Recurrent Head and Neck Cancer: Pembrolizumab Alone or With Chemotherapy at CPS 12
A PD-L1 score that clears the label threshold for the gentler option, against a liver-metastasis burden that may not leave her the time a gentler option's response curve requires.
Denise A. ran a hair salon for close to thirty years before selling it just over a year ago, not long after her first diagnosis of HPV-negative oral cavity cancer, treated then with definitive chemoradiation. Eight months after finishing that treatment she is back, with new lesions in both lungs and liver found on a scan ordered for right upper quadrant pain she'd initially blamed on "eating too fast." The pain has been joined over the past three weeks by early satiety and a ten-pound weight loss she didn't intend, and her performance status has slipped from fully active to needing more rest through the day — not yet housebound, but noticeably slower than the woman her daughter describes from even a month ago. PD-L1 testing on the recurrent tissue returned a combined positive score of 12, positive by any threshold but below the higher cutoff where checkpoint monotherapy's own trial data showed its largest survival advantage.
That number is what the team keeps returning to, because it sits in a real gray zone rather than settling anything cleanly. KEYNOTE-048 showed pembrolizumab monotherapy improved overall survival against the prior standard specifically in patients with a combined positive score of at least 1 — she clears that bar — but the same trial's combination arm, pembrolizumab with platinum and 5-fluorouracil, improved survival across the entire population regardless of PD-L1 status, and did so with a faster, more reliable objective response than monotherapy tends to produce. For a patient with indolent, low-volume disease, that difference might not matter enough to justify the added chemotherapy toxicity. For a patient with symptomatic, growing liver metastases and a performance status already sliding, the gap between "a drug that works, eventually, for many patients" and "a drug combination more likely to shrink this specific tumor burden soon" is not an abstraction — it is the actual question in front of the team today.
First-line planning, recurrent disease
A combined positive score of 12 clears the label threshold for pembrolizumab monotherapy, and KEYNOTE-048 showed a real overall-survival benefit in the CPS-1-and-above population, not just at the higher cutoff. Monotherapy's toxicity profile is meaningfully better than adding chemotherapy — an important consideration given her performance status is already sliding, not stable.
I'd weigh her actual trajectory more heavily than the label threshold. She has symptomatic, growing liver metastases and a performance status moving the wrong direction over just the past month — that's a patient whose disease may not leave her the time a slower monotherapy response curve typically requires. The combination arm's faster, more reliable objective response is the more relevant comparison for her specifically, not the survival curve for the average CPS-positive patient.
I'm not disputing that CPS 12 clears the monotherapy threshold — I'm disputing that clearing a threshold is the same as it being the right choice for someone whose clinical picture is moving this fast.
I'd frame the disagreement slightly differently than either of you. The actual hazard ratios in KEYNOTE-048 show a real overall-survival benefit for monotherapy in the CPS-1-and-above group, but the combination arm's benefit doesn't depend on reaching a higher CPS cutoff the way monotherapy's strongest signal does — it holds across the whole trial population. That means the decision shouldn't really be framed as 'did she clear CPS 1,' it's 'which arm's response-rate and toxicity profile fits a patient whose disease is moving this quickly.'
Read that way, the disease-pace argument is the actual basis for combination therapy here — not that the biomarker permits either option, but that her clinical trajectory specifically favors the faster-acting one.
Agreed: pembrolizumab combined with carboplatin and 5-fluorouracil, with restaging imaging at six weeks rather than the standard nine, given how closely the team wants to track whether the faster response they're aiming for is actually materializing.
Not agreed: whether CPS should have anchored this decision at all. The medical oncologist would still favor monotherapy for a patient at the same CPS with a slower disease pace; the pharmacologist's reframing — that disease trajectory, not the biomarker threshold, should drive the choice — was accepted for today's plan but not adopted as a standing rule for every CPS-positive patient going forward.