CT-388: experimental obesity & type 2 diabetes drug (Roche/Genentech)
A once-weekly injectable being tested for weight loss and type 2 diabetes; it activates two gut hormones at once, like Ozempic but targeting both GLP-1 and GIP pathways. Experimental, not yet an approved drug.
A researcher, an agent, or an algorithm wrote down the sequence and picked a target to hit.
An AI model like OpenFold3 or AlphaFold built a 3D structure and scored how well it fits the binding site.
A second contributor repeated the computation on their own hardware and the scores matched.
A chemistry service or a researcher ordered the sequence, it was manufactured, and mass spectrometry confirmed the right molecule was produced.
A binding or activity measurement confirmed that it actually does what the computer predicted — or didn't.
What this is
CT-388 (also designated RO7795068) is an investigational once-weekly injectable medication being studied for obesity and type 2 diabetes. It activates two gut-hormone receptors at once — the GLP-1 receptor and the GIP receptor — making it a "dual agonist" in the same class as tirzepatide. CT-388 is developed by Roche/Genentech and is not approved by the FDA or any regulatory authority; it is only available to participants in Roche-sponsored clinical trials.
What sets CT-388 apart from tirzepatide is how it engages those two receptors: it is engineered for balanced, equal activation of both GLP-1R and GIPR, whereas tirzepatide favors the GIP receptor by roughly 9:1. CT-388 is also designed with cAMP-biased signaling — a property intended to reduce receptor desensitization over time by minimizing β-arrestin recruitment at both receptors. The stored raw sequence (HAEGTFTSDVSSYLEGQAAKEFIAWLVKGRSSX) is a representative approximation; the full chemistry of the active compound, including any modifications supporting its approximately one-week half-life, has not been disclosed in published literature.
History
CT-388 originated at Carmot Therapeutics, a California-based biotech founded around a platform for engineering peptide multi-receptor agonists with biased signaling profiles (Madsbad and colleagues, Expert Opinion on Investigational Drugs 2025). The molecule was conceived as a deliberate pharmacologic response to tirzepatide's characterized limitations: its imbalanced approximately 9:1 GIP-over-GLP-1 receptor engagement and β-arrestin-driven receptor desensitization, which had been documented in the literature (Willard and colleagues, JCI Insight 2020). Carmot advanced CT-388 through Phase 1 and into Phase 2 before Roche acquired the company in December 2023 for $2.7 billion upfront, with CT-388 — renamed RO7795068 — as the central asset. Roche reported 48-week Phase 2 data in early 2026 and announced the Phase 3 ENITH program.
What it does
CT-388 works by mimicking two gut hormones simultaneously. Activating the GLP-1 receptor suppresses appetite, slows stomach emptying, and stimulates insulin release in proportion to blood sugar levels. Activating the GIP receptor adds an additional boost to the postprandial insulin response and may improve how the body handles fats. Together, these effects reduce food intake and lower blood sugar — the same combination that drives weight loss with tirzepatide.
The added design feature is biased signaling: CT-388 is engineered to preferentially trigger the cAMP-mediated pathway at both receptors while minimizing the β-arrestin pathway, which is associated with receptor internalization and loss of response over time. In preclinical models, the biased-agonism design produced dose-dependent weight reduction, glycemic improvement, and improved metabolic liver pathology (Rodriguez and colleagues, Cell Reports Medicine 2025).
Evidence
- Human: Phase 1 in adults with obesity (including a type 2 diabetes cohort) demonstrated dose-dependent weight loss and glycemic improvement; in the Phase 1b T2D cohort at the 22 mg dose, 50% of participants reached HbA1c below 5.7% (diabetes remission threshold) at approximately 12 weeks (Madsbad and colleagues, Expert Opinion on Investigational Drugs 2025). Phase 2 CT-388-103 (469 adults with obesity, 48 weeks) reported 22.5% placebo-adjusted weight loss at the 24 mg dose (efficacy estimand; 18.3% by treatment-regimen estimand); 87% of participants achieved ≥10% weight loss and 47.8% achieved ≥20%; no weight-loss plateau was observed at study end; treatment-related discontinuation was 5.9%; 73% of prediabetic participants in the CT-388 arm achieved normal blood glucose versus 7.5% in the placebo arm (Madsbad and colleagues, Expert Opinion on Investigational Drugs 2025). A separate Phase 2 trial in approximately 360 adults with type 2 diabetes (CT-388-104) was ongoing at published research date with results not yet available. Phase 3 ENITH1 and ENITH2 were announced to begin Q1 2026; no Phase 3 results were available at published research date.
- Animal: Preclinical studies in rodents and non-human primates showed dose-dependent weight reduction, improved glycemic control, and improved metabolic-associated steatohepatitis (MASH) pathology. The cAMP-biased, β-arrestin-sparing design of dual GLP-1R/GIPR agonists was independently characterized and shown to yield greater glucose-lowering and weight-loss efficacy in preclinical models compared with non-biased agonism (Rodriguez and colleagues, Cell Reports Medicine 2025).
- In vitro: Receptor binding, cAMP-biased signaling, and reduced β-arrestin recruitment were characterized in preclinical assay systems. Spatiotemporal GLP-1R and GIPR signaling and trafficking dynamics induced by dual agonists have been described in the literature (Gasbjerg and colleagues, Physiological Reviews 2026).
Known effects
- Weight loss in adults with obesity — Phase 2 human evidence (single 469-patient trial, 48 weeks; Phase 3 not completed)
- Glycemic improvement in type 2 diabetes — Phase 1b human evidence (small T2D cohort; dedicated Phase 2 T2D trial results pending)
- Prediabetes glycemic normalization — Phase 2 (73% of prediabetic participants reached normal blood glucose at 48 weeks vs 7.5% placebo)
- Metabolic liver improvement (MASH) — Preclinical only (rodent and non-human primate models)
- Reduced receptor desensitization vs. standard dual agonism — Mechanistic/preclinical; clinical translation not yet confirmed
Safety signals
Available safety data derive from Phase 1 and Phase 2 CT-388-103 (469 patients, 48 weeks). The safety profile at available follow-up appears consistent with the GLP-1 receptor agonist class:
- Nausea, vomiting, diarrhea — class-typical gastrointestinal adverse events; described in available sources as mostly mild to moderate, predominantly occurring during the dose-escalation phase
- Decreased appetite — pharmacologically expected from GLP-1R agonism; also part of the intended mechanism
- Treatment-related discontinuation — reported at 5.9% in the 24 mg arm of Phase 2 CT-388-103
The following safety questions remain unanswered because they require multi-year exposure data not yet generated:
- Long-term safety beyond 48 weeks — pancreatitis, gallbladder disease, retinopathy progression, and rare oncologic signals (all identified as Phase 3 unknowns in available literature)
- Cardiovascular outcomes — no cardiovascular outcomes trial has been conducted; MACE-reduction signals documented with semaglutide and tirzepatide do not automatically transfer to CT-388
- Medullary thyroid carcinoma and MEN2 — expected to be labeled contraindications based on the C-cell tumor signal observed across the GLP-1R agonist class in rodent carcinogenicity studies, but CT-388 has no approved label
- Effects in pregnancy and lactation — no human data; available sources describe these as precautionary contraindications consistent with the incretin class
Regulatory status
- US (FDA): Not approved — investigational. Legally administered only to participants in authorized Roche-sponsored clinical trials. Per published sources, not legally compoundable under US 503A or 503B pathways.
- International: Not approved in any jurisdiction. Trial sites span North America, Europe, and other regions; no commercial availability.
- WADA: Not explicitly listed by name on the WADA Prohibited List at published research date. Per available sources, CT-388 falls under WADA S0 (substances not approved by any government regulatory health authority for human therapeutic use). S2 (peptide hormones, growth factors, related substances) is described as the expected classification upon future approval.
Myths and misconceptions
- "CT-388 is available through compounding pharmacies" — No. CT-388 is an investigational Roche product not approved by any regulatory authority and not legally compoundable in the US under 503A or 503B pathways. Anything offered under the CT-388 or RO7795068 name outside a Roche-sponsored trial is not the clinical-grade molecule and cannot be assumed to share its sequence, purity, or pharmacology.
- "CT-388's 22.5% Phase 2 weight loss proves it is more effective than tirzepatide" — Not established. Cross-trial comparisons against tirzepatide's SURMOUNT-1 results use different patient populations, trial durations, estimands, and titration schemes. The Phase 2 number is genuinely strong and the weight-loss curve had not plateaued at 48 weeks, but a fair efficacy comparison requires a head-to-head trial that has not been conducted.
- "Biased agonism is a marketing concept; receptor pharmacology doesn't matter" — Signaling bias is a well-characterized pharmacologic phenomenon; tirzepatide's imbalanced, GIP-favoring profile is documented in peer-reviewed literature (Willard and colleagues, JCI Insight 2020). Whether CT-388's specific design choices produce a clinically meaningful difference in durability or tolerability is an empirical question Phase 3 is designed to answer.
Mechanism
CT-388 is a unimolecular peptide-based dual agonist of the glucagon-like peptide-1 receptor (GLP-1R) and the glucose-dependent insulinotropic polypeptide receptor (GIPR). GLP-1R agonism drives glucose-dependent insulin secretion, suppresses glucagon, activates hypothalamic satiety pathways, and slows gastric emptying. GIPR agonism potentiates the postprandial insulin response and may improve lipid metabolism. This receptor combination is the same mechanistic class as tirzepatide.
CT-388 is distinguished from tirzepatide by two specific design properties. First, balanced equimolar receptor engagement: tirzepatide's receptor potency is approximately 9:1 GIP-over-GLP-1 biased (Willard and colleagues, JCI Insight 2020); CT-388 is designed for parity at both receptors. Second, cAMP signal-biased activation with minimal β-arrestin recruitment at both GLP-1R and GIPR. β-arrestin-mediated receptor internalization is a recognized driver of receptor desensitization and tachyphylaxis in peptide GPCR pharmacology. The biased-agonism design is intended to reduce receptor downregulation and sustain pharmacologic activity over time. Preclinical characterization of dual GLP-1R/GIPR biased agonism is described in Rodriguez and colleagues (Cell Reports Medicine 2025), which showed greater efficacy versus non-biased dual agonism in animal models.
Whether these design differences translate into clinically superior durability or efficacy versus tirzepatide remains an open question. Phase 2 results are consistent with the mechanistic hypothesis; a direct head-to-head trial has not been conducted.
The approximately one-week half-life enabling once-weekly dosing is reported in available sources; the structural modifications (conjugation or formulation strategy) responsible for extended half-life are not individually disclosed in published literature.
Open questions
- Phase 3 efficacy confirmation: Whether the 22.5% Phase 2 weight-loss result will replicate in the Phase 3 ENITH1 and ENITH2 trials at the scale and population diversity required for regulatory approval. Single Phase 2 trials in obesity have historically been variable.
- Biased agonism clinical translation: Whether cAMP-biased, β-arrestin-sparing dual agonism produces a measurably superior durability or efficacy outcome versus tirzepatide in a controlled comparison. The preclinical mechanistic case is characterized as strong in available literature (Rodriguez and colleagues, Cell Reports Medicine 2025); Phase 3 data are required to confirm clinical translation.
- Cardiovascular outcomes: No cardiovascular outcomes trial has been conducted. Whether CT-388 will replicate the MACE-reduction signals documented with semaglutide and tirzepatide is not established, and a dedicated CVOT may be required prior to broad approval.
- Long-term safety beyond 48 weeks: Pancreatitis, gallbladder disease, retinopathy progression, and rare oncologic signals require multi-year exposure data not yet generated by the development program.
- Phase 2 T2D outcome: CT-388-104 (~360 T2D patients) was ongoing at published research date; T2D-specific efficacy and HbA1c endpoint results are not reported.
- Body composition: Sub-studies on lean mass preservation, bone density, and effects of resistance training during CT-388-induced weight loss are not yet published — a gap shared with the broader incretin class.
- Head-to-head comparison: All current efficacy comparisons with tirzepatide and retatrutide are cross-trial, with different populations, estimands, titration schemes, and durations. No prospective head-to-head trial exists at published research date.
- Combination strategies: CT-388 in combination with complementary agents (such as petrelintide, an amylin analog also in the Roche obesity pipeline) is noted as a potential development direction; no combination clinical evidence exists at published research date.
Related peptides
- Tirzepatide — FDA-approved dual GLP-1R/GIPR agonist; CT-388's closest pharmacologic predecessor and benchmark for cross-trial comparison, distinguished by its approximately 9:1 GIP-over-GLP-1 receptor bias
- Retatrutide — triple GLP-1R/GIPR/GCGR agonist from Eli Lilly; Phase 2 results in the same efficacy range as CT-388, used as a cross-trial comparator
- Semaglutide — GLP-1R-only agonist (Ozempic, Wegovy); the established clinical benchmark for GLP-1 class weight loss and cardiovascular outcomes
Research directions for this peptide, selected from the current sources — hypotheses you can explore and model. None of it is proven yet; tap any one to see the full thinking.
Does CT-388 activate its two target receptors equally in all relevant tissues (fat, pancreas, brain), or does the balance shift depending on where in the body it acts?
If the balance varies by tissue, drug developers could optimize dosing strategies to favor brain effects (appetite reduction) or fat effects (lipid mobilization) in different patients, enabling more personalized obesity treatment.
Could CT-388 reverse liver scarring in metabolic fatty liver disease by directly activating a receptor on liver cells, separate from its weight-loss effect?
If true, it would make CT-388 relevant to fatty liver disease, which affects 1 in 4 adults globally and has no approved injectable treatment, potentially giving patients a single drug that addresses both obesity and its most dangerous liver complication.
Does CT-388 avoid the gradual weakening of GIP receptor response that may limit tirzepatide's long-term effectiveness?
If true, patients on CT-388 might continue losing weight at months 6-12 when those on tirzepatide plateau, potentially achieving greater total weight loss without requiring dose escalation, which matters enormously to the hundreds of millions of people with obesity.
Does the long-acting chemical modification of CT-388 shift which receptor it prefers depending on whether it is in the blood or in tissue?
If true, it would mean that carefully tuning where and how CT-388 is modified could allow engineers to precisely control whether the drug preferentially curbs appetite (brain, GLP-1R) or burns fat (adipose, GIPR), enabling a more targeted obesity drug than any currently available.
▸full evidence table2 metrics
| metric | value | tool |
|---|---|---|
| ipTM | 0.7878170013427734 | openfold3-mlx |
| ranking score | 0.84688401222229 | openfold3-mlx |
▸structural qualityopenfold3
| metric | value | note |
|---|---|---|
| gpde | 0.736 | global PDE — lower = better |
| disorder | 0.144 | fraction disordered |
| chain pair ipTM (A, B) | 0.788 | interface quality |
▸3-letter notation
▸recipeopenfold3-mlx 0.3.1
| parameter | value |
|---|---|
| model | openfold3-mlx 0.3.1 |
| weights | — |
| hardware | — |
| mlx version | — |
| python | — |
| random seed | — |
| msa strategy | — |
| diffusion samples | 1 |
| runtime | 580s |
| predicted by | mlx@peptide |
| predicted at | 2026-05-03 |
▸citationbibtex
@peptide{pep10906,
sequence = {HAEGTFTSDVSSYLEGQAAKEFIAWLVKGRSSX},
target = {gipr},
author = {peptidemodel},
year = {2026},
status = {computed}
}