ASCO 2026 preview – no GPC3 glory yet for BeOne

22 May 2026

First human data for BeOne’s BGB-B2033 lack the wow factor.

As GPC3 has emerged as one of the most promising novel oncology targets over the past two years numerous companies have tried to get in on the act, but one of these, BeOne, won’t be celebrating just yet.A newly unveiled ASCO abstract featuring BeOne’s BGB-B2033, a 4-1BB co-stimulated bispecific MAb against GPC3, has shown a meagre 20% confirmed response rate among 59 evaluable patients with GPC3-expressing pretreated liver cancer. This looks meaningfully worse than AbelZeta/AstraZeneca’s Car-T project AZD7003, whose 57% confirmed ORR in a similar setting started off the GPC3 craze.True, it would be unrealistic to expect an antibody approach to match that of a Car-T therapy, and BeOne might argue that an off-the-shelf product without the need for apheresis and lymphodepletion could justify some falloff in efficacy. Whether an ORR number less than half its Car-T rival is acceptable, however, now becomes the big question.First human dataThe ASCO results for BGB-B2033, from a phase 1 trial that started two years ago, represent the first human data for this asset, and indeed they are one of the first human datasets for any GPC3-targeting project outside Car-T therapy. BeOne earlier touted the phase 1 trial as a prelude to starting a pivotal phase 2 liver cancer study later this year.The 20% confirmed ORR comes from 12 partial responses among 59 evaluable patients, and the best BeOne can say about efficacy is that a preliminary dose-response relationship has been seen, and among 38 patients given more than the “predicted target efficacious dose” the ORR rose to 29%.It’s not clear whether the company has scope to dose BGB-B2033 much higher, however. Though the project has been described as generally well tolerated, 8% of patients experienced treatment-related grade 3 or higher adverse events; there was one dose-limiting toxicity, liver enzyme elevation, that resolved after dose reduction. Available data for GPC3-targeting projects in GPC3+ve relapsed liver cancerProjectMechanismCompanyStudyDataAZD7003/ C-CAR031dnTGFbRII-armoured CGPC3 Car-TAstraZeneca/ AbelZetaNCT05155189ORR 57%Ori-C101GPC3 Car-TOriCellChiCTR1900028121ORR 44%EU307GPC3 Car-TEutilexNCT05783570ORR 38%BGB-B2033CPC3 x 4-1BB MAbBeOneNCT06427941ORR 20% (29% at "efficacious" dosesCT011GPC3 Car-TCarsgenNCT03980288ORR 17%ERY974GPC3 x CD3 T-cell engagerChugai (Roche)NCT02748837ORR 0%Source: OncologyPipeline. The lacklustre dataset might be seen as another setback for co-stimulation, where most recently Roche discontinued the anti-CD19 x 4-1BB project englumafusp alfa. BeOne says the design of BGB-B2033, which also includes an engineered Fc region to prevent FcγR binding and increase half-life, is geared towards activating T cells while minimising systemic toxicity. The company is continuing dose escalation in combination with Tevimbra and Avastin, as well as BGB-B2033 monotherapy in dose expansion.For its part, Astra recently doubled down on AZD7003, securing global rights from AbelZeta in preference to its own similarly designed AZD5851. AZD7003, whose 57% ORR reported at ASCO in 2024 marked the first real success for a cell therapy in solid tumours, uses dominant-negative TGFβ receptor armouring in addition to hitting GPC3; perhaps this is the feature that's allowed Astra to succeed where others have underwhelmed.

A newly unveiled ASCO abstract featuring BeOne’s BGB-B2033, a 4-1BB co-stimulated bispecific MAb against GPC3, has shown a meagre 20% confirmed response rate among 59 evaluable patients with GPC3-expressing pretreated liver cancer. This looks meaningfully worse than AbelZeta/AstraZeneca’s Car-T project AZD7003, whose 57% confirmed ORR in a similar setting started off the GPC3 craze.

True, it would be unrealistic to expect an antibody approach to match that of a Car-T therapy, and BeOne might argue that an off-the-shelf product without the need for apheresis and lymphodepletion could justify some falloff in efficacy. Whether an ORR number less than half its Car-T rival is acceptable, however, now becomes the big question.

First human data

The ASCO results for BGB-B2033, from a phase 1 trial that started two years ago, represent the first human data for this asset, and indeed they are one of the first human datasets for any GPC3-targeting project outside Car-T therapy. BeOne earlier touted the phase 1 trial as a prelude to starting a pivotal phase 2 liver cancer study later this year.

The 20% confirmed ORR comes from 12 partial responses among 59 evaluable patients, and the best BeOne can say about efficacy is that a preliminary dose-response relationship has been seen, and among 38 patients given more than the “predicted target efficacious dose” the ORR rose to 29%.

It’s not clear whether the company has scope to dose BGB-B2033 much higher, however. Though the project has been described as generally well tolerated, 8% of patients experienced treatment-related grade 3 or higher adverse events; there was one dose-limiting toxicity, liver enzyme elevation, that resolved after dose reduction.

Available data for GPC3-targeting projects in GPC3+ve relapsed liver cancer

Project	Mechanism	Company	Study	Data
AZD7003/ C-CAR031	dnTGFbRII-armoured CGPC3 Car-T	AstraZeneca/ AbelZeta	NCT05155189	ORR 57%
Ori-C101	GPC3 Car-T	OriCell	ChiCTR1900028121	ORR 44%
EU307	GPC3 Car-T	Eutilex	NCT05783570	ORR 38%
BGB-B2033	CPC3 x 4-1BB MAb	BeOne	NCT06427941	ORR 20% (29% at "efficacious" doses
CT011	GPC3 Car-T	Carsgen	NCT03980288	ORR 17%
ERY974	GPC3 x CD3 T-cell engager	Chugai (Roche)	NCT02748837	ORR 0%

Source: OncologyPipeline.

The lacklustre dataset might be seen as another setback for co-stimulation, where most recently Roche discontinued the anti-CD19 x 4-1BB project englumafusp alfa.

BeOne says the design of BGB-B2033, which also includes an engineered Fc region to prevent FcγR binding and increase half-life, is geared towards activating T cells while minimising systemic toxicity. The company is continuing dose escalation in combination with Tevimbra and Avastin, as well as BGB-B2033 monotherapy in dose expansion.

For its part, Astra recently doubled down on AZD7003, securing global rights from AbelZeta in preference to its own similarly designed AZD5851. AZD7003, whose 57% ORR reported at ASCO in 2024 marked the first real success for a cell therapy in solid tumours, uses dominant-negative TGFβ receptor armouring in addition to hitting GPC3; perhaps this is the feature that's allowed Astra to succeed where others have underwhelmed.