PD-L1 × VEGF-A Target Pair Landscape — Axis Validated, Format Unapproved

1. Executive Snapshot

• The combination axis is clinically validated; the bispecific format is not. PD-L1 blockade + VEGF-A neutralization has an FDA-approved precedent — atezolizumab + bevacizumab, approved May 2020 for first-line unresectable HCC (IMbrave150) ([MedPage](https://www.medpagetoday.com/hematologyoncology/othercancers/86797)). No PD-L1×VEGF bispecific is approved anywhere: do not conflate the two.
• Lead asset: BNT327/PM8002 (pumitamig), an αPD-L1 × αVEGF-A bispecific originated by Biotheus, acquired by BioNTech in Nov 2024 ($800M upfront, up to $950M) and partnered with BMS in Jun 2025 for global co-development/co-commercialization (up to ~$11.1B). ROSETTA-Lung-02 Phase 2 interim (ASCO 2026): 1L NSCLC + chemo ORR ~70% (72.7% squamous), consistent across PD-L1 levels ([BioNTech PR](https://www.biontech.com/int/en/home/mediaroom/news/press-releases/2026/05/Global-Data-for-BioNTech-and-Bristol-Myers-Squibb-s-PD-L1xVEGF-A-Bispecific-Pumitamig-Shows-Encouraging-Efficacy-in-Patients-with-Non-Small-Cell-Lung-Cancer-in-ROSETTA-Lung-02-Trial.html)).
• Landscape: global Ph2/3 (BNT327); China Ph3 ×2 (HB0025); China Ph2/3 (IMM2510); Ph1 (HLX37); IND-stage adjacent variant (JSKN027, PD-L1×VEGFR2 bispecific ADC).
• Negative signals: Instil Bio terminated its ~$2B IMM2510/IMM27M license (Nov 2025); combination-class Ph3 failures (LEAP-002, COSMIC-312, IMmotion151) show dual-target biology does not guarantee outcomes.
• Verdict: 73/100 → Watch (65–79). Axis sound, deals active, but PD-1×VEGF (ivonescimab) holds first-mover mindshare; PD-L1×VEGF is a fast-follower with one large license termination on record.

2. Asset Landscape (Preclinical / Clinical / Approved / Discontinued)

• Approved: none in the bispecific format; the only approved reference is the *combination* atezolizumab + bevacizumab (2020, 1L HCC). Discontinued: IMM2510's US pathway via Instil Bio (2025-11).
• Naming verification: pumitamig = BNT327 = PM8002; palverafusp alfa = IMM2510 = AXN-2510.

3. Companies (Originators / Partners / Licensees)

• BioNTech SE — global rights holder for BNT327; leads the ROSETTA trials; co-develops with BMS ([PR](https://www.biontech.com/int/en/home/mediaroom/news/press-releases/2025/06/biontech-and-bristol-myers-squibb-announce-global-strategic.html)).
• BMS — global pumitamig partner since Jun 2025, co-development & co-commercialization ([BMS](https://news.bms.com/news/corporate-financial/2026/Global-Data-for-BioNTech-and-Bristol-Myers-Squibbs-PD-L1xVEGF-A-Bispecific-Pumitamig-Shows-Encouraging-Efficacy-in-Patients-with-Non-Small-Cell-Lung-Cancer-in-ROSETTA-Lung-02-Trial-2026-TW53rMYYjx/default.aspx)).
• Biotheus (CN) — PM8002 originator; acquired by BioNTech Nov 2024 ($800M upfront / up to $950M) ([Nasdaq](https://www.nasdaq.com/articles/biontech-acquire-biotheus-upfront-consideration-800m)).
• ImmuneOnco Biopharmaceuticals (HK 01541) — IMM2510; licensed to Instil Bio 2024 ($35M upfront, up to ~$2B); license terminated, rights returned 2025-11 ([FierceBiotech](https://www.fiercebiotech.com/biotech/instils-stock-sinks-after-clearing-out-clinical-pipeline-handing-2-drugs-back-immuneonco)).
• Instil Bio (US) — former licensee; exited the pipeline Nov 2025 (negative event).
• Huahai / Huaotaibio (600521) — HB0025, two Ph3 programs ([公告](http://m.epaper.zqrb.cn/html/2025-12/27/content_1208761.htm)); Henlius (02696) — HLX37 ([Henlius](https://www.henlius.com/en/NewsDetails-6049-26.html)); Alphamab (09966) — JSKN027 ([Alphamab](https://www.alphamabonc.com/en/html/news/2742.html)).
• Adjacent comparators (PD-1×VEGF, not GP-02): Akeso/Summit (ivonescimab); LaNova/Merck (LM-299).

4. Clinical & Deal Timeline

5. Risks & Failures

• IMM2510 license termination (largest negative in the pair): Instil Bio terminated its ImmuneOnco license on 2025-11-06 — a deal worth up to ~$2B, with $35M + $5M already paid — clearing its pipeline and returning both drugs. Reason reported: slow US clinical advancement (not a safety signal), exposing cross-border execution risk for smaller companies ([FierceBiotech](https://www.fiercebiotech.com/biotech/instils-stock-sinks-after-clearing-out-clinical-pipeline-handing-2-drugs-back-immuneonco)).
• Combination-class Ph3 failures (checkpoint × anti-angiogenic): LEAP-002 (pembro + lenvatinib, 1L uHCC) missed co-primary OS/PFS endpoints ([ASCO Post](https://ascopost.com/issues/november-10-2022/lenvatinib-plus-pembrolizumab-in-hepatocellular-carcinoma-leap-002-fails-to-meet-co-primary-endpoints/)); COSMIC-312 (cabo + atezo, 1L HCC) no OS benefit vs sorafenib ([OncLive](https://www.onclive.com/view/frontline-cabozantinib-atezolizumab-does-not-improve-os-over-sorafenib-in-advanced-hcc)); IMmotion151 (atezo + bev, 1L RCC) no final OS benefit ([NSTL/JCO](https://jx.nstl.gov.cn/paper_detail.html?id=42fdad731d3b9405cd3e07083f6aefa8)). Implication: the dual-target chemistry works (IMbrave150 positive), but partner choice, indication, dose and stratification decide outcomes — a bispecific does not immunize against these failure modes.
• Class toxicity: anti-VEGF AEs (hypertension, proteinuria, bleeding/thrombosis, GI perforation) plus anti-PD-L1 irAEs; a bispecific must manage additive two-arm toxicity, and no approved PD-L1×VEGF bispecific safety database exists.
• First-mover disadvantage: PD-1×VEGF leader ivonescimab (Summit) runs 10+ registrational Ph3 trials with a global head start; PD-L1×VEGF needs differentiating evidence (e.g., head-to-head vs the approved combination).
• Format & regulatory: no bispecific approval precedent for this pair; PK, immunogenicity, heterodimer manufacturing and regulatory paths are unproven; China-originated asset globalization depends on cross-border execution (Instil case).
• Incomplete disclosure: BNT327 key-trial OS/PFS and safety detail (hypertension/bleeding rates) not yet public; early ORR signals cannot be extrapolated to registrational outcomes.

6. Discovery Questions

• Can a PD-L1×VEGF bispecific beat atezolizumab + bevacizumab head-to-head (or an ivonescimab-class PD-1×VEGF)? No published head-to-head data exist.
• Optimal indication matrix — HCC/RCC (combo validated), NSCLC (large, crowded), TNBC/endometrial (differentiation), cervical (PD-L1+ enriched)?
• Chemotherapy/partner choice and dosing sequence (ROSETTA-Lung-02 uses + chemo)?
• Which design variables — valency (1:1 vs 2:2), Fc effector function, half-life, VEGF-A isoform selectivity (121/165) — drive efficacy/safety?
• Can the PD-L1×VEGFR2 ADC (JSKN027) open a window in PD-L1-low/cold tumors?
• Real PK/tissue-distribution/immunogenicity differences between bispecific vs dual-drug combination (bioanalytical evidence gap)?

7. Why This Matters Now

The axis is proven, deal flow heavy, and a global Ph3 is underway — yet no PD-L1×VEGF bispecific is approved, so the format race is being decided now. BNT327 is in global Ph3 (TNBC; NSCLC vs pembrolizumab) while Chinese assets (HB0025, IMM2510) advance to registration, raising immediate demand for bispecific characterization, free/total PK and ADA/NAb bioanalysis, and comparator testing. The Instil termination and combo-class failures are a reminder that "format" is not "validated outcome": the value sits in differentiating design and head-to-head evidence — exactly where a BsAb-focused platform earns its keep.

8. Sources

• Atezolizumab + bevacizumab 1L HCC approval (IMbrave150): https://www.medpagetoday.com/hematologyoncology/othercancers/86797 (A)
• BioNTech acquires Biotheus: https://www.nasdaq.com/articles/biontech-acquire-biotheus-upfront-consideration-800m; https://www.biospace.com/deals/biontech-takes-aim-at-keytruda-with-potential-950m-biotheus-buy (A)
• BMS × BioNTech collaboration: https://www.biopharminternational.com/view/biontech-to-receive-up-to-11-1-billion-in-deal-with-bms-to-develop-immuno-oncology-agent; https://www.fiercebiotech.com/biotech/bristol-myers-inks-11b-biontech-deal-join-bispecific-gold-rush-leaping-ahead-merck-and (B; upfront figures vary by source — C)
• ROSETTA-Lung-02 interim: https://www.biontech.com/int/en/home/mediaroom/news/press-releases/2026/05/Global-Data-for-BioNTech-and-Bristol-Myers-Squibb-s-PD-L1xVEGF-A-Bispecific-Pumitamig-Shows-Encouraging-Efficacy-in-Patients-with-Non-Small-Cell-Lung-Cancer-in-ROSETTA-Lung-02-Trial.html; https://www.clinicaltrialvanguard.com/news/pumitamig-shows-72-7-response-rate-in-phase-2-squamous-nsclc-trial/ (B)
• ROSETTA registrational trials: https://www.trialfetch.com/trial_match/view_trial/NCT07361510/; https://clinicaltrials.biontech.com/trials/BNT327-05 (A)
• PM8002 cervical/ovarian (ASCO 2024): https://ascopubs.org/doi/10.1200/JCO.2024.42.16_suppl.5524 (B)
• IMM2510 mechanism: https://academic.oup.com/abt/article/9/1/86/8426273 (A, peer-reviewed)
• Instil Bio termination: https://www.fiercebiotech.com/biotech/instils-stock-sinks-after-clearing-out-clinical-pipeline-handing-2-drugs-back-immuneonco; https://cn.immuneonco.com/news/gsdynamics/570.html (B)
• HB0025 (ASCO 2025 + Ph3): https://www.huahaipharm.com/mtbd/1096.html; https://ascopubs.org/doi/abs/10.1200/JCO.2025.43.16_suppl.5602; http://m.epaper.zqrb.cn/html/2025-12/27/content_1208761.htm; https://paper.cnstock.com/html/2026-03/19/content_2189693.htm (B/A)
• HLX37: https://www.henlius.com/en/NewsDetails-6049-26.html; https://www.henlius.com/en/NewsDetails-5690-26.html (A)
• JSKN027: https://www.alphamabonc.com/en/html/news/2742.html (A)
• LEAP-002: https://ascopost.com/issues/november-10-2022/lenvatinib-plus-pembrolizumab-in-hepatocellular-carcinoma-leap-002-fails-to-meet-co-primary-endpoints/ (A)
• COSMIC-312: https://www.onclive.com/view/frontline-cabozantinib-atezolizumab-does-not-improve-os-over-sorafenib-in-advanced-hcc (A)
• IMmotion151: https://jx.nstl.gov.cn/paper_detail.html?id=42fdad731d3b9405cd3e07083f6aefa8 (B)
• BNT327 preclinical (AACR 2025 #6061): https://aacrjournals.org/cancerres/article/85/8_Supplement_1/6061/758678/Abstract-6061-Dual-PD-L1-blockade-and-VEGF-A (A)
• IMM2510 ASCO 2026 poster: https://cn.immuneonco.com/science/strends/meeting/70.html (B)

Research tools this landscape implies: PD-L1/VEGF-A dual-binding kinetics (SPR/BLI) and simultaneous-binding sandwich ELISA; PD-1/VEGFR1-Fc/VEGFR2 competition-blockade assays; PD-1/PD-L1 NFAT-luc reporter and T-cell activation (MLR, IL-2/IFN-γ); HUVEC proliferation, pVEGFR2 and tube-formation assays; ADCC/CDC and FcRn binding; SEC/HIC heterodimer purity and DSF stability; VEGF-A isoform (121/165) binding panel; free-vs-total bispecific PK and ADA/NAb (cell-based neutralizing) bioanalysis; hPD-L1 knock-in syngeneic models with CD31/tumor-infiltrating lymphocyte IHC readouts.