Natural Killer (NK) Cell Transplantation for AML: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Natural Killer (NK) Cell Transplantation for AML (NCT00187096), a non-randomized na study. The primary outcome is Number of Patients Experiencing Grade 3 or 4 Toxicities During Conditioning and up to 100 Days Post-transplant.

2. Clinical Question

3. Trial Design

This was a completed pilot study of haplo-identical natural killer cell transplantation for acute myeloid leukemia, registered as NCT00187096. The sponsor was St. Jude Children's Research Hospital. The registry lists a start of 2005-09, primary completion in 2012-02, enrollment of 49, and phase as NA. These record details describe the study but do not establish how many participants contributed to any particular outcome.

The study was non-randomized and had a single-group model, with no masking. The registry explicitly cautions that the arms are not a like-for-like control comparison. Participants were described in two clinical strata: AML in complete remission, and refractory or relapsed AML or AML with increasing minimal residual disease. The arm descriptions also identify Arm 2a, using cyclophosphamide and fludarabine, and Arm 2b, using clofarabine, etoposide, and cyclophosphamide. These descriptions indicate different treatment contexts, not random assignment to interchangeable alternatives.

For Stratum 1, the record describes cyclophosphamide on Day -7, fludarabine on Days -6 through -2, donor pheresis on Day -1, interleukin-2 beginning on Day -1 and then three times per week for two weeks, and NK cell purification and infusion on Day 0. For Stratum 2, it describes clofarabine, etoposide, and cyclophosphamide on the stated days, donor pheresis on Day -1, and interleukin-2 beginning on Day -1. The registry also lists natural killer cell infusion, the conditioning medicines, interleukin-2, and the CliniMACS System as interventions.

The registry lists ten outcome measures, including engraftment, chimerism, cell activity, relapse-free survival, and overall survival. It reports no analyses in the analysis listing. The record does not provide a randomized contrast or a reported primary-endpoint estimate, so the design supports descriptive interpretation rather than a causal comparison between regimens.

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: 49 participants were enrolled on the study. This report is based on results for 25 patients. 24 were donors and were excluded. All analyses are based on the intent-to-treat principle and therefore include all recipients with data available for the reported endpoint.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Grade 3 and 4 toxicities were assessed using Common Toxicity Criteria V.3.0 The assessment period was from the date on therapy through 100 days post-transplant.

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 Patients Experiencing Grade 3 or 4 Toxicities During Conditioning and up to 100 Days Post-transplant, measured over Beginning at on therapy through 100 days post-transplant.

The primary endpoint counts patients experiencing grade 3 or 4 toxicities during conditioning and through 100 days post-transplant. The registry specifies that toxicity grading used Common Toxicity Criteria V.3.0 and gives the assessment period as beginning on therapy through 100 days post-transplant. The endpoint is a count of participants with an event, rather than a time-to-event measure or a continuous measurement.

In general, a toxicity endpoint is used to describe the burden of serious treatment-related harm observed during a defined follow-up window. A count can communicate how many people met the stated toxicity criterion, but by itself it does not show the timing, duration, recurrence, or clinical course of each event. Nor does a count alone distinguish the contributions of conditioning, cell infusion, underlying disease, or other aspects of care.

The registry record gives the endpoint definition but does not report its observed count, the number assessed for this endpoint, or an uncertainty interval. Consequently, the primary toxicity result cannot be quantified from the registry record. The adverse-event listings are separately reported safety information and should not be substituted for the missing primary-endpoint result: they use arm-level risk sets and serious or other affected counts, without a registry-reported reconciliation to the primary endpoint definition.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

What kind of outcome is the primary endpoint?

The primary endpoint is a binary participant-level outcome: whether a patient experienced grade 3 or 4 toxicity during the specified assessment period. The record describes the result in terms of the number of patients with that event. In general, binary outcomes classify each participant according to whether a defined event occurred. A count is the numerator; interpreting it as a rate or proportion also requires a clearly defined denominator, such as the number of participants assessed for that endpoint.

The registry record does not report the primary endpoint count or its analysis denominator. The study-level enrollment listing is not a substitute for an outcome-specific denominator, because the record does not say that every enrolled participant was assessed for this endpoint. Without both the event count and a denominator, neither an observed proportion nor its precision can be described.

What would an uncertainty interval tell a reader?

In general, a confidence interval expresses the range of values compatible with an estimate under a specified statistical method and its assumptions. It helps distinguish a precise estimate from one that leaves substantial uncertainty. For a binary outcome, the interval depends on the observed count, the denominator, and the method used. The registry record does not report a primary endpoint estimate or confidence interval, and it does not describe a method for calculating one. No interval should be inferred from the safety counts, since those counts do not report the primary endpoint result.

Why does the design matter for interpretation?

The study is described as non-randomized, with a single-group model and no masking. Random allocation, when used, can help make treatment groups comparable on measured and unmeasured factors on average. That protection is absent from a non-randomized comparison. Here, the registry also describes clinically different strata and says the arms are not a like-for-like control comparison. Differences in their observed counts may reflect differences in participant context, treatment regimen, follow-up, or other factors; the record does not provide an analysis that separates those possibilities.

In general, descriptive comparisons can still help characterize what was observed in each group, but they do not by themselves identify a causal effect. The safety counts should therefore remain arm-specific descriptions. They do not establish that one conditioning regimen is safer or more hazardous than another.

What does the registry say about analysis and planning?

The registry lists no analyses. It does not report a statistical analysis plan, adjustment strategy, hypothesis test, confidence interval method, interim analysis, or sample-size justification. These items are not reported in the registry record. Their absence from the record does not establish whether a method was or was not used outside the registry; it means that the record cannot support claims about it.

The registry describes the sample-size illustrations as planning calculations based on an observed effect, using a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. It says the effect to detect is a stated fraction of the observed effect, and that non-time-to-event illustrations are expressed as participants per arm for a specified power. Time-to-event illustrations are described as total event counts using the Schoenfeld method. These are planning illustrations, not observed trial results or observed power. The record does not report a primary effect estimate or numerical illustration results in the registry record for this analysis, so no sample-size value can be interpreted.

How should multiple outcomes be considered?

The registry lists ten outcome measures, spanning toxicity, engraftment, chimerism, laboratory or cell activity measures, relapse-free survival, and overall survival. In general, evaluating many outcomes can create multiplicity: when many statistical tests are conducted, the chance of at least one apparently notable result can increase unless the analysis accounts for the testing strategy. A prespecified hierarchy or adjustment can clarify which results are confirmatory and how uncertainty should be interpreted. The registry record does not report a multiplicity strategy, and it lists no analyses. It is therefore not possible to determine whether multiplicity was addressed.

Outcome availability and population definitions matter as well. A result may be based on all enrolled participants, an outcome-specific subset, or another analysis population. The record does not report analysis populations or explain how missing outcome data were handled. In general, missingness can affect both the denominator and the interpretation of an estimate, particularly if participants with missing measurements differ from those with observed measurements. The posted record does not provide enough information to assess that issue for the primary endpoint.

What do the safety counts contribute?

The adverse-event listing gives arm-specific risk sets and numbers of participants affected for serious and other events. These counts support a basic descriptive reading, but the record does not provide a comparative statistical analysis, uncertainty intervals, or full event-level detail. The listed serious-event terms and percentages add context about the kinds of events recorded, but percentages do not by themselves establish severity distribution, causal attribution, or the number of unique people affected. General statistical interpretation should remain separate from claims about clinical causation, which the registry record does not establish.

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("Arm 1: Conditioning Regimen: Cyclophosphamide/Flud", "Arm 2a: Conditioning Regimen: Cyclophosphamide/Flu", "Arm 2b: Conditioning: Clofarabine/Etoposide/Cyclop")
n      <- c(10, 3, 12)
events <- c(2, 3, 11)
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  Arm 1: Conditioning Regimen: Cyclophosphamide/Flud  10  2  20.0  2.5  55.6  5.7  51.0  6.0  51.8
ARM  Arm 2a: Conditioning Regimen: Cyclophosphamide/Flu  3  3  100.0  29.2  100.0  43.9  100.0  39.8  99.4
ARM  Arm 2b: Conditioning: Clofarabine/Etoposide/Cyclop  12  11  91.7  61.5  99.8  64.6  98.5  64.0  98.1

11. Multiplicity, Missing Data and Analysis Populations

The registry lists ten outcome measures, including the primary toxicity endpoint and secondary measures of NK cell engraftment, chimerism, cell activity, relapse-free survival, and overall survival. It reports no analyses. The registry record does not describe a multiplicity plan, such as a prespecified testing hierarchy or adjustment for evaluating multiple outcomes. In general, testing many outcomes can increase the chance of a chance finding, so the analysis strategy is important when interpreting a set of results. Since no analysis results are listed, there is no basis to assess how multiplicity may have affected reported findings.

The record does not report a statistical analysis plan, analysis populations, or the handling of missing data. It does not state how many participants had primary endpoint data, whether all enrolled participants were evaluated, or how incomplete follow-up was treated. These details are not reported in the registry record. In general, missing outcome data can change the denominator and can bias an estimate if the availability of data is related to a participant’s outcome or characteristics. The registry information does not permit an assessment of that possibility.

The enrollment value of 49 should not be interpreted as the endpoint denominator. The adverse-event listing reports separate risk sets for each arm, but does not identify those risk sets as the primary endpoint analysis populations. The available record therefore supports neither a complete accounting of outcome missingness nor a reconstruction of the primary analysis population.

12. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 0% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: neutrophils/granulocytes (anc/agc) (10.0%), hypotension (8.3%), hepatobiliary/pancreas - other (8.3%), infection (documented clinically or microbiologically) with grade 3 or 4 neutrophils, blood (8.3%), infection (documented clinically or microbiologically) with grade 3 or 4 neutrophils, rectum (8.3%).

The registry lists serious and other adverse-event counts by arm. Arm 1 has a risk set of 10, with 1 participant affected for serious events and 1 for other events. Arm 2a has a risk set of 3, with 0 affected for serious events and 3 for other events. Arm 2b has a risk set of 12, with 2 affected for serious events and 11 for other events. These are arm-level descriptive figures, not results from a randomized comparison.

The listed serious-event terms are neutrophils/granulocytes (ANC/AGC) at 10.0%, hypotension at 8.3%, hepatobiliary/pancreas - other at 8.3%, infection with grade 3 or 4 neutrophils, blood at 8.3%, and infection with grade 3 or 4 neutrophils, rectum at 8.3%. The percentages are presented as registry-listed event-term figures. The record does not explain whether one participant could appear under multiple terms, or how these terms relate to the primary endpoint’s grade 3 or 4 toxicity count.

In general, counts from small groups are descriptive and can be unstable; a difference in counts does not by itself establish a difference in risk. The registry warns that the arms are not like-for-like controls. These adverse-event data should not be used to infer comparative safety or causal effects.

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

The record establishes that a pilot study was conducted to assess safety and efficacy of donor NK cell infusion for AML in remission or after relapse, and that the primary endpoint was the number of patients with grade 3 or 4 toxicities during conditioning and through 100 days post-transplant. It also records arm-specific adverse-event counts and several serious-event terms. Those details describe the study’s purpose, endpoint definition, and posted safety information.

The record does not provide the primary endpoint result. It does not report an estimate of the frequency of grade 3 or 4 toxicity under the endpoint definition, its denominator, or its uncertainty. The adverse-event counts cannot fill that gap because the registry does not report them as the primary endpoint analysis and does not explain their relationship to it. A study enrollment listing is not a substitute for the missing outcome-specific result.

The design also limits what can be concluded from differences across arms. Allocation was non-randomized, the model was single-group, and the registry says the arms are not a like-for-like control comparison. Thus the posted counts do not establish a causal difference between conditioning regimens. Nor do they establish efficacy: although the registry lists secondary outcomes such as relapse-free survival and overall survival, no estimates for those outcomes are reported in the record used.

In general, a pilot study can provide descriptive information and help identify questions for further evaluation, but its interpretation depends on the completeness of outcome reporting, the precision of estimates, and the design. The available record leaves those elements incomplete for the primary endpoint. It supports a careful account of what was registered and what safety counts were listed, but not a definitive estimate of toxicity or a comparative treatment conclusion.

14. Why This Trial Matters Statistically

This trial is statistically notable because it combines a clearly stated safety question with a design that does not support a randomized comparison. The primary endpoint focuses on grade 3 or 4 toxicity during conditioning and through 100 days post-transplant, a defined outcome and follow-up window. The record also describes clinically distinct strata and different conditioning regimens. That context matters: arm-level results should not be read as if participants had been randomly assigned to interchangeable options.

The registry’s record of 49 enrolled participants and its arm-specific adverse-event counts gives some scale to the study, while also underscoring why uncertainty and denominators matter. The primary endpoint count and its outcome-specific denominator are not reported, and no analysis results are listed. Without those elements, the principal statistical question cannot be answered from the registry record. The safety listings offer descriptive information, but they do not substitute for a complete endpoint analysis.

The study also illustrates a broader reporting principle. Naming outcomes such as engraftment, chimerism, relapse-free survival, and overall survival is not the same as reporting their estimates, follow-up, or uncertainty. A transparent account of analysis populations, missing data, statistical methods, and multiplicity would help readers understand how much weight to place on observed results. Those details are not reported in the registry record.

For clinical researchers, the key statistical lesson is to separate protocol intent from reported evidence. A stated aim to assess safety and efficacy does not itself demonstrate either; interpretation depends on the results and their precision. Here, the record supports description of the study design and posted safety counts, while leaving the primary endpoint result and comparative interpretation unresolved.

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 report the primary endpoint count, its denominator, or an uncertainty interval. It therefore cannot quantify the endpoint that was designated as primary. Although enrollment is listed as 49, the record does not state that this value is the analysis population for toxicity or any other outcome.

The study was non-randomized, single-group, and unmasked. Its arms are explicitly not described as like-for-like controls. The strata also represent different disease contexts, and the arm descriptions include different conditioning regimens. These features prevent an unadjusted arm comparison from being interpreted as a causal treatment effect. The registry does not report an adjustment method or other strategy to address differences between groups.

Safety reporting is limited to the arm-level risk sets and affected counts, along with selected serious-event terms and percentages. The record does not provide the event-level detail needed to determine whether participants contributed to multiple terms, the timing or recurrence of events, or how these listings map to the primary endpoint’s toxicity definition. Death counts are listed as not available. The relationship between the adverse-event summaries and the primary endpoint is not reported.

The registry lists secondary outcomes but does not report their estimates in the record used for this analysis. It lists no analyses and does not report analysis populations, missing-data methods, a statistical analysis plan, or a multiplicity strategy. The record also does not provide a primary sample-size justification. These gaps limit assessment of precision, completeness, and the role of chance. They do not establish that analyses or procedures were absent outside the registry; they mean those details are not reported in the registry record.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Natural Killer (NK) Cell Transplantation for AML — NCT00187096.
• Registry record last updated 2014-06-19; status last verified 2013-10.
• Results first posted 2012-11-09.
• 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

NCT00187096 is a completed, non-randomized, unmasked pilot study with a single-group model, sponsored by St. Jude Children's Research Hospital. The registry lists 49 enrolled participants, a start of 2005-09, primary completion in 2012-02, and phase as NA. The study concerned donor NK cell infusion for acute myeloid leukemia, with strata for AML in complete remission and refractory or relapsed disease or increasing minimal residual disease.

The primary endpoint was the number of patients experiencing grade 3 or 4 toxicity during conditioning and through 100 days post-transplant, assessed using Common Toxicity Criteria V.3.0. Its result is not reported in the registry record. Arm-specific adverse-event counts and serious-event terms are listed, but the registry cautions that the arms are not like-for-like controls. It lists ten outcome measures and no analyses; analysis populations, missing-data handling, and a multiplicity strategy are not reported.