Randomized Phase Lll Study of Imatinib Dose Optimization vs Nilotinib in CML Patients With Suboptimal Response to Imatinib: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Randomized Phase Lll Study of Imatinib Dose Optimization vs Nilotinib in CML Patients With Suboptimal Response to Imatinib (NCT00802841), a randomized phase 3 study. The primary outcome is Percentage of Participants With Complete Cytogenetic Response (CCyR).

2. Clinical Question

The clinical question centers on whether the two treatment strategies produced different observed CCyR percentages in this population. The registry record describes the rationale as uncertainty about dose escalation for participants with a suboptimal response and the potential benefit of nilotinib for this group. The reported primary outcome is a response proportion at a specified assessment time, not a time-to-event outcome.

The comparison is between randomized assignment to nilotinib and imatinib dose optimization. It does not directly answer how either strategy compares with continued standard-dose imatinib, because that is not one of the two listed randomized arms. The trial also records secondary outcomes, including molecular response, survival, and other cytogenetic response measures, but the primary question is the 6-month CCyR percentage.

3. Trial Design

This was a randomized, phase III, parallel-group study with no masking listed in the registry record. The trial enrolled 191 participants. The primary outcome population is described as the Full Analysis Set: all participants assigned treatment by randomization. The recorded arm sizes for the primary outcome are 96 for nilotinib and 95 for imatinib.

The assigned regimens were 400 mg nilotinib twice daily and 600 mg imatinib once daily. The primary outcome was assessed at 6 months. CCyR was evaluated using bone marrow samples and defined as having 0% Philadelphia positive chromosome metaphases. The registry lists one analysis for the primary comparison and identifies a Fisher Exact test as the sponsor-reported method.

The study is marked completed. Its listed secondary outcomes include major molecular response, CCyR, time to and duration of CCyR, progression-free survival, event-free survival, and overall survival. The registry lists eight outcome measures and seven posted secondary outcomes. These additional outcomes broaden the clinical questions but do not change the definition of the primary outcome.

The registry record does not report the trial's original sample-size assumptions, the allocation ratio beyond the arm counts, or the reasoning used to select the enrollment target. Details of concealment, adherence, treatment discontinuation, crossover timing, and follow-up procedures are not reported in the registry record. The record identifies the study as unmasked, but does not describe how that feature was handled in outcome assessment or analysis.

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: Cross-over from one arm to the other was allowed for intolerant patients anytime during treatment, patients who failed to achieve CCyR after 6 months of treatment, loss of response, loss of CHR, loss of best achieved cytogenetic response any time during treatment, or other reason approved by the Study Management Committee.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Full Analysis Set: The FAS included all participants to whom study treatment had been assigned by randomization.

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 Percentage of Participants With Complete Cytogenetic Response (CCyR), measured over 6 months.

The primary endpoint is the percentage of participants with complete cytogenetic response at 6 months. CCyR was assessed from bone marrow samples and defined as 0% Philadelphia positive chromosome metaphases in bone marrow. The analysis population was the Full Analysis Set, which includes participants assigned study treatment by randomization.

This endpoint expresses the proportion of participants meeting a specified response definition at a set assessment time. Unlike a time-to-event measure, it does not describe how long it took for a response to occur or how long a response lasted. The registry lists time to CCyR and duration of CCyR separately as secondary outcomes.

As general statistical context, a binary response endpoint can be summarized by counting participants who meet the definition and dividing by the number evaluated in each group. Such a summary is readily interpretable as an event percentage, but its precision depends on the number of participants and the observed response frequency. The posted arm percentages are 50.0% for nilotinib and 42.1% for imatinib. The registry record does not report further details about the timing or handling of missing bone marrow assessments.

7. Results Posted in the Registry

• Imatinib: 42.1% (40/95) versus 50.0% (48/96) on control; risk difference -7.9 percentage points (95% CI -22.0 to 6.2), risk ratio 0.84 (0.62–1.15); unadjusted two-proportion z-test p = 0.274.

8. Statistical Interpretation

9. Statistical Methods Explained

What does the risk difference describe?

For a binary endpoint, a risk difference compares the event proportions in two groups by subtracting one group proportion from the other. The posted recalculation is oriented as imatinib relative to nilotinib, so its estimate is -0.079. A negative value in that orientation corresponds to a lower observed response proportion in the imatinib group. This is an unadjusted contrast based on the posted arm summaries; it is not an estimate adjusted for other participant characteristics.

As general statistical explanation, the risk difference is an absolute measure of contrast on the probability scale. It answers how far apart the group proportions are, rather than expressing a relative ratio. It is useful because the direction and scale of the comparison are explicit, although an estimate by itself does not show how precisely the difference has been measured.

How should the confidence interval be understood?

A confidence interval gives a range of values compatible with the data under the method used to construct it. General statistical interpretation treats a narrow interval as more precise than a wide one, all else being equal. The recalculated interval for this risk difference crosses zero, the value that represents no difference on this scale. Therefore, it does not rule out no difference, and it also allows values in either direction.

The numerical confidence level and the interval construction method are not reported in the registry record. The interval should not be interpreted as assigning a probability to the true value after the data are observed. It instead conveys uncertainty from repeated use of the specified procedure under its assumptions; the registry does not document the interval procedure in detail.

What does the sponsor-reported test assess?

The registry names Fisher Exact as the sponsor-reported test for the primary comparison. As general explanation, Fisher's exact test evaluates the association between group assignment and a binary outcome using the observed counts. It is commonly used for categorical data and does not rely on the same large-sample approximation as some other tests. The sponsor-reported p-value is greater than 0.05.

The registry does not report additional detail about the sponsor's exact testing specification, including whether any covariate adjustment or other modification was used. The recalculated analysis is expressly an unadjusted re-analysis of the posted summary statistics. These analyses should be distinguished rather than treated as interchangeable outputs from a single documented model.

Why can the two p-values differ?

The sponsor-reported test and the recalculated comparison are not identical statistical procedures. Different test statistics, assumptions, or implementation details can produce different p-values from the same underlying counts. The registry record does not explain the reason for the difference, so attributing it to a particular analytic choice would be speculation. The recalculated p-value is 0.274, while the sponsor-reported p-value is greater than 0.05.

Neither p-value is a direct measure of effect magnitude. The group percentages and the risk-difference estimate describe the observed contrast; the interval describes its precision; and the p-value evaluates compatibility with a null hypothesis under a test. These pieces of information answer related but distinct questions.

What is the role of the analysis population?

The primary outcome population is the Full Analysis Set, defined as all participants assigned study treatment by randomization. Using an assigned-group population is consistent with preserving the randomized comparison, but the registry's definition alone does not establish how every participant's outcome was handled. Missing-outcome rules, exclusions after randomization, and imputation procedures are not reported in the registry record.

The primary outcome arm counts are 96 and 95. The registry does not report whether those denominators reflect all randomized participants with available endpoint data or another endpoint-specific handling rule. The record's FAS definition and the arm counts are useful context, but they do not fill in the absent missing-data details.

What can and cannot be concluded from a non-significant test?

As general statistical explanation, a p-value above a chosen significance threshold means the test did not reject its null hypothesis at that threshold. It does not prove that there is no difference, and it does not show that the treatments are equivalent. A demonstration of equivalence or non-inferiority would require a specified margin and an analysis designed for that question. Such a margin or analysis is not reported in the registry record.

Likewise, a non-significant result is not evidence that the trial had inadequate power. The trial's original sample-size calculation and planning assumptions are not reported in the registry record, so its planned ability to detect a particular effect cannot be assessed from the posted result alone. The separate planning illustrations are based on the observed effect and are not the trial's original design calculation.

10. Confidence Intervals and Precision

The posted percentages are 50.0% for nilotinib and 42.1% for imatinib. The unadjusted recalculated risk difference is -0.079 when expressed as imatinib relative to nilotinib. The associated recalculated confidence interval extends across zero, so the estimate is compatible with no difference as well as with differences in either direction. The interval's endpoints include a value reported to more than three decimal places, so the numerical limits are not reproduced.

As general statistical explanation, a confidence interval communicates precision by indicating the range of effect estimates consistent with the observed data under the method used. A range that crosses the no-difference value does not establish that the groups are equal; it indicates that the data do not exclude no difference at the interval's confidence level. The confidence level and construction method are not reported in the registry record.

The arm sizes of 96 and 95 provide context for the amount of information behind the percentages, but the percentages alone do not convey the full uncertainty. The endpoint is binary, and precision depends on both the number of participants and the response frequency. No adjusted interval, subgroup-specific interval, or interval for the sponsor-reported test is reported in the registry record.

The sponsor-reported Fisher Exact p-value and the unadjusted recalculated p-value are both greater than 0.05. These results are consistent with uncertainty about the size and direction of the difference; they should not be taken as proof of equivalence. The interval is more informative about the plausible range of the risk difference than a threshold decision based solely on a p-value.

11. 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 planning figures are participants per arm, not total events. They are illustrative calculations based on the observed effect, with equal allocation between two arms, a two-sided significance level of 0.05, and a normal approximation. Each planning row considers detecting a stated fraction of the observed effect. The original trial's sample-size assumptions are not reported in the registry record.

The 80% power column describes the planned number of participants per arm for a design that would have an 80% probability of detecting the specified effect under the stated assumptions, if that effect were the true one. The 90% power column gives the corresponding per-arm number for a design targeting 90% power. These are planning concepts, not guarantees about any particular trial result.

• For planning at 100% of the observed effect: 625 participants per arm for 80% power and 836 participants per arm for 90% power.
• For planning at 75% of the observed effect: 1,115 participants per arm for 80% power and 1,492 participants per arm for 90% power.
• For planning at 50% of the observed effect: 2,514 participants per arm for 80% power and 3,365 participants per arm for 90% power.

The smaller the effect targeted for detection, the more participants are needed under the same assumptions. Increasing the target power also requires a larger planned sample. These figures are not observed power; they are planning illustrations derived from the observed effect and the stated assumptions. They do not establish what sample size the trial itself was designed to recruit or what power it had for its planned analysis.

12. 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

n is the number of participants in each arm, p₀ and p₁ are the control and treatment proportions, and p̄ is their average.

Worked calculation for this trial

1. Control proportion p₀ = 0.500; treatment proportion p₁ = 0.421; average p̄ = 0.461.
2. First term: z_1−α/2 × √[2p̄(1−p̄)] = 1.96 × 0.7049 = 1.3816.
3. Second term: z_1−β × √[p₀(1−p₀) + p₁(1−p₁)] = 0.842 × 0.7027 = 0.5914.
4. n = (1.3816 + 0.5914)² / (-0.0789)² = 624.5, rounded up to 625 participants per arm 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
p0 <- 0.5                          # control proportion
p1 <- 0.421052631578947                          # treatment proportion
n_per_arm <- function(power, frac = 1) {
  q1 <- p0 + (p1 - p0) * frac               # smaller effect = fraction of the observed difference
  pbar <- (p0 + q1) / 2
  z_b <- qnorm(power)
  ceiling(round((z_a * sqrt(2 * pbar * (1 - pbar)) + z_b * sqrt(p0 * (1 - p0) + q1 * (1 - q1)))^2 / (q1 - p0)^2, 6))
}
for (f in c(1, 0.75, 0.5))
  cat(sprintf("%3.0f%% of observed effect: n per arm = %s (80%% power), %s (90%% power)\n", 100 * f,
              format(n_per_arm(0.80, f), big.mark = ","), format(n_per_arm(0.90, f), big.mark = ",")))

# Check against R's built-in function (unrounded n per arm)
cat("power.prop.test:", round(power.prop.test(p1 = p0, p2 = p1, power = 0.80)$n, 1), "\n")

# Simulation: how often does the z-test reject at the 80%-power sample size?
set.seed(802841)
n    <- n_per_arm(0.80)
reps <- 20000
x0 <- rbinom(reps, n, p0); x1 <- rbinom(reps, n, p1)
ph <- (x0 + x1) / (2 * n)
z  <- ((x1 - x0) / n) / sqrt(ph * (1 - ph) * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(z) > z_a, na.rm = TRUE), reps))

Output from running the code

100% of observed effect: n per arm = 625 (80% power), 836 (90% power)
 75% of observed effect: n per arm = 1,115 (80% power), 1,492 (90% power)
 50% of observed effect: n per arm = 2,514 (80% power), 3,365 (90% power)
power.prop.test: 624.5 
Simulated power at n = 625 per arm: 0.798 (20000 replicates)

The simulation draws many pairs of event counts at the planned sample size and counts how often the two-proportion z-test rejects the null. It should land close to the target power; at very small required sample sizes it can fall below it, because the normal approximation is rough there.

13. Analyses Reported by the Sponsor

The registry reports a sponsor analysis using Fisher Exact, with a reported p-value greater than 0.05. The unadjusted recalculation from the posted response summaries reports a risk difference of -0.079 and a p-value of 0.274. The recalculation is not the sponsor's analysis and should be read as a separate, simple comparison of the posted summary statistics.

The analyses use different reported approaches, and the registry record does not explain the difference between their p-values. The record does not report enough detail to attribute that difference to a particular implementation or analytic decision. Both results do not indicate a statistically significant difference at the 0.05 threshold, but neither is proof that the treatment groups are equivalent.

The sponsor's own analysis is the trial's reported inferential result. The recalculation can help readers see the unadjusted contrast and its uncertainty, but it does not replace the sponsor's test or add adjustment for participant characteristics. The discrepancy flag is false, so no conflict between the posted arm data and a reported sponsor result is identified in the registry record.

14. Multiplicity, Missing Data and Analysis Populations

The registry lists eight outcome measures, including the primary CCyR endpoint and seven secondary outcomes. The secondary outcomes cover major molecular response, CCyR, time to and duration of CCyR, progression-free survival, event-free survival, and overall survival. One primary analysis is listed. The registry record does not report a multiplicity adjustment, a hierarchy for testing outcomes, or a prespecified strategy for interpreting multiple statistical tests.

As general statistical explanation, testing multiple outcomes can increase the chance of at least one apparently notable result when each test is considered separately. Whether and how to control that concern depends on the study's prespecified testing plan. That plan is not reported in the registry record, so the primary finding should be interpreted in the context of the additional outcomes without assuming that a particular correction was or was not applied.

The Full Analysis Set is defined as all participants assigned study treatment by randomization. The registry does not report how missing primary endpoint assessments were handled, whether any randomized participants were excluded from the endpoint analysis, or whether imputation was used. The primary outcome arm counts are listed, but the record does not explain the relationship between those denominators and missing endpoint measurements.

The record also does not report detailed procedures for censoring or missing follow-up in the time-to-event secondary outcomes. Details of interim analyses, stopping rules, and the statistical analysis plan are not reported in the registry record. Their absence from the record means those features cannot be inferred from the posted primary result.

The trial is identified as unmasked, but the registry does not describe whether outcome assessment was blinded or how potential knowledge of assignment was addressed. CCyR was based on bone marrow samples, and the stated response definition is objective in form; however, procedural safeguards for assessment are not reported in the registry record. These omissions limit assessment of the analysis process, not the ability to describe the posted group percentages.

15. 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: tachyarrhythmia (7.7%), thrombocytopenia (2.1%), angina pectoris (2.1%), lymphadenopathy (1.8%), pneumonia (1.8%).

Serious adverse event counts are reported for four groups: 11 of 96 participants in the nilotinib group, 9 of 93 in the imatinib group, 4 of 56 in the cross-over to nilotinib group, and 1 of 13 in the cross-over to imatinib group. The denominators differ across groups. The registry record does not explain the timing, exposure periods, or rules used to define these groups for the safety summaries.

Other adverse events are reported for 80 of 96 in the nilotinib group, 61 of 93 in the imatinib group, 44 of 56 in the cross-over to nilotinib group, and 11 of 13 in the cross-over to imatinib group. The record lists tachyarrhythmia at 7.7%, thrombocytopenia and angina pectoris at 2.1% each, and lymphadenopathy, pneumonia, and blast cell crisis at 1.8% each among the top serious-event terms.

These are descriptive counts and listed percentages, not adjusted comparisons. The registry does not report deaths in these safety summaries. It also does not specify whether each event was considered related to assigned treatment or how event collection and follow-up differed after crossover. The group figures should therefore be read with their denominators and reporting context, not as direct evidence of comparative risk.

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

The primary outcome shows a higher observed CCyR percentage in the nilotinib group than in the imatinib group: 50.0% versus 42.1%. The unadjusted recalculated risk difference is -0.079 for imatinib relative to nilotinib, and its confidence interval crosses zero. The sponsor-reported Fisher Exact test and the recalculated test do not indicate a statistically significant difference at the 0.05 threshold.

This result does not establish that the two strategies are equivalent, nor does it establish that one is superior. A non-significant test means the analysis did not provide sufficient evidence to reject the null hypothesis under the test and threshold used; it does not prove the null hypothesis. The confidence interval also reflects uncertainty that includes no difference and possible effects in either direction.

The endpoint is a response proportion at 6 months, defined by bone marrow cytogenetic findings. It does not alone describe time to response, durability of response, survival, quality of life, or the full safety profile. Some of these outcomes are listed as secondary measures, but their posted results are not included in this primary comparison. Details about missing primary outcome data and the analysis of secondary outcomes are not reported in the registry record.

The data describe randomized groups in this study population and should be interpreted in that context. The record does not provide the original sample-size rationale, a detailed analysis plan, or complete information about endpoint missingness. Those gaps limit what can be concluded about the planned precision and the handling of all randomized participants. The observed percentages are trial results, not individualized predictions or medical advice.

17. Why This Trial Matters Statistically

This trial illustrates why a difference in observed percentages and evidence for a reliable difference are not the same thing. The nilotinib group's posted CCyR percentage is 50.0%, compared with 42.1% for imatinib, but the recalculated interval includes no difference and the reported tests do not cross the 0.05 significance threshold. Looking at the size and uncertainty of the estimate gives a fuller account than reading a p-value alone.

It also highlights the value of specifying the contrast direction. The reported risk difference is for imatinib relative to nilotinib, so its negative sign corresponds to a lower observed response proportion in the imatinib arm. Clear orientation helps prevent readers from assigning the sign to the wrong treatment group.

The sponsor-reported Fisher Exact test and the unadjusted recalculation answer related but not identical statistical questions. Distinguishing the trial's reported analysis from a recalculation helps readers understand that results can vary by method, while avoiding speculation about why they differ. The record's lack of detail on the analysis plan, missing data, and multiplicity further demonstrates why registry results should be interpreted with attention to what is documented and what remains unspecified.

Finally, the sample-size illustrations show how planning depends on the effect one aims to detect: smaller target effects require more participants under the same assumptions. These illustrations are separate from the trial's own design, whose original sample-size rationale is not reported. That distinction prevents a retrospective planning calculation from being mistaken for a description of the trial's planned power.

18. 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.
• Participant counts were reconstructed from posted percentages and arm sizes, so small rounding differences are possible.

The registry record does not report the original sample-size calculation or its assumptions. Consequently, the enrollment of 191 cannot be evaluated against a documented target effect or planned power. The sample-size illustrations based on the observed effect do not substitute for the trial's original design rationale.

The record does not specify the confidence level or construction method for the recalculated interval, and it does not explain why the sponsor-reported Fisher Exact p-value differs from the unadjusted recalculated p-value. The sponsor's testing specification beyond the named method is also not reported. These limits make it important to distinguish reported analysis from retrospective recalculation.

Missing-data handling for the primary endpoint is not reported. Although the primary population is defined as participants assigned treatment by randomization, the record does not explain whether every randomized participant contributed an endpoint measurement or whether missing measurements were excluded or imputed. The details of crossover timing and how crossover participants are represented in safety follow-up are also not reported.

Multiplicity procedures, interim analyses, and the statistical plan are not reported in the registry record. Nor are detailed methods for the secondary time-to-event analyses. The record lists secondary outcomes but does not include their results in the primary comparison. Finally, the unmasked design is recorded, but safeguards for endpoint assessment are not described. These are limits on interpretation of the registry record, not evidence that a particular procedure was or was not used.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: Randomized Phase Lll Study of Imatinib Dose Optimization vs Nilotinib in CML Patients With Suboptimal Response to Imatinib — NCT00802841.
• Registry record last updated 2015-11-16; status last verified 2015-10.
• Results first posted 2015-11-16.
• 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.

22. Record Summary

NCT00802841 is a completed randomized phase III parallel-group study of nilotinib versus imatinib dose optimization in participants with chronic myelogenous leukemia and a suboptimal response to standard-dose imatinib. The study enrolled 191 participants and is listed as unmasked.

The primary endpoint was the percentage of participants with CCyR at 6 months, based on bone marrow samples and defined as 0% Philadelphia positive chromosome metaphases. The Full Analysis Set included participants assigned treatment by randomization. Posted response percentages were 50.0% for nilotinib and 42.1% for imatinib.

The sponsor-reported Fisher Exact test did not indicate a statistically significant difference at the 0.05 threshold. The unadjusted recalculation likewise does not establish a statistically significant difference, and its confidence interval crosses no difference. The registry lists seven secondary outcomes. Original sample-size assumptions, missing-data handling, multiplicity procedures, and several analysis details are not reported in the registry record.