GM-CSF for Reversal of immunopAralysis in pediatriC sEpsis-induced MODS Study: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for GM-CSF for Reversal of immunopAralysis in pediatriC sEpsis-induced MODS Study (NCT03769844), a non-randomized phase 4 study. The primary outcome is Number of Subjects With Restoration of the TNF-alpha Response.

2. Clinical Question

The study was intended to establish an immunologic response pattern across delivery routes and doses in children with sepsis-induced MODS and immunoparalysis. It also listed pharmacokinetic estimation and feasibility of screening, enrollment, drug delivery, and sample collection as aims. The posted primary outcome is a count of participants meeting the restoration criterion, rather than a comparison of average TNF-alpha values. The registry record does not report results for the primary question, pharmacokinetic parameters, or the feasibility aims.

3. Trial Design

The registry describes GRACE as a phase 4, open-label, multi-center interventional study with non-randomized allocation and a sequential model. The status is completed. The record lists enrollment as 75. The listed start date is 2018-12-05, and the listed primary completion date is 2023-12-05. These registry entries describe the study's recorded design and status; they do not supply endpoint results.

Children with sepsis-induced MODS underwent surveillance immune-function testing beginning on Day 2 of MODS. Those meeting the immunoparalysis criterion, a TNF-alpha response less than 200 pg/ml, were assigned to a treatment approach described by route and dose. At 125 mcg/m2/day, GM-CSF was administered IV or SQ for 7 consecutive days. The protocol description then specified sequential dose escalation: a 250 mcg/m2/day IV cohort if the IV 125 mcg/m2/day arm was not successful, and a 250 mcg/m2/day SQ cohort under the stated conditions for the SQ arm and IV escalation.

The design's cohort rule linked immunologic response to a treatment decision. Success required at least 8 out of 10 treated subjects in a cohort to reach the response threshold by the morning after the 3rd dose and retain that response on the morning after the 7th dose. This is a rule for classifying a cohort, not a reported result. Whether any cohort met it is not reported in the registry record.

Because treatment was sequential and non-randomized, route and dose were not assigned in a way that creates a randomized comparison. Changes across cohorts could reflect differences in cohort composition or study circumstances as well as treatment approach. The record does not report the enrollment and treatment flow by cohort, the reasons individual subjects entered particular cohorts, or a statistical plan for comparing them. Masking was none, so the study was open-label.

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: After enrollment, subjects undergo measurement of the TNF response, with ONLY those having immunoparalysis (TNF response \<200pg/ml) going on to get study drug. Many subjects who are consented for immune phenotyping do NOT have immunoparalysis, so the number of subjects assigned to an interventional cohort will be fewer than the number consented. Since an adequate immune response was achieved with the 125 mcg/m2/dose strategy, no subjects were enrolled in the 250 mcg/m2/dose arms.

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 Number of Subjects With Restoration of the TNF-alpha Response, measured over Subjects will be screened for immunoparalysis throughout their first three weeks of sepsis-induced MODS.

The primary outcome is the number of subjects with restoration of the TNF-alpha response. It concerns whole-blood ex vivo LPS-induced TNF-alpha production capacity among subjects with immunoparalysis in the setting of sepsis-induced MODS. The response criterion was a TNF-alpha response of at least 200 pg/ml, reached by the morning after the 3rd scheduled dose and persisting to the morning after the 7th scheduled dose.

This endpoint turns a repeated immune-function measurement into a participant-level response classification: a subject either meets the specified restoration criterion at both stated assessment times or does not. The cohort-level success rule then counts qualifying subjects and sets a threshold of at least 8 out of 10 treated subjects. The registry describes screening for immunoparalysis throughout the first three weeks of sepsis-induced MODS, while the treatment response criterion is anchored to scheduled doses.

Such an endpoint can make a biologic response rule operational for a dose-and-route study, especially when the study aim is to identify a treatment approach that reaches and sustains a prespecified marker threshold. It does not, by itself, describe a patient's overall clinical course or establish a comparison with an untreated group. The registry record does not report the observed number meeting the endpoint, TNF-alpha measurements, or estimates of uncertainty around the response count.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The registry presents a participant-count endpoint: the number of subjects who meet a defined restoration criterion. The underlying response is binary at the participant level, even though TNF-alpha production is measured on a continuous scale. The participant-level classification depends on reaching at least 200 pg/ml at the morning assessment after the 3rd scheduled dose and retaining that response at the morning assessment after the 7th scheduled dose. A count of qualifying participants can describe how often this criterion was met within a treatment cohort.

What does the cohort success rule mean?

The record defines success as at least 8 out of 10 treated subjects in a cohort meeting the sustained-response criterion. This is a prespecified threshold for classifying a cohort, not a reported count and not a p-value. It gives a concrete rule for deciding whether a dose-and-route approach meets the study's stated immunologic target. The registry record does not report the observed numerator for any cohort, so the rule cannot be applied to determine success from the posted results.

A threshold-based endpoint simplifies a biological measurement into a yes-or-no participant outcome. That can be useful when the study's aim is to identify an approach associated with a desired marker level. However, the binary classification does not show how far a participant's response was from the threshold, how measurements changed over time, or whether a response was sustained beyond the specified assessment. Those details are not reported in the registry record.

Why does the sequential design matter?

In a sequential design, later treatment cohorts may depend on earlier cohort results. Here, the registry describes movement to 250 mcg/m2/day if the relevant 125 mcg/m2/day cohort was not successful under the stated conditions. This makes the treatment sequence part of the study's decision structure. It also means that cohorts are not equivalent to concurrently randomized groups. General statistical principle: randomized allocation helps balance measured and unmeasured participant characteristics, on average, whereas sequential non-randomized allocation does not guarantee that balance. A difference between cohorts can therefore have more than one possible explanation.

For the same reason, ordinary comparisons of proportions across the IV and SQ groups would need careful interpretation. A statistical test or confidence interval could quantify uncertainty under stated assumptions, but it would not remove the design limitation or automatically make the groups exchangeable. The registry record reports no primary analysis estimates, confidence intervals, or comparative tests. The statistical plan for such comparisons is not reported in the registry record.

What do uncertainty measures add?

General explanation: a confidence interval describes a range of values compatible with observed data under a statistical model and its assumptions. It helps communicate how precisely an effect or response frequency has been estimated. A point estimate alone does not show that precision. No confidence interval for the primary endpoint is reported in the registry record, and there is no posted primary estimate to interpret.

With participant counts, the estimate of a response frequency is based on the number meeting the criterion relative to the number evaluated. The registry record does not report these cohort-specific outcome counts and denominators. No frequency, ratio, difference, or other comparative quantity can therefore be described from the primary outcome entries. The overall enrollment listing does not substitute for the number treated or assessed in each cohort.

How should planning illustrations be read?

The sample-size assumptions describe planning illustrations based on the observed effect, not the trial's own design. They use a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation; the effect to detect is defined as a stated fraction of the observed effect. The particular fraction is not reported in the registry record. These are planning assumptions, not observed results and not evidence of observed power. Sample-size illustrations can show how a hypothetical future study might be planned under selected assumptions, but they do not change the interpretation of this non-randomized, sequential study.

General explanation: power is a design-stage probability of detecting a specified effect under a statistical model and its assumptions. It is not a measure of whether a completed study's finding is true, and calculating power after observing results does not resolve uncertainty in those results. The posted assumptions are explicitly planning illustrations, so their sample-size values should be understood as prospective planning quantities rather than trial outcomes.

Why does over-dispersion matter?

General explanation: over-dispersion occurs when observed variability is greater than a model expects. It can make standard errors too small if the model assumes less variability than the data contain. This concern often arises with clustered or heterogeneous data, but the registry record does not report an over-dispersion model, a dispersion estimate, or an analysis requiring such a correction. It would be inappropriate to infer that over-dispersion was assessed or handled.

The registered outcome is a count of participants meeting a binary criterion, and the record does not provide statistical analysis details for that count. The distributional assumptions, method for estimating uncertainty, handling of repeated immune-function measurements, and any covariate adjustment are not reported in the registry record. These omissions prevent a more detailed reconstruction of the analysis.

10. Exact Estimates for Each Arm

No two-arm comparison is shown for this trial, but each arm's result can still be summarized with exact interval estimates computed from the posted counts. The Clopper-Pearson interval is exact and conservative; the Wilson interval is a common closer-to-nominal alternative; the Bayesian credible interval uses a uniform Beta(1, 1) prior. For small groups the three can differ noticeably.

R code and output

arm    <- c("IV GM-CSF 125 mcg/m2/Dose", "SQ GM-CSF 125 mcg/m2/Dose")
n      <- c(10, 9)
events <- c(10, 8)
for (i in seq_along(n)) {
  ci <- binom.test(events[i], n[i])$conf.int          # Clopper-Pearson exact interval
  cr <- qbeta(c(0.025, 0.975), 1 + events[i], 1 + n[i] - events[i])   # Bayesian, uniform Beta(1,1) prior
  wi <- (function(x, n) { p <- x / n; z <- qnorm(0.975); d <- 1 + z^2 / n
    c((p + z^2 / (2 * n) - z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d, (p + z^2 / (2 * n) + z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d) })(events[i], n[i])
  cat(sprintf("ARM|%s|%d|%d|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f\n", arm[i], n[i], events[i], 100 * events[i] / n[i],
              100 * ci[1], 100 * ci[2], 100 * wi[1], 100 * wi[2], 100 * cr[1], 100 * cr[2]))
}

ARM  IV GM-CSF 125 mcg/m2/Dose  10  10  100.0  69.2  100.0  72.2  100.0  71.5  99.8
ARM  SQ GM-CSF 125 mcg/m2/Dose  9  8  88.9  51.8  99.7  56.5  98.0  55.5  97.5

11. Multiplicity, Missing Data and Analysis Populations

The registry record lists one primary outcome measure: the number of subjects with restoration of the TNF-alpha response. It also describes additional aims concerning pharmacokinetic parameters and feasibility, but does not list corresponding results in the posted outcome information. The analysis plan, including whether and how those aims were formally tested, is not reported in the registry record.

Multiplicity arises when several outcomes, doses, routes, time points, or subgroup comparisons are examined. General statistical principle: testing many questions can increase the chance of obtaining an apparently notable result by chance unless the analysis plan accounts for the testing strategy. The record describes multiple routes and dose levels and two response assessment times, but it does not report a multiplicity adjustment, hierarchy of tests, or prespecified approach to handling multiple comparisons. It is not possible to determine from the registry record whether such procedures were planned or used.

Missing data are also important for this endpoint because the response definition requires assessment after both the 3rd and 7th scheduled doses. The registry does not report how many subjects had each assessment, whether any observations were missing, how missing measurements were handled, or whether subjects who stopped treatment remained in the primary analysis. The analysis population definition is not reported in the registry record.

The record does not describe an intention-to-treat, per-protocol, or other named analysis population. It also does not provide cohort-specific enrollment, treatment, and assessment flow. Without these details, the relationship between the enrollment listing and the primary endpoint population cannot be established. These are limitations of the information reported in the registry record; they do not establish that the study did or did not use particular procedures.

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: intra-abdominal abscess (11.1%), wound infection (10.0%), bacteremia (10.0%), bacterial infection (10.0%), wound dehiscence (10.0%).

For IV GM-CSF 125 mcg/m2/Dose, the registry lists 10 participants at risk for serious adverse events, with 1 affected, and 10 at risk for other adverse events, with 9 affected. For SQ GM-CSF 125 mcg/m2/Dose, it lists 9 participants at risk for serious adverse events, with 1 affected, and 9 at risk for other adverse events, with 8 affected. The registry lists 0 deaths in each group.

These are descriptive counts in the registry record. They do not indicate whether events were considered related to treatment, how long participants were followed for safety, how many events each affected participant experienced, or whether event severity differed beyond the serious-versus-other classification. Those details are not reported in the registry record. The serious-event terms listed include Intra-abdominal abscess (11.1%), Wound infection (10.0%), Bacteremia (10.0%), Bacterial infection (10.0%), and Wound dehiscence (10.0%). The record does not provide enough information to assign those term percentages to a particular route group.

Because the treatment allocation was non-randomized and sequential, these figures should not be read as a like-for-like safety comparison between routes. The group counts alone do not establish a difference in risk or a causal relationship between GM-CSF and an event. The registry record does not report an inferential safety analysis or uncertainty estimates for these counts.

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

The registry record establishes what the study set out to evaluate: restoration of a TNF-alpha response in children with sepsis-induced MODS and immunoparalysis after GM-CSF delivered by IV or SQ route at specified doses. It defines a response threshold, assessment timing, and a cohort success rule. These details explain the intended immunologic target, but the record does not report whether subjects reached it.

As a result, the primary clinical and statistical question remains unanswered by the posted results. There is no reported count of subjects with sustained restoration, no effect estimate, and no interval describing precision. The record's completed status and enrollment listing do not substitute for those outcome data. Nor does the lack of a reported primary result establish that the study found no response; the result is simply not reported in the registry record.

The study also cannot be interpreted as a randomized comparison of IV versus SQ delivery or of the dose levels. It used non-randomized sequential allocation and no masking. Its cohort success rule may guide the study's stated dose-and-route decision process, but it does not provide a causal comparison by itself. The registry record does not report a statistical method that would support a comparative claim.

The adverse-event counts offer limited descriptive information for the listed 125 mcg/m2/Dose groups. They do not establish whether events were caused by treatment or whether one route has a different safety profile. Safety interpretation would require context that is not reported in the registry record, including event timing, attribution, and a defined comparative analysis.

In general, an immune-function marker can be informative for the biologic question it measures while remaining distinct from broader clinical outcomes. Here, the registered endpoint focuses on ex vivo TNF-alpha production at specified times. The registry record does not report outcomes that would show how this marker response relates to other aspects of illness or recovery. No conclusion about overall clinical benefit or harm can be drawn from the endpoint definition alone.

14. Why This Trial Matters Statistically

This study is statistically notable for linking a biological response threshold to sequential dose-and-route decisions. The primary endpoint is not simply a single TNF-alpha measurement: it requires reaching a target after the 3rd scheduled dose and maintaining it after the 7th. The cohort rule then converts those participant-level responses into a decision about whether a treatment approach meets the study's stated criterion.

That structure makes the distinction between a protocol rule and an observed result especially important. A threshold such as at least 8 out of 10 defines what would count as cohort success; it does not show that success occurred. The registry record reports the rule but not the qualifying subject counts. Statistical interpretation therefore depends on separating the design's decision framework from evidence about the outcome.

The sequential non-randomized allocation also illustrates the challenges of dose-finding and route evaluation without randomized control groups. Moving to a later cohort based on the earlier cohort's status may be practical for the study's objectives, but cohort differences cannot automatically be attributed to dose or route. Analyses must respect the allocation process and its limits. The statistical plan and comparative estimates are not reported in the registry record.

Finally, the registry distinguishes an immunologic endpoint from the study's additional pharmacokinetic and feasibility aims. Those aims represent different kinds of evidence: marker response, drug disposition, and whether study procedures can be carried out. The record describes all three aims but does not report results for them. This makes the registry a useful account of the intended questions and design, while leaving the empirical findings unavailable for statistical interpretation.

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 principal limitation for interpretation is that the registry record does not report the primary endpoint result. It gives the response definition and cohort success rule but no cohort-specific counts of subjects who met the criterion. Consequently, the response frequency and its precision cannot be assessed.

The non-randomized sequential design limits direct comparisons across route and dose cohorts. The record does not describe a randomized control group, and its own design information cautions that the arms are not like-for-like controls. No adjusted comparison or statistical analysis plan is reported. Differences between cohorts therefore cannot be assigned to route or dose using the registry information alone.

The record lists enrollment as 75 but does not report the flow from screening through eligibility, treatment, and endpoint assessment. It also does not state how missing measurements were handled, define the analysis population, or explain whether subjects with incomplete dosing or assessments were included. These details are particularly relevant because the endpoint requires sustained response at two post-dose assessment times.

The study was open-label, with no masking. The record does not explain how this design feature was addressed in outcome assessment or analysis. It also does not report confidence intervals, p-values, covariate adjustment, or multiplicity procedures. The pharmacokinetic and feasibility aims are described, but their results are not reported in the registry record.

Safety information is limited to counts in the listed 125 mcg/m2/Dose groups and serious-event terms with percentages. The registry does not report event attribution, timing, person-time, or comparative uncertainty. These limits make it inappropriate to infer causal safety differences from the descriptive entries. Together, these gaps mean the record is informative about design and endpoint definitions but incomplete for evaluating efficacy, precision, and comparative safety.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: GM-CSF for Reversal of immunopAralysis in pediatriC sEpsis-induced MODS Study — NCT03769844.
• Registry record last updated 2026-05-05; status last verified 2026-04.
• Results first posted 2026-05-05.
• 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 4, open-label, multi-center, non-randomized sequential study of GM-CSF in children with sepsis-induced MODS and immunoparalysis. The listed enrollment is 75. Treatment approaches used IV or SQ delivery and 125 or 250 mcg/m2/day, with higher-dose cohorts described as sequential escalations under specified conditions.

The primary outcome was the number of subjects with sustained restoration of the TNF-alpha response to at least 200 pg/ml at the specified post-dose assessments. Cohort success was defined as at least 8 out of 10 treated subjects meeting that criterion. The primary result, cohort-specific response counts, and statistical analysis estimates are not reported in the registry record.

Descriptive adverse-event counts are listed for the IV and SQ 125 mcg/m2/Dose groups, with 0 deaths in each group. The non-randomized design does not support treating these groups as like-for-like controls. The record also describes pharmacokinetic and feasibility aims but does not report their results.