Inside the phase 2 obesity trial.
A document-level review of the full protocol, final statistical analysis plan, supplementary results, author disclosures, and data-sharing terms—not a summary of the abstract.
Source record
412 PDF pages reviewed
Protocol + statistical analysis plans
351 PDF pages
Supplementary appendix
43 PDF pages
Author disclosures
16 PDF pages
Data-sharing statement
2 PDF pages
Trial facts retain their population, time point, analysis method, and source-page context.
Primary source packet
Four documents, four different jobs
The paper reports the headline result. The protocol explains what was planned, the SAP explains how it was estimated, the appendix exposes subgroup and safety detail, and the disclosure and data-sharing files show the study's governance context.
Protocol + statistical analysis plans
351 PDF pagesOriginal protocol, Amendment b, amendment history, original SAP, and final SAP v4.
Open source PDFSupplementary appendix
43 PDF pagesEligibility, methods, participant flow, efficacy figures, subgroup tables, and detailed safety tables.
Open source PDFAuthor disclosures
16 PDF pagesEmployment, stock, research support, consulting, advisory-board, and speaking relationships.
Open source PDFData-sharing statement
2 PDF pagesConditional access timing, exclusions, independent review, agreement, and secure-environment terms.
Open source PDFPage references use the printed page number inside each document. The protocol file contains the original protocol, Amendment b, the original SAP, and final SAP v4 in one PDF.
Study architecture
A dose-finding trial—not a personalized treatment algorithm
Participants were assigned by a computer-generated random sequence. Sex and baseline BMI category were stratification variables; current weight, goal weight, and pounds to lose did not select a dose.
338
randomized
337 participants were represented in the safety tables.
48 + 4
weeks
48-week treatment period plus a 4-week safety follow-up.
7
randomized groups
Six active dose or escalation groups and one matched placebo group.
28
listed U.S. sites
A U.S.-only phase 2 study; global generalizability is limited.
Randomized active arms
Protocol-controlled escalation paths
Target arm
1 mg
Fixed throughout; no escalation.
Target arm
4 mg
One subgroup escalated at week 4.
Target arm
4 mg
A separate subgroup started and remained at 4 mg.
Target arm
8 mg
Escalation at weeks 4 and 8.
Target arm
8 mg
A separate subgroup escalated at week 4.
Target arm
12 mg
Escalation at weeks 4, 8, and 12.
Randomization
Allocation ratio 2:1:1:1:1:2:2; stratified by sex and BMI below versus at least 36 kg/m².
Lifestyle adjunct
Diet and activity counseling began before randomization and continued through week 44.
Weight nuance
BMI ≤22 triggered a safety dose reduction and BMI ≤19 a stop rule; neither selected a starting dose.
Protocol pp. 150–153, 166, and 171–175; final SAP pp. 308–315. Placebo participants followed matched volume schedules to preserve blinding.
Who was actually studied
Eligibility narrows what the result can mean
The trial enrolled a selected population under active monitoring. Its entry criteria help interpret generalizability; they are not a public screening checklist or evidence that excluded people should follow another schedule.
Core enrollment frame
- Age
- 18–75 years
- BMI
- 30–50 kg/m², or 27–<30 with a qualifying comorbidity
- Diabetes
- Participants with diabetes were excluded
- Readiness
- Able to follow visits, lifestyle counseling, logs, and trained study procedures
- Geography
- United States sites only
Major exclusion domains
The study did not represent every person with obesity
- Type 1 or type 2 diabetes, or screening laboratory values meeting diabetes thresholds
- More than 5 kg of recent weight change; prior or planned bariatric surgery or obesity-device therapy
- eGFR below 45, severe gastric-emptying abnormality, or pancreatitis history
- Uncontrolled hypertension, pulse above 100, recent major cardiovascular event, specified arrhythmias, or NYHA class III–IV heart failure
- Specified liver or gallbladder disease, calcitonin elevation, MTC or MEN2 history, or recent clinically significant malignancy
- Specified unstable psychiatric illness, lifetime suicide attempt, high PHQ-9 score, or recent active suicidal ideation or behavior
- Recent anti-obesity or glucose-lowering medicines, selected weight-promoting medicines, or systemic glucocorticoids
- Other protocol-defined allergy, transplant, hematologic, substance-use, pregnancy, contraception, trial-participation, and investigator-risk exclusions
Mean age
48 years
About 10% were 65 or older.
Sex
48% women
Female enrollment was capped at 60% to increase male representation.
Representation
U.S.-only
The appendix says the study is not representative of the global population.
Protocol pp. 158–165; Supplementary Appendix pp. 4–10 and Table S2, p. 26.
What the protocol says about administration
A supervised liquid study vial—not a public reconstitution recipe
The protocol is specific about the trial's product format, route, abdominal rotation, and training. It is equally important to record what the source packet does not contain.
Documented trial facts
What investigators actually used
Container
Sponsor-provided single-use vial
Study strength
12 mg / 2 mL (6 mg/mL)
Route
Subcutaneous, once weekly
Area
Abdominal wall only
Rotation
Right and left upper and lower quadrants
Equipment
Provided supplies; a new syringe each time
Training and observation were part of the intervention.
Site staff demonstrated vial-and-syringe preparation and injection, observed the first and second injections, reviewed technique at office visits, and could schedule extra training.
Not present in the source packet
What these PDFs cannot establish
- A BAC-water volume for a 10, 15, or 20 mg powder vial
- Vial dimensions, usable capacity, headspace, or a safe fill limit
- Needle gauge or length, syringe capacity, insertion angle, or a fill line
- A U-100 insulin-syringe conversion or individualized draw amount
- A 30-minute room-temperature wait after mixing
- Thigh or upper-arm injection instructions
- Dose selection from height, current weight, ideal weight, or pounds to lose
- An equation predicting how many weeks one person will need to reach a goal weight
A 10, 15, or 20 mg lyophilized research vial is not interchangeable with the protocol's supplied 12 mg / 2 mL liquid study vial.
Protocol pp. 151–153 and 166–168. Full-document searches found no bacteriostatic water, reconstitution, needle specification, U-100 conversion, or post-mixing wait instruction.
Efficacy + analysis
The headline changes when the estimand changes
The primary efficacy estimand asks what average effect would be expected if trial-eligible participants remained on assigned treatment. The hybrid analysis uses different discontinuation assumptions and gives more conservative estimates. Both are population estimates.
Mean body-weight change by randomized group
Efficacy estimand versus hybrid sensitivity analysis
| Randomized group | N | Week 24 efficacy | Week 48 efficacy | Week 24 hybrid | Week 48 hybrid |
|---|---|---|---|---|---|
| Placebo | 70 | −1.6% | −2.1% | −1.8% | −2.6% |
| 1 mg · fixed | 69 | −7.2% | −8.7% | −6.9% | −8.1% |
| 4 mg · initial 2 mg | 33 | −11.8% | −16.3% | −10.7% | −14.7% |
| 4 mg · initial 4 mg | 34 | −13.9% | −17.8% | −13.8% | −15.7% |
| 8 mg · initial 2 mg | 35 | −16.7% | −21.7% | −14.7% | −18.1% |
| 8 mg · initial 4 mg | 35 | −17.9% | −23.9% | −16.2% | −20.5% |
| 12 mg · initial 2 mg | 62 | −17.5% | −24.2% | −16.2% | −20.9% |
Primary endpoint
Percent body-weight change at week 24. Week 48, responder thresholds, BMI, waist, and other outcomes were secondary or exploratory.
No multiplicity adjustment
The four primary dose comparisons used two-sided 0.05 tests without adjustment; secondary measures were not controlled for type I error.
Lifestyle was co-administered
Standardized diet and activity counseling means an individual time-to-goal cannot be inferred from dose and baseline weight alone.
Prespecified exploratory measures
Metabolic and cardiovascular signals
Useful for hypothesis generation; not proof of reduced cardiovascular, kidney, liver, or mortality events.
| Measure | Placebo | Range across active groups |
|---|---|---|
| Fasting glucose · week 48 | +3.1 mg/dL | −2.2 to −10.6 mg/dL |
| Fasting insulin · week 48 | −1.3 mU/L | −3.3 to −8.9 mU/L |
| HbA1c · week 48 | 0.0 percentage points | −0.2 to −0.5 percentage points |
| Triglycerides · week 48 | +1.4% | −17.9% to −43.6% |
| LDL cholesterol · week 48 | −0.3% | −4.7% to −21.7% |
| 24-hour systolic BP · week 36 | −2.9 mm Hg | −4.8 to −11.8 mm Hg |
| 24-hour heart rate · week 36 | +0.9 bpm | +2.6 to +6.7 bpm |
Supplementary Appendix Tables S4–S8 and Figure S3, pp. 16 and 29–33; protocol pp. 195–200; final SAP pp. 306–343.
Safety + tolerability
The trial was monitored far beyond a symptom checklist
Safety conclusions depend on structured surveillance, adjudication, laboratory testing, predefined interruption and reduction rules, and continued follow-up. That infrastructure is part of the evidence.
Protocol surveillance
What researchers watched
- Adverse events, serious adverse events, deaths, and treatment discontinuations
- Sitting and orthostatic vital signs, 24-hour ambulatory blood pressure, and ECGs
- Pancreatic enzymes and adjudication of suspected pancreatitis
- Hepatic tests, gallbladder events, renal function, hydration, and acute kidney events
- Calcitonin, thyroid C-cell events, hypersensitivity, and injection-site reactions
- Antidrug antibodies, pharmacokinetics, glucose, lipids, and mechanistic biomarkers
- PHQ-9, Columbia Suicide Severity Rating Scale, and self-harm monitoring
- Pregnancy testing, contraception requirements, and pregnancy follow-up
Heart-rate signal
+1.7 to +6.0 bpm
Week-48 pulse change across active groups versus +0.4 bpm with placebo.
Treatment persistence
64.5%–84.8%
On target dose at treatment end across active groups; discontinuation ranged from 12.1% to 25.8%.
5.6%
Arrhythmia-related AEs
19 of 337 participants; 21 events, one severe, and two treatment discontinuations.
5.9%
Hyperesthesia-related AEs
20 of 337 participants; 16 resolved, two were resolving, and two remained unresolved at four-week follow-up.
Enzymes
Pancreatic monitoring
Mean amylase and lipase increased in active groups; one acute pancreatitis SAE was adjudicated in the 12 mg arm while the participant was at 2 mg.
Serious-event detail reported in the appendix+
8 mg arm
Intractable vomiting and acute cholecystitis were judged related by the investigator; laparoscopic cholecystectomy was performed.
12 mg arm
A prolonged-QT serious event occurred in a cascade involving severe vomiting, hypokalemia, dehydration, hypotension, and liver injury; it resolved after withdrawal.
12 mg arm · at 2 mg
An acute pancreatitis event on day 14 was judged related and adjudicated by the clinical endpoint committee.
Context
The table also includes serious events judged unrelated and one fatal drowning judged unrelated. Small groups cannot estimate rare-event risk reliably.
Protocol pp. 174–193 and 204–205; Supplementary Appendix Tables S10–S15, pp. 35–41. Adverse-event labels do not prove causation unless attribution or adjudication is stated.
Research integrity
Analysis choices, disclosures, and access terms belong in the evidence
A trustworthy evidence page shows not only the favorable endpoint, but also how missing data were handled, who sponsored the work, what changed between documents, and when participant-level data may become available.
Estimands + missing data
The efficacy estimand excluded post-discontinuation data and used a mixed model under missing-at-random assumptions. The hybrid estimand used reason-specific handling, including reference-based imputation for some discontinuations.
Protocol/SAP change
The amended protocol said no interim analyses were planned. Final SAP v4 documented three later-planned safety and efficacy interims reviewed by an unblinded assessment committee.
Conditional data access
The 2023 statement makes anonymized data access conditional on obesity approval in the U.S. and EU, followed by a six-month delay, independent review, a signed agreement, and Vivli's secure environment.
Author relationships
Disclosures in plain language
- Seven of ten listed authors disclosed Eli Lilly employment; several also disclosed Lilly stock or stock options.
- Other authors disclosed Lilly research support to their institutions and/or consulting or scientific-advisory relationships.
- The sponsor supplied the study intervention and was responsible for the statistical analysis, directly or through its designee.
- These relationships do not erase randomized results, but belong beside methods, uncertainty, and independent replication.
Data-sharing terms
What may be requested later
- Anonymized individual participant data, excluding pharmacokinetic and genetic data
- Protocol, SAP, clinical study report, and blank or annotated case-report forms in the secure environment
- Access only after independent committee approval and a signed data-sharing agreement
- A secure workspace for up to two years per approved proposal
- No fixed expiration was stated once the data become available
Protocol p. 200; final SAP pp. 315–342; Author Disclosures pp. 1–16; Data-Sharing Statement pp. 1–2, posted June 26, 2023.
Bottom line
More evidence, fewer assumptions.
These documents strengthen the trial-design, eligibility, outcome, safety, and governance record. They do not validate a public BAC-water recipe, needle guide, syringe-unit calculator, weight-based dose, or personal time-to-goal prediction.
Supported as trial facts
Design, population, assigned arms, study-supplied vial format, abdominal trial rotation, monitoring, endpoints, results, limitations, and disclosures.
Not supported as public instructions
Reconstitution volume, consumer needle or syringe directions, extra injection sites, personalized escalation, and predicted individual duration.
Research interpretation only. The supplied 2023 source packet describes an investigational study; this page does not provide medical advice, prescribing, preparation, or self-administration directions.