Efficacy and Safety of Benralizumab in EGPA Compared to Mepolizumab.: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Efficacy and Safety of Benralizumab in EGPA Compared to Mepolizumab. (NCT04157348), a randomized phase 3 study. The primary outcome is Number of Subjects Who Achieved Main Remission at Both Weeks 36 and 48.

2. Clinical Question

The trial was randomized, double-blind, active-controlled, and parallel-group. Those design features frame the interpretation: the primary endpoint compares the groups’ recorded remission status at specified assessment times, while the analysis record supplies both a logistic-regression result and figures from which an unadjusted comparison can be recalculated. The two analyses answer closely related but not necessarily identical statistical questions, so their estimates and confidence intervals should be kept distinct.

3. Trial Design

MANDARA was a Phase 3 randomized, double-blind, active-controlled, parallel-group study of benralizumab and mepolizumab for EGPA. The registry describes a 52-week double-blind treatment period, followed by an open-label extension. The extension was intended to allow eligible patients who completed the double-blind period to continue on open-label benralizumab; the registry notes that earlier-enrolled patients could remain in the extension longer than the intended minimum period.

The two arms received different active drugs with matching placebo injections: the benralizumab arm received one benralizumab injection and three placebo-to-mepolizumab injections, while the mepolizumab arm received three mepolizumab injections and one placebo-to-benralizumab injection. The study is described as triple-masked, but the identities of the roles covered by that masking are not reported in the registry record.

The registry lists enrollment as 140 and the primary-outcome arm counts as 70 for benralizumab and 70 for mepolizumab. Enrollment and the analysis population are not interchangeable: the Full Analysis Set is defined as randomized participants who received at least one dose of investigational product, irrespective of adherence or continued participation. The primary outcome counts therefore need to be read in the context of that stated population definition.

The primary outcome was assessed at Week 36 and Week 48 and required main remission at both assessments. Secondary outcomes include accrued time in remission, time to first EGPA relapse, annualized relapse rate, and measures of prednisolone or prednisone dose. The registry lists 32 outcome measures and 17 analyses. The record also includes results for the open-label extension, which should be distinguished from the randomized double-blind comparison when interpreting treatment-group contrasts.

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: All patients completed a screening period of 1-4 weeks during which inclusion/exclusion criteria was assessed, disease activity, corticosteroid medication usage and patient reported outcomes were recorded, medical history and clinical laboratory were taken. All patients who entered open label extension period received benralizumab 30 mg during open label extension period.

Double-blind Treatment Period

Reasons for not completing this period:

Open-label Extension Period

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Full Analysis Set: All patients randomized and receiving at least one (1) dose of IP are included in the full analysis set, irrespective of their protocol adherence and continued participation in the study.

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: All participants who were randomized and received any IP were included in the full analysis set.

6. Primary Endpoint

The registry lists the primary outcome as Number of Subjects Who Achieved Main Remission at Both Weeks 36 and 48, measured over Week 36 and Week 48.

The primary endpoint is the number of participants who achieved main remission at both Week 36 and Week 48. The registry defines main remission as BVAS = 0 together with an oral corticosteroid dose of ≤ 4 mg/day. The endpoint is therefore a binary participant-level outcome: a participant either met the stated remission definition at both specified assessments or did not.

Requiring remission at both assessments makes the endpoint different from remission at a single visit. It records whether the definition was met at the two named time points, rather than describing the duration of remission between visits. The registry lists total accrued duration of remission and sustained remission as separate secondary outcomes, as well as relapse outcomes. Those distinct measures should not be treated as interchangeable with the primary endpoint.

A remission endpoint combines disease activity and corticosteroid dose in a single definition. In general, such a composite can be useful when a clinical question concerns both components together. The observed result consequently applies to the combined definition; it does not, by itself, show how the groups compared separately on BVAS or corticosteroid dose. Separate component results are not included in the primary endpoint information.

The posted primary counts were 40 participants in each arm, with 70 participants listed in each arm. Equal counts do not establish that the endpoint components or other outcomes were equal. They indicate only that the same number in each group met the registry’s stated main-remission criterion at both assessment times.

7. Results Posted in the Registry

• Mepolizumab 300 mg: 57.1% (40/70) versus 57.1% (40/70) on control; risk difference 0.0 percentage points (95% CI -16.4 to 16.4), risk ratio 1.00 (0.75–1.33); unadjusted two-proportion z-test p = 1.000.

8. Statistical Interpretation

9. Statistical Methods Explained

The primary outcome is binary, so the basic comparison concerns the proportion of participants meeting the remission definition in each arm. A risk difference expresses the difference between those proportions on the probability scale. A value of 0.0 means the posted proportions are equal at the point estimate. Unlike a relative measure, the risk difference is expressed as an absolute difference, which can make the direction and magnitude of a group contrast easier to interpret directly.

What does the unadjusted recalculation represent?

The recalculated estimate uses the posted event counts and arm sizes without adjustment for other variables. For this outcome, each arm has a count of 40 among 70 participants. The posted arm proportions are 0.571 and 0.571, and the recalculated risk difference is 0.0. The calculation is a simple comparison of the summary counts; it does not reproduce the sponsor’s logistic-regression analysis.

An unadjusted comparison is useful as a transparent check on the direction and scale of the observed group contrast. It does not account for any covariates or modeling choices. The registry record does not state which covariates, if any, were included in the reported regression model, so a more specific account of why the regression estimate differs is not possible from this record alone.

What does logistic regression contribute?

Logistic regression is a model for a binary outcome. In general, it estimates how the odds of an event vary with group and, where included, other explanatory variables. Results may then be reported on a scale such as a difference in remission rates. The registry reports a logistic-regression analysis with a difference in remission rates of 0.012. The model specification and any covariate details are not reported in the registry record, so the reported estimate should be described as the registry’s regression result rather than as a fully reproducible model.

Model-based estimates and simple unadjusted estimates can differ because they are generated using different statistical procedures and may target different quantities. That general possibility does not identify the cause of the difference in this trial. Here, the sponsor’s result is 0.012 and the unadjusted recalculation is 0.0; both should be quoted with their own uncertainty intervals and methods.

How should the p-values be read?

A p-value describes how compatible the observed data are with a specified null hypothesis under the analysis assumptions. It is not the probability that the null hypothesis is true, and it does not measure the clinical importance of an effect. The unadjusted recalculation has p = 1.0, while the reported regression analysis has p = 0.877. Neither is evidence of a statistically detectable difference at conventional thresholds, but a non-significant result does not prove equality or equivalence.

Confidence intervals add information that a p-value alone does not convey. They show the range of values compatible with the estimate under the method’s assumptions. The recalculated interval is -0.164 to 0.164; the reported regression interval is -0.141 to 0.165. Both include zero and allow effects in either direction. Their widths are important: the point estimates alone would give an incomplete account of the uncertainty.

What is the role of sample-size planning illustrations?

The planning illustrations are based on the observed effect, use a two-sided significance level of 0.05, equal allocation, and a normal approximation. They scale the effect to detect as a stated fraction of the observed effect. For binary outcomes, the listed planning quantities refer to participants per arm. These are planning illustrations, not observed power and not the trial’s original sample-size justification. The registry does not report the trial’s sample-size justification in this record. Planning based on an observed result also depends on assumptions and should not be interpreted as proof that the trial had a particular ability to detect an effect.

Why can over-dispersion matter in general?

For outcomes analyzed as counts or rates, over-dispersion means that variability is greater than a model’s basic assumptions allow. If a model underestimates variability, its standard errors may be too small and its uncertainty intervals too narrow. This is a general consideration for count and rate outcomes; the primary endpoint is a binary participant count, and the registry does not report an over-dispersion issue for this primary analysis.

10. Confidence Intervals and Precision

For the unadjusted recalculation, the risk difference is 0.0 and its 95% confidence interval extends from -0.164 to 0.164. This interval includes zero, which is consistent with no difference in the underlying proportions under the unadjusted analysis. It also includes values below and above zero, so the data do not pinpoint the direction of a possible difference with precision.

The reported logistic-regression estimate is 0.012, with a 95% confidence interval from -0.141 to 0.165. This interval also includes zero and covers negative and positive differences. The two intervals are similar in that both include zero and allow effects in either direction, but they are not identical. They arise from different analyses and should not be merged or substituted for each other.

A confidence interval is not a guarantee that the true effect lies within its endpoints. In general, it reflects the range of estimates compatible with the observed data under the method’s assumptions. Its width communicates uncertainty around the point estimate. Here, equal observed counts and an unadjusted point estimate of 0.0 coexist with an interval that permits nonzero differences; the point estimate alone therefore gives an incomplete picture.

The registry reports a 95% interval for the regression analysis. The recalculated interval is also described as a 95% confidence interval. Neither interval establishes equivalence, because an equivalence conclusion requires a defined equivalence margin and an analysis designed to assess that question. Such a margin is not reported in the registry record. The interval information supports a measured interpretation: the posted analyses do not detect a difference, while their uncertainty remains compatible with effects in either direction.

11. Analyses Reported by the Sponsor

The registry’s reported analysis is a logistic regression for the difference in remission rates. It estimates 0.012, with a 95% confidence interval from -0.141 to 0.165 and p = 0.877. This is the sponsor’s reported analysis in the registry record and should be distinguished from the unadjusted recalculation based directly on the posted counts.

The unadjusted recalculation gives a risk difference of 0.0, with a 95% confidence interval from -0.164 to 0.164 and p = 1.0. It is a simple re-analysis of 40 participants with the endpoint in each arm and 70 participants listed in each arm. The registry record does not give sufficient model detail to reproduce the logistic-regression result or account for the precise difference between the estimates.

The record does not mark a discrepancy. The methods nevertheless differ, so estimates and intervals should be presented with their analysis labels rather than treated as interchangeable. Both confidence intervals include zero, and neither analysis reports a statistically detectable difference. The regression analysis remains the registry’s reported result; the recalculation is a transparent unadjusted comparison, not a replacement for that reported analysis.

12. Multiplicity, Missing Data and Analysis Populations

The registry lists 32 outcome measures, 17 analyses, and 30 posted secondary outcomes. With many outcomes and analyses, the chance of finding at least one apparently notable result can increase if each is interpreted without regard to the others. In general, multiplicity procedures can control or describe this issue by accounting for the number and structure of statistical tests.

A multiplicity strategy for the primary and secondary analyses is not reported in the registry record. The record therefore does not establish whether secondary-outcome p-values were adjusted, whether outcomes were organized into testing families, or whether a hierarchical testing procedure was used. The primary result should be interpreted as the reported analysis for its stated endpoint; unreported multiplicity details should not be inferred.

The primary outcome description defines the Full Analysis Set as all randomized participants who received at least one dose of investigational product, regardless of protocol adherence or continued participation. That definition identifies the intended analysis population, but it does not by itself explain how missing outcome observations were handled. Imputation methods, rules for incomplete assessments, and sensitivity analyses addressing missing data are not reported in the registry record.

Missing outcome information can affect an analysis if the likelihood of an observation being missing is related to treatment or outcome status. In general, methods for handling missing data rely on assumptions, and sensitivity analyses can examine how conclusions change under different assumptions. Because the registry record does not report these details for the primary outcome, the posted counts should be read alongside that limitation rather than treated as a complete account of missing-data handling.

The record also does not report an interim-analysis plan or stopping boundaries. These details should not be presumed from the trial’s phase or design. The study is listed as active, not recruiting; that status does not supply additional information about the analysis plan or the handling of missing results.

13. 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: eosinophilic granulomatosis with polyangiitis (3.2%), adrenal insufficiency (3.2%), prostate cancer (2.9%), appendicitis (1.6%), cellulitis (1.6%).

During the double-blind period, the registry lists serious adverse events for 4 participants in the benralizumab group and 9 in the mepolizumab group, with 70 at risk in each group. The listed counts for other adverse events are 46 and 56, respectively. These are event-category counts as recorded; the registry information quoted does not establish causal attribution or provide adjusted comparisons.

For the open-label extension, the registry lists 10 serious adverse events among 66 at risk in the benralizumab group and 13 among 62 at risk in the group switched from mepolizumab to benralizumab. Other adverse-event counts are 43 and 46, respectively. One death is listed in the switched-to-benralizumab extension group; zero deaths are listed in each of the other groups.

The registry also lists serious-event terms with percentages, including eosinophilic granulomatosis with polyangiitis and adrenal insufficiency at 3.2%, prostate cancer at 2.9%, and appendicitis, cellulitis, and Clostridium difficile infection at 1.6% each. These terms and percentages should not be assigned to a particular randomized arm unless the record does so. The extension groups are not equivalent to a continued randomized comparison, because the registry describes one group as switched to benralizumab. Safety counts are descriptive and do not, by themselves, establish whether a treatment caused an event.

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

The primary result describes how many participants in each randomized arm met the registry’s main-remission definition at both Week 36 and Week 48. The posted counts are equal: 40 participants in each arm, with 70 listed in each arm. The unadjusted recalculation therefore has a risk-difference point estimate of 0.0. The reported logistic-regression estimate is 0.012. Both analyses have confidence intervals that include zero.

This result does not establish that benralizumab and mepolizumab are equivalent. No equivalence margin or equivalence analysis is reported in the registry record. It also does not establish that there is no difference: both intervals include nonzero values in either direction. The appropriate reading is that the reported analyses do not detect a difference in this primary outcome, while the intervals retain uncertainty about the size and direction of a possible difference.

The result concerns a specific combined endpoint: BVAS = 0 and oral corticosteroid dose of ≤ 4 mg/day at both specified assessment times. It does not separately quantify the treatment-group comparison for BVAS, corticosteroid dose, remission duration, or relapse. Those are distinct measures, and the registry lists several of them among the secondary outcomes.

The comparison is between two active treatments given in a randomized, double-blind design. It does not answer how either treatment compares with placebo or with no treatment. Nor should the randomized double-blind result be extended automatically to the open-label extension, where the record describes continued benralizumab and a group switched from mepolizumab to benralizumab.

Finally, a non-significant p-value is not a measure of clinical importance and does not resolve all uncertainty. The reported p-values are 1.0 for the unadjusted recalculation and 0.877 for the regression analysis. Their interpretation depends on their respective methods, and the regression model details and missing-data methods are not reported in the registry record.

15. Why This Trial Matters Statistically

This trial presents a useful statistical comparison because the primary outcome is clinically interpretable as a defined remission state at two named assessment times, and the comparator is an active treatment. The result is not simply a comparison of treatment with no treatment; it addresses the relative remission outcomes recorded for benralizumab and mepolizumab within the randomized, double-blind period.

The matching placebo injections support the blinded comparison while the active treatments use different injection regimens. The randomized parallel-group design provides a framework for estimating a between-group contrast. For interpretation, however, design labels do not replace examination of the outcome definition, analysis population, and uncertainty interval. In this case, the endpoint combines disease activity and corticosteroid dose, and the estimates are close to zero while their intervals still include effects in either direction.

The record also illustrates why a reported model-based result and a simple recalculation should be kept separate. The regression estimate is 0.012, whereas the unadjusted risk difference is 0.0. Both analyses are non-significant, but their estimates and intervals differ. Reporting the method alongside the estimate helps readers understand which statistical question each result addresses, rather than treating every comparison as the same calculation.

Finally, the registry includes multiple secondary outcomes and open-label extension information. Distinguishing the primary randomized comparison from those other results matters: different outcomes measure different aspects of EGPA and treatment, while extension results do not preserve the same randomized comparison described for the double-blind period. The record’s missing details on multiplicity, missing-data handling, and regression specification are also relevant to how fully the estimates can be assessed.

16. 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 registry record provides the primary event counts and a reported regression result, but it does not report the regression model specification or identify any covariates used. That limits the ability to reproduce the reported estimate or explain exactly why it differs from the unadjusted recalculation. The record also does not report the trial’s sample-size justification, so the planning illustrations based on the observed effect should not be mistaken for the original design rationale.

Missing-data methods and sensitivity analyses for the primary endpoint are not reported in the registry record. The Full Analysis Set definition states which participants are included in principle, but does not describe how absent Week 36 or Week 48 measurements were treated. The record also does not report a multiplicity strategy for the many listed outcomes and analyses. These omissions limit assessment of the broader inferential plan.

The primary endpoint is a combined remission definition. It does not show separate treatment-group results for BVAS and corticosteroid dose, and it records remission at both specified visits rather than the duration of remission between them. Although the registry lists remission-duration and relapse outcomes as secondary measures, those outcomes answer different questions.

Safety results include double-blind and open-label extension groups. The extension includes a group switched from mepolizumab to benralizumab, so those figures should not be read as a continuation of the original randomized comparison. The record’s serious-event terms and percentages do not provide arm-specific attribution for each listed term. These limitations call for restrained interpretation of both efficacy and safety findings from the registry record.

17. Related Tutorials

18. Related Statistical Calculators

19. Sources

• ClinicalTrials.gov: Efficacy and Safety of Benralizumab in EGPA Compared to Mepolizumab. — NCT04157348.
• Registry record last updated 2026-08-12; status last verified 2026-07.
• Results first posted 2025-01-14.
• 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.

20. Record Summary

NCT04157348 is a Phase 3 randomized, triple-masked, parallel-group study of benralizumab 30 mg compared with mepolizumab 300 mg for patients with relapsing or refractory EGPA receiving corticosteroid therapy, with or without stable immunosuppressive therapy. The registry describes a 52-week double-blind treatment period and an open-label extension. Enrollment is listed as 140.

The primary endpoint is achievement of main remission at both Week 36 and Week 48, defined as BVAS = 0 and oral corticosteroid dose of ≤ 4 mg/day. The posted count is 40 of 70 in each arm. The unadjusted recalculation gives a risk difference of 0.0, with a 95% confidence interval from -0.164 to 0.164 and p = 1.0. The reported logistic-regression estimate is 0.012, with a 95% confidence interval from -0.141 to 0.165 and p = 0.877.

Both analyses have intervals including zero; neither establishes equivalence. The registry does not report the regression specification, missing-data methods, or multiplicity strategy. Double-blind safety counts and open-label extension counts should be interpreted separately.