This page separates reported registry information from statistical interpretation. The ClinicalTrials.gov record contains a registered primary time-to-event endpoint and selected safety information, but no formal statistical-analysis results.
1. Trial at a Glance
ZENITH was a completed, randomized, parallel, triple-masked phase 3 trial in pulmonary arterial hypertension. The ClinicalTrials.gov record describes 173 enrolled participants assigned to two intervention arms: sotatercept and placebo. Its single registered primary endpoint was time to first confirmed morbidity or mortality event.
| Feature | ZENITH |
|---|---|
| Trial name | ZENITH |
| NCT identifier | NCT04896008 |
| Phase | Phase 3 |
| Status | Completed |
| Therapeutic area | Pulmonology |
| Condition | Pulmonary Arterial Hypertension |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | Triple |
| Primary purpose | Treatment |
| Enrollment | 173 |
| Interventions | Sotatercept; Placebo |
| Primary endpoint type | Time-to-event |
| Results posted | Yes |
| Outcome measures posted | 13 |
2. Clinical Question
The ClinicalTrials.gov record evaluates sotatercept against placebo in participants with pulmonary arterial hypertension who were described in the brief title as WHO FC III or FC IV at high risk of mortality. The primary statistical question is whether the randomized treatment groups differ in the time until the first confirmed morbidity or mortality event.
Population
Participants with pulmonary arterial hypertension who were WHO FC III or FC IV at high risk of mortality.
Intervention
Sotatercept.
Comparator
Placebo.
Primary question
Does randomized assignment to sotatercept versus placebo change the time to the first confirmed morbidity or mortality event?
3. Trial Design
Sotatercept
- Intervention: sotatercept
- Randomized trial arm
- Included in the primary time-to-event comparison
Placebo
- Comparator: placebo
- Randomized trial arm
- Included in the primary time-to-event comparison
The ClinicalTrials.gov record does not provide an allocation ratio between the two arms. They also do not provide arm-specific enrollment counts in the general ClinicalTrials.gov record. Therefore, no allocation ratio or arm-specific randomized sample size is inferred here.
4. Trial Timeline
Trial start
The ClinicalTrials.gov record lists the trial start date as 2021-12-01.
Primary completion
The ClinicalTrials.gov record lists the primary completion date as 2024-07-26.
Completed
the ClinicalTrials.gov record identifies ZENITH as completed and states that results have been posted.
5. Primary Endpoint
| Endpoint | Registered definition | Time frame | Type |
|---|---|---|---|
| Time to First Confirmed Morbidity or Mortality Event | Morbidity or mortality events were defined as all-cause death, lung transplantation, or PAH worsening-related hospitalization of ≥24 hours. All events were adjudicated by a blinded, independent committee of clinical experts. Only adjudication-confirmed lung transplantation and PAH worsening-related hospitalization of ≥24 hours were included in the primary analysis. All deaths that are a first event... | Up to approximately 31 months | Time-to-event |
The endpoint is a composite time-to-first-event outcome. Conceptually, a participant contributes follow-up until the first qualifying morbidity or mortality event, subject to the trial's prespecified event definitions and censoring rules. Because the ClinicalTrials.gov record does not provide the complete endpoint text or censoring rules, those details should not be inferred.
6. Planned Analysis
The ClinicalTrials.gov record identifies the primary endpoint as a time-to-event measure but do not provide any entries under the posted statistical analyses. Consequently, there is no formal treatment-effect estimate, confidence interval, or p-value to report for the primary endpoint from the ClinicalTrials.gov record.
Primary analysis status
The registry data identify the endpoint and its time frame, but the registry-reported statistical-analysis section contains no analysis result.
For a randomized two-arm trial with a time-to-first-event primary endpoint, a conventional analysis pathway would typically describe the event-time distribution using Kaplan-Meier methods and compare treatment groups with a time-to-event hypothesis test such as a log-rank test. A Cox proportional-hazards model is also commonly used to estimate a hazard ratio and its confidence interval. These are general statistical methods for this endpoint type, not reported ZENITH analysis results.
The important distinction is between what the endpoint is and what the trial observed. The ClinicalTrials.gov record establishes that ZENITH was designed around time to the first confirmed morbidity or mortality event, but they do not supply a numerical treatment effect.
A Kaplan-Meier curve, if available from a complete results source, would describe the probability of remaining free of the first event over time. A hazard ratio would instead summarize a relative difference in the instantaneous event rate under a fitted model. Neither quantity is present in the ClinicalTrials.gov record.
7. How a Time-to-Event Analysis Works
Time-to-event analysis is appropriate when both whether an event occurs and when it occurs matter. Participants who have not experienced the event by the end of available follow-up are commonly censored rather than treated as though they experienced the event at the end of follow-up.
Here, di represents the number of events at event time ti, while ni is the number at risk immediately before that time. This is a general explanation of Kaplan-Meier estimation, not a reconstruction of ZENITH data.
The ZENITH endpoint is more specific than a generic mortality endpoint because the first event can arise from several clinically defined components. Composite endpoints require careful interpretation because a treatment effect on one component does not necessarily imply the same effect on every component.
8. Statistical Methodology
Randomization
Randomization assigns participants to the sotatercept or placebo group according to the trial's allocation procedure. Its principal statistical role is to make the randomized groups comparable in expectation with respect to measured and unmeasured prognostic factors, supporting a causal comparison of assigned treatment.
Blinding
The ClinicalTrials.gov record classifies ZENITH as triple-masked. Masking can reduce the potential for knowledge of treatment assignment to influence treatment delivery, participant behavior, outcome assessment, or other aspects of trial conduct. The exact identities of the three masked parties are not provided in the ClinicalTrials.gov record.
Kaplan-Meier estimation
For a time-to-first-event endpoint, Kaplan-Meier estimation provides a way to describe the event-free probability over time while accounting for censoring. A curve falling more rapidly indicates that events are accumulating earlier, although visual separation alone is not a formal statistical test.
Log-rank comparison
A log-rank test is commonly used to compare event-time distributions between randomized groups. Its use would address whether the observed survival or event-free curves differ under the test's assumptions. The registry-reported ZENITH data do not state that a log-rank test was used, so it is presented here only as the typical analysis for this endpoint type.
Cox proportional-hazards model
A Cox model is commonly used to estimate a hazard ratio comparing two treatment groups. A hazard ratio below 1 would conventionally indicate a lower estimated instantaneous event rate in the treatment group, while a value above 1 would indicate a higher estimated instantaneous event rate. The ClinicalTrials.gov record contains no ZENITH hazard ratio.
This ratio is a model-based relative measure. It is not equivalent to a difference in the percentages of participants who experience an event, and it does not by itself state the absolute probability of an event by a particular time.
Confidence intervals
If a hazard ratio were reported, its confidence interval would describe statistical uncertainty around the estimated relative effect under the specified model and sampling framework. A narrow interval generally indicates greater precision than a wide interval. The interval would not describe the range of effects experienced by individual participants.
Proportional-hazards assumption
The usual Cox proportional-hazards interpretation relies on an assumption concerning the relative hazards over time. If the proportional-hazards assumption is inappropriate, a single hazard ratio can become less representative of how treatment effects vary over follow-up. The registry-reported ZENITH data do not report an assessment of this assumption.
9. Statistical Methods Explained
Why is time to the first event different from simply counting events?
A simple event proportion records whether an event occurred during a defined period. A time-to-event endpoint additionally uses information about when the first event occurred and can accommodate participants who have not yet experienced an event when follow-up ends. This makes the timing of morbidity or mortality part of the statistical comparison.
Why use a composite morbidity-or-mortality endpoint?
A composite endpoint combines several clinically defined event types into a single primary outcome. Statistically, this can increase the number of qualifying first events and therefore provide an event-time endpoint that incorporates several clinically important outcomes. Interpretation requires attention to the individual components because the components may differ in frequency, clinical meaning, and treatment sensitivity.
What does “first confirmed” mean?
The registry wording specifies that morbidity or mortality events were adjudicated by a blinded, independent committee of clinical experts and identifies adjudication-confirmed lung transplantation and PAH worsening-related hospitalization of ≥24 hours for the primary analysis. Thus, confirmation and adjudication are part of the endpoint definition rather than merely descriptive features added after analysis.
Why does randomization matter for the primary comparison?
Randomization creates the framework for comparing outcomes between assigned treatment groups without relying solely on adjustment for observed baseline characteristics. In a properly conducted randomized trial, the treatment assignment mechanism is the foundation for causal inference about the randomized intervention.
Why does triple masking matter statistically?
Masking can reduce opportunities for knowledge of treatment assignment to influence participant behavior, clinical management, assessment, or other trial processes. The ClinicalTrials.gov record identifies the trial as triple-masked but does not specify which three parties were masked, so no more specific attribution is made here.
Why does the p-value not measure effect size?
A p-value addresses the compatibility of the observed data with a specified null hypothesis under the statistical testing framework. It does not quantify the magnitude or clinical importance of an effect. For a time-to-event endpoint, effect size is more directly described using measures such as a hazard ratio together with absolute event-free probabilities or other clinically interpretable summaries.
10. Safety Results
the ClinicalTrials.gov record reports serious adverse events by treatment arm. These are the only arm-specific safety results provided in the ClinicalTrials.gov record.
| Safety measure | Sotatercept | Placebo |
|---|---|---|
| Serious adverse events, affected / at risk | 52 / 86 | 64 / 86 |
The reported serious-adverse-event figures are 52/86 for sotatercept and 64/86 for placebo. Because the ClinicalTrials.gov record does not provide a formal statistical analysis of these safety data, they should be treated as descriptive arm-level counts rather than as a reported hypothesis-test result.
The denominators are both 86 for this safety measure, while the overall trial enrollment is 173. This difference is important: the safety figures should not be silently substituted for the full randomized population. The ClinicalTrials.gov record does not explain the relationship between the safety denominator and the overall enrollment, so no additional safety-population assumptions are made here.
11. What the Registry Data Do and Do Not Establish
| Question | What the ClinicalTrials.gov record establishes | What is not reported |
|---|---|---|
| What was tested? | Time to first confirmed morbidity or mortality event. | A numerical treatment-effect estimate. |
| Who was studied? | Participants with pulmonary arterial hypertension who were WHO FC III or FC IV at high risk of mortality. | Detailed baseline characteristics. |
| How was the trial designed? | Randomized, parallel, triple-masked, phase 3. | Detailed randomization and masking procedures. |
| What treatments were compared? | Sotatercept versus placebo. | Detailed dosing or treatment schedules. |
| Were results posted? | Yes; 13 outcome measures were posted. | Formal statistical-analysis entries in the posted results. |
| What safety information is available? | Serious adverse events: 52/86 vs 64/86. | Broader adverse-event tables and formal safety comparisons. |
This distinction is central to a statistical analysis page. A registry can contain posted outcome information without the registry-reported extract containing the complete statistical-analysis results necessary to report an HR, confidence interval, or p-value. The absence of a registry-reported estimate is not evidence of no treatment effect; it simply means that the numerical estimate cannot be responsibly reconstructed from the ClinicalTrials.gov record.
12. Missing Data, Censoring, and Analysis Population
Because the primary endpoint is time-to-event, participants may contribute follow-up without experiencing the event during the observation period. Standard time-to-event methods distinguish such censoring from an observed event.
The ClinicalTrials.gov record does not specify the censoring rules, analysis population definition, handling of protocol deviations, missing event dates, or imputation strategy. These details can materially affect the analysis of a time-to-event endpoint and should be taken from the protocol or statistical analysis plan when available.
13. Multiplicity and Interim Analysis
The ClinicalTrials.gov record identifies one registered primary endpoint. They do not provide a multiplicity strategy, alpha-allocation procedure, interim-analysis schedule, stopping boundary, or statistical power calculation.
What is known
One primary endpoint is identified in the ClinicalTrials.gov record, and the endpoint is a time-to-event outcome measured up to approximately 31 months.
What remains unspecified
The ClinicalTrials.gov record does not state whether an interim efficacy analysis occurred or how type I error was controlled across interim or secondary analyses.
These design features matter because repeated statistical looks at accumulating data can alter the nominal interpretation of a p-value unless the monitoring plan incorporates appropriate error control. Likewise, multiple endpoints or multiple hypothesis tests can require prespecified multiplicity management. No such procedure should be attributed to ZENITH without supporting trial documentation.
14. Non-Inferiority, Bayesian, and Other Specialized Methods
The ClinicalTrials.gov record does not identify a non-inferiority margin, a non-inferiority hypothesis, Bayesian methods, a factorial design, or a crossover procedure.
| Design topic | Information in the ClinicalTrials.gov record |
|---|---|
| Non-inferiority margin | Not reported. |
| Bayesian analysis | Not reported. |
| Factorial design | Not reported; the recorded design model is parallel. |
| Crossover | Not reported. |
| Interim analysis | Not reported. |
| Multiplicity procedure | Not reported. |
| Missing-data/imputation strategy | Not reported. |
| Stratification factors | Not reported. |
Keeping these topics separate from the endpoint analysis is important. They are potentially influential design or analysis features, but their absence from the ClinicalTrials.gov record means they cannot be assigned to ZENITH based on general expectations about phase 3 trials.
15. Understanding the Composite Endpoint
The primary endpoint combines morbidity and mortality events. The registry-reported definition identifies three broad event categories: all-cause death, lung transplantation, and PAH worsening-related hospitalization of ≥24 hours, with adjudication requirements for specified components.
Event 1
All-cause death.
Event 2
Lung transplantation, with adjudication confirmation required for inclusion in the primary analysis.
Event 3
PAH worsening-related hospitalization of ≥24 hours, with adjudication confirmation required for inclusion in the primary analysis.
First event principle
The endpoint is defined around the time to the first confirmed morbidity or mortality event.
A composite endpoint can be statistically efficient because several types of clinically meaningful events can count toward the same endpoint. However, the interpretation of the composite should consider the relative contribution of each component. The ClinicalTrials.gov record does not provide component-specific event counts, so the contribution of each event type cannot be evaluated here.
16. Why This Trial Matters Statistically
ZENITH provides a useful teaching example of how a modern randomized phase 3 trial can be structured around a clinically meaningful composite time-to-event endpoint. Even without a registry-reported numerical treatment-effect estimate, the trial illustrates several important statistical concepts.
| Concept | How it appears in ZENITH |
|---|---|
| Randomization | The trial uses randomized allocation to sotatercept or placebo. |
| Blinding | The registry-reported design is triple-masked. |
| Parallel design | The recorded design model is parallel. |
| Time-to-event analysis | The registered primary endpoint is time to first confirmed morbidity or mortality event. |
| Composite endpoint | Morbidity and mortality events are combined into a primary time-to-first-event outcome. |
| Adjudication | A blinded, independent committee of clinical experts adjudicated events. |
| Censoring | Time-to-event methodology ordinarily requires explicit handling of participants who do not experience the event during observed follow-up. |
| Safety analysis | Serious adverse events are reported by randomized treatment arm in the registry-reported profile. |
| Registry interpretation | Posted outcomes do not necessarily mean that the registry-reported extract contains formal statistical-analysis results. |
17. Limitations
- No formal statistical-analysis results reported: no formal statistical analyses were posted in the ClinicalTrials.gov record, so treatment-effect estimates, confidence intervals, and p-values cannot be reported from the registry record.
- Incomplete endpoint text: the registry-reported primary endpoint definition ends after “All deaths that are a first event,” so the missing remainder cannot be reconstructed.
- Limited baseline information: detailed baseline demographic and clinical characteristics are not included in the ClinicalTrials.gov record.
- No component-level event counts: the ClinicalTrials.gov record does not show how many primary events arose from death, transplantation, or hospitalization.
- No censoring rules: the ClinicalTrials.gov record does not specify how censoring was defined or applied.
- No analysis population definition: the registry-reported profile does not identify the precise efficacy or safety analysis populations beyond the reported serious-adverse-event denominators.
- No multiplicity information: the ClinicalTrials.gov record does not describe control of type I error across primary, secondary, or interim analyses.
- No subgroup results: subgroup estimates cannot be inferred from the overall ClinicalTrials.gov record.
- Safety interpretation is limited: only serious-adverse-event counts of 52/86 and 64/86 are reported, without a broader adverse-event profile or formal statistical comparison.
- General statistical methods are not trial-specific results: Kaplan-Meier, log-rank, and Cox methods are appropriate educational explanations for time-to-event endpoints, but should not be presented as documented ZENITH analyses without supporting evidence.
18. How to Read a Future ZENITH Time-to-Event Result
If a complete statistical-analysis result becomes available, several quantities should be read together rather than in isolation.
| Quantity | What it would tell the reader | What it would not tell the reader |
|---|---|---|
| Hazard ratio | Relative event-rate measure from a time-to-event model. | The absolute probability of an event at a specific time. |
| 95% confidence interval | Uncertainty around the estimated treatment effect under the specified analysis. | The distribution of individual patient effects. |
| P-value | Evidence against a specified null hypothesis under the test framework. | The size or clinical importance of the treatment effect. |
| Kaplan-Meier estimate | Estimated event-free probability at a specified time. | A complete description of the mechanism producing the treatment difference. |
| Median time to event | The time by which the estimated event-free probability reaches 50%, when estimable. | A measure of the treatment effect for every participant. |
The most informative future presentation would combine the relative effect measure with absolute time-specific event estimates and its confidence interval. If a Cox hazard ratio were reported, it should also be accompanied by an assessment of whether its underlying proportional-hazards interpretation is reasonable. The p-value should be read as evidence under the specified testing framework, not as a measure of how large or clinically important the effect is.
19. Serious Adverse Events: Statistical Perspective
The reported serious-adverse-event counts are 52 of 86 participants in the sotatercept arm and 64 of 86 in the placebo arm.
| Arm | Affected | At risk | Descriptive affected fraction |
|---|---|---|---|
| Sotatercept | 52 | 86 | 52/86 |
| Placebo | 64 | 86 | 64/86 |
The table deliberately preserves the registry's reported fractions rather than converting them into newly rounded percentages. A formal comparison of safety rates would require a prespecified definition of the safety estimand, the appropriate analysis population, the event observation period, and the statistical method. Those details are not reported here.
20. Related Tutorials
Learn more about the methods used in this trial:
21. Related Statistical Calculators
22. Sources
- ClinicalTrials.gov: NCT04896008 — ZENITH.
- Linked publication: PubMed record for PMID 40167274.
Continue through Clinical Biostats
Use the broader tutorial and calculator collections to build a deeper understanding of randomized trial design, masking, and statistical analysis.
23. Record Summary
ZENITH is a completed randomized, parallel, triple-masked phase 3 trial enrolling 173 participants with pulmonary arterial hypertension who were described as WHO FC III or FC IV at high risk of mortality. The registered primary endpoint is time to first confirmed morbidity or mortality event, measured up to approximately 31 months, with events defined around all-cause death, lung transplantation, and PAH worsening-related hospitalization of ≥24 hours and subject to blinded independent adjudication as specified in the registry-reported endpoint definition.
The statistical-analysis section in the ClinicalTrials.gov record contains no formal treatment-effect estimates. That limitation is consequential: a responsible statistical analysis must distinguish the existence of a registered time-to-event endpoint from the availability of numerical evidence about the treatment comparison. The appropriate educational framework is therefore to explain how randomized time-to-event analyses are ordinarily performed while preserving the boundary between general methodology and reported ZENITH results.