On 5 November 2026 the G-BA re-assessed zolbetuximab (Vyloy) in first-line gastric and gastro-oesophageal junction (GEJ) cancer and split the label in two. Patients with PD-L1 CPS below 1 got a hint of a minor added benefit. Patients with CPS of 1 or above got "not proven". It replaces the decision of 17 April 2025 and shows what happens to an orphan drug's assumed benefit once sales pass the threshold.

Why an orphan was assessed again

Vyloy is authorised as an orphan medicine, so under German law its added benefit counts as proven by authorisation. That changes when the product's turnover with statutory health insurance passes €30 million in twelve months. The G-BA then requires a full dossier against the appropriate comparator.

The G-BA wrote to the company on 12 February 2026, having found the limit exceeded between December 2024 and November 2025. The company submitted in May 2026, IQWiG published its assessment on 17 August, and the procedure closed on 5 November. The first procedure had ended on 17 April 2025 and is now repealed.

Two populations, two comparators

The G-BA set a different comparator for each PD-L1 group. For CPS of 1 or above it is nivolumab (only at CPS of 5 or above) or pembrolizumab, each with chemotherapy. For CPS below 1 it is a list of chemotherapy regimens, including oxaliplatin plus capecitabine.

The pivotal GLOW trial compared zolbetuximab plus CAPOX with placebo plus CAPOX, so it fits only the second group. IQWiG found no suitable study for CPS of 1 or above, and agreed with the company's own finding of none. With no data, the G-BA could not find a benefit there.

What sits behind the minor benefit

GLOW randomised 507 patients regardless of PD-L1 expression. The relevant subgroup is small: 84 patients on zolbetuximab and 88 on placebo. In that subgroup median overall survival was 15.5 months against 12.1 months, a hazard ratio of 0.68 (95% CI 0.47 to 0.99, p = 0.045), or 3.4 months. Progression-free survival, shown only as supplementary, did not reach significance (HR 0.686, 95% CI 0.460 to 1.021). The G-BA found no relevant differences in symptoms, quality of life or adverse events.

So the benefit rests on one endpoint in one subgroup, with an interval whose upper bound sits close to 1. IQWiG also lowered the certainty of the results:

  • Antiemesis was patchy. Only 77.4% of patients on zolbetuximab plus CAPOX and 67.0% on placebo plus CAPOX received any antiemetic co-treatment, although the product information calls for prophylaxis.
  • Censoring was high and unequal. IQWiG rated bias as high for overall survival because of the share of censored patients, mostly from withdrawn consent, and called it conceivable that poor antiemesis and toxicity contributed.
  • Palliative radiotherapy was restricted to peripheral bone metastases, which does not cover every use seen in care.

The G-BA still reached the rating shown on IQWiG's project page: a hint of minor added benefit for CPS below 1.

The wider evidence base

The GLOW publication reported a hazard ratio of 0.771 for overall survival in the full population, with median survival of 14.39 against 12.16 months. The PD-L1 analysis in that paper was ad hoc and ran on a subset with consented samples, and the text does not report CPS below 1 results. The German subgroup finding therefore stands on data outside the trial's primary design.

What it means for market access teams

  • Plan for the threshold. For an orphan, a €30 million trigger turns a launch-time presumption into a full assessment. Vyloy was first listed on 1 November 2024, the G-BA's request followed in February 2026 and the decision in November 2026. Dossier work should start before sales cross the line.
  • Pre-specify the subgroups the comparator will cut. The G-BA divided the population by PD-L1 and assigned a different comparator to each part. A trial that enrols all comers but was not stratified or powered by those cuts leaves one group with no data.
  • Treat supportive-care protocol choices as HTA evidence. Antiemetic use and radiotherapy rules in a protocol can become the reasons a payer discounts a result.
  • Expect the label to be wider than the benefit. The authorised indication covers both PD-L1 groups, but the assessed benefit covers one. That gap matters in price negotiation and in how sickness funds read the decision.

To discuss how a threshold re-assessment or subgroup split could affect your evidence plan, get in touch.