A Study of Adjuvant Pembrolizumab/Vibostolimab (MK-7684A) Versus Pembrolizumab for Resected High-Risk Melanoma in Participants With High-Risk Stage II-IV Melanoma (MK-7684A-010/KEYVIBE-010): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study of Adjuvant Pembrolizumab/Vibostolimab (MK-7684A) Versus Pembrolizumab for Resected High-Risk Melanoma in Participants With High-Risk Stage II-IV Melanoma (MK-7684A-010/KEYVIBE-010) (NCT05665595), a randomized phase 3 study. The primary outcome is Recurrence-Free Survival (RFS).

2. Clinical Question

The registry describes a randomized, double-blind, active-comparator-controlled comparison. The primary hypothesis was that pembrolizumab/vibostolimab would be superior to pembrolizumab with respect to investigator-assessed recurrence-free survival. The posted result is a hazard ratio of 1.25, with a 95% confidence interval from 0.87 to 1.80. The registry record does not report a p-value or arm-specific median recurrence-free survival estimates.

The question concerns the comparison between these two adjuvant regimens in the stated high-risk resected melanoma population. It is not a comparison with no treatment or with a different cancer population. The registry record lists a total enrollment of 1594, while the primary outcome arm entries each show 701 participants. It does not explain the relationship between these counts. The displayed primary result should therefore be read as the reported relative time-to-event estimate, not as a complete account of each participant's recurrence-free survival experience.

3. Trial Design

This was a Phase 3 randomized, double-blind, parallel-group study with an active comparator. The intervention was pembrolizumab/vibostolimab, and the comparator was pembrolizumab. Both were administered by intravenous infusion on Day 1 of each cycle; the cycle length was 3 weeks, and treatment was described as continuing for up to 17 cycles, or up to approximately 1 year. The registry record lists melanoma as the condition and identifies high-risk resected Stage IIB, IIC, III, and IV melanoma in the study description.

The registry record lists enrollment as 1594. The primary outcome arm entries report 701 participants for pembrolizumab/vibostolimab and 701 for pembrolizumab. The primary efficacy final analysis population is described as all participants randomized by the primary completion data cut-off. The registry record does not explain why the enrollment figure and the arm counts in the primary outcome differ, so no reconciliation should be assumed.

The study status is terminated. The listed start date is 2023-01-19, and the primary completion date is 2024-03-06. The primary outcome timeframe is up to approximately 13 months. The record states that the protocol pre-specified final analysis for recurrence-free survival was conducted with the primary completion data cut-off. The reason for termination is not reported in the registry record.

The interventions are described, but the original sample-size rationale and the trial's own planned statistical design assumptions are not reported in the registry record. The planning illustrations associated with the posted result are separate from that original design information. Their assumptions are described with the sample-size interpretation, and they should not be mistaken for the protocol's original justification or for observed trial outcomes.

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: The protocol-specified futility analysis of the primary outcome measure was completed with a data cut-off of 06-Mar-2024 (Primary Completion Date) and served as the final analysis of the primary outcome measure. Per protocol, 192 participants enrolled after the primary completion date and were analyzed in the End of Trial analysis with a data cut-off of 24-Sep-2025 (Study Completion Date).

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Per protocol, the primary efficacy final analysis population consisted of all participants randomized by the primary completion data cut-off.

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 Recurrence-Free Survival (RFS), measured over Up to approximately 13 months.

Recurrence-free survival (RFS) is defined in the registry as the time from randomization to the first of investigator-assessed recurrence or death due to any cause. Recurrence includes local, locoregional, regional, or distant recurrence. The endpoint therefore combines recurrence and death into a time-to-event outcome: an event occurs when either qualifying event is first recorded.

In general, time-to-event endpoints retain information about how long participants remain free of the event, rather than reducing the outcome to whether an event occurred by one fixed date. A hazard ratio compares event rates over time between groups under the statistical framework used for the analysis. It is a relative measure, not a direct estimate of the number of months gained or lost by an individual participant.

The registry says investigator-assessed RFS was presented for all randomized participants and identifies the primary efficacy final analysis population as all participants randomized by the primary completion data cut-off. It reports a hazard ratio but no arm-specific median RFS estimates. The endpoint is appropriate to the stated trial question because the study aimed to compare the time to recurrence or death after resection. The registry record does not report the detailed operational rules for event confirmation, censoring, or handling of incomplete follow-up.

7. Results Posted in the Registry

• Pembrolizumab/Vibostolimab vs Pembrolizumab: hazard ratio 1.25 (95% CI 0.87–1.80) (as reported in the registry).

8. Statistical Interpretation

9. Statistical Methods Explained

The primary outcome is a time-to-event measure. For each participant, the event is the first investigator-assessed recurrence or death due to any cause after randomization. In general, time-to-event analyses use both whether an event occurred and the time at which it occurred. Participants without a recorded event by the end of their observed follow-up may contribute information up to the point when their outcome is no longer observed; the particular censoring rules used in this trial are not reported in the registry record.

What does the hazard ratio describe?

The posted estimate is a hazard ratio of 1.25 for pembrolizumab/vibostolimab versus pembrolizumab. In general, a hazard is an event rate at a particular point in follow-up among participants who have not yet experienced the event. A hazard ratio compares those rates between groups over time. A value of 1 is the reference for equal hazards; values below or above 1 indicate a relative estimate on either side of that reference. The hazard ratio is not a risk ratio, a difference in the number of events, or a comparison of median survival times.

The registry reports the hazard ratio and its 95% confidence interval, but does not identify the estimation procedure, the model specification, or whether the analysis addressed any departures from model assumptions. Those details are not reported in the registry record. No assumptions about proportional hazards, covariate adjustment, stratification, or the handling of competing events should therefore be attributed to the trial based only on this record.

How should the confidence interval be read?

In general, a confidence interval describes the uncertainty around an estimated effect under the method used to construct it. The reported 95% confidence interval is 0.87 to 1.80. Its lower endpoint is below the reference value of 1 and its upper endpoint is above it. That means the interval includes the reference value as well as relative effects on both sides. The interval is not a range containing every possible true effect, and it is not a statement about the probability that a particular value is correct.

The registry does not report a p-value alongside this result. The confidence interval and point estimate are the posted statistical results to interpret. It would be inappropriate to reconstruct an unreported test or to present a newly calculated value as the sponsor's analysis.

What is known about the analysis population?

The primary outcome description says RFS was presented for all randomized participants. The primary efficacy final analysis population is specified as all participants randomized by the primary completion data cut-off. The registry lists 701 participants in each primary outcome arm, while total enrollment is 1594. The record does not explain that difference. The primary outcome population description should be retained as written rather than replaced by the enrollment total or the safety denominators.

The record says the protocol pre-specified final analysis for this outcome was conducted with the primary completion data cut-off. The stated timeframe is up to approximately 13 months. It does not report the number of RFS events, the amount of follow-up by group, missing outcome counts, censoring details, or a sensitivity analysis. Without those details, the precision and robustness of the analysis cannot be assessed beyond the reported interval and estimate.

Why can over-dispersion matter in general?

In general, over-dispersion means that observed variability is greater than a statistical model expects. When a model does not account for that extra variation, standard errors and intervals can be too optimistic. This general issue is not a reported feature of this time-to-event analysis: the registry record does not describe an over-dispersion assessment or a model that would permit such an assessment. It is therefore context for understanding statistical methods, not a claim about this trial.

The registry identifies the outcome parameter as a median and its dispersion as a 95% confidence interval, but the posted arm-level median values are absent. The reported comparison is instead a hazard ratio with a confidence interval. These are distinct summaries: a median is a time point within a group's event-time distribution, while a hazard ratio expresses a relative comparison of event rates over time. Neither can be recovered from the other using only the posted values.

Finally, the record includes planning illustrations based on fractions of the observed effect. They use a two-sided significance level of 0.05, equal allocation, and a normal approximation. For time-to-event outcomes, the stated planning quantities are total events, not participants per arm. The trial's own original sample-size justification and design assumptions are not reported in the registry record. The planning exercise is therefore a separate illustration and not evidence about the trial's prespecified design or its achieved power.

10. Confidence Intervals and Precision

The reported hazard ratio is 1.25, and its 95% confidence interval is 0.87 to 1.80. The estimate is not very precise in the sense of identifying a narrow range of relative effects: the interval spans values below and above 1. In general, an interval's width reflects statistical uncertainty under the analysis method, while its position shows which effect values are compatible with the reported estimate and method.

Because 1 is within the interval, the result does not rule out equal hazards under the interval's framework. The interval also includes estimates below 1 and above 1, so the posted result allows for different directions of relative effect. The upper endpoint is farther from 1 than the lower endpoint, but the registry does not provide enough methodological detail to explain the interval's shape or to assess the assumptions behind it.

A confidence interval is not an individual prediction, nor does it describe the range of outcomes among participants. It does not give the probability that a treatment effect falls within the interval. In general, the confidence level describes the long-run behavior of a specified interval procedure under its assumptions. The registry does not identify the interval-construction method, so interpretation should remain tied to the reported 95% interval rather than to a more specific method.

The record does not report arm-specific median RFS values, event counts, or a p-value. Consequently, precision cannot be further characterized through those quantities from the registry record. The interval is the available measure of uncertainty for the reported hazard ratio, and it should be considered alongside the fact that the result is relative rather than an absolute difference in recurrence-free time.

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 illustrations use the observed effect as a reference and show scenarios for detecting 100% of that effect, 75% of that effect, and 50% of that effect. For each scenario, the figures are total events required for 80% or 90% power, not participants per arm. The time-to-event event counts follow the stated Schoenfeld approach.

The listed planning assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The effect to detect is the stated fraction of the observed effect. The registry record does not report the trial's original sample-size justification or original design assumptions; these planning assumptions belong to the illustration, not necessarily to the trial protocol.

The 80% power column describes the total number of events in the planning scenario that would correspond to an 80% chance of detecting the specified effect under those assumptions, if that effect were the one to detect. The 90% power column expresses the corresponding planning target for a 90% chance. These are design calculations, not observations about what happened in the trial. The figures are not observed power.

The listed total-event requirements increase as the planning effect becomes smaller: the 100% of the observed effect scenario lists 631 events for 80% power and 845 for 90% power; the 75% scenario lists 1,121 and 1,501; and the 50% scenario lists 2,523 and 3,377. Smaller effects are harder to distinguish from random variation, so more events are needed under the stated assumptions. These event totals should not be confused with the enrollment of 1594 or the arm counts reported for the primary outcome.

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

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 = 1.250, so ln(HR) = 0.2231.
2. d = 4 × (1.96 + 0.842)² / [ln(1.250)]² = 630.5, rounded up to 631 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 <- 1.25                              # 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(5665595)
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 = 631 (80% power), 845 (90% power)
 75% of observed effect: events = 1,121 (80% power), 1,501 (90% power)
 50% of observed effect: events = 2,523 (80% power), 3,377 (90% power)
Simulated power with 631 events: 0.797 (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.

13. Analyses Reported by the Sponsor

The posted comparison is the sponsor-reported hazard ratio of 1.25, with a 95% confidence interval from 0.87 to 1.80. No p-value is reported. The record does not provide enough summary statistics to independently reconstruct a hazard ratio, and no separate recalculated estimate is presented. Accordingly, the sponsor's reported estimate is the relevant result to interpret rather than an unadjusted calculation based on unavailable summary data.

The record does not flag a discrepancy between the sponsor's reported result and a recalculation. This does not establish that every aspect of the analysis has been independently verified; it means only that the posted information does not identify a conflicting recalculated result. The analysis method, model details, and interval construction are not reported in the registry record.

For readers, the practical distinction is between the sponsor's stated analysis and planning illustrations based on the observed effect. The hazard ratio and confidence interval describe the posted comparison. The planning figures explore event totals under stated assumptions and are not an alternative trial result, a validation of the sponsor's analysis, or observed power. The sponsor result should be read with its uncertainty and the limits of the reported methodological detail.

14. Multiplicity, Missing Data and Analysis Populations

The registry record lists one analysis for the primary comparison and two posted secondary outcome measures: the number of participants who experienced at least one or more adverse events, and the number who discontinued study treatment due to an adverse event. The primary endpoint is investigator-assessed recurrence-free survival. The record does not describe a multiplicity adjustment, a hierarchy for testing outcomes, or how any family of hypotheses was defined. Those elements are not reported in the registry record, so no claim about control of the overall type I error rate can be made.

The primary outcome description identifies the final analysis population and says the protocol pre-specified final analysis was conducted with the primary completion data cut-off. It does not provide details on missing RFS assessments, censoring rules, follow-up after a participant's last assessment, or sensitivity analyses for missing data. The analysis population wording should not be expanded into assumptions about how incomplete records were handled.

It is not reported whether interim analyses occurred, whether stopping rules were applied, or how the terminated status relates to the timing or conduct of the final analysis. The reason for termination is not reported in the registry record. The statement that the primary analysis was conducted with the primary completion data cut-off is the available description; it does not answer those other questions.

In general, multiplicity refers to the increased chance of false-positive findings when several statistical hypotheses are tested. Missing-data methods can also affect interpretation when outcome information is unavailable. These are general considerations, not claims that a particular multiplicity problem or missing-data bias occurred in this study. The registry does not report enough detail to assess either issue beyond the posted analysis and its stated population.

15. 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: adrenal insufficiency (4.3%), type 1 diabetes mellitus (1.6%), hypophysitis (1.3%), pneumonia (1.1%), diabetic ketoacidosis (1.0%).

The safety record reports serious adverse events affecting 255 of 797 participants in the pembrolizumab/vibostolimab group and 136 of 796 participants in the pembrolizumab group. It also reports other adverse events affecting 725 of 797 and 693 of 796 participants, respectively. These are posted counts and denominators for the safety entries; they are not the primary outcome arm counts.

The record lists adrenal insufficiency at 4.3%, type 1 diabetes mellitus at 1.6%, hypophysitis at 1.3%, pneumonia at 1.1%, diabetic ketoacidosis at 1.0%, and myocarditis at 0.9% among the top serious adverse event terms. The registry excerpt does not specify the group denominator or whether these term percentages are pooled or group-specific, so they should not be assigned to one treatment group.

Deaths are listed as 23 for pembrolizumab/vibostolimab and 24 for pembrolizumab. The record does not state in these figures whether deaths were attributed to treatment, to an adverse event, or to another cause. The adverse event counts describe reported safety events and do not by themselves establish causation or provide a complete comparison of risk.

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

The registry reports a randomized, double-blind comparison of pembrolizumab/vibostolimab with pembrolizumab in high-risk resected melanoma. The primary posted result is a hazard ratio of 1.25, with a 95% confidence interval of 0.87 to 1.80, for investigator-assessed recurrence-free survival. The interval includes 1 and spans values on both sides of that reference. The estimate therefore has substantial uncertainty about the relative hazard represented by the result.

This result does not establish that either regimen is superior, equivalent, or inferior. The point estimate alone is not enough to make such a claim, and the interval does not exclude equal hazards. No p-value or arm-specific median RFS estimate is reported. The result also does not show absolute recurrence-free survival probabilities, event counts by group, or a difference in months. Those outcomes cannot be inferred from the hazard ratio alone.

The endpoint combines investigator-assessed recurrence at specified anatomical categories with death from any cause. Thus, it does not separately report the relative results for recurrence and death. Nor do the posted death counts in the safety record provide a substitute for the primary endpoint analysis: those counts are listed under safety, and the record does not describe their causes or how they map to RFS events.

The status is terminated, but the reason for termination is not reported in the registry record. The primary analysis is described as conducted at the primary completion data cut-off, with a timeframe up to approximately 13 months. The record does not report later follow-up or additional results. The findings should therefore be understood as the posted analysis at that cut-off and within the limits of the information available in the registry record, not as a complete description of longer-term outcomes.

The safety counts and serious adverse event terms add descriptive information but do not settle the efficacy comparison. The registry record does not state that the reported adverse events were caused by study treatment. No medical decision or individual prognosis can be derived from the group-level estimates alone.

17. Why This Trial Matters Statistically

Statistically, this trial illustrates the difference between a point estimate and the uncertainty around it. The hazard ratio gives a concise relative comparison for a time-to-event outcome, while its 95% confidence interval communicates that the plausible relative effects remain spread across values below and above the reference value of 1. Reading both together is more informative than treating the point estimate as a conclusion.

It also illustrates why the endpoint definition matters. Recurrence-free survival is not simply recurrence: the registry definition includes death from any cause as well as investigator-assessed recurrence. The reported hazard ratio consequently refers to the combined time-to-event endpoint, not to either component separately. The record does not post component-specific estimates or arm-specific median RFS values.

The planning illustrations offer a separate statistical lesson. They show how the total number of events needed for a planning target rises as the effect to detect becomes smaller, under the stated assumptions. These calculations can help readers understand the information requirements of a time-to-event comparison, but they are not the original trial design rationale and are not observed power.

Finally, the registry contains distinct counts for enrollment, the primary outcome arms, and safety denominators. Keeping those populations separate is important when interpreting the result. The available account also leaves key analysis details unreported, including the model specification and missing-data handling. This makes disciplined description especially important: the posted estimate and interval can be interpreted, but unsupported conclusions about its underlying analysis should not be added.

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.

The registry record does not report the reason the study was terminated. It lists a primary completion date and states that the protocol pre-specified final analysis was conducted with the primary completion data cut-off, but does not describe the circumstances of termination or whether any further follow-up was available.

The reported primary outcome comparison contains a hazard ratio and confidence interval, but no p-value, arm-specific median RFS estimates, or RFS event counts. The analysis method, model specification, interval construction, censoring rules, missing-data procedures, and sensitivity analyses are not reported in the registry record. These omissions limit assessment of how the estimate was produced and how robust it is to analysis choices.

The record lists total enrollment of 1594, while the primary outcome arm entries list 701 participants in each group and the safety denominators are 797 and 796. It does not explain the differences among these figures. They should be retained as distinct reported quantities rather than combined or treated as interchangeable.

The primary timeframe is up to approximately 13 months, and the record does not report later follow-up. The result therefore should not be presented as a longer-term estimate. The safety record provides counts and selected serious adverse event terms, but does not specify the attribution of deaths or, for the listed term percentages, the applicable group denominator. The record also does not describe multiplicity control or interim analyses. These are limits of the reported registry information, not evidence that a particular problem occurred.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: A Study of Adjuvant Pembrolizumab/Vibostolimab (MK-7684A) Versus Pembrolizumab for Resected High-Risk Melanoma in Participants With High-Risk Stage II-IV Melanoma (MK-7684A-010/KEYVIBE-010) — NCT05665595.
• Registry record last updated 2026-09-10; status last verified 2026-08.
• Results first posted 2025-04-06.
• 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

NCT05665595 is a Phase 3 randomized, double-blind, parallel-group study of adjuvant pembrolizumab/vibostolimab versus pembrolizumab in high-risk resected melanoma. The registry lists enrollment of 1594, status as terminated, a start date of 2023-01-19, and a primary completion date of 2024-03-06.

The primary outcome is investigator-assessed recurrence-free survival, defined as time from randomization to recurrence or death from any cause. The posted comparison is a hazard ratio of 1.25 with a 95% confidence interval of 0.87 to 1.80. The record does not report a p-value or median RFS estimates by group. It also posts safety counts and planning illustrations, which describe different aspects of the study and should not be confused with the primary efficacy result.