GSK B7-H3 ADC ris-rez first global data: the 61% ORR at 8 mg/kg is 11 of 18 patients, and GU should watch the mCRPC Phase 3

Data as of 2026-10-21 (Taipei time). [Fact] statements are sourced; [Inference] statements are this article's judgment.

1. One-line conclusion

[Fact] At 8 mg/kg the confirmed ORR was 61% (11/18; 95% CI 35.7–82.7) and grade ≥3 TRAEs 42%; at 6 mg/kg the figures were 32% (7/22) and 15% (GSK). [Inference] ris-rez reproduced the direction of the Chinese data in global patients, but the sample is small and the intervals wide; choosing 8 mg/kg for Phase 3 trades higher toxicity for a higher response rate. Whether that trade-off carries over to mCRPC depends on marrow tolerance in older patients.

2. What happened

[Fact] On 2026-10-18 GSK reported data from the Phase Ib dose-optimization part of EMBOLD PanTumour-101 (NCT06551142), to be presented as a poster at ESMO 2026 (GSK).

• Design: global, open-label Phase Ia/b; in Phase Ib, patients with relapsed (second line or later) ES-SCLC were randomized to 6 or 8 mg/kg IV every 3 weeks (GSK).
• Population: 50 patients analyzed, median 2 prior lines, including patients previously treated with a topoisomerase-1 inhibitor or tarlatamab and patients with brain or liver metastases (GSK).
• Efficacy (confirmed ORR, evaluable patients): 61% at 8 mg/kg (11/18; 95% CI 35.7–82.7); 32% at 6 mg/kg (7/22; 95% CI 13.9–54.9) (GSK).
• Safety: median follow-up 100.5 days; grade ≥3 TRAEs 42% at 8 mg/kg and 15% at 6 mg/kg, mostly hematologic; overall ILD incidence 2% (1/50), grade 2 (GSK).
• Dose decision: these data, together with Hansoh's Chinese Phase 3 ARTEMIS-008, support taking 8 mg/kg into the global Phase 3 EMBOLD SCLC-301 (GSK).
• This is a two-dose randomization, not a controlled comparison; there is no HR or p value. DoR, PFS: not found; GSK says longer follow-up is needed (GSK).

Arithmetic check

• 11/18 = 61.1% and 7/22 = 31.8%, consistent with the reported 61% and 32%.
• 18 + 22 = 40, 10 fewer than the 50 analyzed. [Inference] The gap is likely "evaluable" versus "analyzed"; GSK also notes enrollment is ongoing (GSK). Which arms the 10 excluded patients belong to: not found.
• No denominators were given for the 42% and 15% grade ≥3 TRAE rates, so patient counts cannot be derived. The ILD denominator is all 50 patients.

3. Why it matters

• [Inference] Clear dose response, thin evidence: the two CIs overlap (35.7–82.7 vs 13.9–54.9), with only 40 patients and about 3 months of follow-up; 61% will likely be revised down with longer follow-up and more patients.
• [Inference] The toxicity cost: grade ≥3 TRAEs rise from 15% to 42%, mainly hematologic. ES-SCLC patients have poor performance status, so dose interruption and reduction rates in Phase 3 will be critical.
• Low ILD signal: 1/50, grade 2, is good news for a topo-1-payload ADC, but follow-up is only about 100 days. [Fact] In a pooled analysis of nine T-DXd trials, the median time to adjudicated drug-related ILD was 5.4 months after the first dose (ESMO Open). [Inference] ris-rez's ILD risk needs longer follow-up to judge.
• Endorsement from the Chinese Phase 3: [Fact] In Hansoh's Chinese Phase 3 ARTEMIS-008 (relapsed SCLC, 1:1 randomized, open-label), ris-rez 8.0 mg/kg versus topotecan gave an OS HR of 0.46 (95% CI 0.35–0.62; p<0.0001), with median OS 18.5 months (n=230) versus 10.3 months (n=231) (GSK China Phase 3). Cross-trial caveat: this is a single-region Chinese randomized Phase 3 with a topotecan control and both limited- and extensive-stage patients; the global Phase Ib is an uncontrolled two-dose comparison, so the figures cannot be compared directly. [Inference] The point of the global Phase Ib is to show that non-Chinese patients respond similarly, addressing FDA's concerns about single-region data.
• A Project Optimus example: [Inference] GSK's randomized two-dose selection of the RP3D is in line with FDA's Project Optimus dose-optimization approach (FDA).

4. Regulatory and commercial implications

• FDA: [Fact] ris-rez holds US Breakthrough Therapy designation (relapsed or refractory ES-SCLC) and orphan drug designation (GSK). [Fact] On 2022-02-10 FDA's Oncologic Drugs Advisory Committee (ODAC) voted 14:1 that sintilimab's China-only Phase 3 ORIENT-11 required an additional trial applicable to US patients, and FDA stated that multiregional trials are the preferred approach (OncLive, The Cancer Letter). [Inference] GSK may seek accelerated approval with global Phase Ib ORR plus Chinese randomized Phase 3 OS, but given the sintilimab precedent, the global EMBOLD SCLC-301 results are likely to remain decisive.
• NMPA: China rights sit with Hansoh (GSK); Hansoh's China filing status was not checked here.
• Competition: [Inference] Relapsed SCLC already has tarlatamab, and there are other B7-H3 ADCs (e.g., Merck/Daiichi Sankyo's I-DXd); Fierce quotes Jefferies saying ris-rez is "starting to look like the B7-H3 ADC to beat" (Fierce).
• GU: [Fact] GSK's EMBOLD program includes an mCRPC Phase 3 and plans another Phase 3 in chemotherapy-naive mCRPC (GSK). mCRPC Phase 3 design, comparator and dose: not found.
• For the next trial: [Inference] sponsors planning mCRPC ADC trials should settle their comparator and dose strategy before GSK discloses its mCRPC Phase 3 design, to avoid competing with ris-rez for the same later-line patients.

5. Lessons for the next trial design and OncoLattice recommendations

All recommendations below are [Inference]; cited facts carry their own sources.

1. Ask the GU dose question first: if the mCRPC Phase 3 also uses 8 mg/kg, older mCRPC patients with extensive bone metastases and poor marrow reserve may not tolerate 42% grade ≥3 toxicity. Sponsors should run their own randomized dose optimization in mCRPC rather than carrying over the SCLC RP3D.
2. Write hematologic toxicity management into the protocol: prespecify G-CSF rules, transfusion thresholds and dose-reduction steps.
3. ILD monitoring: baseline and periodic chest CT, an adjudication committee, and at least one year of follow-up (in the T-DXd pool, 87.0% of first ILD events occurred within 12 months, ESMO Open).
4. Site selection: mCRPC ADC trials need sites that can monitor marrow toxicity and work with pulmonology on ILD; urology-led sites without medical oncology support carry high risk.
5. Competition for patients: later-line mCRPC already has Lu-PSMA, Ac-225 PSMA, PSMA TCEs and multiple ADCs; assess the number of concurrent competing trials at each site during selection.
6. B7-H3 expression: GSK has not said whether patients were selected by B7-H3 expression; GU trials should at least collect tissue for retrospective analysis.

6. Unknowns and what to watch

• DoR, PFS and per-arm dose modification rates in the ESMO poster.
• Which arms the 10 patients excluded from the efficacy analysis belong to.
• Design details of EMBOLD SCLC-301 and the mCRPC Phase 3.
• ILD incidence with longer follow-up.

Evidence limitations

The efficacy denominator is only 40 patients, median follow-up about 3 months, and the CIs are wide. Although the two dose arms were randomized, the sample is too small for formal comparison. Chinese Phase 3 figures come from GSK's release and Fierce; the full text was not read. The GU section is inference; no ris-rez mCRPC data were found.

Endnote: differences from the original forwarded account

the forwarded account placed "n=50" next to the two ORRs, implying an ORR denominator of 50; the actual efficacy denominator is only 40 evaluable patients. Other figures match GSK's original.