Study of Ruxolitinib in Colorectal Cancer Patients: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Study of Ruxolitinib in Colorectal Cancer Patients (NCT02119676), a randomized phase 2 study. The primary outcome is Overall Survival (OS).

2. Clinical Question

3. Trial Design

This was a phase 2, randomized, double-blind, parallel-group study of ruxolitinib or placebo in combination with regorafenib for people with relapsed or refractory metastatic colorectal cancer. The registry record lists Incyte Corporation as sponsor and an enrollment of 396. The study status is terminated. The registry record does not report the reason for termination.

The results distinguish Substudy 1 and Substudy 2. In each, participants receiving ruxolitinib plus regorafenib were compared with participants receiving placebo plus regorafenib. The posted overall-survival groups contained 87 and 88 participants in Substudy 1, and 110 and 111 in Substudy 2, respectively. The registry record does not report how the two substudies differed, why participants were assigned to them, or whether their populations were defined by particular characteristics. Their results should therefore be read as separate comparisons rather than as evidence that the groups were interchangeable.

The primary analysis population included all randomized subjects in both substudies. The primary outcome was overall survival, analyzed from baseline until death due to any cause, with follow-up described as up to 16 months or the data cut-off on 11 FEB 2016. Participants without a recorded death were censored at the last date known to be alive. The registry record describes Kaplan-Meier estimation of median survival and log-rank comparison between groups.

Other design details—including the original sample-size justification, treatment schedules, allocation ratio, stratification factors, and rules for stopping the study—are not reported in the registry record. The registry record also does not explain whether termination affected follow-up or the planned analyses.

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: Substudy 1; 4 participants were assigned a randomization number but were not given study drug because of clinical deterioration or withdrawal of consent. Substudy 2; 9 participants were assigned a randomization number, but weren't given study drug due to clinical deterioration, withdrawal of consent or not meeting all of the eligibility criteria.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Intent-to-treat (ITT) population included all subjects randomized in Substudy 1 and Substudy 2 of the study.

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: The intent-to-treat (ITT) population consisted of participants that were randomized in the study.

6. Primary Endpoint

The registry lists the primary outcome as Overall Survival (OS), measured over Baseline until death due to any cause; up to 16 months or data cut-off 11 FEB 2016..

Overall survival measures the time from randomization until death from any cause. It is a time-to-event outcome: each participant contributes information about how long they were followed, and participants who were still alive at their last known follow-up are censored rather than treated as having died.

The registry record reports median overall survival estimated with the Kaplan-Meier method. A median is the estimated time by which half of the analyzed population has experienced the event, subject to the observed follow-up and censoring. It is a summary of the survival-time distribution, not a statement that every participant lived for that duration.

Overall survival is used because it records a direct event—death from any cause—and does not depend on judgments about whether a particular scan or symptom represents disease progression. The registry record says the treatment groups were compared using a log-rank test and reports hazard ratios with 95% confidence intervals. The hazard ratio summarizes the relative event rate over follow-up under the analysis model; it is not the ratio of the two median survival times.

The stated observation period extended from baseline until death due to any cause, up to 16 months or the 11 FEB 2016 data cut-off. Follow-up beyond that period, and the completeness of death ascertainment, are not reported in the registry record.

7. Results Posted in the Registry

• Substudy 1: Ruxolitinib + Regorafenib vs Substudy 1: Placebo + Regorafenib: hazard ratio 1.04 (95% CI 0.73–1.49), reported p = 0.588 (as reported in the registry).
• Substudy 2: Ruxolitinib + Regorafenib vs Substudy 2: Placebo + Regorafenib: hazard ratio 0.77 (95% CI 0.48–1.23), reported p = 0.136 (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.

Percentage of Participants Achieving Disease Control

• Endpoint type: Binary: the percentage of participants with the event (counts rebuilt from the posted percentages and group sizes).
• Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
• Result:
  • Substudy 1: Ruxolitinib + Regorafenib: 40.2% (35/87) versus 34.1% (30/88) on control; risk difference 6.1 percentage points (95% CI -8.2 to 20.4), risk ratio 1.18 (0.80–1.74); unadjusted two-proportion z-test p = 0.401.
  • Substudy 2: Ruxolitinib + Regorafenib: 61.8% (68/110) versus 34.1% (30/88) on control; risk difference 27.7 percentage points (95% CI 14.3 to 41.2), risk ratio 1.81 (1.31–2.51); unadjusted two-proportion z-test p = <0.001.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Substudy 1: Ruxolitinib + Regorafenib and the control. The interval gives the range of effects that remain compatible with the data.
  • The 95% confidence interval excludes no difference, so the proportion with the event is statistically higher for Substudy 2: Ruxolitinib + Regorafenib than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

The registry record lists progression-free survival, overall response rate, duration of response, disease control, and treatment-emergent adverse events among the secondary outcomes. Numeric results for disease control are posted, while results for progression-free survival, overall response rate, duration of response, and treatment-emergent adverse events are not reported in the registry record. The overall pattern across the secondary outcomes therefore cannot be assessed comprehensively from the posted results.

The disease-control estimates point in different degrees across the two substudies: the contrast between treatment groups appears more pronounced in Substudy 2 than in Substudy 1. This is a secondary outcome and does not replace the primary overall-survival comparison. A difference in disease control does not by itself establish a difference in survival, and it does not explain the separate survival estimates.

Secondary outcomes can help describe other aspects of disease status and treatment experience, but they carry less weight when their analysis plan, multiplicity handling, or complete results are unavailable. Testing many outcomes can produce apparently notable findings by chance. The registry record does not report a multiplicity adjustment or identify a hierarchy for interpreting secondary outcomes. The disease-control comparisons are unadjusted recalculations rather than sponsor-reported inferential analyses, so they warrant particular caution.

Because results for several named outcomes are absent, silence should not be interpreted as no difference, no events, or a negative finding. Their values and analyses are not reported in the registry record.

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

• Progression Free Survival (PFS)
• Overall Response Rate (ORR)
• Duration of Response
• Percentage of Participants Achieving Disease Control
• Percentage of Participants With Treatment-emergent Adverse Events (TEAEs)

9. Statistical Interpretation

10. Statistical Methods Explained

The primary outcome is time-to-event data. Some participants may have died during follow-up, while others may have been alive at their last recorded follow-up. The latter observations are censored: their exact eventual survival time is unknown, but their observed time contributes information up to the point they were last known to be alive. The registry record says that deaths were counted through the data cut-off and that those without an observed death were censored at their last date known to be alive.

How Kaplan-Meier estimation works

Kaplan-Meier estimation uses the timing of events and censoring to estimate the proportion of participants who remain event-free over time. It accounts for the fact that participants can have different lengths of follow-up. The reported median is located where the estimated survival curve reaches the halfway point. It is a summary of the estimated curve, not an average and not a guarantee about an individual participant's outcome.

Interpretation depends on the follow-up and censoring pattern. If follow-up is limited, the median can be estimated with uncertainty, and the estimate may not describe outcomes beyond the observed period. The registry record gives the maximum stated timeframe and data cut-off, but does not report detailed follow-up distributions or the number censored in each group.

What the hazard ratio represents

The hazard ratio compares the event rates between groups over follow-up. A value of 1 represents equal estimated hazards; a value below 1 indicates a lower estimated hazard in the first group named in the comparison, and a value above 1 indicates a higher estimated hazard, under the comparison's direction. This is a relative, time-to-event summary. It is not a direct comparison of the medians and does not say how many additional participants experienced an event.

The registry record reports hazard ratios from log-rank analyses. The log-rank test compares the survival experience of groups across follow-up, taking event timing into account. The test's p-value addresses compatibility with a null hypothesis of no difference under the test. It does not tell how likely the null hypothesis is to be true.

Why confidence intervals matter

A 95% confidence interval communicates uncertainty around an estimate. In repeated use of the same method under its assumptions, intervals constructed this way would contain the underlying parameter in 95% of repetitions. For a particular trial, it is best read as a range of values reasonably compatible with the data and model. When a hazard-ratio interval includes 1, the posted estimate does not rule out equal hazards at the corresponding confidence level. The interval's breadth also indicates that materially different relative effects remain compatible with the observations.

For Substudy 2, the upper confidence bound for the placebo-plus-regorafenib median is not reported in the registry record. That missing bound does not prevent reading the posted median, but it means the interval cannot be fully described from the record.

Binary secondary outcomes and recalculation

Disease control is reported as a percentage with 95% confidence intervals. It is a binary outcome: participants are classified according to whether they achieved disease control during the stated period. The independently recalculated comparisons use unadjusted differences between the posted group proportions. Such a difference expresses the contrast on the percentage-point scale, while a confidence interval describes uncertainty in that contrast. These recalculations are not the sponsor's reported inferential analyses; the registry record lists no reported analysis for this secondary outcome.

Unadjusted comparisons do not account for any stratification, covariate adjustment, or multiplicity procedure unless one is explicitly specified. The registry record does not report a statistical plan for these secondary comparisons. General statistical practice is to interpret secondary outcomes cautiously when many outcomes or comparisons are assessed, because the chance of at least one apparently notable result can increase when multiple tests are considered without adjustment.

Over-dispersion and model choice

For count outcomes, over-dispersion means that the observed variability is greater than a simple model assumes. In general, ignoring over-dispersion can make standard errors too small and confidence intervals too narrow. The primary endpoint is time-to-event rather than a count outcome, and the registry record does not describe a count model for the primary analysis. This general issue should not be mistaken for a reported feature of this trial's analysis.

The registry record names the analysis methods for overall survival but does not provide model diagnostics, proportional-hazards assessments, covariate adjustments, sensitivity analyses, or details of how missing observations other than censoring were handled. These analytical choices and checks are therefore not reported in the registry record.

11. Confidence Intervals and Precision

The overall-survival estimates have substantial uncertainty, which is reflected in their 95% confidence intervals. For Substudy 1, the hazard-ratio interval runs from 0.73 to 1.49 around the estimate of 1.04. For Substudy 2, it runs from 0.48 to 1.23 around the estimate of 0.77. Both intervals include 1, so neither reported interval excludes equal hazards. They also include values on both sides of 1, meaning the data are compatible with differing directions of the relative hazard.

The p-values, 0.588 and 0.136, are consistent with that uncertainty: neither is evidence against the null at conventional thresholds. They should not be read as measures of effect size or as the probability that a difference is absent. The confidence intervals communicate the range of relative effects compatible with the estimates under the analysis method more directly than the p-values alone.

The median estimates offer a separate description of the time-to-event distribution. Their confidence intervals are reported for the first three arms: 3.5 to 5.4 months for Substudy 1 ruxolitinib plus regorafenib, 4.3 to 6.0 months for Substudy 1 placebo plus regorafenib, and 9.0 to 13.2 months for Substudy 2 ruxolitinib plus regorafenib. For Substudy 2 placebo plus regorafenib, the upper confidence bound is not reported in the registry record; the lower bound is 7.2 months.

Confidence intervals are not guarantees that the true effect lies within a particular realized interval, and they do not account for unreported design or analysis choices. The available estimates should be considered alongside the limited stated follow-up and the fact that the two substudies are reported separately.

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 illustrations describe hypothetical future planning based on fractions of the observed effect; they are not the original trial's design justification. For overall survival, the displayed quantities are total events needed, using the Schoenfeld approach. For the binary disease-control comparisons, the quantities are participants per arm. The distinction matters: an event target is not the same as a participant enrollment target, because not every participant necessarily experiences the event during follow-up.

The listed planning assumptions are a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. The effect to detect is set to 100%, 75%, or 50% of the observed effect. The registry record does not report the original sample-size assumptions for the trial.

The 80% power column gives the number of total events or participants per arm that the illustration estimates would be needed to have an 80% chance of detecting the stated effect, if that effect and the planning assumptions were correct. The 90% power column makes the corresponding illustration for a higher chance of detection. These are planning targets, not guarantees that a study of that size will produce a statistically significant result.

Smaller effects are harder to distinguish from random variation, so the illustrations require more events or participants when the target is reduced from the full observed effect to a smaller fraction. These estimates are not observed power.

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 = 1.040, so ln(HR) = 0.0392.
2. d = 4 × (1.96 + 0.842)² / [ln(1.040)]² = 20,409.7, rounded up to 20,410 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 <- 1.04                              # 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(2119676)
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 = 20,410 (80% power), 27,323 (90% power)
 75% of observed effect: events = 36,284 (80% power), 48,574 (90% power)
 50% of observed effect: events = 81,639 (80% power), 109,292 (90% power)
Simulated power with 20,410 events: 0.796 (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. Analyses Reported by the Sponsor

For overall survival, the registry record reports the sponsor's hazard ratios and log-rank p-values for both substudies. Those reported analyses are the basis for the primary time-to-event interpretation; no alternative recalculation is substituted for them. The reported hazard ratio is 1.04 for Substudy 1 and 0.77 for Substudy 2, with their corresponding confidence intervals and p-values described in the primary results.

The record also posts medians and confidence intervals by group. These descriptive summaries complement the hazard ratios but are not interchangeable with them: the median describes a point on the estimated survival distribution, while the hazard ratio compares relative event rates across follow-up.

For disease control, the registry record includes group percentages and confidence intervals but no reported inferential analysis. Any unadjusted contrasts derived from those summaries are separate recalculations, not sponsor-reported tests. They should not be presented as though they were part of the sponsor's stated analysis plan. The registry record does not report whether the sponsor performed other analyses that were not posted.

15. Multiplicity, Missing Data and Analysis Populations

The registry record reports two primary overall-survival comparisons, one for each substudy, and lists multiple secondary outcomes. It does not report how the statistical plan handled multiplicity across the substudies, outcomes, or repeated analyses. Without such details, the p-values should be read as results for the reported comparisons, not as proof that a family-wide error rate was controlled.

Multiplicity matters because, in general, the more statistical tests that are performed, the greater the chance that at least one will appear notable by chance alone. A prespecified testing hierarchy or adjustment can address this issue, but the registry record does not report either for these results. It also does not report whether analyses were prespecified, whether interim analyses occurred, or whether any decision rules were used.

For the primary outcome, the stated intent-to-treat population includes all subjects randomized in Substudy 1 and Substudy 2. This is an important population definition because it retains participants according to their randomized groups. However, details about analysis exclusions, protocol deviations, and the handling of incomplete data beyond censoring are not reported in the registry record.

For overall survival, participants without an observed death were censored at the last date known to be alive. Censoring is a standard way to use incomplete follow-up, but interpretation depends on the reasons for censoring and whether they are related to outcome. The registry record does not provide censoring counts, reasons, or sensitivity analyses that would clarify this issue. It also does not report the extent of loss to follow-up.

The record lists an intent-to-treat population for the primary endpoint but does not describe analysis populations for the secondary outcomes. It does not report a missing-data strategy for those outcomes, or whether participants with unavailable assessments were excluded or classified in a particular way. These omissions limit assessment of the robustness and completeness of the posted results.

16. 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: abdominal pain (8.1%), sepsis (7.1%), hepatobiliary disorders (5.9%), pneumonia (5.9%), small intestinal obstruction (4.7%).

The registry record lists serious adverse-event counts for participants at risk in each arm. In Substudy 1, 49 of 85 participants at risk in the ruxolitinib-plus-regorafenib group and 42 of 86 in the placebo-plus-regorafenib group were recorded as affected. In Substudy 2, the corresponding counts were 37 of 106 and 37 of 106. These counts describe the posted serious adverse-event information; they do not establish that a study drug caused an event.

The record also lists serious-event terms and percentages, including abdominal pain at 8.1%, sepsis at 7.1%, hepatobiliary disorders at 5.9%, pneumonia at 5.9%, small intestinal obstruction at 4.7%, and dyspnoea at 4.7%. These are listed terms, not a complete account of every adverse event or a comparison of risk between treatment groups.

Deaths in the adverse-event data are not reported in the registry record. Although treatment-emergent adverse events are listed as a secondary outcome, their outcome results are not reported in the registry record. The serious-event counts should therefore be interpreted alongside the missing information about event timing, severity definitions, relatedness, and group-specific event terms.

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

The posted primary results describe overall survival in two separate substudy comparisons. In Substudy 1, the hazard-ratio estimate is close to 1. In Substudy 2, it is below 1, but its confidence interval includes 1. Neither comparison provides a precise estimate that excludes equal hazards at the reported confidence level. The median survival summaries are descriptive and should not be used in place of the time-to-event comparison.

These findings do not demonstrate that the treatments are equivalent. A nonsignificant result does not establish equality, particularly when confidence intervals are compatible with a range of relative effects. Nor do the point estimates alone establish a difference. The posted analyses leave uncertainty about the underlying effects in both substudies.

The results also do not justify combining the two substudies or explaining why their estimates differ. The registry record does not report the basis for the separate substudies or a pooled primary analysis. Their observed differences could reflect the populations, random variation, or other factors that cannot be evaluated from the posted record.

The disease-control outcome offers additional descriptive information, but it is secondary, and its comparisons are not reported as sponsor analyses. Results for several other named secondary outcomes are not reported in the registry record. Those gaps prevent a full assessment of the relationship between survival, progression, response, and treatment experience.

Finally, the results apply to the randomized participants and follow-up described in this study. They do not establish what would occur in populations not represented in the trial, with other treatment regimens, or beyond the stated follow-up. The study status is terminated, but the registry record does not report why; that status alone does not explain the observed outcomes.

18. Why This Trial Matters Statistically

This trial illustrates why a time-to-event result should be read through its estimate, confidence interval, and analysis method together. The hazard ratios summarize relative event rates, the confidence intervals convey uncertainty, and the log-rank p-values address a null comparison. None of these quantities alone gives a complete account of the survival experience.

Reporting results separately for two substudies also highlights the importance of understanding population definitions and analysis plans. The estimates differ, but the registry record does not explain the substudies' relationship or report a combined estimate. Without that context, readers should resist treating one estimate as the definitive overall result or attributing the difference to a particular cause.

The record further demonstrates the distinction between registered outcome names and posted outcome results. Several secondary outcomes are listed, but their numerical results are absent. Missing results are not evidence of no effect; they limit the ability to assess the overall pattern and the possibility of selective reporting.

Finally, the record separates sponsor-reported survival analyses from unadjusted recalculations for disease control. Keeping those sources distinct helps prevent a re-analysis from being mistaken for the trial's own planned analysis. It also reinforces the value of clearly reported analysis plans, multiplicity procedures, missing-data handling, and complete outcome reporting.

19. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

The registry record does not explain why the study was terminated or whether termination changed the planned follow-up or analyses. It does not report the reasons for the separate substudies, their defining characteristics, or a pooled overall-survival estimate. This makes it difficult to understand why the two sets of results should be expected to align or differ.

Follow-up is described as up to 16 months or the data cut-off on 11 FEB 2016, but detailed follow-up duration, number of deaths, censoring patterns, and loss to follow-up are not reported in the registry record. The upper confidence bound for the Substudy 2 placebo-plus-regorafenib median is also not reported. These omissions constrain assessment of precision and completeness.

The registry record does not report the original sample-size justification, a full statistical analysis plan, interim analyses, stopping rules, multiplicity handling, or sensitivity analyses. It also does not explain the rationale for analysis choices beyond naming Kaplan-Meier estimation and log-rank testing for the primary endpoint. The secondary-outcome analysis populations and missing-data rules are not reported.

Several named secondary outcomes lack posted numerical results, including progression-free survival, overall response rate, duration of response, and treatment-emergent adverse events. Safety information is limited to selected serious adverse-event counts and terms; deaths in the adverse-event data and detailed group-specific event descriptions are not reported in the registry record.

These limitations do not invalidate the posted estimates, but they restrict how completely they can be evaluated and how broadly they can be interpreted.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: Study of Ruxolitinib in Colorectal Cancer Patients — NCT02119676.
• Registry record last updated 2018-02-13; status last verified 2018-01.
• Results first posted 2017-06-14.
• 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.

23. Record Summary

This phase 2 randomized, double-blind, parallel-group study compared ruxolitinib plus regorafenib with placebo plus regorafenib in relapsed or refractory metastatic colorectal cancer. The registry record lists an enrollment of 396, identifies Incyte Corporation as sponsor, and records the study status as terminated.

The primary outcome was overall survival, assessed with Kaplan-Meier estimation and compared using log-rank analyses. Results are reported separately for Substudy 1 and Substudy 2. Both hazard-ratio confidence intervals include 1, and the registry record does not provide a pooled survival estimate.

Disease-control results and selected serious adverse-event information are posted. Results for several other listed secondary outcomes and multiple details of the statistical plan are not reported in the registry record.