dominate. The purchase is a wager that the next fight is over what kind of weight comes off.
The Lean-Mass Bill Incretins Keep Sending
Pivotal incretin trials have delivered large mean weight losses: about 14.9% with once-weekly semaglutide in STEP-1, and about 20.9% with tirzepatide 15 mg in SURMOUNT-1. Dual-energy X-ray absorptiometry substudies tell a less tidy story. In STEP-1, lean mass fell 13.2%, or roughly 45% of the kilograms lost. In SURMOUNT-1, lean mass fell 10.9%, or about 26% of lost weight, according to a 2024 review in Diabetes, Obesity and Metabolism. A companion analysis in Circulation reached the same DXA fractions.
Those numbers sit inside a real scientific dispute. Lean mass on a DXA scan is not skeletal muscle. Up to 15% of adipose tissue is fat-free, so large fat losses inflate the lean-loss line. Some MRI work argues that muscle-volume change on GLP-1 medicines is mostly the amount expected for the weight that came off, and that muscle fat infiltration actually improves. The clinical worry does not disappear because the imaging is messy. Older adults, people with low baseline strength, and patients who stop therapy after a large caloric deficit still face sarcopenic obesity, a slower resting metabolism, and weight that returns as fat. FDA weight-management guidance still treats the scale as the primary efficacy endpoint and body composition as a safety add-on in a fraction of Phase 3 volunteers. Drugmakers who can change composition, not only kilograms, are writing their own competitive axis.
A Peripheral Receptor, Not Another Satiety Hormone
HM17321 is a selective agonist of corticotropin-releasing factor receptor 2 (CRFR2). Urocortin-2 is the endogenous ligand. Hanmi describes a non-incretin, peptide mechanism aimed at fat loss with lean-mass preservation. CRFR2 signaling has been tied to insulin sensitivity, skeletal muscle hypertrophy, and adipose catabolism. That is a different circuit from the hypothalamic satiety and delayed gastric emptying that define GLP-1 and dual GIP/GLP-1 drugs.
At the 2025 American Diabetes Association meetings, Hanmi showed diet-induced obese mice on HM17321 monotherapy losing body weight and fat mass while gaining lean mass (886-P). Combined with HM15275, Hanmi's GLP-1/GIP/glucagon triple agonist, fat loss deepened and skeletal muscle weight and function rose. Switching from the triple agonist onto HM17321 kept the fat off and the lean mass intact. A companion poster (843-P) put the analog against semaglutide: slightly less total weight loss than the GLP-1, greater fat-mass loss, increased lean mass, and lower muscle lipid, with pair-fed controls pointing to direct adipose breakdown and muscle anabolism rather than appetite alone. Similar recomposition showed up in obese rats. In monkeys, weight loss was driven by fat.
None of that is a human efficacy readout. The U.S. FDA cleared an investigational new drug application in November 2025. The ongoing study (NCT07219589) is a 90-person, randomized, double-blind, placebo-controlled single- and multiple-ascending-dose trial at Medpace in Cincinnati. Part A doses about 40 healthy adults. Part B gives about 50 adults with obesity once-weekly subcutaneous injections for 12 weeks. Primary endpoint: treatment-emergent adverse events. Secondary endpoints are pharmacokinetic. People with diabetes and recent GLP-1 use are excluded from Part B. Hanmi finishes Phase 1. Genentech owns Phase 2 and beyond.
A Combination Chip Against Lilly and Regeneron
In December 2023 Roche paid $2.7 billion in cash, plus up to $400 million in milestones, to acquire Carmot Therapeutics and its incretin stack: CT-388, a weekly dual GLP-1/GIP agonist; CT-996, an oral small-molecule GLP-1; and CT-868. That release already flagged combinations focused on preserving muscle mass. HM17321 is the non-incretin half of that sentence. As a peptide, it is also a candidate for future fixed-dose combinations, a possibility Hanmi and Genentech flagged and that still needs clinical proof, as a BioXconomy writeup of the deal noted.
Lilly is running bimagrumab, an activin type II receptor antibody, with semaglutide. Regeneron is pairing semaglutide with trevogrumab and garetosmab. Those programs block catabolic muscle signaling. HM17321 tries to turn on an anabolic and lipolytic receptor instead. Roche is paying for optionality: a weekly shot that might stand alone, or sit next to CT-388 or an oral incretin, if CRFR2 agonism is tolerable at doses that recompose people the way it recomposed animal models.