A Study of Nivolumab, Nivolumab Plus Ipilimumab, or Investigator's Choice Chemotherapy for the Treatment of Participants With Deficient Mismatch Repair (dMMR)/Microsatellite Instability High (MSI-H) Metastatic Colorectal Cancer (mCRC): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study of Nivolumab, Nivolumab Plus Ipilimumab, or Investigator's Choice Chemotherapy for the Treatment of Participants With Deficient Mismatch Repair (dMMR)/Microsatellite Instability High (MSI-H) Metastatic Colorectal Cancer (mCRC) (NCT04008030), a randomized phase 3 study. The primary outcome is Progression-free Survival (PFS) by Blinded Independent Review Center (BICR) - Arm B vs. Arm A All Lines Centrally Confirmed MSI-H/dMMR.

2. Clinical Question

The registry’s primary outcome concerns all lines of treatment in the centrally confirmed population, rather than only participants receiving first-line treatment. Its endpoint is progression-free survival (PFS), assessed by a blinded independent review center (BICR). The posted comparison reports a hazard ratio of 0.62, with a 95% confidence interval from 0.48 to 0.81 and p < 0.001. The registry labels the analysis as a superiority analysis.

The clinical question is therefore about the relative timing of progression or death between two randomized treatment groups. It is not a direct comparison of overall survival, objective response, quality of life or every possible outcome of treatment. The study’s broader purpose also includes other comparisons and outcomes, but the primary result addressed is the BICR-assessed PFS comparison between Arm B and Arm A in the centrally confirmed population.

3. Trial Design

This is a Phase 3 randomized, open-label, parallel-group clinical trial. The registry lists an enrollment of 839 and metastatic colorectal cancer as the condition. The study compares nivolumab monotherapy, nivolumab with ipilimumab, and investigator’s choice chemotherapy. Its stated purpose includes comparisons of clinical benefit measured through PFS, objective response rate and overall survival. The primary outcome considered in this analysis is narrower: PFS for Arm B versus Arm A in centrally confirmed MSI-H/dMMR disease.

Arm A is nivolumab monotherapy, Arm B is nivolumab plus ipilimumab, and Arm C is investigator’s choice chemotherapy. The registry lists oxaliplatin, leucovorin, fluorouracil, irinotecan, bevacizumab and cetuximab among the interventions. Participants in Arm C would be allowed to receive nivolumab plus ipilimumab if they progressed. The registry also lists a crossover group in its safety record. The specific chemotherapy choices and their use by individual participants are not reported in the registry record.

Randomization supports a comparison between assigned groups, while the open-label design means participants and treating investigators were not masked to treatment assignment. For the primary outcome, however, disease progression was assessed by a blinded independent review center. Blinded endpoint assessment is relevant because it can reduce the influence of knowledge of assigned treatment on the review of progression, although it does not make the entire trial masked.

The study began on 2019-08-05, and primary completion is listed as 2024-08-28. Its status is active, not recruiting. The primary outcome timeframe is from randomization until documented progression or death, up to approximately 60 months. The allocation ratio, randomization strata, detailed eligibility criteria, treatment schedules, follow-up completeness and the trial’s original sample-size justification are not reported in the registry record.

4. Participant Flow and Accounting

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

Pre-Treatment

Reasons for not completing this period:

Treatment

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: All randomized participants with centrally confirmed DNA mismatch repairdeficient/microsatellite instability-high metastatic colorectal cancer (dMMR/MSI-H mCRC) in Arm A and Arm B only

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 Progression-free Survival (PFS) by Blinded Independent Review Center (BICR) - Arm B vs. Arm A All Lines Centrally Confirmed MSI-H/dMMR, measured over From date of randomization to the date of first objectively documented disease progression or death due to any cause, whichever occurs first (Up to approximately 60 months).

The primary endpoint is BICR-assessed progression-free survival. It measures the time from randomization to the first objectively documented disease progression under RECIST 1.1 or death from any cause, whichever occurs first. The registry describes the population as all randomized participants with centrally confirmed dMMR/MSI-H metastatic colorectal cancer in Arm A and Arm B only. The planned timeframe extends up to approximately 60 months.

PFS is a time-to-event endpoint: each participant contributes information about how long they remain free of the defined event. Progression and death are combined in the event definition, so the endpoint does not distinguish the contribution of each event type by itself. Participants without a recorded event during their observed follow-up may be censored in a time-to-event analysis; the detailed censoring rules are not reported in the registry record.

Time-to-event analysis is useful when participants have different follow-up durations or when not everyone experiences the event during observation. It uses both the timing of events and the information available before follow-up ends. The median PFS is the estimated time by which half of the population has experienced progression or death, under the analysis method. It is a descriptive summary and does not convey the full shape of the event-time distribution.

The registry reports a hazard ratio for the treatment comparison in addition to median estimates. A hazard ratio summarizes the relative event rate over follow-up under the model and assumptions used for the analysis. The registry’s detailed model specification and assessment of the proportional-hazards assumption are not reported in the registry record.

7. Results Posted in the Registry

• Arm A - Nivolumab Monotherapy vs Arm B - Nivolumab + Ipilimumab: hazard ratio 0.62 (95% CI 0.48–0.81), reported p < 0.001 (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.

Progression-free Survival (PFS) by Blinded Independent Review Center (BICR) - Arm B vs. Arm A All Randomized Participants

• Endpoint type: Time to event: compared as a hazard ratio.
• Test: Hazard ratio as reported by the sponsor; it was not recalculated.
• Result:
  • Arm A - Nivolumab Monotherapy vs Arm B - Nivolumab + Ipilimumab: hazard ratio 0.64 (95% CI 0.52–0.79) (as reported in the registry).
• Interpretation:
  • The 95% confidence interval excludes no difference (a ratio of 1), so the hazard of the event is statistically lower for Arm A - Nivolumab Monotherapy vs Arm B - Nivolumab + Ipilimumab than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Progression-free Survival (PFS) by Blinded Independent Review Center (BICR) - 1L Randomized Participants

• Endpoint type: Time to event: compared as a hazard ratio.
• Test: Hazard ratio as reported by the sponsor; it was not recalculated.
• Result:
  • Arm B - Nivolumab + Ipilimumab vs Arm C - Investigator's Choice Chemotherapy: hazard ratio 0.32 (95% CI 0.23–0.46) (as reported in the registry).
• Interpretation:
  • The 95% confidence interval excludes no difference (a ratio of 1), so the hazard of the event is statistically lower for Arm B - Nivolumab + Ipilimumab vs Arm C - Investigator's Choice Chemotherapy than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Progression-free Survival (PFS) by Blinded Independent Review Center (BICR) by Polymerase Chain Reaction (PCR) - Arm B vs. Arm A

• Endpoint type: Time to event: compared as a hazard ratio.
• Test: Hazard ratio as reported by the sponsor; it was not recalculated.
• Result:
  • Arm A - Nivolumab Monotherapy vs Arm B - Nivolumab + Ipilimumab: hazard ratio 0.60 (95% CI 0.45–0.80) (as reported in the registry).
• Interpretation:
  • The 95% confidence interval excludes no difference (a ratio of 1), so the hazard of the event is statistically lower for Arm A - Nivolumab Monotherapy vs Arm B - Nivolumab + Ipilimumab than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

The secondary outcomes extend the primary question across different analysis populations, treatment comparisons and disease-classification methods. The registry lists PFS comparisons among all randomized participants, first-line randomized participants and PCR- or immunohistochemistry-defined groups. It also includes a comparison with investigator’s choice chemotherapy. This breadth can help characterize whether the primary pattern is reflected across related populations and analyses, but the populations and contrasts are not identical.

The posted secondary results include hazard-ratio estimates for the all-randomized comparison of Arm B with Arm A, the first-line comparison of Arm B with chemotherapy, and the PCR subgroup comparison of Arm B with Arm A. The broad pattern of the reported estimates is compatible with the primary result’s indication of different time-to-event experience across treatment groups. These results should be interpreted in their own populations and not as if they were additional estimates of the exact primary estimand. The registry’s secondary entries do not all include p-values or analysis methods.

Secondary analyses generally carry less confirmatory weight when many outcomes or subgroups are examined. Testing multiple questions increases the chance that some results will appear notable by chance, unless the testing strategy accounts for multiplicity. The registry lists many outcomes and analyses, but multiplicity adjustment, a testing hierarchy and the prespecified status of each secondary comparison are not reported in the registry record. A secondary result can therefore be informative and hypothesis-generating without having the same evidential standing as a prespecified primary analysis.

Different subgroups can also have fewer participants and fewer events, making estimates less precise. Subgroup comparisons can be especially vulnerable to chance imbalance and should not be treated as proof that treatment effects differ between subgroups unless an interaction analysis and its interpretation support that claim. The registry does not report a formal interaction analysis for these secondary groups. The secondary pattern provides context around the primary result, but it does not remove the need to distinguish populations, methods and multiplicity considerations.

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

• Progression-free Survival (PFS) by Blinded Independent Review Center (BICR) - 1L Participants
• Progression-free Survival (PFS) by Blinded Independent Review Center (BICR) - Arm B vs. Arm A All Randomized Participants
• Progression-free Survival (PFS) by Blinded Independent Review Center (BICR) - 1L Randomized Participants
• Progression-free Survival (PFS) by Blinded Independent Review Center (BICR) - Crossover and First Line Arm
• Progression-free Survival (PFS) by Blinded Independent Review Center (BICR) by Polymerase Chain Reaction (PCR) - Arm B vs. Arm A
• Progression-free Survival (PFS) by Blinded Independent Review Center (BICR) by Immunohistochemistry (IHC) - Arm B vs. Arm A
• Progression-free Survival (PFS) by Blinded Independent Review Center (BICR) by Polymerase Chain Reaction (PCR) - Arm B vs. Arm C
• Progression-free Survival (PFS) by Blinded Independent Review Center (BICR) by Immunohistochemistry (IHC) - Arm B vs. Arm C

9. Statistical Interpretation

10. Statistical Methods Explained

What does the hazard ratio summarize?

For a time-to-event outcome, the hazard is the event rate at a particular time among participants who have not yet experienced the event. A hazard ratio compares these rates between groups over follow-up, according to the chosen analysis. It is not the same as a ratio of median times, nor does it directly give the proportion of participants who experience progression or death. In general, a hazard ratio of 1 indicates no relative difference in hazards; values on either side of 1 indicate relative differences whose treatment direction depends on which arm is the reference and how the comparison is coded.

The registry reports the primary comparison as a hazard ratio of 0.62 and identifies the method as log rank. The groups are named as Arm A versus Arm B, and the result is labelled superiority. The registry does not give the model’s coding details, so those details should not be reconstructed from the arm ordering. A hazard ratio is usually interpreted as a common relative hazard across follow-up, but that interpretation can be misleading if the relative hazards change substantially over time. The proportional-hazards assessment is not reported in the registry record.

Why analyze time to progression or death?

PFS incorporates both whether and when a participant experiences progression or death. Time-to-event methods can retain partial follow-up information from participants who have not yet had an event when their observation ends. Such participants are commonly censored at their last eligible observation, rather than counted as having experienced an event. The specific censoring rules and how follow-up interruptions were handled are not reported in the registry record.

The endpoint uses BICR assessment of radiologic progression under RECIST 1.1, together with death from any cause. Independent blinded assessment is a feature of endpoint measurement; it is distinct from the trial’s open-label treatment assignment. BICR can reduce the influence of knowledge of treatment assignment on progression review, but it does not by itself resolve issues such as missing assessments, timing of scans or loss to follow-up. The registry record does not report detailed procedures for those issues.

What is the role of the log-rank method?

The log-rank method compares event-time experience between groups across follow-up, taking account of participants at risk at the times events occur. It is a standard approach for comparing survival curves. Its usual interpretation as a test of a consistent treatment difference is strongest when the relative hazards are reasonably stable over time. When effects vary over time, the log-rank result can summarize a more complex pattern and may not communicate when that pattern occurs. The registry names the log-rank method for the primary comparison but does not report further details of its implementation or diagnostic checks.

A reported p-value assesses the compatibility of the observed data with a specified null hypothesis, under the test’s assumptions. The primary p-value is < 0.001. It is not the probability that the null hypothesis is true, nor is it a measure of effect size. The hazard ratio and its confidence interval provide information about the estimated magnitude and its uncertainty; the p-value answers a different question about evidence against the null.

How should the median be read?

The median PFS is a time-point summary: it represents the point where the estimated event-free proportion reaches one half. It does not describe the entire survival curve, and two groups can have different curve shapes even if their medians are similar. A median may be unestimable or not yet reached when follow-up does not include enough events. The registry’s primary entries do not give complete median estimates for both arms: Arm A has a reported value of 39.26 months, while Arm B has no median value listed. A lower bound of 53.82 months is listed for Arm B, but no upper bound is provided. The reason for these entries is not reported in the registry record.

Confidence limits for a median describe uncertainty in its estimated location. Missing limits should not be interpreted as zero uncertainty or as evidence of a particular distribution. Estimates may have wide or one-sided intervals when event information is limited in the relevant portion of follow-up. The registry does not state why the primary median entries are incomplete.

What distinguishes the analysis populations?

The primary outcome population includes all randomized participants with centrally confirmed dMMR/MSI-H disease in Arm A and Arm B. Other posted outcomes use different populations, including all randomized participants and first-line randomized participants. Different eligibility subsets can produce different event-time profiles, so a result from one analysis population cannot automatically be substituted for another. The registry’s secondary outcome list also includes PFS assessments by PCR and immunohistochemistry. The detailed rules for classifying participants across these analyses are not reported in the registry record.

The registry identifies 17 outcome measures and 14 analyses, and lists 15 secondary outcomes as posted. Multiple analyses create opportunities for chance findings, especially when many comparisons are examined. How the trial handled multiplicity, whether outcomes were hierarchically tested, and which comparisons were confirmatory are not reported in the registry record. Those omissions matter when interpreting secondary p-values or treating several results as independent confirmation.

What assumptions affect interpretation?

All time-to-event estimates rely on adequate event ascertainment and appropriate handling of incomplete observation. The interpretation also depends on the defined endpoint, the population analyzed and the analysis method. Open-label treatment assignment can affect care and assessment processes, although BICR was used for the primary progression assessment. Randomization supports a fair comparison of assigned groups at baseline in expectation, but the registry excerpt does not provide the allocation ratio, stratification factors, analysis-set details, or procedures for deviations from assigned treatment.

These general considerations do not establish that a particular assumption was violated. They identify information needed to fully audit an analysis. The registry reports the headline result and selected endpoint details, while the statistical analysis plan, interim-analysis details and missing-data procedures are not reported in the registry record.

11. Confidence Intervals and Precision

The primary hazard-ratio estimate is 0.62, with a 95% confidence interval from 0.48 to 0.81. This interval expresses statistical uncertainty in the estimated relative hazard under the analysis method. Its endpoints are compatible with a range of relative effects; it is not a guarantee that the true value lies inside the interval, nor does it account for every possible source of uncertainty such as measurement limitations, missing data or departures from model assumptions.

The interval excludes 1, the value representing equal hazards under a standard hazard-ratio interpretation. That is consistent with the reported p-value of < 0.001 and the registry’s superiority designation. Statistical compatibility with a difference does not, by itself, determine whether the size of the difference is clinically important. The registry does not report a threshold for clinical importance for this endpoint.

The median information is less complete. Arm A has a reported median PFS of 39.26 months, with a listed lower confidence bound of 2.11 months and no upper bound. Arm B has no median value listed, a lower bound of 53.82 months, and no upper bound. These entries do not support a complete comparison of the median estimates or their precision. The reason the estimates and limits are incomplete is not reported in the registry record.

Confidence intervals are influenced by the amount and timing of event information, not just the number randomized. Participants with short or incomplete follow-up contribute differently from those followed until an event. As a general statistical point, more event information tends to improve precision, but the registry does not provide the event count for the primary comparison or enough detail to reconstruct its interval. The reported interval should therefore be treated as the available precision summary, with the endpoint and analysis-population limits in mind.

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 planning figures are illustrations based on the observed effect, not the trial’s original design justification. For the time-to-event comparisons, the figures are total event counts, not participants per arm. The planning assumptions are a two-sided significance level of 0.05, equal allocation between the two arms and a normal approximation. Each planning scenario uses the stated fraction of the observed effect. The trial’s own design assumptions and sample-size justification are not reported in the registry record.

The 80% power column describes the total number of events that the illustration estimates would be needed for an 80% chance of detecting the chosen effect under those assumptions, if that effect were present. The 90% power column uses a higher target chance of detection and therefore generally requires more event information. Power is a property of a planned design under specified assumptions; it is not the probability that the reported finding is true. These figures are not observed power.

Rows for smaller fractions of the observed effect represent more modest effects to detect. Smaller effects are harder to distinguish from random variation, so they require more events under the same assumptions. In this time-to-event setting, the relevant planning quantity is total events, not the total number randomized. Translating event targets into a recruitment target would require additional assumptions about follow-up, event rates and losses to observation; those assumptions are not reported for the original trial design.

The illustrations do not establish that the trial was designed around the observed result or that it achieved a particular power. They are sensitivity scenarios to show how the event requirement changes when the effect assumption is varied. They should be read separately from the reported hazard ratio and p-value, which describe the posted analysis rather than a prospective design calculation.

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 = 0.620, so ln(HR) = -0.4780.
2. d = 4 × (1.96 + 0.842)² / [ln(0.620)]² = 137.4, rounded up to 138 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 <- 0.62                              # 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(4008030)
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 = 138 (80% power), 184 (90% power)
 75% of observed effect: events = 245 (80% power), 327 (90% power)
 50% of observed effect: events = 550 (80% power), 736 (90% power)
Simulated power with 138 events: 0.793 (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

The sponsor-reported primary analysis gives a hazard ratio of 0.62, a 95% confidence interval from 0.48 to 0.81 and p < 0.001, using a log-rank method. The posted comparison record also carries these same headline values, and no discrepancy is flagged. There is therefore no reported disagreement between the sponsor analysis and the associated comparison estimate that needs to be resolved.

The sponsor’s analysis is the inferential result to prioritize. A recalculation from summary statistics, when available, is necessarily limited by the information contained in those summaries and may not reproduce the trial’s full participant-level analysis, censoring rules or model implementation. It should not replace the sponsor’s reported result when the registry presents that result directly.

The registry also reports median PFS entries, but the primary entries are incomplete: Arm A has a median value, whereas Arm B does not; confidence limits are also missing on the upper side. The hazard ratio and its confidence interval provide the clearest posted comparative result for the primary outcome. The registry does not report the detailed statistical analysis plan or enough information to independently reproduce the sponsor’s analysis.

15. Multiplicity, Missing Data and Analysis Populations

The registry identifies one primary PFS comparison for the centrally confirmed population, alongside numerous secondary outcomes and analyses. It lists 17 outcome measures and 14 analyses, while 15 secondary outcomes are posted. Whenever investigators examine multiple endpoints, populations or subgroups, some apparently strong results can arise by chance even if there is no true effect for those particular questions. Statistical procedures such as a prespecified testing hierarchy or adjustment for multiple comparisons can help control this risk.

The registry record does not report whether secondary hypotheses were tested in a fixed sequence, whether multiplicity adjustments were applied, or which analyses were designated confirmatory. It also does not report details of interim analyses, stopping rules or alpha allocation. These omissions prevent a full account of how the overall type I error was controlled. The primary p-value should be distinguished from secondary p-values, where available, because the latter may be less protected against the consequences of multiple testing.

The registry defines the primary population as all randomized participants with centrally confirmed dMMR/MSI-H metastatic colorectal cancer in Arm A and Arm B. The exact analysis-set rules, handling of treatment discontinuations and crossovers, and missing-data procedures are not reported in the registry record. Participants in the chemotherapy arm could receive nivolumab plus ipilimumab after progression, making treatment received after the initial comparison relevant to some later outcomes, particularly overall survival. The effect of crossover on those outcomes cannot be determined from the information.

In general, time-to-event analysis requires rules for censoring participants whose event status is not observed throughout follow-up. The registry gives the event definition and timeframe but does not report detailed censoring rules, missing assessment handling or sensitivity analyses. Nor does it report how centrally confirmed disease status was established for every analysis population. These are not evidence that the analyses were inappropriate; they are information gaps that limit independent assessment of robustness.

The distinction between randomized assignment and actual treatment received also matters in crossover settings. An analysis based on original assignment can preserve the randomized comparison, while later treatment changes may influence outcomes after progression. The registry record does not provide enough detail to characterize the analysis approach for crossover or its effect on the primary and secondary comparisons.

16. 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: malignant neoplasm progression (10.3%), abdominal pain (5.2%), intestinal obstruction (5.2%), diarrhoea (5.0%), ileus (3.5%).

The safety record lists serious adverse event counts among participants at risk in each named group. In Arm A, 161 of 351 participants at risk were recorded as affected by serious adverse events; in Arm B, the corresponding counts are 174 of 352. Arm C lists 58 of 115, and the crossover nivolumab plus ipilimumab group lists 38 of 60. These counts describe the registry’s safety entries and are not a direct comparison of risk adjusted for follow-up time or other factors.

The most frequently listed serious adverse event terms are malignant neoplasm progression (10.3%), abdominal pain (5.2%), intestinal obstruction (5.2%), diarrhoea (5.0%), ileus (3.5%) and pneumonia (3.5%). The registry also lists deaths by group: 150 in Arm A, 103 in Arm B, 62 in Arm C and 24 in the crossover group. These are reported counts and should not be assumed to represent deaths caused by treatment; the registry excerpt does not specify cause attribution for these group totals.

Safety counts require context about exposure duration, event definitions, collection windows and attribution. The registry record does not provide those details in the information. The crossover group is distinct from the randomized arms, and its count should not be treated as a randomized-arm comparison. These figures describe reported safety data; they do not alone establish comparative safety or individual risk.

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

The posted primary analysis estimates a difference in BICR-assessed PFS between nivolumab plus ipilimumab and nivolumab alone among randomized participants with centrally confirmed dMMR/MSI-H metastatic colorectal cancer. The reported hazard ratio is 0.62, its 95% confidence interval is 0.48 to 0.81, and the reported p-value is < 0.001. The registry labels the comparison as a superiority analysis. Together, these statistics indicate that the primary analysis found evidence of differing progression-or-death hazards between the named treatment groups under its analysis.

The result does not mean every participant assigned to one group experienced a better outcome than every participant assigned to the other. It does not quantify the change in an individual participant’s outcome, guarantee a particular duration without progression, or establish the effect on overall survival. A hazard ratio is a relative time-to-event measure, not a percentage-point difference or a direct difference in median months. The incomplete median entries in the primary record also limit direct interpretation of median PFS.

The result applies to the specified endpoint, centrally confirmed population and comparison. It should not automatically be extended to participants with other disease classifications, different treatment settings or other outcomes. The wider trial includes chemotherapy and crossover, and the registry lists additional populations and subgroup analyses, but these do not change the definition of the primary result.

Statistical evidence against a no-difference hypothesis is not the same as a judgment about clinical importance, treatment choice or a person’s care. The registry result alone does not resolve the balance of benefits and harms, the relevance of adverse event profiles to an individual, or the implications of treatment after progression. Such judgments require clinical context beyond the statistical comparison. No medical advice follows from these estimates.

Finally, interpretation depends on study methods and completeness of reporting. The trial is randomized and uses blinded independent review for progression assessment, but it is open label. Important details about missing data, multiplicity, analysis populations and model assumptions are not reported in the registry record. The posted result is meaningful within its defined scope, while those limits should remain visible in any broader reading.

18. Why This Trial Matters Statistically

This trial illustrates how a randomized, open-label treatment study can use a blinded independent review center for a radiologic time-to-event endpoint. The design separates two features that are sometimes conflated: participants and investigators were not masked to assignment, while the primary progression assessment was performed by a blinded review center. That distinction matters when evaluating the potential for assessment knowledge to influence endpoint measurement.

The primary analysis also demonstrates why a hazard ratio, a confidence interval and a p-value answer related but different questions. The hazard ratio describes the comparative event rate under the analysis, the confidence interval conveys uncertainty around that estimate, and the p-value assesses compatibility with a null comparison. The registry’s primary estimate is reported with all three, but the median entries are incomplete, underscoring that different summaries of the same time-to-event outcome can have different levels of completeness.

The trial includes multiple arms, populations and secondary assessments, including subgroup results by PCR and immunohistochemistry. These broaden the scientific questions but make multiplicity and analysis-population definitions important. A primary result should not be conflated with secondary findings simply because they use a related endpoint. Likewise, a planning illustration based on an observed effect is not a substitute for the original design rationale or evidence of achieved power.

Overall, the registry offers a clear headline comparison while leaving important methodological details unreported. Statistical interpretation therefore requires both attention to what the randomized comparison estimates and restraint about questions that the registry record does not resolve. This is a useful example of how effect estimates, precision, endpoint measurement, safety reporting and multiplicity all contribute different pieces to an evidence assessment.

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.

Several limitations are specific to the posted record. The primary endpoint is defined for the centrally confirmed population in Arm A and Arm B, while other posted analyses use all randomized participants, first-line participants, PCR-defined groups, immunohistochemistry-defined groups or chemotherapy comparisons. These populations and contrasts differ, so the estimates cannot be treated as one common analysis.

The primary median entries are incomplete: the combination arm has no median value listed, and confidence limits are missing for both arms on the upper side. This makes the median summaries difficult to compare or assess for precision. The registry does not explain the incomplete entries. The hazard-ratio result is more fully reported, but the detailed model specification and assessment of proportional hazards are not reported in the registry record.

The trial is open label, which may affect treatment behavior or some aspects of outcome ascertainment. BICR assessment of the primary progression endpoint is a mitigating design feature for review of progression, but details about missing scans, censoring, follow-up completeness and sensitivity analyses are not reported. The allocation ratio, stratification variables, treatment schedules and original sample-size justification are also not reported in the registry record.

There are many secondary outcomes and analyses. The registry record does not report a multiplicity plan, hierarchy of testing or which secondary analyses were confirmatory. Some secondary results lack p-values or named analysis methods. These gaps limit the strength of conclusions that can be drawn from secondary patterns. The chemotherapy arm’s permitted crossover to nivolumab plus ipilimumab is also relevant to interpretation of subsequent outcomes, but the record does not provide enough information to assess its impact.

The safety counts are not adjusted for exposure duration, and the record excerpt does not detail event collection windows or attribution rules. The death counts are not identified as treatment-related deaths. Finally, a registry result is not a complete report of all trial methods or participant-level data. These limitations define the scope of interpretation; they do not establish that any particular result is biased or invalid.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: A Study of Nivolumab, Nivolumab Plus Ipilimumab, or Investigator's Choice Chemotherapy for the Treatment of Participants With Deficient Mismatch Repair (dMMR)/Microsatellite Instability High (MSI-H) Metastatic Colorectal Cancer (mCRC) — NCT04008030.
• Registry record last updated 2026-08-20; status last verified 2026-08.
• Results first posted 2025-09-22.
• 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.

23. Record Summary

NCT04008030 is a Phase 3 randomized, open-label, parallel-group study of nivolumab, nivolumab plus ipilimumab, and investigator’s choice chemotherapy for metastatic colorectal cancer with MSI-H/dMMR status. The registry lists enrollment of 839 and an active, not recruiting status.

The primary outcome is BICR-assessed PFS in centrally confirmed participants randomized to Arm A or Arm B. The registry reports a hazard ratio of 0.62, with a 95% confidence interval from 0.48 to 0.81 and p < 0.001, using a log-rank analysis and identifying the result as superiority. The primary median entries are incomplete.

Secondary outcomes cover additional populations, treatment comparisons and biomarker-defined groups. Safety counts are posted for the randomized arms and crossover group. Important details, including the original sample-size justification, multiplicity procedures, missing-data handling and several analysis specifications, are not reported in the registry record.