Pembrolizumab + BCG

KEYNOTE-676 tests whether adding pembrolizumab (Keytruda) to intravesical BCG beats BCG alone in high-risk non-muscle invasive bladder cancer (HR NMIBC). Merck wants to move its PD-1 blockbuster earlier in the bladder cancer treatment sequence, into a population currently treated with a decades-old attenuated TB vaccine (BCG) and little else [1]. The company already holds a pembrolizumab approval for BCG-unresponsive HR NMIBC with carcinoma in situ (KEYNOTE-057, 2020), but that is a small salvage-setting market [2][3]. HR NMIBC overall is a much larger population, roughly 60,000 new U.S. cases annually. Current first-line HR NMIBC spend in the U.S. is on the order of $1-2B per year, dominated by BCG induction/maintenance (approximately $10,000-15,000 per patient-year including cystoscopy surveillance). Layering pembrolizumab at roughly $150,000-180,000 per patient-year across even half of that population would push the addressable spend into the $5-10B range before accounting for duration, discounting, and payer pushback. A positive readout would extend Keytruda's bladder franchise into first-line NMIBC and help defend revenue ahead of the drug's 2028 loss-of-exclusivity for the primary composition-of-matter claims. Keytruda generated $29.5B in 2024 revenue and is on track for a similar or higher 2025 number per Merck's 10-K and quarterly updates [4][5].

Pembrolizumab was approved for BCG-unresponsive high-risk NMIBC with carcinoma in situ in January 2020 on the strength of KEYNOTE-057's single-arm complete response data (41% CR at 3 months, median duration 16.2 months) [2][3]. KEYNOTE-676 is the Phase 3 label-expansion trial testing pembrolizumab plus BCG versus BCG alone across two cohorts: Cohort A enrolls persistent or recurrent disease after BCG induction (distinct from the fully BCG-unresponsive population where pembrolizumab monotherapy is already approved), and Cohort B enrolls BCG-naive patients [1][11]. The trial was initiated in 2018 and is listed as active, not recruiting on ClinicalTrials.gov [1]. No new FDA designations apply because pembrolizumab is approved; this is a straightforward label-expansion study. The primary endpoint is event-free survival (EFS), a composite of high-grade recurrence, progression, or death. Topline readout was previously guided toward 2024-2025 but has drifted as events accrue slowly in a mostly curative population. Importantly, Merck did not present a KEYNOTE-676 primary analysis at ASCO GU 2026 and has issued no new readout guidance; this absence is itself informative - it signals continued event-accrual delay rather than an imminent data drop [12]. A pragmatic readout window is 2027-2028.

PD-1 is a brake pedal on T cells. When an immune cell recognizes a tumor, engagement of PD-1 by its ligand PD-L1 (which many tumors overexpress as a defense) shuts the T cell down. Pembrolizumab is a monoclonal antibody that blocks PD-1, releasing the brake and letting T cells attack [6]. BCG is a weakened strain of Mycobacterium bovis originally developed as a tuberculosis vaccine. It was first reported effective in bladder cancer by Morales et al. in 1976 and FDA-approved for HR NMIBC in 1990 (TICE BCG, Organon) - the standard of care for over three decades [13][14]. Instilled directly into the bladder through a catheter, it triggers a strong local inflammatory response, pulling T cells and macrophages into the bladder wall, which collaterally kills cancer cells. The combination logic: BCG recruits immune cells to the tumor bed, and pembrolizumab keeps them active by preventing PD-1-mediated exhaustion. The pembrolizumab mechanism is deeply validated across tumor types (lung, melanoma, head and neck, metastatic urothelial, MSI-H cancers) and specifically in bladder cancer at both the metastatic (KEYNOTE-045) and BCG-unresponsive NMIBC settings [2][6]. The combination-specific hypothesis rests on translational data showing BCG induces PD-L1 upregulation in bladder tumors and that BCG-recurrent patients have an exhausted T cell phenotype that checkpoint blockade might rescue [7].

Phase 3 randomized open-label trial, pembrolizumab plus BCG versus BCG alone, in HR NMIBC [1]. Enrollment target is approximately 975 patients across two parallel cohorts. Cohort A enrolls patients with persistent or recurrent disease after BCG induction, a population that overlaps with but is distinguished from BCG-unresponsive disease where pembrolizumab monotherapy is already approved. Cohort B enrolls BCG-naive patients with HR NMIBC [1][11]. Primary endpoint is event-free survival, defined as time to high-grade recurrence, progression, or death from any cause. Key secondary endpoints include complete response rate in the CIS subpopulation, bladder-intact survival, time to cystectomy, and safety. The design is reasonable but carries a mechanical weakness: EFS in HR NMIBC is a slow endpoint because many patients on BCG alone remain disease-free for extended periods, so event accrual takes years. Enrollment is complete; the trial is in follow-up [1]. KEYNOTE-676 does not use PD-L1 expression as an enrollment criterion; whether pre-specified PD-L1 subgroup analyses are planned is not disclosed in the current ClinicalTrials.gov record [1]. This is a genuine scientific limitation - KEYNOTE-045 showed differential benefit by PD-L1 status in metastatic urothelial disease, and if the KEYNOTE-676 primary endpoint misses, PD-L1 enrichment analysis would be the obvious next question but may not be adequately powered. The two-cohort structure is strategically important because it lets Merck pursue approval in the broader BCG-naive population while generating supportive data in the recurrent setting.

Our model estimates a 50% chance this drug is eventually approved. It starts from the historical base rate for Phase 3 drugs in this area (about 48%), then adjusts using ten facts about the trial and sponsor. What moves the number most: it is helped by more secondary endpoints than usual, larger-than-typical enrollment for this phase, and its light or open-label blinding; it is held back by weak or limited earlier-phase results. The other facts land near average for this stage, so they leave the estimate roughly where the base rate put it.

Efficacy is the main failure mode. Pembrolizumab works well in BCG-unresponsive disease because those patients have no other immune-based options; adding it on top of still-active BCG in a largely BCG-sensitive population is a higher statistical bar. If BCG monotherapy produces 2-year EFS in 60-65% of patients, the combination has to show a durable incremental benefit, and the delta might be small. Safety: pembrolizumab carries immune-related adverse events including colitis, pneumonitis, hepatitis, and endocrinopathies. These are manageable in metastatic disease where life expectancy is measured in years, but they are a harsher tradeoff in a mostly curable disease where many patients would otherwise be cured by BCG alone. BCG itself causes cystitis and rare systemic reactions. Combination could worsen urinary toxicity and lead to higher discontinuation. BCG supply remains chronically constrained - Merck's TICE BCG has been in intermittent shortage for over a decade - and the combination arm's net BCG demand (sparing via dose optimization vs additive load across a bigger population) will affect formulary uptake and real-world penetration regardless of trial outcome [15]. Commercial risk is sharp. The BCG-unresponsive setting is already crowded: nadofaragene firadenovec (Adstiladrin, Ferring, approved 2022) [8], nogapendekin alfa inbakicept plus BCG (Anktiva, ImmunityBio, approved 2024) [9], TAR-200 (gemcitabine intravesical system, now branded Inlexzo, J&J, FDA-approved in 2026 based on an 82% CR rate in SunRISe-1) [10], and cretostimogene grenadenorepvec (CG0070, CG Oncology, an intravesical oncolytic adenovirus with ~75% CR in BOND-003 and a filed BLA) all target that space with intravesical delivery and lower systemic toxicity than a PD-1 antibody [16]. KEYNOTE-676's commercial case lives or dies in the BCG-naive population; payers will push back on adding six-figure IV antibody therapy to a $150 generic bladder instillation.

Watch, but calibrate expectations. The specific signal to look for is the event-free survival hazard ratio in Cohort B (BCG-naive); a hazard ratio below 0.65 would be commercially meaningful and support a label expansion that reshapes first-line HR NMIBC. For readers unfamiliar with hazard ratios: a hazard ratio below 1.0 means the combination arm had fewer events (recurrences or deaths) than the BCG-alone arm, and 0.65 means roughly a 35% reduction in event rate - clinically meaningful and historically the kind of magnitude that supports a label change. Trials in this setting with hazard ratios above 0.80 have generally failed to gain approval even when nominally statistically significant. The BCG-recurrent cohort (Cohort A) is scientifically more interesting (exhausted T cell phenotype plausibly rescuable by PD-1 blockade) but a smaller market because it overlaps with existing pembrolizumab, Anktiva, Adstiladrin, and now Inlexzo (TAR-200) approvals [10]. For Merck, this is a Keytruda lifecycle management bet. Keytruda generated $29.5B in 2024 and faces biosimilar competition starting in 2028 for the primary composition claim [4][5]; a new NMIBC indication could carry its own regulatory exclusivity window, and Merck's subcutaneous pembrolizumab (Keytruda Qlex) formulation layered on top may extend franchise durability past biosimilar IV entry - but neither mechanism rescues the primary composition patent. Readout window is 2027-2028; check back after ESMO or ASCO GU in those years, or watch for a pre-announcement in Merck's quarterly updates. A clean miss would constrain the Keytruda bladder franchise to the metastatic and salvage settings and leave the much larger first-line NMIBC market to BCG monotherapy and the intravesical competitors (Anktiva, Adstiladrin, Inlexzo, cretostimogene). A clean win would be one of the more consequential NMIBC data drops of the decade.