Statistical Breakdown of 2026 Phase 3 Trial Endpoints: Volumetric Mass Evaluation in Triple-Agonist Registrations
Source-linked interpretation of TRIUMPH-1 and TRIUMPH-2 week-80 endpoints, estimand differences, A1C units and the planned regulatory submission.
TRIUMPH trial records: scope and terminology
This reference examines sponsor-reported Phase 3 findings available on 8 October 2026. The numerical tables below use the efficacy estimand; they are not interchangeable with results calculated under a treatment-regimen estimand. The arm labels identify randomized trial groups, not laboratory preparation instructions.
Although the article title uses “volumetric mass evaluation,” the reported endpoint is percentage change in body weight, a mass measurement. It is not a volumetric assay. A1C change is expressed in percentage points, not relative percent reduction or statistical standard deviations.
Separate populations, separate interpretations
TRIUMPH-1, registered as NCT05929066, evaluated a population without diabetes. TRIUMPH-2, NCT05929079, evaluated a population with type 2 diabetes. These independent trial populations do not form a randomized comparison against one another. Differences between their aggregate results must not be attributed solely to receptor pharmacology.
Week-80 endpoint matrix
Signed values describe change from baseline: negative values indicate a decrease. TRIUMPH-1 A1C values are marked as unreported in the cited release rather than inferred from the separate diabetes trial.
| Trial node ID | Cohort variant | Body-weight change at week 80 (%) | A1C change at week 80 (percentage points) |
|---|---|---|---|
| TRIUMPH-1 | 4 mg | −19.0% | Not reported in cited release |
| TRIUMPH-1 | 9 mg | −25.9% | Not reported in cited release |
| TRIUMPH-1 | 12 mg | −28.3% | Not reported in cited release |
| TRIUMPH-1 | Placebo | −2.2% | Not reported in cited release |
| TRIUMPH-2 | 4 mg | −12.7% | −1.4 percentage points |
| TRIUMPH-2 | 9 mg | −19.1% | −1.6 percentage points |
| TRIUMPH-2 | 12 mg | −20.8% | −1.5 percentage points |
| TRIUMPH-2 | Placebo | −4.0% | −0.2 percentage points |
Sources: Lilly’s 21 May 2026 TRIUMPH-1 release and 29 September 2026 TRIUMPH-2 release. The 25.9% and 28.3% figures belong to different TRIUMPH-1 arms; they do not describe a range achieved by the 12 mg arm across both trials.
Statistical interpretation: identify the estimand before comparing values
An estimand defines the effect being estimated, including the population, endpoint, handling of intercurrent events and summary measure. A result without this context is an incomplete analytical record.
- Efficacy estimand
- An estimate under assumptions about remaining on the assigned intervention and avoiding specified additional interventions. Consult the trial’s analysis documentation for the precise definition.
- Treatment-regimen estimand
- An alternative framework addressing outcomes under the assigned regimen with its stated handling of discontinuation and other events. It answers a different analytical question.
- Point estimate
- A summarized estimate, not an uncertainty interval or an individual prediction.
- Between-arm contrast
- A comparison within a trial, interpreted alongside uncertainty, the prespecified analysis and multiplicity controls.
Arithmetic contrasts are not reconstructed model outputs
Subtracting the displayed placebo change from an active-arm change yields a descriptive difference in percentage points. That arithmetic does not recover an adjusted model estimate, confidence interval, standard error or probability value. Those require the original statistical output.
Similarly, a decrease of 1.5 percentage points in A1C cannot be relabelled as a 1.5% relative decrease. Relative change would require an explicit denominator and a separate calculation. Preserve the unit attached to every result.
Missing values should remain missing
An unreported A1C field is neither zero nor evidence that the endpoint was unchanged. In a data adapter, store an unavailable value as null with its reporting status. Keep the presentation label separate from the numeric field so subsequent calculations cannot accidentally treat an empty string as zero.
Data-model integrity and limitations
- Retain provenance: store the trial identifier, publication date, original URL and table version for each extracted value.
- Preserve dimensions: distinguish percentage body-weight change, mass in kilograms and A1C change in percentage points.
- Do not pool unlike populations: separate eligibility criteria, baseline characteristics and endpoint definitions before considering a combined analysis.
- Track missingness: identify absent observations, discontinuations and the statistical assumptions used to handle them.
- Separate exploratory findings: retain whether an analysis was prespecified and how multiple comparisons were controlled.
- Review safety with efficacy: adverse events and discontinuations are necessary to interpret a clinical dataset; efficacy tables alone do not establish an overall benefit–risk conclusion.
The interpretation checks above are an editorial data-governance framework. They do not reconstruct participant-level data, establish a causal comparison between trials or substitute for an independent review of the full protocol and statistical analysis plan.
Regulatory pipeline: planned submission and distinct decision stages
Lilly’s July 2026 announcement states that it is completing the Chemistry, Manufacturing, and Controls package and plans a Biologics License Application submission in Q1 2027. The September update reiterates that plan. This is a sponsor-stated future milestone, not evidence of an accepted application or FDA approval.
- Submission planning
- A proposed filing milestone that can change as the evidence package develops.
- Application submission
- A distinct event requiring confirmation that the application has actually been filed.
- Regulatory review
- Assessment by the regulator; neither an intended filing quarter nor trial completion establishes its outcome.
- Approval status
- A separate decision that must be confirmed from an authoritative regulatory record.
For a maintained reference library, display the source date next to the milestone and update the status only when a new primary record establishes the next event. Avoid converting forward-looking statements into completed regulatory facts.
Primary sources and reference maintenance
- Lilly: TRIUMPH-1 results, 21 May 2026
- Lilly: detailed TRIUMPH-2 results, 29 September 2026
- Lilly: additional Phase 3 findings and planned BLA submission, 23 July 2026
Reference review date: . Source-linked trial findings do not authenticate a separately supplied laboratory material, establish its batch identity or demonstrate equivalence to an investigational clinical formulation.