Pevonedistat Plus Azacitidine Versus Single-Agent Azacitidine as First-Line Treatment for Participants With Higher-Risk Myelodysplastic Syndromes (HR MDS), Chronic Myelomonocytic Leukemia (CMML), or Low-Blast Acute Myelogenous Leukemia (AML): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Pevonedistat Plus Azacitidine Versus Single-Agent Azacitidine as First-Line Treatment for Participants With Higher-Risk Myelodysplastic Syndromes (HR MDS), Chronic Myelomonocytic Leukemia (CMML), or Low-Blast Acute Myelogenous Leukemia (AML) (NCT03268954), a randomized phase 3 study. The primary outcome is Event-Free Survival (EFS).

2. Clinical Question

This is a time-to-event question: the analysis considers both whether an event occurred and how long participants remained event-free. The primary outcome was the median EFS in each treatment group, accompanied by a 95% confidence interval, and a reported hazard ratio from a log-rank analysis. These statistics address different aspects of the event-time distributions and should be interpreted together rather than as interchangeable measures.

The trial also recorded overall survival and several other outcomes, including early mortality and measures of remission. Those outcomes provide additional context, but they do not replace the prespecified primary question. The registry record does not report the original statistical design assumptions or a multiplicity strategy for interpreting multiple outcomes.

3. Trial Design

This was a randomized, controlled, open-label, parallel-group phase 3 study sponsored by Takeda. The registry record lists 454 participants as enrolled and the study status as completed. The study began on 2017-11-28, and primary completion was 2021-05-28. Participants were assigned to azacitidine alone or pevonedistat plus azacitidine. Masking was listed as none, so the study was open-label.

The azacitidine-alone group received azacitidine 75 mg/m^2 by intravenous or subcutaneous injection on Days 1 to 5 and Days 8 and 9 of 28-day cycles. The combination group received the same azacitidine schedule plus pevonedistat 20 mg/m^2 by intravenous infusion on Days 1, 3, and 5 of each 28-day cycle. In both groups, treatment continued until disease progression or unacceptable toxicity, or for up to a maximum of 63 cycles.

The primary analysis used the ITT population, which the registry record defines as all randomized participants. The posted primary results list 227 participants in each treatment group. A randomized comparison is intended to reduce systematic differences between groups at assignment; the registry record does not provide enough information to assess baseline balance or the details of the randomization procedure.

The registry record does not report the original sample-size justification, the planned treatment-effect assumptions, or the full statistical analysis plan. It also does not report a detailed account of interim analyses. The planning illustrations associated with the posted results use stated assumptions for illustration and should not be mistaken for the trial’s original design rationale.

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: Participants diagnosed with myelomonocytic, and myelogenous leukemia were randomized into two groups in 1:1 ratio to receive single-agent azacitidine or azacitidine + pevonedistat.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: ITT Population included all participants who were randomized.

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.

Population: ITT Population included all participants who were randomized.

6. Primary Endpoint

The registry lists the primary outcome as Event-Free Survival (EFS), measured over From randomization until transformation to acute myeloid leukemia, or death due to any cause: up to approximately 42 months.

The primary endpoint, event-free survival (EFS), measures the time from randomization to the first qualifying event. In participants with MDS or CMML, an event was death from any cause or transformation to AML, whichever occurred first. The registry record defines AML transformation using the World Health Organization classification: greater than 20 % blasts in the blood or marrow and an increase in blast count by 50%. For participants with low-blast AML, an EFS event was death.

Time-to-event endpoints are useful when the timing of an event matters and participants may be followed for different lengths of time. A median gives a central point on the estimated event-time distribution: it is the time by which the estimated proportion remaining event-free has reached one-half. It is not the average time to an event, and it does not describe every participant’s experience.

The registry record reports median EFS with 95% confidence intervals and a hazard ratio from a log-rank analysis. The hazard ratio summarizes the relative event rate over follow-up under the assumptions of that analysis; it is not a direct difference between the medians. EFS also combines distinct events for participants with MDS or CMML, whereas death alone defines the event for low-blast AML. The endpoint therefore reflects the registry’s stated disease-specific definitions, not a single identical event type for every participant.

7. Results Posted in the Registry

• Azacitidine 75 mg/m^2 vs Pevonedistat 20 mg/m^2 + Azacitidine 75 mg/m^2: hazard ratio 0.97 (95% CI 0.76–1.24), reported p = =0.557 (as reported in the registry).

8. Secondary Efficacy Endpoints

The registry also posts results for secondary outcomes. For each outcome below, the endpoint type, the test applied to the posted summary statistics, the result and its interpretation are given in turn. Secondary endpoints are usually not adjusted for multiple testing, so they support rather than confirm the primary finding.

Overall Survival (OS)

• Endpoint type: Time to event: compared as a hazard ratio.
• Test: Hazard ratio as reported by the sponsor (Log Rank); it was not recalculated.
• Result:
  • Azacitidine 75 mg/m^2 vs Pevonedistat 20 mg/m^2 + Azacitidine 75 mg/m^2: hazard ratio 0.89 (95% CI 0.71–1.11), reported p = =0.152 (as reported in the registry).
• Interpretation:
  • The 95% confidence interval includes no difference (a ratio of 1), so these data do not show a statistically significant difference between Azacitidine 75 mg/m^2 vs Pevonedistat 20 mg/m^2 + Azacitidine 75 mg/m^2 and the control. The interval gives the range of effects that remain compatible with the data.

Thirty-Day Mortality Reported as Number of Participants Who Died Up to Day 30

• Endpoint type: Binary: the number of participants with the event, compared as proportions.
• Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
• Result:
  • Pevonedistat 20 mg/m^2 + Azacitidine 75 mg/m^2: 2.2% (5/223) versus 2.7% (6/220) on control; risk difference -0.5 percentage points (95% CI -3.4 to 2.4), risk ratio 0.82 (0.25–2.65); unadjusted two-proportion z-test p = 0.743.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Pevonedistat 20 mg/m^2 + Azacitidine 75 mg/m^2 and the control. The interval gives the range of effects that remain compatible with the data.

Sixty-Day Mortality Reported as Number of Participants Who Died Up to Day 60

• Endpoint type: Binary: the number of participants with the event, compared as proportions.
• Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
• Result:
  • Pevonedistat 20 mg/m^2 + Azacitidine 75 mg/m^2: 6.3% (14/223) versus 6.8% (15/220) on control; risk difference -0.5 percentage points (95% CI -5.1 to 4.1), risk ratio 0.92 (0.46–1.86); unadjusted two-proportion z-test p = 0.818.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Pevonedistat 20 mg/m^2 + Azacitidine 75 mg/m^2 and the control. The interval gives the range of effects that remain compatible with the data.

The secondary outcomes add context but do not overturn the primary EFS interpretation. Overall survival is the other posted time-to-event comparison with a reported hazard ratio. Its interval includes the no-relative-difference value, and its p-value does not provide clear evidence of a between-group difference. The early mortality recalculations also have confidence intervals that include no difference. Across these posted comparisons, the estimates do not establish a clear separation between the treatment groups.

The registry lists additional secondary outcomes, including six-month and one-year survival estimates, AML transformation, remission measures, and mortality at other time points. Results for several listed secondary outcomes are not reported in the registry record. Their presence in the outcome list should not be confused with posted numerical findings.

Secondary endpoints can help describe different aspects of a trial, but they generally carry less weight than the primary endpoint for the main treatment question. Multiple outcomes create multiple opportunities for apparently notable findings, especially when no multiplicity adjustment or hierarchy is reported. The registry record does not report a multiplicity strategy, so the strength of any unadjusted secondary signal cannot be evaluated in that context.

Secondary results also differ in definition, timing, and statistical scale. Overall survival records death over time, while early mortality is a count within a defined period. Remission and transformation outcomes address other clinical events. They are not interchangeable, and an apparent pattern across them does not by itself establish a common treatment effect. The primary EFS comparison remains the central result for the question stated in the trial description.

Secondary outcomes registered for this trial, for reference (only those above were recalculated):

• Overall Survival (OS)
• Kaplan-Meier Estimates of Six-Month Survival Rate
• Kaplan-Meier Estimates of One-Year Survival Rate
• Thirty-Day Mortality Reported as Number of Participants Who Died Up to Day 30
• Sixty-Day Mortality Reported as Number of Participants Who Died Up to Day 60
• Time to Acute Myelogenous Leukemia (AML) Transformation in Higher-Risk Myelodysplastic Syndromes (HR MDS), Higher-Risk Chronic Myelomonocytic Leukemias (HR CMML) and HR MDS/CMML Participants
• Number of Participants With Complete Remission (CR) and CR+ Complete Remission With Incomplete Blood Count Recovery (CRi)
• Number of Participants With CR and Marrow CR

9. Statistical Interpretation

10. Statistical Methods Explained

How should the median EFS be read?

The median is a summary of the estimated time-to-event distribution, rather than a count of events or a mean. It marks the time at which the estimated event-free proportion reaches one-half. In this trial, the registry reports medians of 15.7 months for azacitidine and 17.7 months for the combination. The values alone do not account for uncertainty or establish a treatment contrast. Their confidence intervals communicate uncertainty around each group’s estimated median, while the reported hazard ratio addresses the between-group comparison.

What does the hazard ratio describe?

A hazard is an instantaneous event rate among participants who have not yet had the event, conditional on remaining at risk up to that time. A hazard ratio compares those rates between groups under the analysis. A value of 1 is the no-relative-difference reference; values on either side represent different directions of relative event rates, depending on the comparison’s group ordering. The registry labels the comparison as azacitidine versus the combination and reports a hazard ratio of 0.968. It is safest to report that estimate and its uncertainty as stated rather than infer a direction from the order or reverse it.

A hazard ratio is not a ratio of median survival times, an absolute change in months, or the proportion of participants who experienced an event. It is a summary of relative event rates over follow-up. Its interpretation can be less straightforward if the relative rates change over time; the registry record does not report a proportional-hazards assessment or other diagnostics for this assumption.

What do the log-rank analysis and p-value contribute?

The registry reports a log-rank analysis for the primary EFS comparison. In general, the log-rank test compares observed and expected event patterns over follow-up, accounting for the ordering of event times and censoring. The reported p-value is 0.557. A p-value describes how unusual data at least as incompatible with a null hypothesis would be under that hypothesis and the analysis assumptions; it is not the probability that the null hypothesis is true, nor the probability that the finding occurred by chance.

The p-value should be considered alongside the effect estimate and confidence interval. Here, the reported hazard ratio is 0.968 and its 95% confidence interval is 0.757 to 1.238. The interval includes 1, so it does not rule out the no-relative-difference value at the corresponding confidence level. It also leaves room for effects in either direction. The result is therefore better described as inconclusive about a difference than as proof of no effect.

How do confidence intervals add information?

A confidence interval expresses uncertainty in an estimate under a statistical procedure and its assumptions. It is not a range containing a known fraction of individual participant outcomes, nor does it guarantee that the true value lies within the interval. For a hazard ratio, the interval’s location relative to 1 helps show whether the data are compatible with no relative difference and how broad the range of compatible estimates is.

The median EFS intervals are 14.42 to 19.68 months for azacitidine and 13.63 to 20.24 months for the combination. They describe uncertainty in each median separately. Comparing the endpoints of these separate intervals is not the same as estimating a confidence interval for a difference between medians. The reported hazard-ratio interval is the direct interval for the stated relative comparison.

How are the early-mortality comparisons different?

Early mortality is reported as participant counts by Day 30 and Day 60. For these outcomes, the posted recalculations use unadjusted risk differences based on the group-level counts and denominators. A risk difference is an absolute contrast between two event proportions; it differs from a hazard ratio because it does not use event timing. The Day 30 recalculated risk difference is -0.005, with a 95% confidence interval of -0.034 to 0.024 and p = 0.743. The Day 60 recalculated risk difference is -0.005, with a 95% confidence interval of -0.051 to 0.041 and p = 0.818.

These are explicitly unadjusted recalculations from posted summary counts, not sponsor-reported analyses. Their confidence intervals include zero, the no-risk-difference value. The registry record does not report a sponsor analysis for these comparisons. Because the endpoints are secondary and the estimates are imprecise, they are best treated as descriptive evidence rather than definitive tests of a treatment difference.

Why can over-dispersion matter?

In general, over-dispersion occurs when observed variability exceeds what a model assumes. If a model ignores it, standard errors and confidence intervals may be too narrow, and tests may overstate precision. The registry record does not report a model-based count analysis for the reported primary time-to-event endpoint, nor does it provide a specific over-dispersion assessment. The concept is relevant when interpreting statistical models generally, but it should not be used to infer an unreported analysis in this trial.

Finally, the registry’s primary population is ITT, defined as all randomized participants. That definition anchors the primary result to assigned groups. The registry record does not report further detail on censoring rules, missing outcome handling, or sensitivity analyses, so those aspects cannot be assessed from the posted information.

11. Confidence Intervals and Precision

Precision is reflected in the width of an estimate’s confidence interval. The primary hazard-ratio estimate is 0.968, and its 95% confidence interval runs from 0.757 to 1.238. The interval includes 1 and permits values on either side of that reference. This means the result is compatible with a range of relative effects; it is not precise enough to isolate a clear direction of difference from the interval alone.

The median EFS confidence intervals also show uncertainty in each group’s estimated median. The interval is 14.42 to 19.68 months for azacitidine and 13.63 to 20.24 months for the combination. They overlap, but overlap is not a formal comparison, and these separate intervals are not a confidence interval for the difference between medians. The reported hazard-ratio interval is the direct interval for the primary relative comparison.

Precision depends on factors such as the amount of event information, variability, and follow-up. The registry record lists the primary timeframe as from randomization until an EFS event, up to approximately 42 months. It does not report the event count used for the primary analysis, detailed censoring information, or a precision target for the original design. Those omissions limit further assessment of why the interval has its observed width.

Confidence intervals should not be interpreted as showing that every value inside is equally likely, or that values outside are impossible. They quantify uncertainty under the method’s assumptions. The primary interval does not support a definitive claim of benefit, harm, or equivalence; it shows that the data remain compatible with multiple possible relative effects.

12. Planning a Similar Trial: Sample Size

This is a planning illustration, not a re-evaluation of the original trial. It asks: if a new study wanted to detect a difference like the one observed, or a smaller one, how large would it need to be? It is deliberately not “observed power” of the finished trial, which is not a meaningful quantity.

The sample-size figures are planning illustrations based on the observed effects, not the trial’s original design assumptions. The registry record does not report the original sample-size justification or the assumptions used to plan enrollment. The illustrations use a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. They consider detecting the stated fraction of the observed effect.

For time-to-event outcomes, the figures are the total number of events needed, not participants per arm. For other outcomes, the figures are participants per arm. This distinction matters: an event target is not interchangeable with an enrollment target, because participants may be followed for different lengths of time and not all will have an event during follow-up.

The 80% power column represents the planned chance of rejecting the no-difference null hypothesis if the assumed effect is real and the analysis assumptions hold. The 90% power column represents a higher target chance under the same conditions. Neither column describes what power the completed trial actually achieved. The listed figures are not observed power.

Rows for planning for 100%, 75%, and 50% of the observed effect show how the target effect size changes the planning requirement. A smaller effect is harder to distinguish from random variation, so it generally requires more events or participants to reach the same power target. These rows are illustrations for hypothetical future planning, not evidence that the trial was designed around the observed effect or that any listed target was reached.

The curve should be read in the same spirit: it displays how the illustrated planning requirement varies with the assumed effect size. It does not alter the interpretation of the trial’s estimates, confidence intervals, or reported p-values. Readers should keep the endpoint scale in mind when moving between the event-based time-to-event figures and the per-arm participant figures.

13. How the Sample Size Is Derived

Power is the probability that a study of a given size rejects the null hypothesis when a specific true effect exists. It depends on four things: the effect to detect, the variability of the outcome, the significance level (here a two-sided α of 0.05, so z_1−α/2 = 1.960) and the desired power (z_1−β = 0.842 for 80% and 1.282 for 90%).

Formula

d is the total number of events needed (Schoenfeld's formula) with 1:1 allocation and proportional hazards.

Worked calculation for this trial

1. Hazard ratio HR = 0.968, so ln(HR) = -0.0325.
2. d = 4 × (1.96 + 0.842)² / [ln(0.968)]² = 29,681.2, rounded up to 29,682 events in total for 80% power.

The smaller-effect rows in the table above repeat the same calculation with the effect multiplied by 0.75 and 0.5 (for ratios, applied to the logarithm of the ratio).

R code

alpha <- 0.05
z_a   <- qnorm(1 - alpha / 2)          # two-sided
hr <- 0.968                              # observed hazard ratio
events_needed <- function(power, frac = 1) ceiling(round(4 * (z_a + qnorm(power))^2 / log(hr^frac)^2, 6))   # Schoenfeld, 1:1 allocation
for (f in c(1, 0.75, 0.5))
  cat(sprintf("%3.0f%% of observed effect: events = %s (80%% power), %s (90%% power)\n", 100 * f,
              format(events_needed(0.80, f), big.mark = ","), format(events_needed(0.90, f), big.mark = ",")))

# Simulation: given d total events under proportional hazards, events in the treatment arm are Binomial(d, hr / (1 + hr)).
# The score (log-rank type) statistic is (D1 - d / 2) / sqrt(d / 4).
set.seed(3268954)
d    <- events_needed(0.80)
reps <- 20000
D1 <- rbinom(reps, d, hr / (1 + hr))
z  <- (D1 - d / 2) / sqrt(d / 4)
cat(sprintf("Simulated power with %s events: %.3f (%d replicates)\n", format(d, big.mark = ","), mean(abs(z) > z_a), reps))

Output from running the code

100% of observed effect: events = 29,682 (80% power), 39,735 (90% power)
 75% of observed effect: events = 52,767 (80% power), 70,640 (90% power)
 50% of observed effect: events = 118,725 (80% power), 158,939 (90% power)
Simulated power with 29,682 events: 0.802 (20000 replicates)

The simulation uses the fact that, with d events in total and proportional hazards, the number of events in the treatment arm is binomial. It is an approximation that conditions on the total number of events, and it should land close to the target power.

14. Group-Sequential Design Analysis

Many time-to-event trials, especially in oncology, stop or report early when interim data are overwhelming, so a single fixed-sample number understates the real design problem. The calculation below is an illustrative group-sequential design built from this trial's posted control median and hazard ratio. The registry record does not state the planned number of looks or the spending function, so the design is a planning illustration, not a reconstruction of the protocol. Assumed: one-sided α = 0.025, 90% power, three equally spaced looks on information, Lan-DeMets O'Brien-Fleming spending, 24 months of accrual, 60 months in total and a 1% monthly dropout hazard.

Under these assumptions the design needs about 53,895 participants in total, with the boundaries below.

Fixed design versus group-sequential alternatives

Group-sequential designs cost a small increase in maximum sample size and events compared with the fixed design, in exchange for the chance to stop early for overwhelming efficacy. O'Brien-Fleming spending keeps early boundaries very conservative and the final boundary close to the fixed-design value, whereas Pocock spending spends more alpha early.

Sensitivity to a smaller true effect

Planning for exactly the observed effect is optimistic, because observed effects in published trials tend to be somewhat larger than the true effect. The table repeats the main design (three looks, O'Brien-Fleming spending, 90% power) for smaller true hazard ratios.

For context, this trial enrolled 454 participants.

Cross-check, upper boundaries: gsDesign 3.71, 2.511, 1.993 | rpact 3.71, 2.511, 1.993. Agreement between the two packages confirms the boundary calculation.

R code and output

suppressWarnings(suppressMessages({ library(gsDesign); library(rpact) }))
med_control <- 15.7          # control-arm median, months (converted from the posted unit)
hr          <- 0.968          # hazard ratio used for planning
lambdaC     <- log(2) / med_control
eta         <- 0.01               # dropout hazard per month (assumption)
acc_months  <- 24                 # accrual duration (assumption)
study_months<- 60                 # total study duration (assumption)

design <- function(k, sfu, beta, h = hr) gsSurv(k = k, test.type = 1, alpha = 0.025, beta = beta, sfu = sfu,
  lambdaC = lambdaC, hr = h, hr0 = 1, eta = eta, gamma = 1, R = acc_months, T = study_months, minfup = study_months - acc_months)

# Main design: 3 looks, Lan-DeMets O'Brien-Fleming spending, one-sided alpha 0.025, 90% power
x <- design(3, sfLDOF, 0.10)
cat(sprintf("N_TOTAL|%s\n", format(ceiling(sum(x$gamma * x$R)), big.mark = ",")))
for (i in 1:3) cat(sprintf("GS|%d|%.1f|%.0f|%.3f|%.5f|%.5f\n", i, x$T[i], x$n.I[i], x$upper$bound[i], pnorm(-x$upper$bound[i]), cumsum(x$upper$spend)[i]))
cat("\n"); print(gsBoundSummary(x))

# Comparison of designs
cmp <- list(list("Fixed design, one analysis", 1, sfLDOF), list("Two looks, O'Brien-Fleming", 2, sfLDOF),
            list("Three looks, O'Brien-Fleming", 3, sfLDOF), list("Three looks, Pocock", 3, sfLDPocock))
cat("\n")
for (b in c(0.20, 0.10)) for (d in cmp) {
  y <- design(d[[2]], d[[3]], b)
  cat(sprintf("DES|%s|%d|%.0f|%s|%.1f\n", d[[1]], round(100 * (1 - b)), tail(y$n.I, 1), format(ceiling(sum(y$gamma * y$R)), big.mark = ","), tail(y$T, 1)))
}

# Sensitivity: what if the true effect is smaller than the observed one? (3 looks, O'Brien-Fleming, 90% power)
cat("\n")
for (fr in c(1, 0.75, 0.5)) {
  y <- design(3, sfLDOF, 0.10, exp(log(hr) * fr))
  cat(sprintf("SENS|%.2f|%.3f|%.0f|%s\n", fr, exp(log(hr) * fr), tail(y$n.I, 1), format(ceiling(sum(y$gamma * y$R)), big.mark = ",")))
}

# Independent cross-check of the boundaries with rpact
r <- getDesignGroupSequential(kMax = 3, alpha = 0.025, beta = 0.10, sided = 1, typeOfDesign = "asOF", informationRates = x$timing)
cat(sprintf("\nCross-check, upper boundaries: gsDesign %s | rpact %s\n", paste(round(x$upper$bound, 3), collapse = ", "), paste(round(r$criticalValues, 3), collapse = ", ")))

Method: LachinFoulkes 
      Analysis               Value Efficacy
     IA 1: 33%                   Z   3.7103
      N: 43630         p (1-sided)   0.0001
 Events: 13401        ~HR at bound   0.9379
     Month: 19    P(Cross) if HR=1   0.0001
               P(Cross) if HR=0.97   0.0338
     IA 2: 67%                   Z   2.5114
      N: 53896         p (1-sided)   0.0060
 Events: 26802        ~HR at bound   0.9698
     Month: 31    P(Cross) if HR=1   0.0060
               P(Cross) if HR=0.97   0.5603
         Final                   Z   1.9930
      N: 53896         p (1-sided)   0.0231
 Events: 40203        ~HR at bound   0.9803
     Month: 60    P(Cross) if HR=1   0.0250
               P(Cross) if HR=0.97   0.9000



Cross-check, upper boundaries: gsDesign 3.71, 2.511, 1.993 | rpact 3.71, 2.511, 1.993

15. Analyses Reported by the Sponsor

For EFS, the registry reports the sponsor’s log-rank hazard ratio as 0.968, with a 95% confidence interval of 0.757 to 1.238 and p = 0.557. The primary outcome is a time-to-event endpoint, and the posted interpretation uses the sponsor-reported hazard ratio rather than a newly calculated estimate from summary medians. A hazard ratio cannot be reconstructed reliably from the medians alone.

The registry record also reports a sponsor log-rank analysis for overall survival: hazard ratio 0.888, 95% confidence interval 0.709 to 1.113, and p = 0.152. These are the sponsor’s reported statistics for that secondary time-to-event outcome. No separate recalculation from the medians is used to replace them.

The early-mortality comparisons are different. The risk differences and associated intervals and p-values are unadjusted recalculations from the posted counts and denominators, and the registry record does not list sponsor-reported analyses for those comparisons. They should therefore be identified as recalculations, not attributed to the sponsor.

No discrepancy is flagged between the posted sponsor analyses and the values used for the time-to-event interpretation. Where a discrepancy exists, the sponsor’s own analysis takes precedence; that situation is not indicated for these results. The distinction between reported analyses and simple recalculations remains important because the latter do not reproduce every feature of a trial’s full analysis.

16. Multiplicity, Missing Data and Analysis Populations

The registry record lists a primary endpoint and numerous secondary outcomes, but does not report a multiplicity-adjustment procedure or a formal hierarchy for testing them. When many statistical comparisons are examined, the chance of at least one apparently unusual result can increase. In general, multiplicity procedures are used to control the interpretation of multiple tests or to allocate statistical error across them. Without a reported plan, secondary p-values should be regarded as exploratory or descriptive rather than treated as if each were the single primary test.

The reported primary EFS comparison is distinct from those secondary comparisons. Its reported log-rank p-value is 0.557. The record does not state whether any interim testing or alpha-spending procedure was used, so those design features are not reported in the registry record. No assumptions about them can be made from the final estimate alone.

The primary population is described as the ITT population and includes all randomized participants. This is a clear population definition for the primary endpoint, but the record does not provide detailed information on how missing outcome data were handled, the censoring rules, or sensitivity analyses addressing missingness. Time-to-event analyses often use censoring to account for participants whose event time is not observed during follow-up; the particular procedures used in this trial are not reported in the registry record.

The listed enrollment and analysis group sizes do not answer questions about completeness of follow-up, reasons for censoring, or missing secondary measurements. Nor do they establish that every secondary endpoint was measured for every randomized participant. Without those details, readers cannot assess the impact of missingness or evaluate whether alternative assumptions would change the conclusions.

Open-label conduct is also relevant context: participants and study personnel were not masked. The primary endpoint includes death and AML transformation, but the registry record alone does not show how outcome assessment was organized or whether adjudication was independent. These details, as well as the statistical plan for multiple endpoints, are not reported in the registry record.

17. 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: febrile neutropenia (19.3%), pneumonia (14.8%), pyrexia (7.3%), anaemia (5.8%), septic shock (4.9%).

The registry reports serious adverse events in 145 of 220 participants at risk in the azacitidine group and 156 of 223 at risk in the combination group. The most frequently listed serious adverse-event terms were febrile neutropenia (19.3%), pneumonia (14.8%), pyrexia (7.3%), anaemia (5.8%), septic shock (4.9%), and sepsis (4.5%). These percentages are posted for the listed terms; the record excerpt does not specify additional detail needed to interpret event timing or recurrence.

Other adverse events were recorded for 207 of 220 participants at risk in the azacitidine group and 210 of 223 at risk in the combination group. The adverse-event section also lists deaths as 153 and 150, respectively. These are descriptive safety counts and are not, by themselves, estimates of treatment-attributable risk. The registry record does not establish that every death was caused by treatment.

Safety counts should be read with their denominators and the registry’s definitions in mind. Serious adverse events, other adverse events, and deaths are distinct summaries and should not be combined. These data do not support a causal conclusion about comparative safety without fuller information on definitions, exposure, event attribution, and follow-up.

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

The primary result answers a focused statistical question: whether the posted EFS analysis shows evidence of a difference between pevonedistat plus azacitidine and azacitidine alone in the randomized population. The reported hazard ratio is close to the no-relative-difference reference, its confidence interval includes that reference, and the log-rank p-value is 0.557. The registry results therefore do not provide clear statistical evidence of an EFS difference.

That conclusion is not the same as saying the treatments are identical or equivalent. The confidence interval remains compatible with relative effects in either direction, and no equivalence margin is reported. A non-significant result can reflect an estimate near the null, uncertainty around the estimate, or both; it does not prove that a clinically meaningful difference is impossible.

The medians describe the estimated time at which half the population is estimated to have experienced the endpoint, not an expected individual outcome. The hazard ratio summarizes relative event rates under the analysis and is not a difference in months. The EFS definition also differs by disease category: MDS and CMML include death or AML transformation, while low-blast AML uses death. These features matter when translating the statistic into a plain-language account of what was measured.

Secondary outcomes provide additional descriptive information, but the reported secondary comparisons do not establish a clear difference either. Their interpretation is further limited by the number of outcomes and the absence of a reported multiplicity strategy. Several listed outcomes do not have results reported in the registry record.

These findings describe the posted trial results and their statistical uncertainty. They do not establish a treatment recommendation, predict an individual participant’s outcome, or replace clinical judgment. The registry record also leaves some methods details unreported, including missing-data handling and the original design assumptions, which limits how fully the analysis can be independently assessed.

19. Why This Trial Matters Statistically

This trial illustrates why a treatment comparison cannot be reduced to comparing two medians. EFS medians provide an accessible description of each group’s event-time distribution, but the hazard ratio, confidence interval, and log-rank test address the comparative time-to-event question more directly. The measures agree in pointing to substantial uncertainty about a between-group difference rather than a clear statistical separation.

It also demonstrates the importance of reading estimates with their intervals. A point estimate alone can invite an overly definite interpretation; its confidence interval shows how much uncertainty remains and whether the no-difference reference is compatible with the data. In this study, the hazard-ratio interval crosses 1, so the estimate does not exclude either direction of relative effect.

The trial’s endpoint definition is another important statistical feature. EFS combines death and AML transformation for participants with MDS or CMML, but death alone for participants with low-blast AML. A composite endpoint can capture more than one clinically relevant event, but the components need not have identical meaning or frequency. Readers should understand the stated definition before interpreting the summary statistic.

Finally, the trial makes clear why secondary findings, safety summaries, and planning illustrations should be kept in their proper roles. Secondary outcomes can complement the primary analysis but raise multiplicity concerns. Safety counts are descriptive and require context about exposure and event attribution. Planning figures based on observed effects are hypothetical illustrations, not achieved power or evidence about the original design. Clear separation of these concepts supports a more calibrated reading of a complex clinical-trial record.

20. 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.

Several trial-specific limitations affect interpretation of the posted results. The study was open-label, with no masking listed. The registry record does not report how EFS events were assessed or whether AML transformation was reviewed independently. Because EFS includes transformation as well as death for MDS and CMML, details of outcome ascertainment would help readers judge the potential for assessment differences.

The EFS event definition is not identical across all diagnostic groups: transformation or death is counted for MDS and CMML, whereas death is counted for low-blast AML. The registry record does not provide subgroup-specific primary estimates in the registry record for this comparison. The overall estimate therefore should not be assumed to describe each disease group separately.

The registry record does not report the original sample-size assumptions, missing-data procedures, censoring rules, interim-analysis plan, or multiplicity strategy. Those omissions limit assessment of design choices and the robustness of secondary comparisons. Nor are results reported for several listed secondary outcomes, restricting the scope of conclusions about the broader outcome set.

The primary timeframe is described as up to approximately 42 months, while overall survival is listed as up to approximately 6.9 years. Different follow-up horizons can affect the amount of information available for distinct endpoints. Detailed follow-up completeness and event counts are not reported, so the precision of the estimates cannot be linked to those factors in depth.

Lastly, the serious adverse-event summaries are counts and listed terms, not a complete causal comparison. The registry record does not provide the information needed to evaluate exposure-adjusted rates, recurrent events, or attribution of deaths. These limitations qualify interpretation without changing the numerical results reported.

21. Related Tutorials

22. Related Statistical Calculators

23. Sources

• ClinicalTrials.gov: Pevonedistat Plus Azacitidine Versus Single-Agent Azacitidine as First-Line Treatment for Participants With Higher-Risk Myelodysplastic Syndromes (HR MDS), Chronic Myelomonocytic Leukemia (CMML), or Low-Blast Acute Myelogenous Leukemia (AML) — NCT03268954.
• Registry record last updated 2025-10-29; status last verified 2025-10.
• Results first posted 2022-09-22.
• 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.

24. Record Summary

NCT03268954 was a completed, randomized, controlled, open-label, parallel-group phase 3 study sponsored by Takeda. It compared first-line pevonedistat plus azacitidine with azacitidine alone in participants with HR MDS, CMML, or low-blast AML. The registry lists enrollment of 454 and 227 participants in each primary analysis group.

The primary endpoint was EFS, defined from randomization to death or AML transformation for MDS or CMML, and to death for low-blast AML. Median EFS was 15.7 months with azacitidine and 17.7 months with the combination. The reported hazard ratio was 0.968, with a 95% confidence interval of 0.757 to 1.238 and p = 0.557.

The primary comparison does not provide clear statistical evidence of a difference, while its confidence interval remains compatible with effects in either direction. Overall survival and early mortality were also posted; several other listed secondary outcomes do not have results reported in the registry record. Details including the original sample-size rationale, multiplicity strategy, and missing-data handling are not reported in the registry record.