favezelimab/pembrolizumab

Merck is testing whether stacking a LAG-3 antibody on top of Keytruda extends the franchise into tumors where PD-1 monotherapy stalls. MK-4280A puts favezelimab (an investigational anti-LAG-3 antibody) and pembrolizumab into a single fixed-dose vial, currently in Phase 2 as the KEYFORM-010 basket study (NCT06036836, n≈160) across two indications: neoadjuvant resectable cutaneous squamous cell carcinoma with pathologic complete response as endpoint, and previously-treated mismatch-repair-proficient (pMMR) advanced endometrial cancer with a lenvatinib backbone and ORR as endpoint [1]. The strategic context matters. Merck's Phase 3 KEYFORM-007 trial of the same combo (without lenvatinib) in PD-L1+ second-line colorectal cancer (NCT05064059) missed its primary overall survival endpoint, which closed the CRC indication [4]. KEYFORM-010 is the program's second attempt to find a tumor type where LAG-3 inhibition adds meaningfully to pembrolizumab, with Keytruda's biosimilar cliff in 2028 putting pressure on Merck to extract value from every defensible add-on mechanism. Bristol Myers Squibb's Opdualag (relatlimab + nivolumab) is the only approved LAG-3 inhibitor, and only in melanoma. Whether MK-4280A can do better in endometrial or cSCC is genuinely uncertain. After correcting the structured factors (validated target rather than first-in-class; prior Phase 3 miss as negative signal), our PoS estimate lands at ~32% - slightly above the Phase 2 oncology base rate.

MK-4280A is a fixed-dose coformulation, not a novel single agent. Pembrolizumab needs no introduction - Keytruda generated roughly $29.5B in 2024, around 45% of Merck's total revenue [2]. Favezelimab is the unknown half: a Merck-developed anti-LAG-3 antibody, first dosed in humans in a Phase 1 study in microsatellite-stable colorectal cancer that showed modest activity (ORR ~6.3% in the favezelimab/pembrolizumab arm) [3]. The coformulation makes commercial sense if it works: one infusion, simpler logistics, easier payer story than two separately priced biologics. The focal trial is KEYFORM-010 (NCT06036836), a Phase 2 basket study with two parallel cohorts in neoadjuvant cSCC and previously-treated pMMR endometrial cancer; estimated enrollment ~160, with the endometrial arm ~80 patients (40 per arm) [1]. No breakthrough or fast track designations have been granted, and Merck has not asked for them, which itself is a tell about internal conviction. The earlier Phase 3 KEYFORM-007 in PD-L1+ second-line metastatic colorectal cancer (n=441) reported its OS readout and did not meet its primary endpoint [4], effectively shutting down the CRC indication for the combo. Beyond KEYFORM-007 and KEYFORM-010, the favezelimab development program has substantially narrowed since the CRC failure; Merck has not announced new Phase 3 starts. Expected readouts for the Phase 2 endometrial and cSCC cohorts are likely 2026-2027, depending on event accrual. There is no Phase 3 in either indication yet, so even in a best-case scenario commercial launch is no earlier than 2029.

T cells have multiple brake pedals that prevent them from attacking healthy tissue. Pembrolizumab releases one brake (PD-1, programmed death-1). Favezelimab releases another, called LAG-3 (lymphocyte activation gene 3), an inhibitory receptor that ramps up on exhausted T cells inside tumors and dampens their killing function when engaged by MHC class II (a protein displayed on antigen-presenting cells that normally helps T cells recognize threats) or FGL1 (a liver-secreted protein that also suppresses T cells via LAG-3) [5]. The hypothesis is straightforward: some tumors evade PD-1 blockade by leaning on LAG-3 as a backup off-switch, so releasing both brakes should generate immune responses where releasing just one fails. The mechanism has one major proof point and several question marks. Bristol Myers Squibb's relatlimab + nivolumab (Opdualag) became the first approved LAG-3 inhibitor in 2022 for first-line metastatic melanoma, after the RELATIVITY-047 trial showed median PFS of 10.1 vs 4.6 months versus nivolumab alone [6]. That is real validation, but the absolute benefit is modest and the OS advantage was less convincing. LAG-3 inhibition has yet to prove itself outside melanoma. Why pick endometrial and cSCC after the CRC miss? The biological rationale is thin but defensible. Endometrial tumors - especially pMMR - have measurable LAG-3 expression on TILs and represent a setting where pembrolizumab alone is weak, so the marginal-additivity bar is low. cSCC has a very high UV-driven mutational burden and is highly immunogenic, with PD-1 response already strong; the bet is that combining LAG-3 blockade in a neoadjuvant window may deepen pathologic responses through enhanced TIL activation. Honest read: Merck has not published a strong translational dataset arguing specifically for these indications. The selection looks more pragmatic (tumors where Keytruda is approved or active, with manageable trial logistics) than translationally rigorous. The biology says maybe. The clinical data so far says in melanoma yes, in CRC no.

KEYFORM-010 / NCT06036836 is a Merck-sponsored Phase 2 basket study with two parallel cohorts [1]. Cohort A enrolls patients with resectable cutaneous squamous cell carcinoma in a neoadjuvant setting; the primary endpoint is pathologic complete response (pCR) - no residual tumor cells in the surgical specimen - comparing MK-4280A against pembrolizumab monotherapy. Cohort B enrolls previously-treated pMMR advanced endometrial cancer patients on a lenvatinib backbone, randomizing 1:1 (≈40 per arm, ~80 total) between MK-4280A + lenvatinib and pembrolizumab + lenvatinib, with overall response rate as the primary endpoint. Total enrollment is approximately 160 across both cohorts. Note this corrects a common misread: the pCR endpoint sits on the cSCC arm, not endometrial. The cSCC neoadjuvant design is the more scientifically clean arm. pCR is an objective readout with paired pre- and post-treatment tissue, ideal for biomarker work in a defined timeframe. But the comparator pressure is high: cemiplimab and pembrolizumab already produce roughly 40-50% ORR in advanced cSCC as monotherapy, and neoadjuvant cemiplimab has reported pCR rates in the 50% range. MK-4280A needs to beat those numbers meaningfully - a steep bar in a tumor where PD-1 alone works well. The endometrial arm asks a different question: does LAG-3 inhibition add to PD-1 in pMMR disease, where pembrolizumab monotherapy is underwhelming and the standard 2L+ regimen is pembrolizumab + lenvatinib (KEYNOTE-775)? The use of a lenvatinib backbone means MK-4280A is being tested as an additive over an already-established doublet, raising the bar for incremental benefit. MSI-H/dMMR endometrial tumors already respond well to Keytruda monotherapy, so the entire commercial rationale here lives or dies on pMMR signal. Competing in this race: Regeneron's fianlimab/cemiplimab LAG-3+PD-1 combo, which is in Phase 3 in melanoma but has not yet entered endometrial or cSCC - a watch item but not a near-term threat to MK-4280A in these specific indications.

Our model estimates a 7% chance this drug is eventually approved. That starting point comes from the historical approval rate for Phase 2 drugs in this area, which runs around 13%. The estimate is pulled up by the sponsor's strong track record, but pulled down by weak earlier-phase results, heavier-than-usual blinding, and a randomized trial design. The remaining factors are close to average for this stage, so they leave the final number near where it started.

The biggest efficacy risk is mechanism-additivity. LAG-3 inhibition has not produced clear wins outside melanoma, and KEYFORM-007 failed in colorectal cancer. The honest read is that LAG-3 may be a weaker checkpoint than PD-1, contributing meaningfully only when the tumor immunology happens to align (as in cutaneous melanoma). Endometrial and cSCC are not obviously the right next targets, and Merck has not published a strong translational case for these specific indications. Safety risk is manageable but real. Combining checkpoint inhibitors raises immune-related adverse event rates. Opdualag's safety profile is acceptable but not clean, with grade 3+ irAEs running higher than nivolumab alone. The fixed-dose coformulation creates a small wrinkle: you cannot separately hold favezelimab if a patient develops what looks like a LAG-3-specific adverse event, though clinically irAEs are rarely target-specific. Execution risk is low. Merck runs hundreds of pembrolizumab combination trials and the enrollment infrastructure is mature. Competitive risk is moderate. Regeneron's fianlimab/cemiplimab (Phase 3 in melanoma) is the most credible PD-1/LAG-3 challenger; if that combo reads out positively in melanoma and Regeneron moves quickly into cSCC (where cemiplimab is already first-line standard), MK-4280A could face a faster competitor in the indication where Merck's own data is being generated. Commercial risk is the sleeper. Even if MK-4280A hits its primary endpoints, the payer case is harder than it looks. Pembrolizumab is the comparator in both arms and the more expensive component of the combo. To justify MK-4280A over Keytruda alone (cSCC) or Keytruda+lenvatinib (endometrial), the absolute benefit needs to be substantial. Market sizing for context: the US advanced/recurrent endometrial cancer population is roughly 8,000-12,000 patients per year (a rough majority pMMR based on registry data showing ~28% pMMR among MMR-tested cases, with the true fraction likely higher given the larger pMMR base in unselected populations) [8]; advanced/metastatic cSCC eligible for systemic therapy is roughly 15,000-25,000 patients per year (derived from ~700K-1.3M annual US cSCC cases, ~2-5% metastasis rate) [9]. Peak sales potential is modest in both indications relative to existing Keytruda revenue - likely $500M-$1.5B combined at best-case penetration, well short of the billions needed to meaningfully offset biosimilar erosion. These are rough estimates and should be treated as directional only.

Worth a low-priority watch, not a high-conviction signal. Merck has every reason to keep developing this. They need to extend Keytruda's franchise before biosimilars arrive in 2028, and a fixed-dose combo with a second mechanism is one of the cleaner ways to do that. But the program has already shown it can miss in Phase 3 (KEYFORM-007), and LAG-3 outside melanoma is unproven. The specific data points worth watching: (1) pCR rate in the cSCC neoadjuvant cohort relative to the pembrolizumab comparator, and (2) ORR delta in the pMMR endometrial cohort over pembro+lenvatinib. If MK-4280A produces a clearly higher pCR rate than pembrolizumab alone in cSCC, or a meaningful ORR lift over the already-approved pembro+lenvatinib doublet in pMMR endometrial cancer, those are real signals that LAG-3 inhibition can extend PD-1 benefit beyond melanoma. Anything restricted to MSI-H/dMMR endometrial is uninteresting because Keytruda already works there. When to check back: ESMO Congress 2026 in October or SITC December 2026 for the first cohort readouts. If Merck does not announce a Phase 3 plan within six months of those data, the program is probably winding down quietly. Merck's $65B in annual revenue means they can afford to keep multiple second-life programs running, but this one needs to show something specific to justify a Phase 3 commitment. Given the modest addressable populations (~10K pMMR endometrial + ~20K advanced cSCC annually in the US) and the high comparator bar in both, even a clinical win is unlikely to meaningfully blunt the biosimilar cliff.