A Study to Find Out Whether BI 1015550 Improves Lung Function in People With Idiopathic Pulmonary Fibrosis (IPF): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study to Find Out Whether BI 1015550 Improves Lung Function in People With Idiopathic Pulmonary Fibrosis (IPF) (NCT05321069), a randomized phase 3 study. The primary outcome is Absolute Change From Baseline in Forced Vital Capacity (FVC) [mL] at Week 52.

2. Clinical Question

The study enrolled 1177 participants. The registry describes the study as completed and lists idiopathic pulmonary fibrosis as its condition. The primary endpoint was analyzed across repeated lung-function measurements, with the reported least-squares means referring to Week 52. Secondary outcomes included event-based measures such as exacerbation, hospitalization, and death, as well as symptoms and other measures of lung function. These outcomes address different aspects of disease experience and progression, so they complement but do not replace the primary comparison.

3. Trial Design

This was a phase 3, randomized, double-blind, placebo-controlled, parallel-group trial sponsored by Boehringer Ingelheim. The registry lists quadruple masking. Participants were assigned by chance to one of three groups: nerandomilast 9 mg twice daily, nerandomilast 18 mg twice daily, or matching placebo twice daily. The study description says participants could continue nintedanib or pirfenidone if they were already taking either medicine for IPF.

The registry lists enrollment of 1177 participants. The primary endpoint results include 391 participants in the placebo group, 390 in the 9 mg group, and 392 in the 18 mg group. Those are the analysis counts reported with the primary outcome; the endpoint population also required a baseline and at least one post-baseline measurement. The separate time-to-death endpoint lists 393 participants in placebo, 392 in the 9 mg group, and 392 in the 18 mg group. The distinct counts reflect the populations recorded for those analyses and should not be assumed to be interchangeable.

The study description says participants could remain in the study for up to two and a half years. Site visits were planned 10 times during the first year and every 3 months thereafter, with regular lung-function testing and health checks. The primary endpoint timeframe specifies measurements from baseline, defined as Week -8 to Week -1, and at Week 2, Week 6, Week 12, Week 18, Week 26, Week 36, Week 44, and Week 52. The primary result is the least-squares mean at Week 52 from a longitudinal model using those measurements.

The registry lists the study start as 2022-10-06, primary completion as 2024-08-16, and status as completed. Further details about the randomization procedure, allocation concealment, planned sample-size justification, or any interim analyses are not reported in the registry record.

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 participants were screened for eligibility prior to participation in the trial. Participants attended a specialist site which ensured that they (the participants) strictly met all inclusion and none of the exclusion criteria. Participants were not to be allocated to a treatment sequence if any of the entry criteria were violated.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Full Analysis Set (FAS): This patient set includes all randomized patients who received at least one dose of study drug. only patients with baseline measurement and at least one post-baseline measurement are included.

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: Full Analysis Set (FAS): This patient set includes all randomized patients who received at least one dose of study drug.

6. Primary Endpoint

The registry lists the primary outcome as Absolute Change From Baseline in Forced Vital Capacity (FVC) [mL] at Week 52, measured over The MMRM model is a longitudinal analysis, and it incorporated FVC measurements from baseline (Week -8 to Week -1) and Week 2, Week 6, Week 12, Week 18, Week 26, Week 36, Week 44 and Week 52. The data represent the Least Squares Mean at Week 52.

The primary endpoint is absolute change from baseline in forced vital capacity (FVC), measured in milliliters, at Week 52. FVC is the volume of air that can be forcefully exhaled after a full inhalation. In IPF research, it serves as an objective measure of lung function. Measuring change from baseline focuses the comparison on how the recorded lung-function value changed during the study, rather than comparing only the final readings.

The registry reports least-squares means, which are model-based group estimates rather than simple arithmetic averages. The primary model used repeated FVC measurements over time and reported the estimated Week 52 value. This approach uses the longitudinal measurement pattern in the analysis, while accounting for the model’s stated treatment, baseline antifibrotic use, baseline FVC, and visit terms. The endpoint description identifies an unstructured within-participant covariance structure.

The reported mean changes were negative in all three groups: -183.5 mL for placebo, -138.6 mL for nerandomilast 9 mg BID, and -114.7 mL for nerandomilast 18 mg BID. A negative change indicates that the estimated FVC at Week 52 was lower than baseline on average. The between-group contrasts quantify how far those estimated changes were apart. A contrast favoring a smaller decline is not itself evidence about every participant’s experience, nor does it directly describe symptoms, acute events, or survival. Those are different outcomes, some of which appear among the trial’s secondary endpoints.

FVC is used because it provides a standardized, quantitative measure of respiratory function that can be followed repeatedly. Its interpretation is strongest when considered with the analysis methods, uncertainty intervals, other outcomes, and the limitations of the registry record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The primary endpoint is continuous: it measures change in FVC in milliliters. The registry reports least-squares means and a mixed-model mean difference for each active-treatment comparison with placebo. A least-squares mean is an estimate derived from a statistical model, adjusted for the terms included in that model. It need not equal the raw mean calculated directly from all observed values. The posted primary group estimates are -183.5 mL for placebo, -138.6 mL for nerandomilast 9 mg BID, and -114.7 mL for nerandomilast 18 mg BID.

What does the repeated-measures model do?

The primary analysis used a restricted maximum likelihood (REML)-based mixed model with repeated measures, commonly abbreviated MMRM. The model incorporated FVC measurements at baseline and follow-up visits through Week 52. Treating visit as a repeated measure allows the analysis to use the pattern of observations across visits, rather than treating the Week 52 measurement as the only relevant observation. The registry states that treatment and baseline antifibrotic use were included as categorical effects at each visit, baseline FVC was included as a continuous effect, and visit was treated as a repeated measure. An unstructured covariance structure was used for within-participant variability.

In general, repeated measurements from the same participant tend to be related rather than independent. A covariance structure describes that within-participant relationship. An unstructured structure allows the model to estimate the relationships among visits without imposing a simpler pattern. This flexibility can accommodate varied correlations across visits, although it also depends on adequate data and model estimation. The registry does not provide the fitted covariance estimates or model diagnostics.

REML is a method for estimating model variance components. In general, mixed models can use participants with incomplete repeated measurements without requiring every participant to have observations at every visit. That feature does not eliminate the need to consider why measurements are missing. The registry record does not state the missing-data mechanism assumed, give detailed missingness counts, or describe sensitivity analyses for missing data. Therefore, whether departures from the model’s assumptions affected the result cannot be assessed from the registry record alone.

How should the confidence intervals and p-values be read?

A confidence interval describes the uncertainty around an estimated contrast under the model and its assumptions. The registry reports 95% confidence intervals for both primary comparisons. The interval for 9 mg BID versus placebo is 6.44 to 83.33 mL, around a mean difference of 44.89 mL. The interval for 18 mg BID versus placebo is 30.26 to 107.39 mL, around a mean difference of 68.83 mL. Because each interval is wholly above zero, each reported contrast is compatible with a positive mean difference under the stated analysis. The intervals are not ranges containing a specified proportion of individual treatment effects.

The p-values, 0.022 and 0.001, quantify how unusual results at least as incompatible with a zero contrast would be under a null model and the analysis assumptions. A p-value is not the probability that the null hypothesis is true, nor does it measure the size or clinical importance of an effect. The estimated differences and confidence intervals provide information about magnitude and precision, while the p-values address compatibility with the null model.

What does multiplicity change?

Multiplicity arises when several hypotheses are tested, such as comparisons between two active doses and placebo, or analyses across many outcomes. If tests are treated as independent opportunities to declare a result, the chance of at least one false positive across a family of tests can be greater than for a single test. Statistical plans may address this by defining a testing hierarchy, adjusting significance thresholds, or specifying which outcomes are confirmatory and which are exploratory. The registry record does not report whether or how the primary comparisons or secondary endpoints were adjusted for multiplicity. The posted p-values therefore should be interpreted as reported, without assuming a correction or a particular testing sequence.

General statistical principle: over-dispersion is relevant when observed variability exceeds what a model’s assumed distribution allows. It can make standard errors and resulting uncertainty assessments unreliable if not modeled appropriately. The primary outcome is analyzed with a continuous repeated-measures model, and the registry does not report a separate over-dispersion assessment. This concept should not be taken to imply that over-dispersion was present in this analysis.

What population does the primary estimate represent?

The registry defines the Full Analysis Set as randomized participants who received at least one dose of study drug, with inclusion in this endpoint analysis requiring a baseline measurement and at least one post-baseline measurement. The primary group counts were 391 for placebo, 390 for nerandomilast 9 mg BID, and 392 for nerandomilast 18 mg BID. Thus, these estimates pertain to the defined analysis set, not necessarily every person randomized. The registry does not report a separate intention-to-treat analysis, detailed participant flow for this endpoint, or the number excluded for each criterion. Those details would help readers understand how the analyzed population relates to all randomized participants.

The registry’s endpoint description also lists baseline antifibrotic use among model terms. This matters because participants could continue nintedanib or pirfenidone if already taking them. The description does not give the number using either background medicine in each group or report subgroup-specific primary estimates. The overall treatment contrasts should therefore be read as the modeled comparisons for the analyzed population, not as evidence about a particular background-treatment subgroup.

10. Analyses Reported by the Sponsor

The registry’s sponsor-reported primary analysis used a mixed-model mean difference for each dose compared with placebo. For nerandomilast 9 mg BID, the reported difference was 44.89 mL, with a 95% confidence interval of 6.44 to 83.33 mL and p=0.022. For nerandomilast 18 mg BID, the reported difference was 68.83 mL, with a 95% confidence interval of 30.26 to 107.39 mL and p< 0.001. These are model-based contrasts, consistent with the registry’s description of an MMRM analysis of repeated FVC measurements.

Any recalculated values based on posted group summaries are simple, unadjusted calculations and are not a substitute for the sponsor’s model. The sponsor’s own analysis takes precedence when a recalculation differs, because the sponsor’s model uses the stated repeated-measures structure and covariates, while a summary-statistic calculation cannot reproduce all model details. The registry marks discrepancy as false, so no discrepancy between the reported analysis and a recalculation is flagged.

Even when results are consistent, the sponsor-reported analysis and a simple recalculation answer somewhat different statistical tasks. The former reflects the specified longitudinal model; the latter can only use the posted summary values and does not account for the full participant-level measurement pattern or model covariance. Neither the p-values nor the confidence intervals remove the need to consider missing-data assumptions, multiplicity, and the defined analysis population. Details of any additional sponsor analysis beyond the posted registry results are not reported in the registry record.

11. Multiplicity, Missing Data and Analysis Populations

Two active doses were compared with placebo for the primary FVC endpoint, and the registry also lists a range of secondary outcomes and analyses. When several comparisons are made, multiplicity becomes relevant: the probability of at least one false-positive result can rise if each test is judged without regard to the others. A statistical plan may control that risk through an ordered testing strategy, adjusted thresholds, or another prespecified procedure. Whether such a strategy was used, and how the family of tests was defined, is not reported in the registry record. The displayed p-values should not be assumed to be adjusted or unadjusted beyond what the record explicitly states.

The primary analysis population was the Full Analysis Set, described as randomized participants who received at least one dose of study drug. For this endpoint, the record further requires a baseline FVC value and at least one post-baseline measurement. These criteria mean the primary analysis does not necessarily include every randomized participant. The registry record does not provide detailed counts of missing measurements by visit, reasons for missingness, or the number of participants excluded under each criterion. It also does not report a separate sensitivity analysis showing how results might change under alternative missing-data assumptions.

The MMRM uses repeated observations and can incorporate incomplete measurement histories, but the ability to analyze incomplete data does not prove that missingness is harmless. General statistical principle: an analysis can be biased if the probability of missing data depends on unobserved outcomes in ways not handled by the model. The registry description does not specify the missing-data mechanism or describe sensitivity analyses; no conclusion about the direction or size of any potential bias can be drawn from that absence.

The registry specifies model terms, REML estimation, repeated visit measurements, and an unstructured covariance structure. It does not report diagnostics, the exact handling of protocol deviations, a detailed estimand, or a sample-size justification. These details matter for evaluating how the planned analysis relates to the trial question and how robust the results are to assumptions. The registry record also does not report interim analyses or stopping rules. Their absence from the record should not be interpreted as evidence that particular procedures were or were not used.

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: condition aggravated (8.7%), pneumonia (5.6%), idiopathic pulmonary fibrosis (3.3%), pulmonary hypertension (3.1%), respiratory failure (2.3%).

The safety entries report serious adverse events in 181 of 393 participants in the placebo group, 159 of 392 participants in the nerandomilast 9 mg BID group, and 165 of 392 participants in the nerandomilast 18 mg BID group. These counts describe participants with serious adverse events over the registry’s safety reporting period; they are not the number of distinct event episodes. The listed serious-event terms include condition aggravated at 8.7%, pneumonia at 5.6%, idiopathic pulmonary fibrosis at 3.3%, pulmonary hypertension at 3.1%, respiratory failure at 2.3%, and pneumothorax at 2.0%. The registry does not provide exposure-adjusted rates for these entries in the registry record.

Other adverse events were reported in 316 of 393 placebo participants, 319 of 392 participants receiving 9 mg BID, and 337 of 392 participants receiving 18 mg BID. Death counts in the adverse-event summary were 42, 36, and 26, respectively. These are not the same counts as those in the separate time-to-death endpoint, which reports 28, 26, and 21. The two sets belong to different registry entries and analysis definitions; they should not be combined or treated as directly comparable.

Counts alone do not establish whether treatment caused an event. Safety comparisons can depend on exposure duration, event definitions, baseline health, and how events were collected and coded. The registry record does not give enough detail to assess all such factors or to make a causal assessment from the counts alone. The figures describe reported safety outcomes and should be read alongside the study design and the limits of the recorded information.

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

The primary result describes the average modeled change in FVC at Week 52 for the trial’s analysis population. The two nerandomilast groups had less negative estimated mean changes than placebo, and the reported mean differences were positive with 95% confidence intervals above zero. This is evidence about a group-level lung-function measure under the specified analysis; it is not a prediction that every participant had the same change or experienced an improvement.

The result does not by itself establish what an individual participant would experience, whether symptoms changed, or whether the risk of an acute exacerbation, hospitalization, or death differed. Those outcomes were assessed separately, and their analyses have distinct measures and uncertainty. Nor does a Week 52 result answer questions about outcomes after the follow-up represented in the record. The registry describes a study lasting up to two and a half years, but the primary endpoint is explicitly a Week 52 measure.

A statistically distinguishable group contrast is not automatically a measure of clinical importance. The registry result quantifies the difference in FVC change and its statistical uncertainty, but the registry record does not state a threshold for clinical importance or provide a patient-level interpretation of the difference. The result also depends on the model, analysis population, and assumptions about incomplete repeated measurements. The record does not report a detailed missing-data sensitivity analysis or the multiplicity approach used across comparisons and outcomes.

These findings should therefore be understood narrowly: among the participants included in the primary analysis, the modeled mean FVC decline by Week 52 was smaller in each nerandomilast group than in placebo. They do not show that all participants benefited, do not settle the interpretation of other endpoints, and do not substitute for a complete assessment of uncertainty, safety, and the trial’s full analysis plan. The registry information supports description of the posted results, not individualized medical decisions.

14. Why This Trial Matters Statistically

This trial is statistically informative because it uses a randomized, placebo-controlled, parallel-group design to compare two doses with a common comparator. Random allocation is intended to make treatment groups comparable on average at the start of a trial, reducing confounding as an explanation for between-group differences. Masking can also reduce the influence of expectations on care and outcome assessment. The registry identifies quadruple masking, though it does not define the roles covered by that label.

The primary outcome is based on repeated FVC measurements rather than a single isolated visit. The MMRM incorporates observations across baseline and scheduled follow-up, producing modeled Week 52 means and contrasts. This approach makes use of longitudinal data and allows the estimates to reflect the pattern across visits, subject to the model assumptions and available observations. It also makes clear why a simple comparison of raw endpoint values would not necessarily reproduce the sponsor’s analysis.

The posted results offer estimates and confidence intervals, not only significance labels. That matters because the size and precision of the contrast are central to interpretation. The primary contrasts were 44.89 mL and 68.83 mL, with confidence intervals that quantify uncertainty under the model. The record also includes secondary event-based and symptom outcomes, which address questions that FVC alone cannot answer. Their value depends on careful attention to analysis populations, event definitions, and multiplicity.

The trial also illustrates the importance of separating efficacy estimation from safety description and of distinguishing endpoint-specific counts. The safety summary and time-to-death endpoint report different death counts, underscoring that data labels and analysis definitions matter. More broadly, the registry record contains useful information about design and primary modeling, while leaving some inferential details unreported. A statistically careful reading acknowledges both what the data estimate and the boundaries of what can be concluded.

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 registry record does not provide a full statistical analysis plan, so it is not possible to verify the prespecified hierarchy of outcomes, multiplicity adjustment, or the planned role of each secondary analysis. It also does not report the sample-size justification, interim analyses, or stopping rules. Those omissions constrain assessment of how the posted analyses compare with the original plan; they do not establish that such planning or procedures were absent.

The primary analysis set has specific inclusion criteria: randomized participants had to receive at least one dose, and the endpoint analysis required baseline and at least one post-baseline FVC measurement. The record does not give detailed participant flow or missing-measurement patterns by treatment group and visit. While the MMRM description gives the repeated-measures structure, the missing-data mechanism and sensitivity analyses are not reported in the registry record. This leaves uncertainty about the robustness of the primary result to alternative assumptions about missing outcomes.

The registry reports model-based means and contrasts but does not supply participant-level data, fitted model diagnostics, or enough detail to reproduce the analysis independently. The record identifies baseline antifibrotic use as a model term but does not provide subgroup estimates or group-specific counts for background treatment. It also does not state a threshold for clinical importance of the observed FVC differences, limiting direct assessment of their clinical meaning from the registry alone.

Some secondary endpoints are time-to-event outcomes, while others use thresholds or symptom scores. Their analysis populations, event definitions, and follow-up can differ. The registry does not report a multiplicity procedure, so the confirmatory status of secondary findings cannot be established from the record. Safety counts also need careful interpretation: the deaths listed in the safety summary differ from those in the separate time-to-death endpoint, and the registry entries do not provide enough detail to reconcile the definitions fully. These limitations argue for endpoint-specific interpretation rather than treating all posted results as one overall test.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: A Study to Find Out Whether BI 1015550 Improves Lung Function in People With Idiopathic Pulmonary Fibrosis (IPF) — NCT05321069.
• Registry record last updated 2025-10-20; status last verified 2025-10.
• Results first posted 2025-09-12.
• 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 describes a completed phase 3 randomized, double-blind, placebo-controlled, parallel-group trial of nerandomilast in adults with idiopathic pulmonary fibrosis. Enrollment was 1177. Participants were assigned to nerandomilast 9 mg BID, nerandomilast 18 mg BID, or placebo; the record lists quadruple masking. The primary endpoint was absolute change from baseline in FVC at Week 52, analyzed using a REML-based MMRM with repeated measurements through Week 52.

The reported least-squares mean changes were -183.5 mL for placebo, -138.6 mL for 9 mg BID, and -114.7 mL for 18 mg BID. The sponsor-reported mean differences versus placebo were 44.89 mL and 68.83 mL, respectively, with 95% confidence intervals above zero. Secondary outcomes cover events, lung-function thresholds, and symptoms; safety entries report serious adverse events and other adverse events by group.

The record gives the central design and primary model details, but multiplicity handling, detailed missing-data analyses, and the sample-size justification are not reported in the registry record.