Inhaled GM-CSF Therapy of Autoimmune PAP: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Inhaled GM-CSF Therapy of Autoimmune PAP (NCT00901511), a randomized phase 2 study. The primary outcome is Time to Rescue WLL.

2. Clinical Question

The trial addresses a treatment strategy question rather than a comparison of GM-CSF with no lavage: baseline WLL was scheduled for all participants, with inhaled GM-CSF added in one group. The primary endpoint is time to the first rescue WLL after that common baseline procedure. The registry reports a median of 18 months for the control group and 30 months for the GM-CSF group. It also reports a Wilcoxon (Mann-Whitney) comparison, with a median difference of 12 and p = 0.008.

These results describe the posted endpoint and its statistical comparison in this trial. They do not by themselves establish how the intervention would perform in other populations or under other treatment strategies. The registry record does not report a confidence interval for the between-group result, and the posted results do not provide participant-level outcome histories. The available record therefore supports a focused reading of the time-to-rescue endpoint, not a broader assessment of all possible benefits and harms.

3. Trial Design

The registry identifies this as a randomized, open-label, parallel-group phase 2 study of inhaled granulocyte/macrophage-colony stimulating factor following WLL in people with autoimmune pulmonary alveolar proteinosis. Enrollment is listed as 18. The record gives a study start of 2009-07, primary completion of 2016-06, and status of completed. Masking is listed as none, so the trial was not masked according to the registry record.

Both groups were scheduled to receive bilateral WLL at baseline, identified as study month 0. The GM-CSF group was then assigned inhaled treatment beginning 1 week after that procedure. The described schedule is 250 mcg daily for 7 consecutive days every other week for 12 weeks. The description also includes a washout, without further detail in the record. The control description specifies baseline WLL and unscheduled rescue WLL if disease progression occurred. The complete control-arm progression criteria are not reported in the registry record.

The main endpoint was time from baseline WLL to the first unscheduled rescue WLL during 30 months. The record lists 9 participants in each group for this endpoint. In the reported results, the median was 18 months in control and 30 months in the GM-CSF group. The reported dispersion is expressed as an interquartile range: 6 to 27 months for control and 30 to 30 months for GM-CSF.

Random allocation and a parallel-group structure support a comparison between assigned groups, while the open-label design means treatment assignment was not masked. The registry record does not report the randomization procedure, allocation concealment, or details of how endpoint timing was assessed. It also does not report a sample-size justification or a detailed statistical analysis plan. Those omissions limit what can be concluded about the design beyond the features explicitly recorded.

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.

Overall 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.

6. Primary Endpoint

The registry lists the primary outcome as Time to Rescue WLL, measured over 30 months.

The primary endpoint is time to rescue WLL. It measures the interval in months between scheduled baseline WLL at study month 0 and the first administration of an unscheduled rescue WLL during the 30 months after baseline. In practical terms, it describes when a participant first underwent an additional lavage under the study’s rescue approach, rather than the number of lavages or a direct measurement of lung function.

A time-to-event endpoint can retain information about when an event occurs, rather than reducing the outcome to whether it occurred by a single fixed date. The median is the point at which the observed time distribution is divided into two halves; it is a measure of the distribution’s center, not an average. The registry reports medians of 18 months in control and 30 months in the GM-CSF group. It gives interquartile ranges of 6 to 27 months and 30 to 30 months, respectively.

The endpoint is relevant to the study’s focus on whether participants proceeded to rescue lavage after baseline treatment. However, it is a procedure-based outcome and should not be treated as interchangeable with oxygenation, symptoms, or other measures of disease status. The registry lists several secondary outcomes concerning gas exchange, lung function, imaging, and serum markers, but the primary result addressed is time to rescue WLL. The record does not report participant-level event and censoring details or explain how participants without a rescue event were handled in the endpoint analysis.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The registry reports a Wilcoxon (Mann-Whitney) test for the primary between-group comparison. This is a rank-based method: broadly, it compares the relative ordering of observations across groups rather than requiring the endpoint values to follow a normal distribution. The test is often used when a comparison is framed around distributions or ranks. It should not automatically be interpreted as a test of medians alone unless assumptions about the distributions support that interpretation.

What does the median describe?

A median is the middle of a distribution in the sense that half of the observations lie on either side, allowing for ties. It is less influenced by unusually large or small values than an arithmetic mean. For a time-to-event endpoint, the reported median is a summary of event timing, not a statement that each participant experienced the event at that time. The registry lists 18 months for control and 30 months for GM-CSF. The endpoint’s stated observation window is 30 months, and the GM-CSF median is reported as 30 months.

The interquartile range describes the central half of a distribution, extending from the first to the third quartile. The record gives 6 to 27 months for control and 30 to 30 months for GM-CSF. This is descriptive spread, distinct from a confidence interval for a group estimate or a between-group effect. The equal reported quartiles in the GM-CSF group do not, on their own, explain the underlying event pattern. Participant-level event times and censoring information are not reported in the registry record.

How should the p-value be read?

The reported p-value is 0.008. In general, a p-value measures how unusual data at least as incompatible with a specified null model would be if that model and the test assumptions held. It does not measure the size of an effect, the chance that the null is correct, or the probability that the result is due to chance in an everyday sense. Here, the p-value belongs to the reported Wilcoxon (Mann-Whitney) comparison, and it should be interpreted alongside the endpoint summaries and the limitations of the record.

The analysis is identified as a superiority comparison. The registry does not report the detailed hypothesis statement, test implementation, handling of ties, or a statistical analysis plan. It also does not report a confidence interval for the median difference. Without an interval, readers cannot use the reported comparison to assess the range of effects compatible with the data through that standard form of uncertainty summary.

Why does time-to-event analysis need special care?

Time-to-event data can include participants whose event has not been observed by the end of follow-up or whose event timing is otherwise incomplete. Such observations are commonly handled as censored, but the registry record does not report the number of censored observations, the censoring pattern, or the method used to handle them for the primary result. The endpoint is described as time to the first rescue WLL during 30 months, and the summaries are reported as medians and interquartile ranges. The record does not give enough detail to reconstruct an event-time analysis independently.

General statistical practice distinguishes a rank-based comparison of observed values from methods that explicitly model event timing and censoring. The record identifies the test as Wilcoxon (Mann-Whitney), but does not describe any survival model or report a hazard ratio. The kind of outcome is time-to-event; no sponsor-reported hazard ratio is included among the primary result details. The displayed p-value and median summaries should therefore not be recast as a hazard ratio or as a modeled effect on an event rate.

In general, confidence intervals communicate the range of effect estimates that remain reasonably compatible with the data under a specified method. They also indicate precision. No confidence interval is reported for the primary between-group median difference, so the uncertainty around that contrast cannot be characterized from an interval in the registry result. The interquartile ranges do not fill this role: they describe variation among participant outcomes rather than uncertainty about the treatment contrast.

What do recalculations contribute?

Simple unadjusted recalculations from posted summary statistics can help readers examine a result from a different statistical perspective, but they may not reproduce the sponsor’s analysis. For this primary endpoint, the registry result is a time-to-event outcome and the record specifies that only sponsor-reported hazard ratios are used for such endpoints. No recalculated effect estimate or confidence interval is reported for this comparison. The sponsor’s reported Wilcoxon result is therefore the relevant inferential result in the registry record.

Planning illustrations based on an observed effect are not evidence of the trial’s actual power. They describe hypothetical future designs under stated assumptions, rather than the probability that this completed trial detected an effect. The registry record does not report the trial’s sample-size justification, and no planning illustration should be mistaken for a design feature or an observed result.

10. Analyses Reported by the Sponsor

The sponsor-reported primary analysis compares the groups using a Wilcoxon (Mann-Whitney) test. It reports a median difference of 12 and p = 0.008, and classifies the comparison as superiority. The group medians are 18 months for control and 30 months for GM-CSF. No confidence interval accompanies the reported median difference.

No recalculated between-group estimate is reported for this time-to-event endpoint. Recalculations from group summary statistics would be unadjusted and would not necessarily reproduce the sponsor’s method or account for event-time features such as censoring. For this type of endpoint, the sponsor’s own reported analysis takes precedence over a simple recalculation. The absence of a recalculated estimate is not evidence that the sponsor result is incorrect; it reflects the distinction between the reported analysis and a summary-statistic recalculation.

The sponsor’s result should still be read with the limitations of the registry record in view. The reported p-value does not substitute for an effect confidence interval, and the record does not describe participant-level event timing, censoring, missing endpoint data, or the full statistical analysis plan. These omissions constrain independent assessment of the analysis and its precision.

11. Multiplicity, Missing Data and Analysis Populations

The registry lists 14 outcome measures and 13 secondary outcomes posted. It also records 103 analyses. The primary endpoint is time to rescue WLL, while the secondary outcomes include rescue WLL occurrence and group differences in oxygenation, lung function, imaging, and serum markers. These outcomes cover different clinical and biological dimensions and should not be conflated with the primary time-to-event result.

When several outcomes or analyses are examined, the chance of finding at least one apparently notable result can increase unless multiplicity is addressed. The registry record does not report whether or how the primary and secondary analyses were adjusted for multiple comparisons, whether outcomes were organized in a testing hierarchy, or which analyses were prespecified. The reported p-value should therefore be understood as the result of the stated comparison, not as evidence that all analyses across the record share a single error rate or confirmatory status.

The record does not report a detailed statistical analysis plan, interim analyses, or a sample-size justification. It also does not report the analysis population definition for the primary endpoint, such as whether the comparison followed intention-to-treat principles, nor does it describe exclusions or protocol deviations. These are unreported details, not evidence that a particular procedure was or was not followed.

Missing endpoint data and censoring are especially relevant for time-to-event outcomes. The registry describes follow-up through 30 months but does not report how many participants had an observed rescue event, how many were censored, the reasons for any incomplete follow-up, or the method used to account for incomplete observation. The GM-CSF interquartile range is reported as 30 to 30 months, but that summary alone cannot identify the individual event or censoring pattern.

In general, multiplicity and missing-data methods affect how confidently results can be interpreted, especially when many outcomes and analyses are reported. Because the corresponding methods are not reported in the registry record, the primary comparison can be described as posted, but its position within a broader prespecified analysis framework cannot be established from this record.

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 0% reporting threshold.

The posted adverse-event summary lists 9 participants at risk for serious adverse events in each group. It reports 0 participants affected by serious adverse events in the control group and 0 in the GM-CSF group. The record also lists 9 participants at risk for other adverse events in each group and 0 affected in either group. Deaths are reported as 0 in both groups.

These figures are counts within the adverse-event summary. They do not establish that no adverse events of any kind occurred outside the categories or collection procedures represented in that summary. The registry lists no top serious adverse-event terms. Further detail on event definitions, ascertainment, timing, severity gradation, or the duration of safety follow-up is not reported in the registry record.

With a small number of participants listed in each arm, zero observed serious adverse events should be read as a description of the posted trial data, not as proof that a treatment strategy has no serious risks. The registry record does not provide a confidence interval for adverse-event rates or other information that would quantify uncertainty around these counts. The safety summary also does not establish comparative safety beyond the trial and its recorded follow-up.

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

The primary result describes longer reported median time to rescue WLL in the GM-CSF group than in the control group: 30 months versus 18 months. The registry reports a median difference of 12 and a Wilcoxon (Mann-Whitney) p-value of 0.008. The interquartile ranges are 30 to 30 months for GM-CSF and 6 to 27 months for control. These summaries describe the posted comparison; no confidence interval for the between-group difference is reported.

The result does not compare a strategy with baseline WLL against a strategy without it. Both groups were scheduled to receive baseline bilateral WLL, and the treatment distinction was the addition of inhaled GM-CSF in one group. Therefore, the primary result concerns the timing of rescue WLL after baseline lavage under the trial’s assigned strategies. It does not isolate the effect of GM-CSF in the absence of baseline WLL.

The endpoint is a procedure-based measure. It does not by itself demonstrate changes in oxygenation, lung function, symptoms, imaging, or serum markers. Although the registry lists those as secondary outcomes, their results are not included in the primary endpoint details. The primary comparison also does not establish effects beyond the stated 30-month endpoint window.

The small group sizes, open-label design, and absence of a confidence interval are important to the precision and interpretation of the result. The record does not report a detailed account of event timing, censoring, missing data, analysis populations, multiplicity handling, or the statistical analysis plan. Those missing details limit the ability to assess the analysis independently and to determine how the primary comparison fits within the full set of reported analyses.

In general, a statistically notable result can coexist with substantial uncertainty about its magnitude, generalizability, and robustness. The p-value addresses compatibility with a null model for the specified comparison; it does not settle those broader questions. The posted data support describing the observed group summaries and sponsor-reported test, while leaving wider clinical and methodological conclusions beyond what the registry record establishes.

14. Why This Trial Matters Statistically

This trial illustrates why the endpoint definition matters. Time to rescue WLL is not simply a count of procedures or a measure of disease physiology. It captures when the first unscheduled procedure occurred after a scheduled baseline lavage, within a defined follow-up window. Interpretation depends on understanding that both groups received baseline WLL and that the primary contrast concerns strategies with or without the added inhaled treatment.

It also illustrates the difference between a descriptive summary and an inferential result. The medians and interquartile ranges describe the group distributions as posted; the Wilcoxon (Mann-Whitney) p-value expresses the result of the reported comparison. Neither component alone gives a complete account of the effect’s precision. No confidence interval is reported for the median difference, and the registry record does not provide details needed to reconstruct the time-to-event analysis.

The endpoint has a time-to-event structure, which raises questions about censoring and incomplete follow-up. Although such methods are important in general, the registry record does not say how censoring was handled or provide participant-level event histories. This limits independent assessment of the relationship between the reported medians and the statistical test. The recorded interquartile range of 30 to 30 months in the GM-CSF group is a descriptive summary, not a substitute for those details.

Finally, a randomized, open-label phase 2 study with 9 participants listed per arm provides a setting for a focused comparison, but its result should not be treated as a universal estimate. The record’s missing information about multiplicity, analysis populations, and the prespecified statistical plan also matters when interpreting a primary p-value alongside numerous listed outcomes and analyses. The statistical lesson is to keep the target outcome, design, test, precision, and limitations distinct.

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 result is based on 9 participants in each group, and the study is identified as phase 2. The registry describes the trial as randomized and open-label, with no masking. The record does not report the randomization method, allocation concealment, or whether outcome assessment was independently standardized. Because the endpoint involves the timing of an unscheduled procedure, the open-label design is relevant to how rescue decisions may have been made; the record does not provide enough detail to assess that process.

The primary result has no reported confidence interval for the median difference. The sponsor reports a Wilcoxon (Mann-Whitney) p-value, but the registry does not give the detailed statistical analysis plan, handling of censored observations, or participant-level event times. It also does not report the number or reasons for incomplete endpoint observations. These omissions limit assessment of precision and independent reconstruction of the analysis.

The control-arm description is incomplete in the registry record: it states that rescue WLL could occur for progression, but the available text does not provide the full progression definition. The treatment description specifies the GM-CSF schedule and includes a washout, but details of the washout are not reported. The registry does not report adherence to the assigned regimen or reasons for any treatment discontinuation.

The record lists many outcome measures and analyses, but does not report how multiplicity was handled, whether interim analyses occurred, or which analyses were prespecified. A sample-size justification is not reported in the registry record. The safety summary lists zero affected by serious adverse events in both groups, but does not provide detailed event ascertainment or a confidence interval for those counts.

These trial-specific limitations do not negate the posted result; they define the boundaries of what can be inferred from the available record. The result is a comparison within this study’s baseline-lavage framework and follow-up, rather than a complete account of treatment effects across outcomes or settings.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Inhaled GM-CSF Therapy of Autoimmune PAP — NCT00901511.
• Registry record last updated 2024-07-12; status last verified 2024-07.
• Results first posted 2023-06-26.
• 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

NCT00901511 is a completed, randomized, open-label, parallel-group phase 2 study of inhaled GM-CSF following baseline WLL in autoimmune pulmonary alveolar proteinosis. The registry lists enrollment as 18 and a primary endpoint of time from baseline WLL to first rescue WLL during 30 months.

The reported median times are 18 months for control and 30 months for GM-CSF, with interquartile ranges of 6 to 27 months and 30 to 30 months, respectively. The sponsor reports a Wilcoxon (Mann-Whitney) median difference of 12 and p = 0.008. No confidence interval for that difference is reported.

Both groups were scheduled to receive baseline bilateral WLL; inhaled GM-CSF was added in the intervention group. The registry reports 0 serious adverse events and 0 deaths in each group. Important details concerning censoring, missing data, multiplicity, analysis populations, and the statistical plan are not reported in the registry record.