Monoclonal Antibody 3F8 and Sargramostim in Treating Patients With Neuroblastoma: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Monoclonal Antibody 3F8 and Sargramostim in Treating Patients With Neuroblastoma (NCT00072358), a non-randomized phase 2 study. The primary outcome is Efficacy at Completion of Treatment.

2. Clinical Question

The central clinical question is therefore the study’s intended efficacy question, not a question that can be answered from its posted primary-outcome data. The record identifies an intervention and a target population, but does not provide an observed response estimate, a comparison estimate, or an uncertainty interval for the primary endpoint. The secondary aims include comparing granulocyte activation after short-term versus prolonged daily exposure to sargramostim after 4 courses and simplifying treatment with a consequent reduction in cost. The registry does not report results for those aims either.

3. Trial Design

This is a Phase II, non-randomized, unmasked, parallel study sponsored by Memorial Sloan Kettering Cancer Center. The registry lists enrollment as 291 and the study status as completed. The recorded start is 2003-07, and primary completion is dated 2021-01-15. These registry entries describe the study record; they do not establish when every participant received treatment or completed follow-up.

The intervention is anti-GD2 murine IgG3 monoclonal antibody 3F8 in combination with GM-CSF, identified in the record as sargramostim. The arm descriptions concern patients with refractory bone marrow disease and patients with no evidence of disease. Both descriptions state that the study assesses minimal residual disease response and aims to help establish the optimal way to use GM-CSF. Because allocation is non-randomized, the groups should not be interpreted as randomized treatment and control arms. Their disease-status descriptions also indicate that they may represent clinically different groups; the record does not specify an adjustment strategy that would make them exchangeable.

Unmasked means that the registry does not record masking. It does not, by itself, establish how outcome assessments were conducted. The record describes a parallel model but does not report details of treatment assignment, treatment duration, adherence, follow-up procedures, or the statistical analysis plan. Those details are not reported in the registry record.

The listed enrollment is not a result for the primary endpoint and should not be read as the number contributing to every outcome. The primary outcome population is explicitly recorded as data not collected. The record also lists 4 outcome measures and 0 analyses. The implications are that the registry supports a description of the design and aims, but not a quantitative assessment of primary efficacy.

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

Analysis population for the primary outcome, as defined in the registry: Data were not collected

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 Efficacy at Completion of Treatment, measured over 3 years.

The primary outcome is titled “Efficacy at Completion of Treatment,” and its recorded timeframe is 3 years. The study description gives context for that aim: the trial was intended to assess response of minimal residual disease in high-risk neuroblastoma. The record does not specify the primary outcome’s measurement unit, parameter, dispersion measure, or a detailed definition of what counted as a response. Those details are not reported in the registry record.

In general, a prespecified primary endpoint gives a trial a principal outcome against which its central question can be assessed. A clear endpoint definition is important because the result depends on what is measured, when it is measured, and how participants are classified. For a response outcome, the operational criteria and assessment procedures determine which observations count as responses. Without those definitions and the recorded endpoint data, readers cannot verify how the intended outcome was measured or interpret an estimated response rate.

The registry lists the primary outcome population as “Data were not collected.” It therefore supplies no result, effect estimate, or precision measure for this endpoint. The stated timeframe does not remedy that absence: it identifies the recorded assessment horizon but does not establish that outcome data were available through that horizon. The endpoint’s clinical purpose can be described from the trial aims, but its observed statistical result cannot be evaluated from the registry record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

General statistical context: A trial’s design determines what kinds of conclusions its data can support. Random allocation is one way to make treatment groups comparable on average; a non-randomized study does not have that protection. In a non-randomized design, group differences may reflect the reasons people entered each group as well as treatment or other factors. This matters particularly when groups are described by different disease status, as in this record.

What does non-randomized allocation imply?

The record identifies allocation as non-randomized and the model as parallel. Parallel describes groups observed in the same study framework; it does not mean that participants were randomly assigned. Since the arm descriptions distinguish refractory bone marrow disease from no evidence of disease, a direct group contrast would need careful clinical and statistical interpretation. The registry record does not report an adjustment method, a propensity approach, or another method for addressing confounding. Such methods should not be assumed.

General statistical context: Confounding occurs when a factor related to both group membership and outcome can distort an association. Adjustment can sometimes address measured confounders, but cannot guarantee removal of unmeasured confounding. Randomization, when used appropriately, helps balance measured and unmeasured factors in expectation. None of these general principles supplies a missing analysis for this study.

What would an endpoint estimate communicate?

For an outcome such as response, an estimate would describe the observed result under a defined measurement rule and in a defined analysis population. A confidence interval would express the statistical uncertainty around that estimate under the model used. It would not capture every source of uncertainty, such as bias from non-random allocation, incomplete outcome assessment, or an unclear endpoint definition. The registry record has no primary endpoint result or confidence interval, so neither the observed response nor its statistical precision can be evaluated.

The record lists a primary outcome timeframe of 3 years, but does not provide an endpoint measurement definition or the data. In general, a timeframe is useful for understanding when outcomes are intended to be assessed. It does not tell readers how missing observations were handled, whether follow-up was complete, or whether all participants contributed outcome information. Those details are not reported in the registry record.

What do the analysis listings tell readers?

The registry record lists 4 outcome measures and 0 analyses. This indicates that analyses are not reported in the registry record. It does not establish whether analyses were conducted elsewhere or whether the posted outcome information is exhaustive. No test statistics, adjusted estimates, subgroup analyses, or primary-outcome calculations are available in this record. There are also no recalculated results to compare with a sponsor analysis.

General statistical context: Statistical significance and precision answer different questions. A p-value, when reported, relates the observed data to a specified null hypothesis and model. A confidence interval gives a range of values compatible with the data and assumptions at a stated confidence level. Neither can be reconstructed responsibly without suitable outcome data and an analysis specification. A non-significant result, if one existed, would not automatically demonstrate equivalence or absence of an effect; a statistically significant result would not by itself establish clinical importance. No such result is reported for the primary outcome.

How should sample-size planning be read?

General sample-size calculations are planning tools. They describe the amount of information that might be needed under stated assumptions, rather than the probability that a completed study has demonstrated an effect. The planning illustration associated with this record is based on an observed effect, assumes a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. It describes effects to detect as stated fractions of the observed effect. For time-to-event outcomes, such illustrations give the total number of events; other illustrations give participants per arm.

These assumptions are not reported as the trial’s own design or statistical plan. The illustration is explicitly not the trial’s design calculation. It should not be described as observed power, and it does not replace missing primary-outcome results. In general, planning estimates depend on the chosen effect, variability or event assumptions, significance level, power target, allocation, and analysis method. If assumptions differ from the conditions observed in a study, the planning illustration may not describe the information actually available.

General statistical context: Over-dispersion occurs when outcome variability is greater than expected under a chosen model. If it is ignored, standard errors may be too small and confidence intervals too narrow. The registry does not report a dispersion measure for the primary outcome, and its primary outcome data are recorded as not collected. Therefore, no assessment of over-dispersion or model fit can be made for that endpoint.

Overall, the statistical record is defined more by what it does not report than by an inferential result. The non-randomized design, the lack of primary outcome data, and the absence of analyses prevent readers from assessing an efficacy effect or its precision. General statistical methods explain what would be needed, but cannot fill those evidentiary gaps.

10. Multiplicity, Missing Data and Analysis Populations

The registry lists 4 outcome measures, including a primary outcome titled “Efficacy at Completion of Treatment” and secondary aims concerning granulocyte activation and treatment simplification with consequent cost reduction. The record lists 0 analyses. It does not report the full outcome definitions, analysis hierarchy, statistical plan, or which comparisons were intended for formal inference. Those details are not reported in the registry record.

General statistical context: Multiplicity arises when researchers assess multiple outcomes, time points, groups, or comparisons. The more opportunities there are to declare a result, the greater the chance of at least one apparently positive finding under a null model if no adjustment or prespecified hierarchy is used. Common approaches include identifying a primary outcome in advance, controlling an error rate, or treating secondary results as exploratory. The registry information does not establish which approach, if any, was used. No multiplicity adjustment should be inferred.

Missing data are a separate issue from multiplicity. The primary outcome population is recorded as “Data were not collected.” This is stronger than merely lacking a posted estimate: the record states that primary outcome data were not collected. It does not explain why, whether data existed outside the registry, or how missing observations were handled for other outcomes. Missing-data assumptions and methods are not reported in the registry record.

The record also does not define analysis populations, such as whether participants were analyzed according to treatment received, intended treatment, or outcome availability. Enrollment of 291 should not be taken to mean that the same participants contributed to each analysis. No outcome-specific denominators or analysis-population rules are reported for the primary endpoint. Consequently, readers cannot assess how exclusions, incomplete follow-up, or deviations from planned treatment might affect an efficacy estimate.

11. 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: fever (41.6%), infection w.out neutropenia (19.6%), rigors, chills (6.2%), vomiting (5.5%), hypertension (4.1%).

The registry reports a safety group titled “Participants With Refractory Bone Marrow Disease and High Risk of Recurrence.” It lists 291 at risk for serious adverse events and 153 affected. For other adverse events, it lists 291 at risk and 287 affected. The record also reports 135 deaths. These counts should be read as descriptive registry entries, not as estimates of treatment-attributable risk.

The listed serious adverse-event terms include Fever (41.6%), Infection w.out neutropenia (19.6%), Rigors, chills (6.2%), Vomiting (5.5%), Hypertension (4.1%), and Allergic Reaction/Hyper (3.8%). These percentages are reported as event-term values. The registry record does not give additional information about event timing, duration, recurrence, severity grading, or whether a participant could contribute to more than one term. They should not be added together or treated as mutually exclusive categories.

General statistical context: An “at risk” count identifies the group represented in a safety summary; an affected count describes the registry’s reported count for the event category. Neither count alone establishes causation. Interpreting safety comparatively requires a suitable comparator and clear information about exposure and event ascertainment. This study is non-randomized, and the record does not provide a like-for-like control comparison or explain the relationship of the reported deaths to treatment. No causal attribution should be made from these figures alone.

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

The record establishes that a Phase II study of anti-GD2 3F8 with sargramostim was conducted for high-risk neuroblastoma, with aims involving minimal residual disease response and the use of GM-CSF. It also records non-randomized allocation, an unmasked design, a parallel model, and completed status. These registry entries describe the study structure and intent.

The record does not establish an observed primary efficacy result. Although the primary outcome is titled “Efficacy at Completion of Treatment” and has a recorded timeframe of 3 years, the primary outcome population is marked “Data were not collected.” There is no response estimate or uncertainty interval to interpret. The study’s rationale and purpose cannot be treated as findings, and completed status does not imply that the registry contains evaluable efficacy results.

The record also does not support a causal comparison between the arm descriptions. The study is non-randomized, and the groups are characterized by different disease status. A difference between such groups, if reported, could reflect factors associated with group membership rather than an intervention effect. The record contains no efficacy analysis to assess in any case.

Descriptive safety information is available, including counts, deaths, and selected serious adverse-event terms with percentages. Those figures can indicate what the registry reports, but they do not establish that the intervention caused an event or that the event frequency differs from an alternative treatment. The information needed for those conclusions, including a comparable control group and detailed exposure and event context, is not reported in the registry record.

In general, a trial result is interpreted by combining the endpoint definition, estimate, precision, design, and analysis assumptions. Here, the intended primary question is identifiable, but the primary result and its analysis are absent. The appropriate conclusion is limited: the registry documents the study and selected safety summaries, but does not permit a quantitative conclusion about primary efficacy or a causal treatment effect.

13. Why This Trial Matters Statistically

This record illustrates why a trial’s question, design, and results must be kept distinct. The study title and aims identify a clinically focused question about 3F8 with sargramostim and minimal residual disease response in high-risk neuroblastoma. The design details indicate that allocation was non-randomized and that the groups were described by disease status. Those details constrain how any comparison could be interpreted.

The primary outcome is listed with a 3 years timeframe, but the record states that primary outcome data were not collected. This makes the registry a useful example of the difference between a prespecified or intended endpoint and an analyzable result. Naming an endpoint does not provide an estimate; completing a study does not, by itself, provide a reportable efficacy analysis. Without endpoint observations, readers cannot evaluate response, precision, or consistency with the stated aim.

The safety summary also demonstrates the distinction between descriptive reporting and causal inference. Counts of affected participants, deaths, and adverse-event terms convey what the registry records. They do not automatically identify an intervention effect, particularly in a non-randomized study without a like-for-like control comparison. Statistical interpretation depends not only on the size of a count, but also on the population, ascertainment, exposure, comparison, and analysis behind it.

Finally, the record lists 4 outcome measures and 0 analyses. That gap makes transparency about endpoint definitions, missing data, analysis populations, and multiplicity especially important. General statistical methods can clarify what information would be needed, but cannot recover unreported results. The statistical importance of this study record is therefore as much about evidentiary limits as about the questions the trial set out to address.

14. 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 limitation for interpretation is that the registry records the primary outcome population as “Data were not collected.” No primary efficacy estimate, confidence interval, or analysis is available. The timeframe of 3 years does not establish that outcome measurements were obtained or that follow-up was complete. The endpoint’s unit, parameter, dispersion, and detailed response definition are also absent.

The non-randomized allocation limits causal interpretation. The arm descriptions distinguish patients with refractory bone marrow disease from those with no evidence of disease, so the groups are not presented as like-for-like control groups. The registry does not report a statistical method for addressing differences between groups. No treatment-effect comparison should be inferred from the group descriptions.

The registry lists 4 outcome measures and 0 analyses. It does not report a statistical analysis plan, outcome hierarchy beyond the primary designation, multiplicity approach, missing-data method, or analysis-population definitions. It also does not give details of treatment exposure, adherence, outcome assessment, or the timing and context of the reported safety events and deaths. These details are not reported in the registry record.

Safety counts and percentages are descriptive and do not establish causation. The record does not provide a randomized comparator or other basis for a like-for-like safety contrast. Its enrollment figure should not be assumed to be the denominator for every outcome, and the posted primary outcome explicitly lacks collected data. These limitations prevent a quantitative appraisal of efficacy and restrict the interpretation of safety information to what the registry reports.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: Monoclonal Antibody 3F8 and Sargramostim in Treating Patients With Neuroblastoma — NCT00072358.
• Registry record last updated 2022-05-16; status last verified 2022-04.
• Results first posted 2022-05-16.
• 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.

18. Record Summary

This completed Phase II study evaluated anti-GD2 murine IgG3 monoclonal antibody 3F8 with GM-CSF, also called sargramostim, for high-risk neuroblastoma. It used non-randomized allocation, had no masking, and followed a parallel model. The arm descriptions distinguish patients with refractory bone marrow disease and patients with no evidence of disease. The study aimed to assess minimal residual disease response and inform use of GM-CSF.

The primary outcome is titled “Efficacy at Completion of Treatment,” with a recorded timeframe of 3 years. Its population is listed as “Data were not collected,” and no primary efficacy result is reported. The registry lists 4 outcome measures and 0 analyses. Descriptive safety information includes serious and other adverse-event counts, selected serious adverse-event terms with percentages, and deaths. These data do not establish causation or a comparative treatment effect.