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
CADENCE is a completed, randomized, parallel-group, triple-masked phase 2 treatment study of sotatercept in pulmonary hypertension due to HFpEF. The registry reports 164 enrolled participants and three study arms, with pulmonary vascular resistance measured by right heart catheterization as the primary endpoint at Week 24.
| Feature | CADENCE |
|---|---|
| Phase | Phase 2 |
| Status | Completed |
| Condition | Hypertension, Pulmonary |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | Triple |
| Primary purpose | Treatment |
| Enrollment | 164 |
| Number of arms | 3 |
| Start date | 2021-12-29 |
| Primary completion date | 2025-09-05 |
| Lead sponsor | Acceleron Pharma, Inc., a wholly-owned subsidiary of Merck & Co., Inc., Rahway, NJ USA |
| Sponsor type | Industry |
| ClinicalTrials.gov | NCT04945460 |
2. Clinical Question
The clinical question is whether sotatercept can alter pulmonary vascular resistance in patients with pulmonary hypertension due to HFpEF, as assessed from baseline to Week 24 by right heart catheterization.
Population
Participants in a phase 2 study for pulmonary hypertension due to HFpEF.
Interventions
Sotatercept 0.3 mg/kg, or sotatercept 0.3 mg/kg escalating to 0.7 mg/kg.
Comparator
Placebo.
Primary question
What is the change from baseline in pulmonary vascular resistance at Week 24?
Because the registry identifies three randomized arms, the statistical question is not simply a two-group comparison. A three-arm design permits comparisons involving two sotatercept dosing strategies and placebo. The precise confirmatory comparison structure is not reported in the available registry information, so the primary endpoint should be interpreted first as a three-arm randomized comparison rather than assuming a particular pairwise testing hierarchy.
3. Trial Design
Sotatercept 0.3 mg/kg
- Sotatercept 0.3 mg/kg
- Randomized study arm
Placebo
- Placebo
- Randomized comparator arm
Sotatercept escalation
- Sotatercept 0.3 mg/kg escalating to 0.7 mg/kg
- Randomized study arm
Why the three-arm structure matters
A three-arm randomized trial contains more information than a simple treatment-versus-placebo comparison. Here, the two sotatercept strategies differ in dosing approach: one remains at 0.3 mg/kg, while the other escalates from 0.3 mg/kg to 0.7 mg/kg. This creates a natural statistical question about whether the endpoint differs across the three randomized groups and, depending on the prespecified analysis plan, whether particular sotatercept regimens differ from placebo or from one another.
The distinction is important because the number and order of statistical comparisons affect interpretation. If several pairwise hypotheses are tested, the overall false-positive error rate can rise unless the testing strategy accounts for multiplicity.
4. Trial Timing and Status
Study start
The registry lists 2021-12-29 as the study start date.
Primary completion
The registry lists 2025-09-05 as the primary completion date.
Current registry status
The study status is listed as completed.
The study's status and completion date establish that the registered trial has reached completion. They do not, by themselves, establish the direction or magnitude of the primary endpoint result.
5. Primary Endpoint
| Endpoint | Registry definition | Time frame | Measurement |
|---|---|---|---|
| Change From Baseline in Pulmonary Vascular Resistance (PVR) at Week 24 | PVR, a hemodynamic variable of pulmonary circulation, is measured by right heart catheterization (RHC). | Baseline and Week 24 | Change from baseline in PVR |
The primary endpoint is a continuous change-from-baseline hemodynamic measure, rather than a time-to-event endpoint such as overall survival or progression-free survival. That distinction determines the statistical framework that would normally be appropriate.
What is being compared?
For each participant, the conceptual quantity is the Week 24 PVR measurement relative to the participant's baseline PVR. The treatment comparison then concerns how the distribution of these changes differs across the randomized groups.
The registry defines the endpoint as change from baseline in PVR at Week 24. The sign and clinical interpretation of an observed change depend on the scale and direction of PVR and on the prespecified statistical analysis.
6. Statistical Methodology
Analysis of a continuous endpoint
Because PVR change is a continuous outcome, an analysis of means or another appropriate model for continuous data would generally be considered. In a randomized three-arm study, the primary model would ordinarily account for the treatment group while evaluating differences in the Week 24 change from baseline.
An analysis of covariance (ANCOVA) is a common approach for this type of endpoint when a baseline measurement is available. Rather than analyzing only the raw Week 24 values or only the change scores, ANCOVA can model the Week 24 outcome while adjusting for baseline PVR. This can improve precision when baseline and follow-up measurements are correlated.
Here, the treatment terms represent differences among randomized groups after accounting for baseline PVR. This is a conceptual representation of a common analysis framework; the registry information does not specify the actual statistical model used for CADENCE.
Three-arm treatment comparison
With three randomized groups, a model can first evaluate whether there is evidence of any overall treatment-group difference. If the overall test is informative, prespecified pairwise comparisons can then address specific questions such as a sotatercept regimen versus placebo.
Overall treatment effect
Asks whether the mean or adjusted mean outcome differs somewhere among the three randomized groups.
Pairwise treatment effect
Asks whether one specific sotatercept regimen differs from placebo or another study arm.
Baseline adjustment
Can account for the baseline PVR measurement when modeling the Week 24 outcome.
Multiplicity
Multiple pairwise comparisons require an appropriate interpretation of type I error and confidence intervals.
Least-squares means
When ANCOVA is used, adjusted group means are often expressed as least-squares means, also called estimated marginal means. These represent model-based treatment-group means at a common value of the baseline covariate. Their differences can provide an interpretable estimate of the treatment effect while accounting for baseline imbalance in the statistical model.
Confidence intervals
A confidence interval around a treatment contrast quantifies statistical uncertainty in the estimated difference. A narrow interval indicates greater precision than a wide interval, all else equal. The interval does not describe the range of individual patient responses.
Randomization and causal interpretation
Randomization is central to the treatment comparison. When the analysis follows randomized assignment and the prespecified statistical framework, differences between groups can be interpreted as treatment effects under the assumptions of the trial design. Baseline adjustment can improve statistical precision, but it does not replace randomization as the principal source of causal identification.
7. Statistical Methods Explained
Why is pulmonary vascular resistance a continuous endpoint?
PVR is a measured hemodynamic quantity rather than a binary event. The primary endpoint therefore contains more information than simply classifying participants as improved or not improved. A continuous analysis can use the magnitude of each participant's change, which can improve efficiency when the measurement behaves appropriately under the chosen model.
Why measure change from baseline?
Participants can begin a study with different baseline PVR values. A change-from-baseline endpoint focuses on the within-participant difference between baseline and Week 24. This can make the treatment comparison more directly reflect how the measured hemodynamic state changed during the study.
Why might ANCOVA be preferable to a simple comparison of changes?
ANCOVA can use the baseline measurement as a covariate when analyzing the Week 24 outcome. If baseline and follow-up PVR are correlated, this can reduce residual variability and produce a more precise treatment comparison. The advantage depends on the characteristics of the data and the assumptions of the model.
Why does the three-arm design matter statistically?
The three arms create more than one possible treatment contrast. A statistical analysis therefore needs to distinguish an overall test across all three groups from individual pairwise comparisons. If multiple hypotheses are tested, the interpretation of nominal P-values and confidence intervals depends on the prespecified multiplicity strategy.
What does a treatment difference in PVR mean?
A treatment contrast describes the difference in the estimated change or adjusted Week 24 outcome between two randomized groups. It is a population-level comparison, not a statement that every participant experiences the same magnitude of change.
Why is the baseline measurement statistically important?
Baseline PVR provides information about each participant before randomized treatment exposure. Adjusting for baseline can account for some of the natural variation among participants and may increase precision. Importantly, adjustment should be based on a prespecified model rather than selected after seeing the treatment results.
8. Planned Analysis
The registry identifies Change From Baseline in Pulmonary Vascular Resistance (PVR) at Week 24 as the primary endpoint, measured at baseline and Week 24 by right heart catheterization. Results have not been posted on ClinicalTrials.gov.
For an endpoint of this type, a typical statistical analysis would compare the three randomized groups using a continuous-outcome model, with ANCOVA being a common framework when baseline PVR is available. The model could estimate adjusted treatment-group means and prespecified treatment contrasts, with confidence intervals describing the precision of those contrasts.
The three-arm structure makes the multiplicity strategy particularly important. An overall comparison among all three groups and individual comparisons such as each sotatercept regimen versus placebo are different statistical questions. A prespecified hierarchy, adjustment procedure, or other error-control strategy would determine how multiple comparisons should be interpreted.
What a complete primary analysis would ordinarily contain
| Component | Purpose |
|---|---|
| Baseline PVR | Provides the pretreatment covariate or baseline value for the analysis. |
| Week 24 PVR | Provides the follow-up outcome used to assess the primary endpoint. |
| Treatment-group effect | Estimates differences among the three randomized groups. |
| Prespecified contrasts | Addresses specific treatment comparisons, such as a sotatercept regimen versus placebo. |
| Confidence intervals | Quantifies uncertainty around treatment-effect estimates. |
| Multiplicity adjustment | Controls interpretation when multiple confirmatory comparisons are made. |
| Model diagnostics | Assesses whether the assumptions of the selected continuous-outcome model are reasonable. |
9. Interpreting a Continuous Treatment Effect
If a model estimates a difference between a sotatercept arm and placebo, that difference represents the estimated separation between the two randomized groups on the primary PVR endpoint. Its interpretation depends on the direction of the PVR scale and the exact contrast defined in the statistical analysis.
A 95% confidence interval around the treatment contrast describes uncertainty about the population-level estimate under the model and sampling framework. It does not indicate that 95% of individual participants experienced effects within that interval.
A P-value addresses the compatibility of the observed data with a specified null hypothesis under the statistical model. It does not measure the size, clinical importance, or probability of a treatment effect.
For a continuous endpoint, it is particularly useful to report the estimated treatment contrast together with its confidence interval rather than relying on a P-value alone. The estimated magnitude tells the reader how far apart the groups are on the measured scale, while the confidence interval communicates precision.
10. Assumptions Behind a Continuous-Outcome Model
Any model-based analysis of PVR requires assumptions. The precise assumptions depend on the selected method, but a conventional ANCOVA framework generally relies on an appropriate functional relationship between baseline PVR and the Week 24 outcome, a suitable residual structure, and a treatment effect that is represented adequately by the model.
Residual behavior
Model residuals should be reasonably compatible with the assumptions used for inference. Strong departures can affect standard errors and confidence intervals.
Baseline relationship
If baseline PVR is included as a covariate, the relationship between baseline and follow-up outcome should be represented appropriately.
Influential observations
Extreme PVR measurements can have disproportionate influence in a continuous model and should be evaluated diagnostically.
Missing Week 24 measurements
Incomplete follow-up can affect the analysis population and inference. The appropriate handling depends on the missing-data mechanism and prespecified analysis strategy.
11. Missing Data and Analysis Populations
The primary endpoint requires both a baseline measurement and a Week 24 assessment to characterize individual change directly. In a real clinical-trial dataset, some participants may have incomplete measurements, withdrawals, or other reasons for not contributing a Week 24 value.
The registry information does not specify an imputation method, missing-data model, or detailed analysis-population definitions. Those details are important because different approaches to incomplete Week 24 measurements can produce different estimates and uncertainty.
Why missingness matters
If missing Week 24 measurements are related to treatment response, disease status, tolerability, or other factors, simply analyzing participants with observed Week 24 values may not preserve the original randomized comparison in the same way as a prespecified model that appropriately accounts for missingness. The validity of any missing-data approach depends on assumptions that should be stated explicitly.
12. Multiplicity in a Three-Arm Trial
Multiplicity is especially relevant when a trial has three treatment groups. Suppose the scientific questions include whether each sotatercept regimen differs from placebo. That creates two treatment-versus-control contrasts. A separate comparison between the two sotatercept regimens would create another hypothesis.
| Potential question | Statistical contrast | Why it matters |
|---|---|---|
| Any difference among groups? | Overall three-group test | Addresses the global treatment-group question. |
| Sotatercept 0.3 mg/kg vs placebo | Pairwise contrast | Addresses one specific regimen-versus-control question. |
| Sotatercept 0.3 mg/kg escalating to 0.7 mg/kg vs placebo | Pairwise contrast | Addresses the other regimen-versus-control question. |
| Escalating regimen vs 0.3 mg/kg regimen | Pairwise contrast | Addresses a dose-strategy question. |
These are distinct hypotheses. If several are tested confirmatorily, an analysis plan needs to specify how the type I error rate is controlled. Common approaches include hierarchical testing or multiplicity-adjusted simultaneous inference, but the registry information does not identify a particular strategy for CADENCE.
13. Randomization, Masking, and Bias Control
CADENCE is described in the registry as randomized and triple-masked. Randomization provides the basis for comparing outcomes between treatment assignments, while masking can reduce the influence of treatment knowledge on behavior, assessment, and reporting.
The use of a hemodynamic measurement obtained through right heart catheterization also distinguishes the primary endpoint from outcomes based only on subjective assessment. Nevertheless, measurement quality, timing, protocol adherence, and missing observations remain relevant to the validity of a continuous endpoint.
14. What the Registry Reports — and What It Does Not
| Element | Registry status |
|---|---|
| Trial phase | Reported: Phase 2 |
| Enrollment | Reported: 164 |
| Number of arms | Reported: 3 |
| Randomization | Reported: randomized |
| Design | Reported: parallel |
| Masking | Reported: triple |
| Primary endpoint | Reported: change from baseline in PVR at Week 24 |
| Endpoint measurement | Reported: PVR measured by right heart catheterization |
| Formal posted statistical analysis | Not posted on ClinicalTrials.gov |
| Primary endpoint estimate | Not posted on ClinicalTrials.gov |
| Confidence interval | Not posted on ClinicalTrials.gov |
| P-value | Not posted on ClinicalTrials.gov |
| Secondary endpoint results | No secondary endpoint information is reported in the registry information available here. |
| Serious adverse events by arm | No arm-specific serious adverse-event results are reported in the registry information available here. |
This distinction is statistically important. A completed randomized trial and a posted numerical efficacy result are not the same thing. Trial completion describes the status of the study; an estimated treatment effect requires an analysis of the observed endpoint data.
15. Limitations
- No posted primary analysis: the registry does not provide a formal statistical analysis, effect estimate, confidence interval, or P-value for the primary endpoint.
- Three-arm interpretation: the registry identifies three randomized groups but does not specify the confirmatory comparison hierarchy in the information available here.
- Missing-data strategy: the registry information does not specify how incomplete Week 24 PVR measurements are handled statistically.
- Model specification: the registry identifies the endpoint and its measurement method but does not specify the statistical model used for the primary endpoint.
- Multiplicity: the registry information does not specify how multiple treatment comparisons are controlled.
- Analysis populations: detailed efficacy analysis-population definitions are not reported in the registry information available here.
- Precision cannot be assessed: without an effect estimate and confidence interval, the statistical precision of the primary treatment comparison cannot be characterized.
- Clinical magnitude cannot be assessed: the registry endpoint definition identifies what was measured but does not establish how much PVR changed in any treatment group.
16. Why This Trial Matters Statistically
CADENCE is a useful statistical teaching case because its primary endpoint is fundamentally different from the time-to-event outcomes that dominate many clinical-trial analyses. Instead of estimating a hazard ratio, the central question concerns a continuous hemodynamic measurement observed at baseline and Week 24 across three randomized groups.
| Concept | How it appears in CADENCE |
|---|---|
| Randomization | Participants were randomized to three study arms. |
| Parallel-group design | The three randomized groups were evaluated concurrently. |
| Triple masking | The registry classifies the study as triple-masked. |
| Continuous endpoint | PVR change is a quantitative hemodynamic outcome. |
| Baseline adjustment | Baseline PVR provides a natural covariate for a continuous-outcome analysis. |
| ANCOVA | A common framework for comparing Week 24 outcomes while accounting for baseline. |
| Pairwise contrasts | Three treatment arms create multiple possible treatment comparisons. |
| Multiplicity | Multiple confirmatory contrasts require an appropriate type I error strategy. |
| Confidence intervals | Would quantify uncertainty around estimated treatment differences. |
| Missing data | Incomplete Week 24 measurements can affect inference and require a prespecified strategy. |
| Measurement methodology | PVR is measured by right heart catheterization. |
17. Statistical Interpretation of the Primary Endpoint
The most important distinction in CADENCE is between measurement and inference. The measurement is pulmonary vascular resistance obtained by right heart catheterization. The inferential question is whether the change observed in the randomized sotatercept groups differs from the change observed in the placebo group.
A statistically meaningful analysis therefore needs more than a statement that PVR was measured at Week 24. It needs an estimated treatment contrast and a measure of uncertainty around that contrast. A P-value can address a prespecified null hypothesis, but it cannot substitute for the estimated magnitude of the treatment difference.
An estimated difference between randomized groups summarizes how the groups differed on the primary PVR endpoint at the population level. If the analysis is baseline-adjusted, the contrast reflects the fitted model's comparison after accounting for baseline PVR.
A group-level treatment contrast does not mean that every participant experienced the same change. Individual responses can vary substantially around the estimated population effect.
The confidence interval communicates the precision of the estimated treatment contrast under the chosen statistical framework. Greater width indicates greater uncertainty in the estimate; narrower width indicates greater statistical precision.
A P-value is not the probability that the treatment works, the probability that the null hypothesis is true, or a measure of clinical importance. It is tied to a particular null hypothesis and statistical model.
18. Clinical Interpretation vs Statistical Interpretation
Statistical interpretation
The primary endpoint is a continuous change-from-baseline measure evaluated across three randomized groups at Week 24. A suitable analysis would estimate treatment contrasts with corresponding uncertainty measures.
Clinical interpretation
The clinical meaning of the result depends on the magnitude and direction of the observed PVR change and how that change relates to pulmonary hemodynamics in the studied population.
These two levels of interpretation should remain distinct. Statistical significance, if established, would describe compatibility with a prespecified statistical hypothesis. Clinical interpretation requires consideration of the magnitude and direction of the hemodynamic difference as well as the clinical context.
19. Trial History and Design Context
Study begins
CADENCE begins according to the registry's listed start date.
Randomized treatment comparison
The study evaluates three parallel randomized arms: sotatercept 0.3 mg/kg, placebo, and sotatercept 0.3 mg/kg escalating to 0.7 mg/kg.
Primary completion
The registry lists 2025-09-05 as the primary completion date.
Registry status
The study is listed as completed.
20. Key Statistical Takeaways
1. This is not a survival endpoint
PVR change is continuous, so the central analysis concerns differences in quantitative measurements rather than hazards or survival curves.
2. Baseline matters
The primary endpoint explicitly compares Week 24 with baseline, making baseline PVR central to interpretation and potentially to model adjustment.
3. Three arms create multiple questions
The two sotatercept strategies and placebo generate an overall group comparison as well as possible pairwise treatment contrasts.
4. Multiplicity cannot be ignored
If multiple pairwise hypotheses are tested, the interpretation of nominal P-values requires consideration of the prespecified error-control strategy.
5. Effect size matters
A P-value alone would not communicate the magnitude of the PVR difference. The estimated treatment contrast and confidence interval are essential.
6. Completion is not an efficacy result
The completed status establishes study status, but it does not by itself establish the direction or magnitude of the primary endpoint.
21. Sources
- ClinicalTrials.gov: CADENCE, NCT04945460.
- PubMed: PMID 41904795.
Continue through the Clinical Biostats clinical-trial library
Explore additional clinical-trial analyses and statistical methods across randomized study designs, endpoints, and analysis frameworks.
22. Record Summary
CADENCE is a completed phase 2, randomized, parallel, triple-masked treatment study with 164 participants and three study arms. The primary endpoint is change from baseline in pulmonary vascular resistance at Week 24, with PVR measured by right heart catheterization. Statistically, this makes CADENCE a continuous-outcome, three-arm comparison rather than a conventional time-to-event trial. A typical analysis would focus on baseline-adjusted comparisons among the randomized groups, potentially using ANCOVA or a related continuous-outcome model, with prespecified contrasts and appropriate handling of multiplicity and missing data.
The registry establishes the design, intervention groups, enrollment, endpoint definition, measurement method, and study timing. It does not post a formal statistical analysis or numerical primary endpoint result. The key statistical lesson is therefore the structure of the analysis: a quantitative baseline-to-Week-24 outcome, randomized comparison across three arms, and the need to distinguish overall treatment effects from individual pairwise contrasts.