A GARP antibody reached Phase 2/3 in four tumour types on a single-arm response rate, then every trial closed

What was tried

AbbVie ran NCT06487559, study M24-695, an open-label, non-randomised, sequential Phase 1b study of livmoniplimab with budigalimab in Chinese adults with locally advanced or metastatic Child-Pugh A hepatocellular carcinoma. Eligibility required Barcelona Clinic Liver Cancer stage B or C disease, ECOG performance status 0 or 1, and progression after a first-line regimen that included an immune checkpoint inhibitor. Two sequential cohorts were planned, each taking one livmoniplimab dose level with budigalimab every three weeks.

The primary endpoints were adverse event rate, dose-limiting toxicities, and maximum plasma concentration and area under the curve for both antibodies. Response by RECIST v1.1, duration of response, progression-free survival and overall survival sat in the secondary set.

The study opened on 11 September 2024. Actual primary completion and overall completion are both recorded as 18 September 2026, with three participants enrolled. The record was updated to terminated on 21 October 2026 and the posted reason reads "Strategic considerations". No results have been posted.

The biological hypothesis

Livmoniplimab, also known as ABBV-151 and originally ARGX-115, is a monoclonal antibody that binds the GARP:TGF-beta1 complex. GARP, encoded by LRRC32, holds latent TGF-beta1 on the surface of regulatory T cells and platelets and controls its activation and release, which is how Open Targets describes the protein. Binding the complex rather than the free cytokine was meant to switch off TGF-beta1 only where it is presented, leaving the rest of TGF-beta signalling intact and avoiding the cardiac toxicity that ended earlier pan-TGF-beta programmes.

Budigalimab, also known as ABBV-181, is an Fc-modified anti-PD-1 antibody. Pairing the two in checkpoint-experienced liver cancer rested on TGF-beta1 as a driver of immune exclusion and of checkpoint resistance. The supporting tumour biology is real. An immune-specific molecular class of hepatocellular carcinoma was described in Gastroenterology in 2017, and TGF-beta1-induced SOX18 was shown there in 2024 to raise PD-L1 transcription in this tumour.

What the hypothesis lacked was human genetic support. Open Targets scores the LRRC32 association with hepatocellular carcinoma at 0.0064, with a genetic association datatype score of 0 and the whole signal coming from literature text mining at 0.052, and records no clinical-stage drug candidate against LRRC32 at all.

What actually happened

The three participants in NCT06487559 produced no public data, and the decision that closed the study was taken above the trial.

The first-in-human study, NCT03821935, enrolled 364 participants and published its dose-escalation stage in Frontiers in Oncology in 2024. Fifty-seven patients were escalated, 23 on livmoniplimab alone and 34 on the combination with budigalimab at a fixed 500 mg dose. Doses ran from 3 mg to 1500 mg every two weeks, no maximum tolerated dose was reached, and the 1500 mg maximum administered dose went into expansion. Blood biomarker data showed saturation of the GARP:TGF-beta1 complex on platelets at doses within the linear pharmacokinetic range, so the antibody engaged what it was built to engage. Monotherapy produced no objective responses. The combination escalation produced an objective response rate of 15 percent with a median duration of response of 8.4 months in heavily pretreated patients across mixed solid tumours.

On that basis AbbVie opened four further trials: Phase 2 and Phase 2/3 in hepatocellular carcinoma (NCT05822752, 130 enrolled, and NCT06109272, 89 enrolled), Phase 2/3 in non-small cell lung cancer (NCT06236438, 167 enrolled) and Phase 2 in urothelial carcinoma (NCT06632951, 150 planned). AbbVie's February and April 2026 pipeline updates both carried livmoniplimab in the active oncology pipeline, with a Phase 2 first-line lung readout flagged among anticipated early-stage events. The July 2026 update lists livmoniplimab in non-small cell lung cancer under discontinued clinical programmes. As of 25 October 2026 all five of those trials are closed to enrolment and the sixth is terminated.

Failure mechanism, best guess

The archetype is translational mismatch. The posted reason for this trial is strategic, which is accurate at the level of the trial. The programme-level pattern is what that decision followed.

Three published facts sit together. Target engagement worked, with measured saturation on platelets. The monotherapy arm produced nothing. The only efficacy signal came from a 34-patient single-arm combination escalation containing an active anti-PD-1 antibody, so the 15 percent response rate cannot be attributed to GARP blockade rather than to PD-1 blockade. Nothing in the published design was built to separate the two.

ApexOnco made a related observation on 6 February 2026, writing that AbbVie had pushed livmoniplimab into Phase 3 on what it called rather limited phase 1 backing. The underlying error predates this trial. An antibody with no genetic association to its indications and no clinical precedent at its target was advanced into randomised Phase 2/3 work across four tumour types on the strength of an uncontrolled response rate.

How to prevent this next time

No endpoint-level results were posted for any livmoniplimab trial, so no reanalysis is possible here and none is offered. The levers that applied are design levers, available before the Phase 2/3 commitment.

The first is a randomised contribution-of-components step. The question was whether adding livmoniplimab to budigalimab beats budigalimab alone, and a randomised Phase 2 with a budigalimab monotherapy arm answers it at a fraction of the cost of four parallel Phase 2/3 trials. The published escalation could not answer it, and the programme scaled regardless.

The second is base-rate adjustment on the target. Open Targets scores the LRRC32 association with hepatocellular carcinoma at 0.0064 with zero genetic evidence and records no other clinical candidate against the target. A first-in-class target with no genetic support and no precedent carries a higher prior probability of failure than a Phase 1 response rate suggests, and that prior should raise the bar for Phase 3 entry rather than lower it.

The third is a red team on the pharmacodynamic claim. Saturation on platelets shows that the antibody binds its complex in blood. It does not show that latent TGF-beta1 activation fell inside the tumour, which is where the hypothesis lives. A tumour-tissue pharmacodynamic endpoint, paired with the immune-class stratification already described for this tumour, would have tested the mechanism in the compartment that mattered.

The single highest leverage change would have been running one randomised contribution-of-components Phase 2 against budigalimab alone before opening any Phase 2/3 trial.

What this means for similar programs

Latent-complex targeting is not disproved here, because no trial was designed to isolate its contribution, and the mechanism question stays open. What generalises is that a combination escalation containing an active agent cannot license Phase 3 entry on its own.

Within the Claidex graph this is the first recorded failure against LRRC32. The modified risk score for the target is 34, in the yellow band, and its composition is unusual. Phase burden contributes 3.5 of 30 and saturation 2.7 of 15, both low because the target is new and uncrowded. Genetic deficit contributes 14.9 of 15, which is the whole story in one number.

Open questions

Does blocking the GARP:TGF-beta1 complex reduce TGF-beta1 activation inside human tumours, as distinct from on circulating platelets? Will the first-line lung Phase 2 readout flagged in AbbVie's February and April 2026 updates be published now that the programme is discontinued, and will the five trials closed to enrolment post results at all?

Sources

1. • ClinicalTrials.gov, NCT06487559, livmoniplimab with budigalimab in Chinese participants with hepatocellular carcinoma, record accessed 25 October 2026: https://clinicaltrials.gov/study/NCT06487559 - ClinicalTrials.gov, NCT03821935, first-in-human study of livmoniplimab, accessed 25 October 2026: https://clinicaltrials.gov/study/NCT03821935 - ClinicalTrials.gov, NCT05822752, NCT06109272, NCT06236438 and NCT06632951, livmoniplimab Phase 2 and Phase 2/3 trials, accessed 25 October 2026: https://clinicaltrials.gov/study/NCT06236438 - Shimizu T and colleagues, First-in-human phase 1 dose-escalation results with livmoniplimab, an antibody targeting the GARP:TGF-beta1 complex, as monotherapy and in combination with the anti-PD-1 antibody budigalimab in patients with advanced solid tumors, Frontiers in Oncology, 2024,- Corrigendum to the above, Frontiers in Oncology, 2024,- Metelli A and colleagues, Immunoregulatory functions and the therapeutic implications of GARP-TGF-beta in inflammation and cancer, Journal of Hematology and Oncology, 2018,- Sanjabi S and colleagues, Transforming growth factor-beta1 in regulatory T cell biology, Science Immunology, 2022,- Teixeira AF and colleagues, Targeting immunosuppression by TGF-beta1 for cancer immunotherapy, Biochemical Pharmacology, 2021,- Sia D and colleagues, Identification of an immune-specific class of hepatocellular carcinoma, based on molecular features, Gastroenterology, 2017,- TGF-beta1-induced SOX18 elevation promotes hepatocellular carcinoma progression and metastasis through transcriptionally upregulating PD-L1, Gastroenterology, 2024,- A fully human afucosylated anti-GARP antibody PBA-0111 drives anti-tumor immunity through combined ADCC induction and TGF-beta1 pathway blockade, Frontiers in Immunology, 2026,- AbbVie, Pipeline Update, July 2026, livmoniplimab NSCLC listed under discontinued clinical programs: https://investors.abbvie.com/static-files/de1828c0-47ed-42cb-8573-24fe42ceb748 - AbbVie, Pipeline Update, February 2026 and April 2026, livmoniplimab in the active R and D pipeline: https://investors.abbvie.com/static-files/a88f5dbd-340e-4cfc-ba72-39bf41ec207e - ApexOnco, AbbVie scales back its GARP ambitions, 6 February 2026, subscription required: https://www.apexonco.com/abbvie-scales-back-its-garp-ambitions - Open Targets Platform, LRRC32 (ENSG00000137507) and hepatocellular carcinoma (MONDO_0007256) association, tractability and clinical candidate records, queried 25 October 2026: https://platform.opentargets.org - ChEMBL, livmoniplimab CHEMBL5314989 and budigalimab CHEMBL4297858, queried 25 October 2026: https://www.ebi.ac.uk/chembl - openFDA drug adverse event endpoint, queried 25 October 2026 for livmoniplimab and budigalimab: https://api.fda.gov/drug/event.json.