This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record.
1. Trial at a Glance
AMPLITUDE-O was a randomized, parallel-group, quadruple-masked phase 3 trial evaluating efpeglenatide in participants with type 2 diabetes mellitus. The primary endpoint was a time-to-event cardiovascular composite analyzed for non-inferiority, with a secondary superiority analysis of the same composite.
| Feature | AMPLITUDE-O |
|---|---|
| Phase | Phase 3 |
| Condition | Type 2 Diabetes Mellitus |
| Therapeutic area | Cardiovascular |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | Quadruple |
| Primary purpose | Treatment |
| Enrollment | 4,076 |
| Interventions | Efpeglenatide (SAR439977); placebo |
| Primary endpoint type | Time-to-event |
| Primary method | Log-rank test, with hazard ratio estimated from a Cox proportional hazards model |
| Hypothesis types | Non-inferiority; superiority |
| ClinicalTrials.gov | NCT03496298 |
| Lead sponsor | Sanofi |
2. Clinical Question
The central statistical question was whether pooled efpeglenatide treatment at 4 mg and 6 mg could be compared with placebo for the time to first occurrence of major adverse cardiovascular events, using a prespecified non-inferiority framework and subsequently a superiority analysis.
Population
Participants with type 2 diabetes mellitus enrolled in the randomized phase 3 AMPLITUDE-O trial.
Intervention
Efpeglenatide at 4 mg or 6 mg. The primary analysis pooled the two efpeglenatide doses.
Comparator
Placebo.
Primary question
Does pooled efpeglenatide meet the prespecified non-inferiority criterion for first occurrence of MACE compared with placebo?
3. Trial Design
Efpeglenatide 4 mg
- Efpeglenatide (SAR439977)
- One of the two efpeglenatide dose groups
- Included in the pooled 4 mg + 6 mg efficacy comparison
Efpeglenatide 6 mg
- Efpeglenatide (SAR439977)
- One of the two efpeglenatide dose groups
- Included in the pooled 4 mg + 6 mg efficacy comparison
Placebo
- Placebo
- Comparator arm for the primary pooled analysis
Efpeglenatide 4 mg + 6 mg vs placebo
- Primary efficacy comparison
- Analysis performed in the intent-to-treat population
- Pooled efpeglenatide population was prespecified in the protocol
4. Trial Timeline
Study start
The trial began on April 27, 2018.
Primary completion
The registry lists December 10, 2020 as the primary completion date.
Up to 31.5 months
The primary MACE endpoint was assessed from Day 1 until the first adjudicated and confirmed major cardiovascular event, with a maximum duration of up to 31.5 months.
5. Primary Endpoint
| Endpoint | Definition and time frame | Analysis |
|---|---|---|
| Time to first occurrence of MACE | Time to first adjudicated and confirmed occurrence of a major cardiovascular event from Day 1, with a maximum duration of up to 31.5 months. | Event rate per 100 participant-years; Kaplan-Meier curves; log-rank comparison; hazard ratio from a Cox proportional hazards model. |
The registered composite endpoint was the first occurrence of CV death, non-fatal myocardial infarction, or non-fatal stroke. All MACE events positively adjudicated by the clinical endpoint committee were used in the composite analysis.
The registry describes the event rate as the number of events per 100 participant-years, calculated from the number of participants with events and the sum of time at risk across participants.
6. Secondary Endpoints
| Endpoint | Time frame | Hypothesis | Effect measure |
|---|---|---|---|
| First occurrence of MACE | From Day 1 until first adjudicated and confirmed major CV event; maximum duration up to 31.5 months | Superiority | Hazard ratio |
| First occurrence of expanded MACE composite events | From Day 1 until first adjudicated and confirmed major CV event; maximum duration up to 31.5 months | Superiority | Hazard ratio |
| First occurrence of composite renal endpoint | From Day 1 until confirmed occurrence of composite renal endpoint; maximum duration up to 31.5 months | Superiority | Hazard ratio |
7. Statistical Methodology
Intention-to-treat analysis
The primary and secondary analyses were performed in the intent-to-treat (ITT) population. The analysis compared participants according to their randomized treatment assignment, with the efpeglenatide 4 mg and 6 mg groups pooled for the prespecified efficacy comparison against placebo.
The ITT principle is particularly important in randomized trials because maintaining participants in their randomized groups preserves the treatment comparison created by randomization. It also means that the resulting estimate describes the effect associated with assignment to the randomized strategy rather than simply the effect among participants who remained continuously exposed to treatment.
Kaplan-Meier estimation
The primary endpoint was a time-to-event outcome. Kaplan-Meier curves were used to depict the cumulative event experience over time by treatment group.
where di is the number of events at time ti and ni is the number at risk immediately before that time.
For a time-to-event endpoint, this approach allows participants who have not experienced the event by the end of their observed follow-up to contribute information up to their censoring time rather than requiring every participant to have an event.
Log-rank test
The registered statistical method was the log-rank test. The log-rank framework compares the observed and expected numbers of events across treatment groups over the observed follow-up and is commonly paired with Kaplan-Meier estimation for randomized time-to-event comparisons.
Cox proportional hazards model
The hazard ratio and its 95% confidence interval were obtained from a Cox proportional hazards model. The model included region, randomization stratum of SGLT2 inhibitor use, and treatment as fixed-effect factors.
Region was categorized as North America, Latin America, Europe, or other. The SGLT2 inhibitor-use stratum was current use, potential future use, or neither current nor potential future use.
Event rates per 100 participant-years
The registry also reports the endpoint in terms of an event rate per 100 participant-years. This expresses the number of events relative to accumulated time at risk and is different from a simple proportion of participants who experience an event.
The registry defines the reported event-rate measure using the number of participants with events and the summed time at risk across participants.
8. Non-Inferiority Framework
The primary analysis was explicitly a non-inferiority analysis. The registry states that non-inferiority of efpeglenatide 4 mg + 6 mg versus placebo was to be claimed if the upper bound of the two-sided 95% confidence interval for the hazard ratio met the prespecified criterion described in the record.
Primary non-inferiority criterion
The registry wording states that non-inferiority was to be claimed if the upper bound of the two-sided 95% CI of the hazard ratio for the incidence met the stated 1.8 and 1.3 criteria.
Non-inferiority reasoning differs from a conventional superiority test. The question is not simply whether the observed hazard ratio differs from 1. Instead, the confidence interval is compared with a prespecified margin that represents the largest effect compatible with the non-inferiority objective.
9. Primary Result: First Occurrence of MACE
The primary analysis compared pooled efpeglenatide 4 mg + 6 mg with placebo in the ITT population. The endpoint was time to first adjudicated and confirmed occurrence of CV death, non-fatal MI, or non-fatal stroke.
Primary hazard ratio
95% CI: 0.583–0.918 · P < .0001
Efpeglenatide 4 mg + 6 mg vs placebo
| Element | Primary analysis |
|---|---|
| Analysis population | Intent-to-treat population |
| Comparison | Efpeglenatide 4 mg + 6 mg vs placebo |
| Endpoint | Time to first occurrence of MACE |
| Effect measure | Hazard ratio |
| Estimate | 0.732 |
| 95% CI | 0.583–0.918 |
| P-value | <.0001 |
| Method | Log-rank test; Cox proportional hazards model for HR and CI |
| Hypothesis | Non-inferiority |
The hazard ratio of 0.732 indicates that, under the fitted Cox model, the estimated instantaneous rate of first MACE was approximately 73.2% of the corresponding rate in the placebo group over the analyzed follow-up. Expressed as a simple relative interpretation, this corresponds to an estimated hazard about 26.8% lower for pooled efpeglenatide than placebo.
The hazard ratio does not mean that 26.8% of participants avoided MACE, nor does it represent an absolute risk reduction. It is a relative time-to-event measure based on the fitted model.
The 95% confidence interval of 0.583–0.918 describes uncertainty around the estimated hazard ratio. It does not describe the range of individual treatment effects among participants.
The p-value of <.0001 summarizes the evidence under the statistical hypothesis associated with the reported test. It does not measure the size or clinical importance of the treatment effect.
For the non-inferiority question, the key statistical feature is the upper confidence bound relative to the prespecified non-inferiority criterion. The model also depends on the time-to-event framework and its assumptions, including the proportional-hazards structure underlying the Cox hazard ratio. Censoring and the ITT population are therefore important parts of interpretation.
10. Secondary Result: Superiority Analysis of MACE
The same pooled efpeglenatide 4 mg + 6 mg versus placebo comparison was evaluated for superiority as a secondary analysis.
Superiority hazard ratio
95% CI: 0.583–0.918 · P = 0.0069
Efpeglenatide 4 mg + 6 mg vs placebo
| Element | Superiority analysis |
|---|---|
| Analysis population | Intent-to-treat population |
| Comparison | Efpeglenatide 4 mg + 6 mg vs placebo |
| Endpoint | Time to first occurrence of MACE |
| Effect measure | Hazard ratio |
| Estimate | 0.732 |
| 95% CI | 0.583–0.918 |
| P-value | 0.0069 |
| Method | Log-rank test; Cox proportional hazards model for HR and CI |
| Hypothesis | Superiority |
The superiority analysis uses the same estimated hazard ratio, 0.732, with a two-sided 95% confidence interval of 0.583–0.918. The registry states that superiority was claimed when the upper bound of the two-sided 95% confidence interval was less than 1.
The estimate therefore describes a lower estimated hazard of first MACE for pooled efpeglenatide relative to placebo under the fitted model. It does not establish that every participant had a lower event risk, nor does it translate directly into an absolute reduction in the probability of MACE.
The confidence interval provides the principal measure of statistical precision around the hazard-ratio estimate. The p-value of 0.0069 summarizes the evidence for the superiority hypothesis; it is not a measure of treatment-effect magnitude.
Because the endpoint is time-to-event, interpretation also depends on censoring and the Cox-model framework. The analysis population is ITT, and the comparison is based on the prespecified pooled efpeglenatide population rather than a separate estimate for each efpeglenatide dose.
11. Secondary Result: Expanded MACE Composite
The registry also reports a superiority analysis for the time to first occurrence of the expanded major adverse cardiovascular events composite.
Expanded MACE hazard ratio
95% CI: 0.65–0.96 · P = 0.02
Efpeglenatide 4 mg + 6 mg vs placebo
| Element | Expanded MACE analysis |
|---|---|
| Analysis population | Intent-to-treat population |
| Comparison | Efpeglenatide 4 mg + 6 mg vs placebo |
| Endpoint | Time to first occurrence of expanded MACE composite events |
| Effect measure | Hazard ratio |
| Estimate | 0.79 |
| 95% CI | 0.65–0.96 |
| P-value | 0.02 |
| Method | Log-rank test; Cox proportional hazards model for HR and CI |
| Hypothesis | Superiority |
The registry states that superiority was claimed when the upper bound of the two-sided 95% confidence interval for the hazard ratio was less than 1.
12. Secondary Result: Composite Renal Endpoint
The registry reports a superiority analysis for the time to first occurrence of the composite renal endpoint.
Renal endpoint hazard ratio
95% CI: 0.574–0.794 · P < .0001
Efpeglenatide 4 mg + 6 mg vs placebo
| Element | Composite renal analysis |
|---|---|
| Analysis population | Intent-to-treat population |
| Comparison | Efpeglenatide 4 mg + 6 mg vs placebo |
| Endpoint | Time to first occurrence of composite renal endpoint |
| Effect measure | Hazard ratio |
| Estimate | 0.675 |
| 95% CI | 0.574–0.794 |
| P-value | <.0001 |
| Method | Log-rank test; Cox proportional hazards model for HR and CI |
| Hypothesis | Superiority |
The registry again states that superiority was claimed when the upper bound of the two-sided 95% confidence interval for the hazard ratio was less than 1.
13. Comparing the Reported Time-to-Event Estimates
| Endpoint | HR | 95% CI | P-value | Hypothesis |
|---|---|---|---|---|
| First MACE — non-inferiority | 0.732 | 0.583–0.918 | <.0001 | Non-inferiority |
| First MACE — superiority | 0.732 | 0.583–0.918 | 0.0069 | Superiority |
| Expanded MACE | 0.79 | 0.65–0.96 | 0.02 | Superiority |
| Composite renal endpoint | 0.675 | 0.574–0.794 | <.0001 | Superiority |
The estimates are all below 1, but they should not be treated as interchangeable measures. Each corresponds to a different endpoint definition and hypothesis. The primary MACE analysis also has both a non-inferiority and a superiority interpretation, whereas the expanded MACE and renal analyses are reported as superiority analyses.
The chart is a visual representation of the reported hazard-ratio estimates; it is not a reconstructed survival curve and does not display absolute event probabilities.
14. Statistical Methods Explained
Why was a log-rank test used?
The primary and secondary outcomes are time-to-event endpoints. A log-rank test is designed to compare event-time distributions between randomized treatment groups while accounting for the timing of events and censoring. It therefore uses more information than a simple comparison of the proportion of participants who experienced an event by a fixed date.
What does a hazard ratio of 0.732 mean?
A hazard ratio of 0.732 means that the estimated instantaneous event rate under the fitted model was 0.732 times the corresponding rate in the comparator group. A convenient descriptive translation is a 26.8% lower estimated hazard, but this should not be mistaken for a 26.8% absolute reduction in the probability of MACE.
Why is the confidence interval central to the non-inferiority analysis?
Non-inferiority is defined relative to a prespecified margin. The confidence interval shows the range of effect estimates compatible with the statistical model and data at the stated confidence level. The upper confidence bound is therefore directly relevant to determining whether the treatment remains within the allowed non-inferiority boundary.
Why are there two analyses of the same MACE endpoint?
The registry reports both a non-inferiority analysis and a superiority analysis. These are different hypotheses even though they use the same estimated hazard ratio and confidence interval. Non-inferiority asks whether the treatment is not unacceptably worse according to the prespecified criterion; superiority asks whether the treatment effect is statistically below the reference value of 1.
Why was the pooled 4 mg + 6 mg population used?
The registry states that the primary and secondary efficacy analyses were planned for the pooled population of participants who received efpeglenatide at either 4 mg or 6 mg. Consequently, the reported hazard ratios describe the prespecified pooled comparison rather than separate primary estimates for the two efpeglenatide doses.
What is the role of the Cox proportional hazards model?
The Cox model provides the hazard-ratio estimate and its confidence interval while incorporating the treatment factor together with the reported region and SGLT2 inhibitor-use randomization strata. Its proportional-hazards structure means that interpretation of a single HR should remain connected to the model assumptions and the observed follow-up.
15. Intention-to-Treat Analysis
All four reported statistical analyses were conducted in the ITT population. The registry specifically states that the data were planned to be collected and analyzed for the pooled efpeglenatide population as prespecified in the protocol.
Why ITT matters
Analysis according to randomized assignment preserves the comparison created by randomization and avoids redefining treatment groups based on later events or exposure.
What ITT does not guarantee
ITT does not eliminate every source of bias. Interpretation still depends on follow-up, censoring, adherence, subsequent treatment, and the prespecified analysis model.
For this trial, ITT is especially relevant because the endpoint is a time-to-event outcome. Participants can contribute information for different lengths of time, and the analysis incorporates their observed time at risk.
16. Stratification and Model Adjustment
The Cox model used for the hazard-ratio estimates included three classes of fixed-effect factors: region, the randomization stratum for SGLT2 inhibitor use, and treatment.
| Model factor | Registry categories |
|---|---|
| Region | North America; Latin America; Europe; other |
| SGLT2 inhibitor-use randomization stratum | Current use; potential future use; neither current nor potential future use |
| Treatment | Efpeglenatide 4 mg; efpeglenatide 6 mg; placebo |
The inclusion of these factors means that the reported hazard ratio is a model-based treatment estimate rather than a simple ratio calculated from two crude event proportions.
17. Serious Adverse Events
The registry reports serious adverse events by randomized arm using the number affected and the number at risk.
| Arm | Affected | At risk | Reported affected / at risk |
|---|---|---|---|
| Placebo | 298 | 1,355 | 298/1355 |
| Efpeglenatide 4 mg | 312 | 1,360 | 312/1360 |
| Efpeglenatide 6 mg | 275 | 1,358 | 275/1358 |
These are reported as serious adverse events by arm. They should be considered separately from the time-to-event efficacy analyses because safety summaries and efficacy analyses answer different questions and may use different analysis conventions.
18. Statistical Interpretation of the Primary Analysis
The primary hazard ratio of 0.732 indicates a lower estimated hazard of first MACE for pooled efpeglenatide relative to placebo under the fitted Cox model. The corresponding relative-hazard interpretation is approximately 26.8% lower estimated hazard.
The two-sided 95% confidence interval extends from 0.583 to 0.918. The interval communicates uncertainty around the hazard-ratio estimate and is the critical quantity for the registry's non-inferiority framework.
The reported non-inferiority p-value is <.0001, while the superiority analysis reports 0.0069. A p-value describes evidence against a specified null hypothesis; it does not quantify how large or clinically important an effect is.
The hazard ratio is not an absolute risk measure. It does not tell the reader the percentage of participants who experienced MACE, nor does it provide the absolute difference in event probability at a particular time point.
The hazard ratio comes from a Cox proportional hazards model with region, SGLT2 inhibitor-use stratum, and treatment as fixed-effect factors. The interpretation therefore depends on the time-to-event framework, censoring, and the proportional-hazards structure of the model.
19. Why Non-Inferiority Is Different From Superiority
| Question | Non-inferiority | Superiority |
|---|---|---|
| Core question | Is the treatment effect sufficiently close to the comparator that the prespecified non-inferiority criterion is met? | Is the treatment-associated hazard statistically below the comparator hazard? |
| Key reference | Prespecified non-inferiority margin | Hazard ratio of 1 |
| Reported AMPLITUDE-O criterion | Upper bound of the two-sided 95% CI evaluated against the stated 1.8 and 1.3 criteria | Upper bound of the two-sided 95% CI less than 1 |
| Primary role in the registry | Primary MACE analysis | Secondary MACE analysis |
The same estimated hazard ratio can therefore be considered under two different statistical hypotheses. The interpretation depends not only on the point estimate but also on the confidence interval and the prespecified decision rule.
20. Multiplicity and Multiple Endpoints
The registry reports one primary endpoint and three additional statistical analyses identified as secondary. The primary endpoint was analyzed under both non-inferiority and superiority hypotheses, while the expanded MACE and composite renal endpoint were analyzed for superiority.
| Endpoint | Role | Hypothesis |
|---|---|---|
| First occurrence of MACE | Primary | Non-inferiority |
| First occurrence of MACE | Secondary analysis | Superiority |
| Expanded MACE composite | Secondary | Superiority |
| Composite renal endpoint | Secondary | Superiority |
Multiple endpoint analyses raise an important statistical issue: each additional hypothesis test creates another opportunity for a nominally significant finding. The interpretation of secondary p-values therefore depends on the prespecified multiplicity strategy. The registry data identify the hypothesis type for each analysis but do not provide a separate multiplicity procedure in the information reported here.
21. Censoring and Time-to-Event Interpretation
Time-to-event analyses depend on information collected over varying amounts of follow-up. Participants who do not experience the endpoint during their observed period can remain informative through their censoring time.
This is one reason a hazard ratio cannot be interpreted as if it were simply the ratio of two percentages. The Cox model uses the timing of events and the risk sets formed during follow-up. The Kaplan-Meier approach similarly uses the observed event and censoring times to estimate the event-time distribution.
Censoring
A participant without an observed event can still contribute information until the end of the period for which the participant is known to be event-free.
Risk sets
The Cox model compares event rates using the participants who remain under observation and at risk at relevant event times.
22. What the Hazard Ratio Does — and Does Not — Mean
Under the fitted model, the estimated instantaneous rate of first MACE associated with pooled efpeglenatide was 0.732 times that of placebo. A descriptive translation is approximately a 26.8% lower estimated hazard.
It does not mean that 26.8% of participants avoided MACE, that the absolute probability of MACE was reduced by 26.8 percentage points, or that each individual participant experienced the same proportional reduction in risk.
The 95% CI of 0.583–0.918 communicates the statistical uncertainty surrounding the estimated hazard ratio. It is not an interval containing the effects experienced by individual participants.
Hazard ratios are relative measures. A complete clinical interpretation would ordinarily consider absolute event probabilities or event rates alongside the relative treatment effect. The registry information reported here provides the event-rate framework but does not provide arm-specific event-rate estimates in the posted analyses summarized above.
23. Important Limitations and Interpretation Issues
- Early termination: the registry states that the study was terminated early by the Sponsor, but not due to any safety concerns. Early termination can affect the amount of information accumulated for statistical estimation.
- Time-to-event interpretation: the primary analysis is based on event timing and censoring rather than a simple fixed-time proportion.
- Hazard-ratio assumptions: the Cox proportional hazards model provides a single relative hazard estimate under its modeling framework. Interpretation should therefore remain connected to the proportional-hazards assumption.
- Non-inferiority margin: the non-inferiority conclusion depends on the prespecified margin, not merely on whether the p-value is small.
- Pooled treatment comparison: the primary and secondary analyses pool the 4 mg and 6 mg efpeglenatide groups. The reported hazard ratios therefore do not represent separate primary estimates for each dose.
- Multiple analyses: the registry reports a primary endpoint and several secondary analyses. Interpretation of the secondary p-values depends on the prespecified multiplicity strategy.
- Registry scope: the ClinicalTrials.gov record provides the registered endpoint definitions and posted statistical analyses. Additional statistical detail is not included unless reported in the record.
24. Why This Trial Matters Statistically
AMPLITUDE-O is a useful teaching case because it combines randomized treatment assignment with a prespecified non-inferiority framework, subsequent superiority testing, adjudicated cardiovascular time-to-event outcomes, pooled treatment groups, and Cox-model hazard ratios.
| Concept | How it appears in AMPLITUDE-O |
|---|---|
| Randomization | Randomized, parallel-group phase 3 design with 3 arms. |
| Blinding | Quadruple masking. |
| Intention-to-treat | Primary and secondary analyses performed in the ITT population. |
| Time-to-event endpoint | Time to first adjudicated and confirmed cardiovascular or renal event. |
| Kaplan-Meier estimation | Used to depict cumulative event rates over time for MACE. |
| Log-rank test | Registry-listed method for the time-to-event comparisons. |
| Hazard ratio | Primary effect measure for the reported analyses. |
| Cox model | Used to obtain hazard ratios and 95% confidence intervals. |
| Non-inferiority | Primary MACE hypothesis evaluated against a prespecified criterion. |
| Superiority | Reported for MACE, expanded MACE, and the composite renal endpoint. |
| Stratification | Model included region and the SGLT2 inhibitor-use randomization stratum. |
| Safety analysis | Serious adverse events reported separately by treatment arm. |
25. Planned Statistical Logic for a Time-to-Event Non-Inferiority Trial
The statistical logic of the primary analysis can be summarized as a sequence of linked decisions rather than as a single p-value.
This structure illustrates why non-inferiority analysis cannot be reduced to the statement that a treatment had a statistically significant p-value. The point estimate, confidence interval, prespecified margin, analysis population, censoring, and model assumptions all contribute to the interpretation.
26. Related Tutorials
Learn more about the methods used in this trial:
27. Related Statistical Calculators
28. Sources
- ClinicalTrials.gov: NCT03496298.
- PubMed: PMID 34215025.
- PubMed: PMID 39963952.
- PubMed: PMID 36802715.
- PubMed: PMID 34775781.
- PubMed: PMID 33026143.
Continue through the Clinical Biostats knowledge graph
Connect randomized trial design, time-to-event endpoints, non-inferiority methods, and survival-analysis concepts to deeper statistical tutorials and calculators.
29. Record Summary
AMPLITUDE-O provides a compact example of how a randomized clinical trial can use a time-to-event endpoint for both non-inferiority and superiority questions. The primary MACE analysis used the ITT population and compared pooled efpeglenatide 4 mg + 6 mg with placebo. The reported hazard ratio was 0.732, with a two-sided 95% CI of 0.583–0.918. The same estimated hazard ratio was reported for the secondary superiority analysis, with a p-value of 0.0069. Secondary analyses reported hazard ratios of 0.79 for the expanded MACE composite and 0.675 for the composite renal endpoint.
Statistically, the important lesson is that these numbers cannot be interpreted independently of their endpoint definitions and hypotheses. The hazard ratio provides a relative time-to-event measure, the confidence interval describes statistical precision, the p-value addresses a specified hypothesis, and the non-inferiority framework requires comparison with a prespecified margin. Together, these elements provide a more complete description of the evidence than any single statistic alone.