Residual Triple-Negative Disease: A Regimen Combination Neither Trial Ever Tested
She has residual triple-negative disease after pembrolizumab-based neoadjuvant therapy — a population where CREATE-X supports adding capecitabine and KEYNOTE-522 supports continuing pembrolizumab. Nobody has ever tested giving her both agents together.
Camille D., a 39-year-old emergency dispatcher, took her diagnosis the way she takes a bad overnight shift — methodically, one call at a time, scheduling her chemotherapy infusions around her rotation rather than the other way around. Her triple-negative tumor was treated per KEYNOTE-522, pembrolizumab layered onto carboplatin, paclitaxel, and then anthracycline-cyclophosphamide, and her post-surgical pathology shows real but incomplete response: a residual 1.4cm invasive focus in the breast, with the nodes clear. That residual disease places her in exactly the population CREATE-X was designed for — and exactly the population where nobody has ever formally tested what she is now effectively being asked to receive: capecitabine added on top of continued adjuvant pembrolizumab.
CREATE-X randomized patients with residual invasive breast cancer after neoadjuvant chemotherapy to six to eight cycles of adjuvant capecitabine versus observation, and found a real disease-free and overall survival benefit concentrated in triple-negative disease specifically — a trial run before checkpoint inhibition was part of the standard regimen, so its population never received pembrolizumab at all. KEYNOTE-522's own protocol, on the other side, continues adjuvant pembrolizumab for nine cycles regardless of the surgical pathology result, including in patients with residual disease like Camille's, but that trial's protocol never added capecitabine to the adjuvant phase either. The two trials each answer their own question cleanly and completely; neither one says anything at all about combining both drugs at once, which is the actual decision in front of the team, not a extrapolation from either trial alone. Capecitabine's major toxicities — hand-foot syndrome, diarrhea, mucositis — don't obviously interact with pembrolizumab's immune-related adverse-event profile at a receptor or organ-system level, but the absence of a described interaction is not the same evidence as a trial that actually gave both drugs together and reported what happened.
Medical oncology, adjuvant planning
Her residual disease is textbook CREATE-X — triple-negative, invasive residual after neoadjuvant chemo. I want to add capecitabine. The survival benefit in that population is real and specific to exactly her situation; the fact that the trial predates checkpoint inhibition doesn't make her tumor biology different.
I don't disagree that CREATE-X's finding should still apply to her tumor biology — the concern isn't whether capecitabine works, it's that we have zero prospective safety data on giving it alongside nine cycles of ongoing pembrolizumab. Combining two active regimens each validated alone is not automatically equivalent to validating the combination.
That's a real gap, but it's a data gap, not a signal of actual harm — there's no described mechanistic interaction between a fluoropyrimidine and checkpoint blockade, unlike, say, combining two drugs that both suppress the same organ system. Absence of evidence here is closer to "untested" than "unsafe."
Both points are fair, and I don't think this resolves into a confident answer either way tonight — which is itself useful information for how we monitor her, not just what we prescribe. If we combine them, I'd want more frequent liver-function and symptom checks than either drug's label alone requires, specifically because we're watching for an interaction pattern neither trial was designed to detect.
Agreed: capecitabine added alongside continued pembrolizumab, with enhanced monitoring — liver function and symptom review at each cycle rather than the standard interval for either drug alone.
Explicitly not resolved, and named as such in the chart rather than left implicit: whether this combination carries any real added toxicity risk beyond either drug alone remains genuinely unknown, and the group agreed any new or unusual finding over the next several cycles should be treated as potentially interaction-related rather than attributed reflexively to whichever drug is more familiar.