Direct answer. A response percentage describes the group analyzed for that endpoint. The 2025 CBL-514 study randomized 76 people but 38 of 75 treated participants completed the study; the 2026 abstract reports responder denominators smaller than its randomized total.
Follow the participant flow before the percentage
In the 2025 randomized Phase II report, 76 participants were randomized: 50 to CBL-514 and 26 to placebo. One assigned participant did not receive treatment, leaving 75 in the safety population. Of those treated, 38 completed the study and 37 did not. The authors report a 49.3% non-completion rate; listed reasons include loss to follow-up, withdrawal, weight instability, schedule non-compliance and adverse events. Its per-protocol group comprised 30 people. The publication discusses both intention-to-treat and per-protocol analyses, which answer related but not identical questions.
The 2026 Phase 2b abstract says 108 adults were randomized, yet its quoted primary responder counts are 23/30 and 7/37. Those fractions must be reported with the authors’ analysis definition; they are not percentages of every randomized participant. The current registry record lists actual enrollment as 107. Do not guess why the record and publication totals differ. Consult record history and the full report before reconciling the figures.
Evidence check
- 2025 randomized: 76; treated: 75; study completers: 38.
- 2025 per-protocol population: 30, as defined in the paper.
- 2026 abstract primary endpoint counts: 23/30 versus 7/37.
What missing follow-up changes in an interpretation
A participant who stops a study may have a different outcome from one who completes every visit. Excluding that person can change the population described by a result. It also matters why follow-up was missing: an adverse event, a protocol requirement and loss of contact have different implications. The 2025 paper attributes many discontinuations to strict study requirements, but the observed non-completion still belongs beside the efficacy result. A large difference between randomized and evaluable groups warrants attention to the missing-data method and sensitivity analyses.
The FDA-hosted ICH E9(R1) addendum asks investigators to specify the treatment effect of interest and address data not collected after relevant events. It supplies a framework, not an alternative calculation for CBL-514. An editorial summary should therefore name the paper’s actual population and model instead of silently reusing a subgroup percentage as though every randomized participant contributed an observed endpoint.
Evidence check
- Identify randomized, treated, safety and endpoint-analysis populations.
- List reasons for missing follow-up where the report provides them.
- Look for prespecified handling and sensitivity analyses in the source.
How to quote the result without overstating it
Use a sentence with the study, endpoint, time point, numerator, denominator and comparator. For example, the 2026 abstract reports at least one-grade clinician-rated improvement in 23 of 30 participants in its CBL-514 analysis group and 7 of 37 in the placebo group at 12 weeks. This states what the abstract reports without implying a population-wide success rate. Link the original publication and, where available, the registry record. A percentage alone is too easy to detach from its design.
Keep the limitations in the same content block, not hidden in a distant disclaimer. The 2025 report had substantial non-completion and short follow-up, and the 2026 abstract provides less methodological detail than a full report. Do not turn either clinical result into a guarantee, a claimed approval or a property of a supplier powder. Readers assessing a research material need a different set of evidence: batch documentation and a defined specification.
Evidence check
- Name the exact numerator and denominator.
- State comparator and assessment week.
- Link the paper and distinguish trial evidence from batch evidence.
Reviewed 23 October 2026. This editorial article describes published research and documentation boundaries, not treatment or administration advice. The clinical development program is independent of this website.
Frequently asked questions
Why is 76.7% not necessarily 76.7% of all randomized participants?
The 2026 abstract reports 23 responders among 30 participants in the stated analysis group. It also says 108 adults were randomized. Those are different denominators; the percentage must be tied to the group used for that endpoint.
What does intention-to-treat mean in the 2025 CBL-514 paper?
The paper defines its ITT population as all randomized participants, while the per-protocol population is a subset meeting specified completion and protocol conditions. Read the paper’s methods for how each outcome was analyzed.
Does non-completion invalidate the 2025 trial?
Non-completion is a material limitation that needs examination, not an automatic verdict. Review its reasons, missing-data approach and agreement between the reported analysis populations.
Why do the 2026 paper and registry show different totals?
The abstract describes 108 randomized adults; the current public registry lists actual enrollment as 107. The available summary does not establish the reason for the difference. Check record history and the full publication before interpreting it.