HER2-Positive Gastric Cancer: A Triplet Regimen Narrowed by Its Own Subgroup Data
The trial that added a third drug to the HER2-positive standard also found that third drug's benefit depended on a biomarker this patient doesn't have.
G.S., a 68-year-old man, retired from the postal service four years ago and has spent most of that time restoring an old wooden sailboat in his driveway, work he's had to slow down considerably since developing progressive difficulty swallowing solid food over the past two months. Endoscopy found a gastroesophageal junction mass, biopsy-confirmed adenocarcinoma, with staging CT showing liver metastases — unresectable, metastatic disease from the start. Tumor testing returned HER2 IHC 3+, strongly positive, and PD-L1 combined positive score of 0, tested by the same immunohistochemistry panel used to make first-line treatment decisions in this disease.
Trastuzumab added to platinum-fluoropyrimidine chemotherapy has been the first-line standard for HER2-positive metastatic gastric and GEJ adenocarcinoma since ToGA established a real overall survival benefit for that combination. KEYNOTE-811 tested adding pembrolizumab as a third agent on top of that standard and found an improved response rate and progression-free survival overall — but the benefit was concentrated in patients with PD-L1 CPS of 1 or higher, and the regulatory approval for the triplet was subsequently narrowed specifically to that CPS-positive population after interim data suggested the benefit did not clearly extend to CPS-negative tumors like G.S.'s. That narrowing is itself the finding driving today's decision: the newer, more aggressive-sounding three-drug regimen is not simply "more treatment" for every HER2-positive patient — for a PD-L1-negative tumor specifically, it is a regimen whose own evidence base argues against adding it rather than for it. A CPS of 0 is not a borderline result that might be read either way; it is the floor of the scale, as far from the approval's threshold as the assay can place him.
First treatment planning visit after staging
I'd treat with trastuzumab and chemotherapy, without pembrolizumab. KEYNOTE-811 (Janjigian et al., Lancet 2023) is the trial behind the triplet, and its own regulatory approval was specifically narrowed to PD-L1 CPS of 1 or higher after the data showed the benefit wasn't clearly present in CPS-negative tumors. G.S.'s CPS is 0. Adding a drug outside the population its own evidence supports isn't a cautious extra — it's treatment without the specific justification we'd need to expose him to added toxicity.
I take the approval scope seriously, and I'm not going to argue we should override a regulatory decision made on the trial's own data. But the overall intention-to-treat population in KEYNOTE-811 did show benefit, and pembrolizumab's toxicity profile is generally manageable for most patients. I want it on the record that "the subgroup wasn't statistically significant" isn't the same claim as "there is definitely no benefit for this individual patient" — it's a real, important distinction even if it doesn't change what I'd actually recommend today.
You're right that the regulatory narrowing reflects the specific data, and I'm not disputing that it should be followed here — I just don't want the reasoning collapsed into "proven not to work," which the subgroup analysis didn't actually establish either.
Worth naming for the record: PD-L1 CPS scoring does carry real interobserver and sampling variability, and a single biopsy result is a snapshot, not a certainty. That said, I don't think that uncertainty is large enough, or clinically actionable enough today, to justify a repeat biopsy purely to re-test CPS. I'd follow the approved indication as it stands and revisit only if a future biopsy, done for an independent clinical reason, happens to show something different.
Agreed: trastuzumab plus platinum-fluoropyrimidine chemotherapy as first-line therapy, following the original ToGA regimen without pembrolizumab, consistent with the CPS- scoped approval for the triplet.
The pharmacologist's caveat — that a non-significant subgroup result is not the same as proven absence of benefit — was recorded in the chart as a genuine, unresolved epistemic point, not treated as grounds to add the drug outside its evidence-supported population. If a future biopsy, obtained for an independent clinical reason, shows a different PD-L1 result, the pembrolizumab question would be reopened at that time.