BMS-986205

Linrodostat (BMS-986205) is Bristol-Myers Squibb's oral IDO1 inhibitor tested in combination with nivolumab across solid tumors, an asset BMS acquired from Flexus Biosciences in an $800M upfront deal in 2015 [7]. The featured trial (NCT03854032) is an investigator-sponsored Phase 2 window-of-opportunity study at Thomas Jefferson and Ohio State testing neoadjuvant nivolumab with or without linrodostat in resectable stage II-IV head and neck squamous cell carcinoma, with final pathologic and clinical outcomes already presented at ASCO 2022 (no primary publication to date) [4][8]. The commercial program collapsed after Incyte's epacadostat failed the ECHO-301 melanoma Phase 3 in 2018, and BMS subsequently removed the linrodostat arm from its Phase 3 muscle-invasive bladder cancer trial ENERGIZE [5][6]. What remains is scattered academic follow-on work: the HNSCC study, a Memorial Sloan Kettering endometrial Phase 2 (NCT04106414), a Northwestern newly diagnosed glioblastoma Phase 1 (NCT04047706), plus a combined Phase 1/2 dose-finding publication in 571 patients [1][3]. BMS has not signaled renewed commercial investment and no regulatory submission is planned in any indication.

Linrodostat is investigational everywhere. No approvals. No public FDA breakthrough, fast-track, or orphan designations. The featured HNSCC study (NCT03854032) is Phase 2, active but not recruiting, ~42 patients enrolled across Thomas Jefferson and Ohio State [4]. Final pathologic and clinical outcomes were presented at ASCO 2022 [8] but a peer-reviewed primary publication has not appeared in the four years since, which is itself a signal: window-of-opportunity studies with clean positive signals rarely sit unpublished this long. The registrational-scale program was ENERGIZE (NCT03661320), a Phase 3 in muscle-invasive bladder cancer that ultimately enrolled 855 patients; BMS removed the linrodostat arm during the trial after the IDO1 field cratered in 2018 [5][2]. Final Phase 3 results from the surviving ENERGIZE arms (chemotherapy alone vs. chemotherapy plus nivolumab) have not yet been publicly reported as of early 2026 [2]. The most recent peer-reviewed data came from the Phase 1/2 combined dose-finding study covering 571 patients across multiple tumor types (Luke et al., Clin Cancer Res 2025) [1] and a small newly diagnosed glioblastoma Phase 1 combining linrodostat with nivolumab and radiation (Lukas et al., Clin Cancer Res 2026) [3]. The endometrial Phase 2 at Memorial Sloan Kettering (NCT04106414) is active but not recruiting, and no Phase 3 has been announced or committed to by BMS or its Japanese partner Ono Pharmaceutical. Timeline for any registrational submission: none disclosed. The program is best described as academic wind-down rather than a live commercial pipeline.

IDO1 is an enzyme that breaks down tryptophan, one of the amino acids cells need to build proteins. Tumors exploit this in a specific way: by burning through tryptophan and producing kynurenine, they starve nearby T cells (the immune cells that kill cancer) and generate a metabolite that actively suppresses them. Block IDO1, the theory went, and T cells recover just as checkpoint inhibitors like nivolumab and pembrolizumab release their brakes. In animal models this combination worked. In early-phase human data it looked plausible: response rates in melanoma reached 55-60% when epacadostat was added to pembrolizumab in single-arm cohorts. Then ECHO-301, the confirmatory Phase 3, showed nearly identical outcomes with or without epacadostat, no benefit on progression-free or overall survival [6]. Multiple explanations followed: IDO1 was not the dominant immunosuppressive pathway in most tumors, compensatory enzymes (TDO2, IDO2) filled the gap, or the animal models had misled the field. Linrodostat is a more potent and more selective IDO1 inhibitor than epacadostat, so the target-engagement story is cleaner, but the same biological question, whether IDO1 blockade actually reprograms the tumor microenvironment (the mix of immune cells, blood vessels, and signaling molecules surrounding a tumor) enough to matter in humans, was never answered in linrodostat's favor before commercial investment left the target [1].

NCT03854032 is a Phase 2 open-label window-of-opportunity study of neoadjuvant nivolumab with or without linrodostat in resectable stage II-IV head and neck squamous cell carcinoma, run out of Thomas Jefferson University and Ohio State with roughly 42 patients enrolled (31 in the combination arm, 11 in the nivolumab-alone arm) [4][8]. The reported primary endpoint on ClinicalTrials.gov is objective response rate (ORR, the fraction of patients whose tumors shrink by a defined threshold); the ASCO 2022 presentation reported final pathologic and clinical outcomes [8]. There is no pembrolizumab-plus-chemotherapy control arm, which is now the first-line standard of care in recurrent or metastatic HNSCC for PD-L1-positive patients, so any interpretation is against historical benchmarks rather than a modern comparator. The larger study of note, ENERGIZE (NCT03661320), was designed as neoadjuvant chemotherapy alone versus chemo plus nivolumab versus chemo plus nivolumab plus linrodostat in muscle-invasive bladder cancer, with pathological complete response (pCR - no residual invasive cancer found in the surgically removed bladder) as primary endpoint [2][5]. BMS dropped the linrodostat arm mid-study following the epacadostat Phase 3 failure, and final Phase 3 results comparing the surviving chemo and chemo+nivolumab arms have not yet been publicly reported [2]. The design was informed by earlier BLASST-1 data showing a 49% pCR rate for neoadjuvant nivolumab plus gemcitabine/cisplatin, but that signal has not yet been formally confirmed against a chemo-alone control in a Phase 3 readout. The Phase 1/2 combined dose-finding analysis (Luke et al., Clin Cancer Res 2025) covered 571 patients across multiple tumor types and reported ORRs in the roughly 6-30% range depending on cohort, insufficient to justify further registrational development on their own [1]. The academic follow-ons in glioblastoma and endometrial cancer are exploratory Phase 1/2 studies without power to support an approval path.

Our model estimates a 5% chance this drug is eventually approved. It starts from the historical base rate for Phase 2 drugs in this area (about 13%), then adjusts using ten facts about the trial and sponsor. What moves the number most: it is helped by its light or open-label blinding, more secondary endpoints than usual, and the sponsor's strong record of getting drugs approved; 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 risk dominates. The class was invalidated in Phase 3 by a structurally similar compound against a matched checkpoint-inhibitor combination, and no post-hoc biomarker-selected rescue analysis has produced a signal strong enough to fund a new registrational trial [6]. If IDO1 blockade added even a real 5-percentage-point ORR bump on top of pembrolizumab, ECHO-301 would have detected it. The rest of the class has fared no better: navoximod (Roche/NewLink) reported only 9% partial response rates in combination with atezolizumab and was dropped, and indoximod (dual IDO1/TDO) has completed most of its trials with no development path forward [9]. The defensible counter-argument for linrodostat is potency and selectivity: it inhibits IDO1 at lower doses than epacadostat and does not touch TDO2, so the target-engagement hypothesis remains formally alive [1]. That argument has been on the table for eight years without producing a positive Phase 3. Safety risk is low relative to the class. The Phase 1/2 combined data showed adverse events dominated by the nivolumab profile, no unexpected on-target liabilities, no serious hepatic or neurotoxicity signal [1]. Execution risk is high because the featured HNSCC study is investigator-initiated at academic centers with no follow-on Phase 3 committed by any sponsor, and the four-year gap between ASCO 2022 presentation and a primary publication is not encouraging [8]. Commercial risk is close to total: even a positive Phase 2 signal in a small cohort would require a sponsor willing to fund Phase 3 into a field payers and clinicians have written off. Pembrolizumab-plus-chemotherapy owns first-line HNSCC, and nivolumab loses patent exclusivity in the early 2030s. BMS appears to retain worldwide commercial rights (Ono Pharmaceutical is the historical Japanese partner on nivolumab but has not announced independent linrodostat development), and no out-licensing or divestiture of the linrodostat IP has been disclosed publicly.

Noise, with one narrow exception. The IDO1 story is a case study in how a plausible mechanism, an elegant preclinical rationale, and encouraging Phase 1 data can all be wrong at once. Bristol-Myers Squibb spent $800M upfront on Flexus Biosciences in 2015 largely to acquire linrodostat, and the asset has been walked away from in every indication that mattered commercially [7]. What would flip this from noise to signal: a positive objective response rate readout in the Thomas Jefferson/Ohio State HNSCC Phase 2 (NCT03854032) that clearly exceeds nivolumab monotherapy's CheckMate-141 benchmark of 13.3% ORR in second-line HNSCC [10], combined with a defined biomarker (IDO1 immunohistochemistry - IHC, a lab stain that detects protein levels in tumor tissue - or kynurenine-to-tryptophan ratio) that identifies which patients responded [4]. The trial's ASCO 2022 results are already public but not peer-reviewed, and the failure to publish a primary paper in four years is itself a soft-negative signal [8]. The newly diagnosed glioblastoma Phase 1 (Lukas et al. 2026) is worth tracking as a cheap pointer because checkpoint monotherapy has been a complete failure in GBM and any signal there would reopen the mechanism debate [3]. Check the 2027 ASCO/SITC cycle for updated HNSCC and GBM follow-up. Absent both a biomarker and a publication-quality efficacy signal, linrodostat becomes a historical artifact of a target that never worked in humans, and the listed sponsor entries correspond to legacy IP ownership rather than active commercial development.