A Study of Sotatercept for the Treatment of Pulmonary Arterial Hypertension (MK-7962-003/A011-11)(STELLAR): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study of Sotatercept for the Treatment of Pulmonary Arterial Hypertension (MK-7962-003/A011-11)(STELLAR) (NCT04576988), a randomized phase 3 study. The primary outcome is Change From Baseline in 6-Minute Walk Distance (6MWD) at Week 24.

2. Clinical Question

The trial also recorded other outcomes, including pulmonary vascular resistance, NT-proBNP, World Health Organization functional class, risk score, PAH-SYMPACT domain scores, and time to death or a clinical-worsening event. These additional measures address different aspects of disease status and patient experience, so they do not answer precisely the same question as the primary walking-distance endpoint. Their findings should be interpreted as complementary evidence rather than as interchangeable measures of the primary outcome.

3. Trial Design

This was a Phase 3, randomized, double-blind, placebo-controlled study with a parallel-group model. The registry record describes quadruple masking. Participants were assigned to sotatercept plus background PAH therapy or placebo plus background PAH therapy. Enrollment was 324. The study status is completed; the recorded start date is 2021-01-25 and the primary completion date is 2022-08-26.

The treatment contrast is add-on treatment versus placebo, with background PAH therapy in both groups. The randomized comparison therefore concerns the additional assigned study treatment in the context of background therapy; it is not a comparison of background therapy against no therapy. The intervention description gives a sotatercept starting dose of 0.3 mg/kg and target dose of 0.7 mg/kg, administered subcutaneously every 21 days. Placebo was also administered subcutaneously every 21 days.

The primary endpoint was change from baseline in 6MWD at Week 24, assessed using the 6-minute walk test. The primary analysis population is described as randomized participants who received at least one dose and had a baseline 6MWD value. The reported outcome groups contain 163 participants in the sotatercept group and 160 in the placebo group. The registry record does not explain how the primary analysis population relates to the overall enrollment count.

The stated objectives included evaluating efficacy and safety at 24 weeks. The registry record also lists secondary outcomes measured at Week 24 and a time-to-event outcome followed for up to approximately 18 months. The allocation method beyond randomization, the details of the masking procedures, and the sample-size justification are not reported in the registry record. The record also does not state whether interim analyses were planned or conducted.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Before assignment: Per protocol, not all participants from the double-blind placebo controlled (DBPC) period entered the long-term double blind (LTDB) period due to clinical worsening or consent withdrawal after DBPC period.

Double Blind Placebo Controlled (DBPC)

Reasons for not completing this period:

Long Term Double Blind (LTDB)

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: All randomized participants who received at least one dose of study treatment and had a baseline value of 6MWD.

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

Population: Of the 324 participants randomized in the study, one participant was randomized in error in placebo plus background PAH therapy arm and no data was collected on this participant. Per protocol, this participant was excluded from the study analysis.

6. Primary Endpoint

The registry lists the primary outcome as Change From Baseline in 6-Minute Walk Distance (6MWD) at Week 24, measured over Baseline and Week 24.

The primary endpoint is change from baseline in 6-minute walk distance at Week 24. The 6-minute walk test measures the distance a participant walks during a fixed test period. In this trial, 6MWD is identified as a measure of functional capacity. Expressing the outcome as change from baseline focuses the comparison on movement from each participant’s starting measurement rather than only on the Week 24 distance.

Walking distance is a continuous measure, but the registry record reports the group outcomes using medians and full ranges. The sotatercept group’s median change was 34.4 meters, with a full range from 32.5 to 35.5. The placebo group’s median change was 1.0 meters, with a full range from -1.0 to 5.0. These summaries describe the distributions as reported; the range is not a confidence interval and does not quantify uncertainty in the estimated group effect.

A treatment comparison is also reported using an aligned rank stratified Wilcoxon method. That analysis gives a treatment difference of 40.8 meters, with a 95% confidence interval from 27.53 to 54.14 and p < 0.001. This is a separate inferential summary and should not be treated as a simple subtraction of the two reported medians. General statistical context: rank-based methods compare the ordering of observations and can be useful when an analysis is not based on a normal-distribution model for the raw measurements. The registry record does not provide further detail on the exact estimand or stratification variables.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

General statistical context: a randomized comparison is designed to make assigned groups comparable, on average, with respect to both measured and unmeasured characteristics at the point of assignment. Randomization does not guarantee that the groups will be identical, and the registry record does not give the method used to generate or conceal the allocation sequence. The parallel-group design means the assigned groups were followed as distinct groups rather than participants crossing between treatments in the primary comparison.

Why report medians and ranges?

The primary group summaries use medians and full ranges. A median is the central ordered observation: half of the observations lie at or below it and half at or above it, subject to ties and the number of observations. A full range describes the smallest and largest reported values. It is sensitive to extreme observations and does not describe how values are distributed between the endpoints. The registry record does not give a mean, standard deviation, or interquartile range for the primary outcome. The reported full ranges should not be mistaken for confidence intervals around the medians.

General statistical context: medians can be useful when a distribution is skewed or when an analysis is based on ranks. Rank procedures compare the relative ordering of observations rather than requiring the comparison to be expressed as a difference between arithmetic means. The registry record specifies an aligned rank stratified Wilcoxon method for the treatment comparison. It does not identify the stratification factors or provide enough detail to reconstruct the analysis from the arm summaries.

How should the treatment estimate be read?

The reported treatment difference is 40.8 meters, with a 95% confidence interval from 27.53 to 54.14. A confidence interval reflects sampling uncertainty under the analysis and its assumptions. General statistical context: narrower intervals indicate greater precision than wider intervals on the same scale, while an interval’s location describes the range of effect values compatible with the data and method at the stated confidence level. The interval is not a prediction interval for individual participants. The registry record does not state whether additional adjustment was used beyond the named rank-stratified method.

The reported p-value is p < 0.001. General statistical context: a p-value is calculated under a specified null hypothesis and test procedure. It describes how surprising data at least as extreme as those observed would be if that null hypothesis and the test assumptions held. It is not the probability that the result occurred by chance, and it does not indicate the magnitude or importance of an effect. Effect estimates and confidence intervals provide information that a p-value alone does not.

What does the time-to-event method contribute?

The registry record reports a Log Rank analysis for time to death or the first occurrence of a clinical-worsening event. The reported hazard ratio is 0.163, with a 95% confidence interval from 0.076 to 0.347 and p < 0.001. General statistical context: the Log Rank test compares event-time experience across groups while accounting for the timing of events and censoring. A hazard ratio summarizes relative event rates over time under the analysis framework; it is not an absolute risk, an event count, or a direct estimate of time gained. The record gives no median time to event in either group, and the underlying number of events is not stated among the reported results in the registry record.

The planning illustration associated with this outcome is based on the observed effect, not on the trial’s own design. It assumes a two-sided significance level of 0.05, equal allocation, and a normal approximation. The event-count illustrations give total events, not participants per arm: using 100% of the observed effect, the totals are 10 for 80% power and 13 for 90% power; using 75% of the observed effect, they are 17 and 23; using 50% of the observed effect, they are 39 and 52. General statistical context: such illustrations show how planning requirements can depend on the effect size assumed. They are not observed power, do not establish the study’s actual design assumptions, and should not be interpreted as the trial’s own sample-size justification.

Why can over-dispersion matter?

General statistical context: over-dispersion occurs when observed variability exceeds what a statistical model assumes. If a model understates variability, uncertainty can be understated and intervals can be too narrow. The registry record’s primary outcome analysis is rank-based, and the provided information does not establish that an over-dispersed count model was used. Over-dispersion is therefore a general consideration, not a stated feature of this primary analysis.

What analysis details remain unclear?

The primary population is defined as randomized participants who received at least one dose and had baseline 6MWD. This definition excludes, by its terms, participants who did not meet those conditions. The registry record does not explain how missing Week 24 measurements were handled, whether imputation was used, or whether sensitivity analyses were conducted. It also does not report an interim-analysis plan, the sample-size justification, or the statistical plan’s full details. These omissions limit independent assessment of all assumptions and analysis choices.

10. Analyses Reported by the Sponsor

The registry record reports the primary treatment comparison using an aligned rank stratified Wilcoxon analysis. Its estimate is 40.8 meters, with a 95% confidence interval from 27.53 to 54.14 and p < 0.001. This reported analysis should be distinguished from a simple unadjusted recalculation based on posted group summaries. The arm summaries are medians with full ranges, not means with standard deviations, and the registry record does not provide the participant-level data needed to reproduce the rank-based comparison.

The two group medians, 34.4 meters and 1.0 meters, do not by themselves recreate the reported treatment difference. A rank-based treatment estimate can differ in meaning from a contrast formed by subtracting medians, and the method is described as stratified without the stratification details. No recalculated primary comparison is available from the posted summary statistics in this record. There is therefore no recalculated estimate to use as a competing result.

The sponsor’s reported inferential analysis is the relevant result for describing the trial’s primary comparison. Its confidence interval and p-value belong to that method and should not be attached to a separate contrast derived from the arm summaries. The registry record does not provide further statistical-plan details that would allow an independent reconstruction or fuller examination of the analysis.

11. Multiplicity, Missing Data and Analysis Populations

The primary analysis population is described as all randomized participants who received at least one dose of study treatment and had a baseline 6MWD value. This is a restricted analysis population defined by treatment exposure and availability of a baseline measure. The registry record does not describe an intention-to-treat analysis of all randomized participants, explain how participants outside the stated population were handled, or give the number excluded for any reason.

The registry record does not report how missing Week 24 measurements were handled. It does not specify whether missing values were imputed, whether observed cases alone were analyzed, or whether sensitivity analyses examined alternative assumptions about missing outcomes. Missing-data handling matters because an analysis can be affected when outcome availability is related to treatment assignment or participant status. General statistical context: methods that rely on assumptions about missingness can yield different estimates when those assumptions do not fit the data.

The record lists 12 outcome measures, 10 analyses, and 9 secondary outcome measures. It does not explain how these counts relate to the prespecified analysis hierarchy. The statistical plan’s approach to multiplicity is not reported in the registry record. In the absence of that information, the reported secondary analyses should not be assumed to have been adjusted for the number of comparisons, and isolated secondary p-values would warrant more cautious interpretation than a single prespecified primary test.

The registry record also does not report whether interim analyses were planned or conducted, whether stopping rules were used, or how the sample-size target was justified. The event-count planning illustration is not the trial’s own design justification. No full protocol-level description of analysis populations, missing-data assumptions, or multiplicity control is available in the registry record. These gaps constrain assessment of the robustness and confirmatory status of the estimates, without changing the values reported for the analyses that are documented.

12. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

Deaths

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: hypervolaemia (1.4%), pulmonary arterial hypertension (1.4%), gastrointestinal haemorrhage (1.3%), vascular device occlusion (1.3%), covid-19 (1.3%).

For the double-blind, placebo-controlled period, the registry record lists 23 participants with serious adverse events among 163 at risk in the sotatercept-plus-background-therapy group, and 36 among 160 at risk in the placebo-plus-background-therapy group. It reports 0 deaths in the sotatercept group and 6 in the placebo group for that period. These are event summaries for the recorded groups; they do not by themselves establish whether an event was caused by study treatment.

The serious-event terms listed as most frequent include Hypervolaemia and pulmonary arterial hypertension at 1.4% each, and gastrointestinal haemorrhage, vascular device occlusion, COVID-19, and catheter site infection at 1.3% each. These terms are presented as recorded event categories, not as a complete account of every serious event or an estimate of comparative risk for each category.

The registry record also gives adverse-event summaries for a long-term double-blind period, with separate group counts. Those figures concern a different period from the primary double-blind comparison and should not be combined with its counts. General statistical context: safety counts are descriptive and depend on the period, population at risk, event definitions, and reporting rules. The registry entries given do not supply enough information to make a causal assessment or a comprehensive comparison of specific harms.

13. What the Result Does, and Does Not, Mean

The primary result describes a difference in the Week 24 6MWD outcome between randomized groups receiving different assigned add-on treatments while continuing background PAH therapy. The group medians and the reported rank-based comparison indicate that the observed distributions differed in this trial. The confidence interval and p-value quantify aspects of statistical uncertainty for the specified analysis; they do not settle the clinical importance of the result or its relevance to every patient.

The result does not mean that every participant’s walking distance changed by the group median, nor does it predict an individual outcome. The full ranges show that individual changes varied within each group, but ranges alone do not reveal the distribution’s shape or how much individual-level overlap there was. The reported treatment difference is not a direct difference between the two medians, and its interpretation is tied to the aligned rank stratified Wilcoxon analysis.

The time-to-event analysis is a distinct result over follow-up of up to approximately 18 months. Its hazard ratio does not give an absolute event probability or the number of weeks by which an event was delayed. Median event times and event counts are not reported in the registry record. The result therefore cannot be translated from the available information into an absolute benefit or risk for an individual.

Statistical significance is not the same as clinical importance, and neither the primary nor secondary result establishes effects beyond the trial’s recorded population, treatment context, endpoints, and follow-up. The trial evaluated sotatercept as an addition to background therapy, not as a replacement for it. The registry record does not include the detailed information needed to evaluate all missing-data assumptions, multiplicity procedures, or the full statistical analysis plan. These limits should accompany any reading of the estimates.

14. Why This Trial Matters Statistically

This trial illustrates why a reported group summary and a treatment-effect analysis need not be the same quantity. The primary arm outcomes are medians with full ranges, whereas the treatment comparison is an aligned rank stratified Wilcoxon result. Treating the treatment estimate as a subtraction of the medians would conflate distinct summaries and could misstate what the inferential analysis evaluated.

It also shows the value of reading an estimate together with its uncertainty. The reported treatment difference is paired with a 95% confidence interval and a p-value. Each contributes different information: the estimate gives the analysis’s central comparison, the interval communicates uncertainty on the same scale, and the p-value expresses compatibility with a specified null under the test assumptions. None replaces the others or supplies a complete judgment of clinical relevance.

The secondary time-to-event analysis demonstrates that outcome types require different interpretations. A hazard ratio summarizes event-time comparisons, not a change in distance or an absolute event count. The planning illustration likewise depends on assumptions about the fraction of the observed effect to detect and reports total events for the time-to-event design. It is an illustration rather than evidence of observed power or the trial’s actual design rationale.

Finally, the registry record makes clear how interpretation depends on analysis details beyond headline estimates. The primary population is restricted by treatment receipt and baseline measurement, while handling of missing follow-up data and multiplicity procedures is not reported. Recognizing these details helps distinguish what the posted results support from what remains uncertain, without treating a statistically significant finding as a complete account of the trial.

15. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

The primary outcome population is defined as randomized participants who received at least one dose and had a baseline 6MWD value. The registry record does not explain how this population differs from the overall enrollment, or why the reported arm counts are 163 and 160. It also does not state how participants with missing Week 24 measurements were handled. These omissions limit assessment of possible effects of exclusions and incomplete follow-up.

The primary outcome is summarized by medians and full ranges, and the treatment analysis uses an aligned rank stratified Wilcoxon method. The stratification variables and other implementation details are not reported in the registry record. Without participant-level data, the reported rank-based estimate cannot be independently reproduced from the summary values. The full range gives limited information about distributional shape, while the median summaries do not establish the size of a mean difference.

The registry record does not report a sample-size justification, full statistical analysis plan, interim-analysis details, or multiplicity procedure. It lists 12 outcome measures and 10 analyses but does not explain the counts. Detailed results are not reported for the other listed secondary outcomes in these registry entries, limiting assessment of the overall pattern across outcomes. For the time-to-event endpoint, median times and event counts are not reported, even though a hazard ratio and its confidence interval are given.

The safety summaries separate the double-blind, placebo-controlled period from the long-term double-blind period. Those periods should not be conflated, and the listed serious-event terms do not establish treatment causality. The registry record does not provide full event narratives or the details needed to evaluate all safety comparisons. These limitations are specific to the information recorded and do not, by themselves, determine whether the underlying study data contain further analyses.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: A Study of Sotatercept for the Treatment of Pulmonary Arterial Hypertension (MK-7962-003/A011-11)(STELLAR) — NCT04576988.
• Registry record last updated 2024-09-19; status last verified 2024-09.
• Results first posted 2023-08-23.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

19. Record Summary

The registry record describes a completed Phase 3 randomized, quadruple-masked, parallel-group, placebo-controlled study of sotatercept added to background PAH therapy versus placebo added to background therapy. Enrollment was 324. The primary endpoint was change from baseline in 6MWD at Week 24 among randomized participants who received at least one dose and had a baseline value.

The reported median changes were 34.4 meters for sotatercept and 1.0 meters for placebo. The aligned rank stratified Wilcoxon analysis reported a treatment difference of 40.8 meters, with a 95% confidence interval from 27.53 to 54.14 and p < 0.001. A secondary Log Rank analysis of time to death or first clinical-worsening event reported a hazard ratio of 0.163, with a 95% confidence interval from 0.076 to 0.347 and p < 0.001.

Important analysis details, including missing-data handling, multiplicity procedures, interim analyses, and the sample-size justification, are not reported in the registry record.