OPC-167832 in DBOS and PBOS regimens: the 3-month tuberculosis target slips again

What was tried

Gates Medical Research Institute ran NCT05971602 (TBD06-201), a Phase 2b/c multicenter, two-stage, open-label, duration-randomized trial of four-drug regimens in adults with drug-sensitive pulmonary tuberculosis. Stage 1 randomized adults to one of three options: DBOS (Delamanid, Bedaquiline, OPC-167832, Sutezolid) for 17 weeks, PBOS (Pretomanid, Bedaquiline, OPC-167832, Sutezolid) for 17 weeks, or standard 2HRZE/4HR for 26 weeks. Stage 2 was designed to take the winning XBOS regimen and randomize duration across 9, 11, 13, 15, and 17 weeks against the same 26-week comparator, with the objective of identifying a regimen of three months or less. The novel agent in DBOS and PBOS is OPC-167832 (quabodepistat, ChEMBL5095080), Otsuka's oral DprE1 inhibitor at Phase 2. Enrollment closed at 93 participants. The trial was terminated with primary completion dated 6 February 2025. The public statement is that data generated do not support the investigational regimens achieving the 3-month-or-less objective.

The biological hypothesis

DprE1 (decaprenylphosphoryl-beta-D-ribose 2'-oxidase, ChEMBL3804751) is essential for arabinogalactan synthesis in the mycobacterial cell wall. Inhibition is bactericidal and synergizes with bedaquiline (ATP synthase) and sutezolid (mycobacterial ribosome) by hitting an orthogonal pathway. OPC-167832 showed early bactericidal activity in 14-day Phase 2a EBA studies and preclinical synergy with bedaquiline and delamanid or pretomanid (Saito et al., 2020). DBOS and PBOS were therefore designed around a defensible all-oral four-drug combination that, on paper, should have shortened therapy through additive bactericidal activity and a non-overlapping resistance pattern. The clinical literature on shorter TB regimens has been consistent: pretomanid plus bedaquiline plus linezolid (BPaL) shortened MDR-TB therapy to 6 months in ZeNix and Nix-TB. The question DBOS and PBOS asked was whether adding OPC-167832 and substituting sutezolid for linezolid could push the duration in drug-sensitive disease below three months.

What actually happened

Stage 1 ran. Stage 2 did not progress as designed. The two primary endpoints were the proportion of DS-TB participants reporting Grade 3 or higher treatment-emergent adverse events and the proportion of DS-TB participants with unfavorable outcome status (defined as failure to achieve or maintain stable culture conversion, treatment discontinuation, or unable to be assessed). Gates MRI announced termination because the Stage 1 data did not justify advancing any XBOS combination into the duration-randomized Stage 2. Gates MRI did not publicly disclose granular numbers, but the program conclusion is unambiguous: across the two investigational regimens and standard-of-care, the combinations did not show the bactericidal or sterilizing kinetics needed to cut TB therapy to 90 days. With 93 patients enrolled across three arms, statistical power was always going to be limited for non-inferiority against 2HRZE/4HR. The read appears directional rather than statistically conclusive.

Failure mechanism, best guess

The most parsimonious explanation is that adding a DprE1 inhibitor to a BPaL-style backbone does not produce the supra-additive sterilizing activity that ultra-short regimens require. Standard TB therapy works because the regimen kills persister populations slowly over six months. Hitting cell-wall synthesis with a DprE1 inhibitor is bactericidal against replicating bacilli but provides limited additional sterilization of slow-growing persister and intracellular populations, which are the rate-limiters for relapse-free cure. A secondary contributor is the inter-individual variability in OPC-167832 and sutezolid exposure seen in earlier Phase 1 PK data, which inflates the share of patients on sub-therapeutic exposure and erodes cure at any duration shorter than four months. Population mix matters too: the trial included both DS-TB and an observational RR/MDR-TB cohort, and early outcomes in the resistant cohort plausibly diluted the signal Gates MRI needed for Stage 2.

How to prevent this next time

Future ultra-short TB regimen trials should use explicit Bayesian stopping rules and pharmacokinetic-anchored enrichment. With a prior on 12-week cure centred at 75% based on extrapolation from BPaL-style 6-month data and a decision threshold of 85% required to declare non-inferiority to 26-week 2HRZE/4HR, the posterior probability of success can be evaluated continuously:

PP(success)=∫θ​P(final success∣θ,interim data)⋅π(θ∣interim data)dθ

Adaptive randomization should be wired into the protocol so that arms whose interim posterior crosses a futility threshold of 0.10 close automatically rather than waiting for a Data Monitoring Committee discretionary call. Biomarker enrichment by baseline cavitary disease, time-to-culture-conversion at week 4, and Mycobacterium tuberculosis lineage would have improved the discrimination of any short-duration arm and reduced the number-needed-to-screen. A historical base-rate adjustment using ZeNix, Nix-TB, and TBTC Study 31 would have anchored the prior at a 6-month, not 3-month, duration. The single highest leverage change would have been mandatory pharmacokinetic-guided dosing of OPC-167832 to keep the proportion of subjects below 80% of the target AUC under 10%, evaluated at the end of Stage 1 before any Stage 2 advance.

What this means for similar programs

The DprE1 inhibitor class is still credible but the ultra-short TB hypothesis is now substantially weaker. Macozinone (PBTZ169), BTZ-043, and TBA-7371 remain in Phase 2 evaluation. Programs running DprE1 inhibitors should expect Gates MRI's read to recalibrate funder expectations toward 4 to 6 month regimens. The next class-level question is whether DprE1 inhibition adds enough beyond bedaquiline plus pretomanid plus linezolid to justify a fifth orthogonal target, or whether resource should flow to sterilizing-activity targets such as mycobacterial respiratory chain components.

Open questions

What were the per-arm culture-conversion rates and time-to-culture-conversion distributions in Stage 1? Did the safety signal differ between DBOS and PBOS, particularly for QT prolongation and hepatotoxicity given the bedaquiline-pretomanid backbone? What proportion of subjects achieved target plasma exposure for OPC-167832 and sutezolid, and how did exposure correlate with outcome? Will Otsuka and Gates MRI continue OPC-167832 in 4 to 6 month regimens, or is the asset effectively shelved?

Sources

1. ClinicalTrials.gov, NCT05971602, full record retrieved 2026-05-22. ChEMBL CHEMBL5095080 (quabodepistat / OPC-167832), small molecule, max phase 2. ChEMBL3804751 (M. tuberculosis H37Rv DprE1, single protein, Decaprenylphosphoryl-beta-D-ribose oxidase). OpenFDA FAERS bedaquiline (counts): QT prolonged 407, peripheral neuropathy 324, anaemia 315, vomiting 247, death 222, hepatotoxicity 114. OpenFDA FAERS pretomanid: death 114, anaemia 72, peripheral neuropathy 64, QT prolonged 34, hepatotoxicity 22. Open Targets Platform v25.06, disease lookup EFO_1000049 (pulmonary tuberculosis), retrieved 2026-05-22. Annual progress in chemotherapy for tuberculosis in 2025, Zhonghua Jie He He Hu Xi Za Zhi 2026,Benzoselenazinone-based DprE1 inhibitors, J Med Chem 2026,Advancing Tuberculosis Treatment with Next-Generation Drugs and Smart Delivery Systems, Pharmaceutics 2026,Discovery of potent DprE1-targeted antitubercular agents based on PBTZ169 and TBA-7371, Mol Divers 2025,2-Benzyl-2,7-diazaspiro[3.5]nonane benzothiazinones with broad-spectrum antimycobacterial activity, J Med Chem 2025,. Available from: https://clinicaltrials.gov/study/NCT05971602.