ESMO 2026 Opens in Madrid: Five Chinese Phase III Trials Take the Presidential Stage, Four ADCs and a Bispecific Face Global Scrutiny

ESMO 2026 runs 23-27 October in Madrid with roughly 30 Chinese studies among the late-breaking abstracts - about one third of the total and a record high. China holds five of the twelve Presidential Symposium slots, all Phase III: HutchMed's SAFFRON and Akeso's HARMONi-GI1 both hit PFS and OS, Hansoh's B7-H3 ADC ARTEMIS-011 delivered the first positive Phase III in osteosarcoma in a decade, Kelun-Biotech's TROP2 ADC TroFuse-005 met dual primary endpoints, while KeyMed/Lepu's CLDN18.2 ADC met OS but missed PFS.

The 2026 Congress of the European Society for Medical Oncology (ESMO) runs from 23 to 27 October in Madrid. Regular abstracts went live on 19 October, with late-breaking abstracts (LBAs) read out during the meeting. Of the roughly 90 LBAs, about 29-32 come from China - the difference in broker counts reflecting whether Chinese molecules led by overseas partners are included - representing about one third of the total and a record high.

Position matters more than volume. The Presidential Symposium, the congress's highest-profile session, has twelve slots globally; China holds five, all randomised Phase III trials. The LBA bar is high by design: only randomised Phase II or Phase III studies whose results could change clinical practice or reshape how a disease is treated qualify. Most of the roughly 30 Chinese LBAs are therefore registration-grade data, not early proof of concept.

The five Presidential studies: four ADCs, one bispecific, one targeted combination

SAFFRON (LBA6, HutchMed/AstraZeneca): savolitinib plus osimertinib in advanced EGFR-mutant, MET-overexpressed or amplified NSCLC after progression on osimertinib. Both PFS and OS were positive - the first Phase III evidence that a MET-TKI plus third-generation EGFR-TKI combination improves both endpoints in a resistant population.

CLARITY-Gastric01 (LBA7, KeyMed Biosciences/Lepu Biopharma, licensed to AstraZeneca): the CLDN18.2 ADC CMG901/AZD0901 in previously treated advanced gastroesophageal cancer. The OS co-primary endpoint was met at third-line and beyond, but overall PFS did not reach statistical significance - the only split read among the five. For an ADC, an OS win carries more clinical weight than PFS (regulators view OS as harder to flatter through trial design), yet a PFS miss will be read by overseas partners as a signal of insufficient efficacy magnitude.

HARMONi-GI1 (LBA8, Akeso): ivonescimab plus chemotherapy versus durvalumab plus chemotherapy in first-line advanced biliary tract cancer. Both OS and PFS were positive - another indication expansion for the PD-1/VEGF bispecific and key evidence for pushing IO 2.0 from lung cancer into cold tumours.

ARTEMIS-011 (LBA9, Hansoh Pharma, licensed to GSK): the B7-H3 ADC ris-Rez/HS-20093 in heavily pretreated relapsed or refractory osteosarcoma met its PFS primary endpoint - the first positive Phase III in this tumour type in about a decade. Osteosarcoma has long lacked effective systemic therapy; the clinical significance does not map one-to-one onto commercial scale, but it establishes regulatory-grade evidence for B7-H3 outside its core solid-tumour indications.

TroFuse-005 (LBA11, Kelun-Biotech/Merck): the TROP2 ADC sac-TMT in previously treated advanced or recurrent endometrial cancer met both OS and PFS primary endpoints. This is the second registration-grade readout for sac-TMT in gynaecological oncology and the most direct valuation test for the asset since Merck licensed it.

Four of the five are ADCs or bispecifics, one is a targeted combination. The collective signal: the evidence tier for Chinese assets has graduated from early proof of concept to randomised, practice-changing confirmation.

Why these five days matter more than an earnings season

Over the past six months the way Chinese assets are priced has changed. Roche bought Hansoh's ADC platform, Merck bought Kelun-Biotech's TROP2 asset, GSK bought Hansoh's B7-H3, Lilly bought JITai's delivery platform - all paid for platforms rather than single molecules. The pricing anchor moved from what this molecule is worth to what this platform can keep producing.

Whether a platform can keep producing can only be proven by successive Phase III readouts. That is why this ESMO is treated in investment circles as a collective defence of Chinese innovation: if individual studies disappoint, the damage is not confined to one company's single asset but resets the pricing coefficient for an entire platform. Conversely, the double-positive SAFFRON and HARMONi-GI1 raise the negotiating baseline for future out-licensing of comparable platforms.

Capital flows corroborate the shift. China's National Development and Reform Commission said at its October press conference that outbound technology licensing reached about USD 130B in the first three quarters, close to all of last year. PharmaCube counted 209 deals worth over USD 140B in total with more than USD 8B in upfronts as of 17 October, while Changjiang Securities put Q3 outbound BD upfronts at USD 2.3B (+27.0%) and milestones at USD 26.6B (+6.3%).

Three transmission channels into the CXDMO industry

First, funnel expansion. A Phase III-positive molecule entering global registrational trials and commercial supply planning releases process development, scale-up, stability and GMP batch orders over the following 12-24 months. ADCs are especially heavy: high-potency capacity for toxins and linkers, batch consistency through conjugation scale-up, and cross-border supply chain management for segregated DS/DP manufacturing are all high-value CXDMO services.

Second, capacity attribution. Four of the five Presidential assets have ex-China rights licensed to AZ, GSK and MSD, and the geographic location of commercial manufacturing becomes part of the negotiating terms. Once Phase III data validate the China-molecule, global-development path, production allocation will be constrained simultaneously by clinical data, inspection records and geopolitical policy. That is why WuXi Biologics' no-notice FDA inspection at Wuxi Mashan with zero major observations on 20 October carried extra weight in timing.

Third, CMC execution has become the condition for milestone realisation. Out-licensing milestones are tied to clinical nodes, approvals and first commercial sales - all dependent on the certainty of process scale-up and batch delivery. For a CDMO, whether it can deliver inside a client's externally committed window has moved from a service-quality issue to a contractual risk issue. Novo Nordisk's 18 October decision to source peptide APIs externally and the 19 October extension of the Samsung Biologics injunction to nine core manufacturing stages are, at root, the same topic.

Two background variables deserve to be read alongside the congress. On 16 October, MOFCOM and MOST published Announcement No. 48 proposing five biopharma technologies for the export control catalogue while explicitly excluding licence-outs of self-owned IP, outsourcing services and outbound licensing - the practical effect is boundary clarification rather than tightening. And on 22 October, A-share and H-share results season brought Hengrui, BeiGene and Innovent into the same week as the academic readouts.

How to follow the next five days: first, the complete OS curves and safety data for the five Presidential studies, particularly how CLARITY-Gastric01's PFS miss is explained; second, whether any of the other 22-24 Chinese LBAs produce an unexpected indication or combination; third, whether follow-on out-licensing announcements appear within a week of the meeting. Historically, the four to eight weeks after positive ESMO data is a high-density window for BD announcements.