Gastric Cancer Below the Claudin 18.2 Cutoff That Defined Its Own Drug
A biomarker that qualifies a tumor to be called positive by one reasonable definition and negative by the one the pivotal trial actually used, in a patient whose stomach is already struggling to do its job.
H.A., a 61-year-old woman, has run a neighborhood bakery for eighteen years and prided herself on never missing a morning bread delivery, until early satiety and repeated vomiting after meals forced her to hand the ovens to her nephew three weeks ago. Endoscopy found a large, partially obstructing antral mass, biopsy-confirmed adenocarcinoma, HER2-negative, with staging CT showing peritoneal disease — unresectable, metastatic gastric cancer. Claudin 18.2 immunohistochemistry, now routine at diagnosis for HER2-negative gastric and GEJ adenocarcinoma, returned moderate-to- strong membranous staining in 65% of tumor cells. She has lost eleven pounds in the past month and is tolerating only small volumes of liquids and soft food, a direct consequence of the tumor's location and size rather than the cancer's general burden.
Zolbetuximab, added to platinum-fluoropyrimidine chemotherapy, improved progression-free and overall survival in SPOTLIGHT and GLOW, but both trials defined Claudin 18.2 positivity as moderate-to-strong staining in at least 75% of tumor cells — a specific, pre-registered threshold, not a looser "detectable expression" standard. H.A.'s 65% falls short of that cutoff, technically making her Claudin 18.2 expression real but below the population the drug was actually tested and approved in. Compounding the question is zolbetuximab's own toxicity profile: because Claudin 18.2 is also expressed on differentiated gastric mucosal cells, the drug causes substantial on-target nausea and vomiting in the trials that established its benefit — a toxicity that would compound, not just add to, the oral intake problem her own obstructing tumor is already causing. Eleven pounds down in a month, tolerating only liquids, and already hypokalemic from vomiting, H.A. would be absorbing that on-target emesis on top of a stomach that cannot empty — the trial populations were required to be ECOG 0 or 1, and her ECOG 2 is nutritional, which means the drug's defining toxicity acts on the very deficit that disqualifies her from the trials that measured it.
First-line treatment planning, after biomarker results returned
SPOTLIGHT (Shitara et al., Lancet 2023) and GLOW (Shah et al., Nature Medicine 2023) both defined Claudin 18.2 positivity as moderate-to-strong staining in at least 75% of tumor cells, and that's the population where the survival benefit was actually shown. H.A.'s result is 65%. I'd treat with chemotherapy alone — the threshold isn't arbitrary paperwork, it's where the evidence base for this drug actually starts.
I understand leading with the trial's own cutoff, and I wouldn't call 65% strongly positive by the trial's definition either. But thresholds like this are usually chosen for enrollment efficiency and statistical power in the pivotal population, not because the underlying biology switches off cleanly at that exact number. Sixty-five percent moderate-to-strong staining is substantial target expression, not a trace finding — I don't think we should treat 75% as a biological law just because it's where the trial happened to draw its enrollment line.
I'll concede directly: no trial has actually shown the drug works below 75%. I'm arguing this is worth an honest conversation with her, not that the data settle it either way.
Whatever we decide about the biomarker, I don't think either chemotherapy regimen is safely deliverable today. She's vomiting from mechanical obstruction and down eleven pounds in a month. Zolbetuximab's defining toxicity in both trials was nausea and vomiting from on-target expression in normal gastric mucosa — adding that to a partially obstructed stomach is a real, specific risk neither trial population had to face in the same way. I'd address the obstruction and her nutritional status first, and revisit the biomarker threshold question once she can actually tolerate treatment.
Agreed: obstruction managed with gastrojejunostomy this week, followed by chemotherapy without zolbetuximab once she is tolerating an adequate oral or enteral intake, consistent with her Claudin 18.2 result falling outside the pivotal trials' defined population.
Not settled as a closed question: the pharmacologist's argument that 65% expression may carry some real, untested biological activity was recorded explicitly rather than dismissed, and the team agreed to revisit zolbetuximab eligibility if a future line of therapy is needed and either her obstruction has fully resolved or updated trial data address the subthreshold population directly.