Selumetinib in Treating Patients With Recurrent or Refractory Acute Myeloid Leukemia: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Selumetinib in Treating Patients With Recurrent or Refractory Acute Myeloid Leukemia (NCT00588809), a na phase 2 study. The primary outcome is Response Rate for Subjects Without FLT3 ITD Mutation.

2. Clinical Question

3. Trial Design

This was a Phase 2 study of selumetinib in patients with recurrent or refractory acute myeloid leukemia. The registry lists enrollment of 47 participants and identifies the National Cancer Institute as the sponsor. The study started in 2007-12, had primary completion in 2012-04, and is marked completed in the registry record.

The design is recorded as single-group, with allocation listed as not applicable and masking listed as none. Participants received selumetinib orally twice daily on days 1–28, with treatment repeating every 4 weeks if disease progression or unacceptable toxicity did not occur. A control arm, random assignment, and blinding are not described. This means the study can describe observations among treated participants, but it cannot isolate a treatment effect by comparing outcomes against a concurrent control group.

The stated enrollment is the number recorded for the study as a whole. It is not necessarily the denominator for the primary response outcome: that outcome includes only participants without FLT3 ITD mutation, and the size of that subgroup is not reported in the registry record. The record also does not report the number included in the endpoint analysis or explain how missing endpoint information was handled.

The primary outcome is a response proportion, measured over up to 52 weeks. Five secondary outcomes concern the proportions of subjects with baseline p-ERK activation, NRAS mutation, KRAS mutation, FLT3 ITD mutation, and KIT mutation. The registry record does not provide a statistical analysis for these outcomes. The absence of a concurrent comparator and the lack of posted endpoint estimates limit what can be concluded from the available record.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Before assignment: A Simon, optimal two-stage design was employed for subjects without FLT3 ITD mutations. The study also included a second cohort of subjects with FLT3 ITD mutations.

Overall Study

Analysis population for the primary outcome, as defined in the registry: Analysis only includes subjects without FLT3 ITD mutation.

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 Response Rate for Subjects Without FLT3 ITD Mutation, measured over Up to 52 weeks.

The primary endpoint is the proportion of subjects without FLT3 ITD mutation who experienced a complete response, partial response, minor response, or unconfirmed minor response. The registry states that these categories use standard criteria developed by an International Working Group. The timeframe is up to 52 weeks. The number of participants without FLT3 ITD mutation and the observed response proportion are not reported in the registry record.

A response proportion is a binary-outcome summary: each participant is classified according to whether they met the endpoint definition, and the proportion describes the share who did. As a general statistical concept, this kind of endpoint can summarize whether a specified response category was recorded in a study population. Combining several response categories into one endpoint can give a broader measure of response than a single category alone; the registry's definition specifically combines complete, partial, minor, and unconfirmed minor response.

The endpoint is relevant to the study question because the trial evaluates response during treatment in a population with recurrent or refractory disease. However, an endpoint definition alone is not an outcome result. Without a posted count, proportion, or uncertainty estimate, the extent and precision of the observed response cannot be assessed. The registry record also does not report details of endpoint assessment or how participants with unavailable response information were treated in the analysis.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The statistical information in the registry record is limited. It describes the study as single-group and defines a binary response endpoint, but does not report an analysis plan, an analysis result, or an estimate of uncertainty. The following explains how such quantities are generally interpreted and separates those concepts from what this record actually reports.

What does a response proportion describe?

As a general statistical concept, a response proportion is the number of participants who meet a defined outcome divided by the relevant analysis population. For this trial, the qualifying outcome combines complete response, partial response, minor response, and unconfirmed minor response. The relevant population is subjects without FLT3 ITD mutation. The registry does not state how many subjects were in that population or how many met the response definition, so no response proportion can be interpreted from the posted results.

The endpoint is binary at the participant level: a subject either meets the combined response definition or does not. This does not mean that all participants had identical disease histories, follow-up, or opportunity for response. Those details can matter when interpreting a proportion, especially when outcome observation extends over a specified period. The registry lists the timeframe as up to 52 weeks but does not describe the endpoint assessment schedule or the handling of participants whose outcome status was unavailable.

What would an uncertainty interval add?

In general, an estimate based on a sample has uncertainty because another sample from a similar population could yield a different estimate. A confidence interval gives a range of values compatible with the data and the method's assumptions. It is commonly used to communicate precision alongside a point estimate; it does not remove limitations in study design or guarantee that the true value lies in the interval. No confidence interval or point estimate for the primary endpoint is reported in the registry record. Consequently, neither the size nor precision of the response proportion can be assessed.

A statistical test and a confidence interval answer related but different questions. A test evaluates data against a specified null hypothesis under stated assumptions, while an interval displays a range of estimated values and their uncertainty. The registry record does not report a null hypothesis, test, p-value, or inferential analysis for the primary outcome. No statistical conclusion should be attributed to the trial record on that basis.

What does the single-group design allow?

In general, a single-group study can document outcomes in participants who receive an intervention. Without a concurrent control group, it cannot directly compare those outcomes with a group assigned to a different intervention or control condition. Random assignment, when used in a comparative trial, helps balance measured and unmeasured prognostic factors between groups on average. This study has no randomized comparison, so that design-based protection is not part of the record. The registry's allocation entry is not applicable, consistent with the single-group description.

The study is also listed as unmasked. In general, lack of masking means treatment knowledge is not concealed in the way it can be in a masked study. Depending on the outcome and its assessment, knowledge of treatment can affect reporting or assessment. The registry does not give further details about response assessment procedures, so the implications for this endpoint cannot be specified beyond noting the listed design.

How should planning illustrations be read?

Any sample-size figures associated with planning illustrations are not observed trial outcomes and are not evidence of the study's actual power. The stated illustration assumptions use a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation, with the effect to detect expressed as a fraction of the observed effect. Such assumptions describe a hypothetical planning exercise rather than the design documented for this single-group trial. The registry does not report the study's own sample-size justification, so it is not possible to determine whether those assumptions were used to plan enrollment.

In general, a sample-size calculation depends on a specified effect, variability or event frequency, significance level, power target, allocation, and analysis method. Changing those assumptions changes the calculated planning requirement. A calculation based on an observed effect is especially distinct from a prespecified design calculation: it describes a hypothetical future planning scenario using an effect estimate, rather than demonstrating that the completed study had a particular probability of detecting an effect. The registry record does not report observed power, and a planning illustration should not be interpreted as observed power.

For a non-time-to-event outcome, a planning display expressed as participants per arm refers to a hypothetical two-arm design. That format does not turn the actual study into a two-arm comparison. The trial record remains single-group, and no control arm is described. Since no primary response estimate is reported in the registry record, the basis for any observed-effect planning illustration is not available in the registry description itself.

Why can over-dispersion matter?

In general, over-dispersion occurs when the variability in observed data is greater than a statistical model assumes. If a model ignores that extra variability, it can understate uncertainty and produce confidence intervals or tests that are too optimistic. This concept is relevant when data are modeled with distributions whose variance is tied to their mean, or when outcomes are clustered or heterogeneous. The registry record does not report an over-dispersion assessment or a model for the primary response endpoint. Because the endpoint is defined as a participant-level proportion and no analysis is posted, there is no basis to say whether over-dispersion was assessed or affected the reported results.

Overall, the registry provides endpoint definitions and broad design features, but not the statistical procedures needed to reproduce or evaluate an endpoint analysis. The analysis population size, missing-data approach, uncertainty estimates, and inferential results are not reported in the registry record.

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("Selumetinib")
n      <- c(36)
events <- c(6)
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  Selumetinib  36  6  16.7  6.4  32.8  7.9  31.9  8.0  32.0

11. Multiplicity, Missing Data and Analysis Populations

The registry lists one primary outcome and five secondary outcomes: proportions with baseline p-ERK activation and with NRAS, KRAS, FLT3 ITD, and KIT mutations. It does not report a multiplicity strategy, such as a prespecified adjustment across multiple tests, a testing hierarchy, or an explanation of whether the secondary outcomes were descriptive. The statistical analysis plan and inferential results are not reported in the registry record. Without that information, the extent to which multiple outcomes were considered in statistical interpretation cannot be determined.

Missing-data handling is also not reported in the registry record. The record does not state how participants without an evaluable response, incomplete follow-up, or missing mutation status were handled. It does not give the number of participants in the primary population without FLT3 ITD mutation, the number with available outcome assessments, or any method for imputing or otherwise addressing missing observations. These omissions matter because the endpoint is restricted to a biomarker-defined population and measured over up to 52 weeks.

The primary analysis population is defined as subjects without FLT3 ITD mutation. The overall enrollment is 47, but the size of this restricted population is not reported. The record does not explain whether all enrolled participants were screened for this mutation or how unknown mutation status was treated. No analysis is listed in the registry record. Thus, neither the population included in an analysis nor the handling of missingness can be reconstructed from the posted information.

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 5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: infections and infestations - other (23.4%), death nos (21.3%), febrile neutropenia (17.0%), dehydration (10.6%), infection with grade 3 or 4 neutrophils (anc <1.0 x 10e9/l): blood (8.5%).

The registry lists 47 participants at risk for serious adverse events, with serious adverse events affecting 38 participants. It also lists 47 participants at risk for other adverse events, with other adverse events affecting 47 participants. These are safety summaries for the selumetinib group; the registry does not include a comparator group for interpreting differences in risk.

The listed serious adverse event terms include Infections and infestations - Other (23.4%), Death NOS (21.3%), febrile neutropenia (17.0%), dehydration (10.6%), infection with Grade 3 or 4 neutrophils (ANC <1.0 x 10e9/L): Blood (8.5%), and intracranial hemorrhage (8.5%). These are event-term percentages as recorded, not estimates of comparative risk. The registry does not report a count of deaths in the deaths field; the term Death NOS in the serious-event listing should not be treated as a separate total-deaths count.

In general, serious adverse event summaries and summaries of other adverse events refer to different reporting categories. The listed terms do not establish that selumetinib caused an event. The registry record does not provide enough detail to assess timing, duration, recurrence, or individual attribution.

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

The record establishes that a Phase 2, single-group study of selumetinib was conducted in recurrent or refractory acute myeloid leukemia, and it specifies a primary response endpoint for subjects without FLT3 ITD mutation. It also records enrollment of 47 and lists safety summaries for the selumetinib group. These registry entries describe the trial's design and recorded outcomes, but the primary response estimate itself is not reported in the registry record.

The endpoint definition tells readers which response categories were to be counted and the timeframe was up to 52 weeks. It does not tell readers how many participants met those criteria. Because the subgroup size, response count, response proportion, and confidence interval are absent, the frequency and precision of the primary outcome cannot be determined from the registry record.

The design does not include a randomized comparator. In general, a single-group observation cannot distinguish an intervention-associated outcome from changes that might have occurred for other reasons, nor can it estimate a comparative difference against another treatment. The posted record therefore does not support a comparative claim. The primary outcome results are absent as well, so no conclusion about the observed response rate can be drawn from these data.

Safety figures describe counts of participants affected and percentages attached to listed serious adverse event terms. They do not, by themselves, establish causation or allow a comparison with another group. The serious-event listing also should not be confused with a complete account of all clinical outcomes or with a total-deaths figure.

In general, interpreting a clinical trial requires considering the outcome definition, population, precision, missing information, and design together. Here, several of those elements are not reported in the registry record, including the analysis population size, missing-data approach, and statistical analysis. The defensible reading is therefore limited: the registry describes a study and its endpoint and safety summaries, but does not provide the primary result or the information needed for a comparative estimate.

14. Why This Trial Matters Statistically

This study illustrates why a clearly defined endpoint and a reported endpoint result are both necessary for statistical interpretation. The registry specifies a combined response outcome and identifies the population to which it applies: subjects without FLT3 ITD mutation. Those details make the intended question understandable. However, without a response count, proportion, denominator, and uncertainty estimate, the endpoint cannot be evaluated quantitatively from the registry record.

The trial also highlights the distinction between describing outcomes and estimating a treatment contrast. In a single-group study, outcomes can be recorded among participants receiving the intervention, but there is no concurrent control group against which to estimate a difference. This distinction is important when readers encounter treatment-focused language: a study's purpose does not replace the comparative evidence that a control group would provide.

Another statistical issue is the relationship between the overall enrolled population and a restricted endpoint population. Enrollment is 47, but the primary endpoint excludes subjects with FLT3 ITD mutation. Without the size of the eligible subgroup and a description of missing mutation or response data, readers cannot determine the endpoint denominator or assess how representative the analyzed population might be.

The record's safety summaries offer another lesson in careful interpretation. Counts and listed percentages can describe events recorded in a treated group, but without a comparison group they do not quantify relative risk. Nor do event listings alone establish causation. Statistical reporting needs to distinguish descriptive frequency from comparative inference, and the available registry information leaves that distinction especially important.

Finally, planning calculations are not substitutes for trial results or a documented design rationale. A hypothetical sample-size illustration based on an observed effect is not evidence of observed power and is not a record of the study's original planning. This trial therefore matters statistically as an example of how design details, outcome definitions, missing analysis information, and clear reporting jointly determine what readers can conclude.

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 of the registry record for statistical interpretation is that it does not report the primary response result. The response count, proportion, confidence interval, and test result are absent. The number of subjects without FLT3 ITD mutation, the primary endpoint population, is also not reported. Although enrollment is listed as 47, that total cannot be assumed to be the denominator for the restricted primary outcome.

The study is single-group, with no randomized allocation or control arm. This limits interpretation to description of outcomes among participants in the study and prevents a direct comparison with another treatment or control condition. The record lists no masking, and it does not describe further response-assessment procedures that might help readers assess the role of outcome ascertainment.

The registry record does not report an analysis plan, missing-data methods, endpoint-analysis population, multiplicity approach, or statistical tests. It also does not report results for the five secondary outcomes. These gaps prevent assessment of how the planned outcomes were analyzed, whether all relevant participants were included, or how uncertainty and multiple outcomes were handled.

The safety summaries provide participant counts and listed serious adverse event percentages, but do not supply a comparator or detailed information about event timing, duration, or attribution. The Death NOS event term is listed, while the deaths field contains no count. The record therefore does not support interpreting that term as a complete mortality summary.

The registry also does not report the study's own sample-size justification. Any planning illustration using an observed effect must be understood as hypothetical rather than as evidence of the study's original design assumptions or observed power. Together, these limitations mean that the available record describes the study structure and selected safety information, but does not permit a full statistical evaluation of the primary endpoint.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Selumetinib in Treating Patients With Recurrent or Refractory Acute Myeloid Leukemia — NCT00588809.
• Registry record last updated 2015-08-05; status last verified 2013-02.
• Results first posted 2015-08-05.
• 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

NCT00588809 is a completed Phase 2, single-group study sponsored by the National Cancer Institute. It enrolled 47 participants and evaluated selumetinib in recurrent or refractory acute myeloid leukemia. Treatment was administered orally twice daily on days 1–28 and scheduled to repeat every 4 weeks in the absence of progression or unacceptable toxicity.

The primary outcome was the proportion of subjects without FLT3 ITD mutation who experienced complete, partial, minor, or unconfirmed minor response, with a timeframe of up to 52 weeks. The registry record does not report the response estimate, its denominator, or an analysis. It also lists five secondary outcomes concerning baseline p-ERK activation and NRAS, KRAS, FLT3 ITD, and KIT mutations, without reporting their results.

Safety summaries list 38 of 47 participants affected by serious adverse events and 47 of 47 affected by other adverse events. The study has no concurrent comparator, and the record does not report missing-data handling, multiplicity methods, or the study's sample-size justification.