Phase 3 Trial in Subjects With Metastatic Melanoma Comparing 3 mg/kg Ipilimumab Versus 10 mg/kg Ipilimumab: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Phase 3 Trial in Subjects With Metastatic Melanoma Comparing 3 mg/kg Ipilimumab Versus 10 mg/kg Ipilimumab (NCT01515189), a randomized phase 3 study. The primary outcome is Overall Survival (OS).

2. Clinical Question

The comparison concerns dose, not ipilimumab versus no treatment or an alternative drug. The registry record describes a randomized, double-blind Phase 3 study and lists melanoma as the condition. The stated population includes people with disease that was unresectable or metastatic and permits either previously treated or untreated disease. The record does not report further eligibility details, such as the specific treatment history required or criteria governing re-induction.

Overall survival is a time-to-event outcome: it uses the timing of deaths as well as whether a participant had died by the analysis cutoff. This is different from comparing only the number of deaths at a single time. The reported median describes the time by which the estimated survival experience reaches its midpoint; the hazard ratio summarizes the relative event rate over follow-up under the reported analysis. These measures address the same broad survival question but describe it in different ways.

3. Trial Design

This was a randomized, quadruple-masked, parallel-group Phase 3 study. The official title describes it as a randomized double-blind study, while the registry design fields list quadruple masking. Both descriptions indicate masking, but the registry record does not identify which participant, care, or assessment roles were masked. It also does not report the allocation sequence method or whether allocation was concealed.

The study compared two ipilimumab doses: 10 mg/kg and 3 mg/kg. In both arms, the described schedule was intravenous administration once every 3 weeks for 4 doses. Each arm description states that re-induction was an option and that treatment continued until disease progression or unacceptable toxicity. The registry record does not give further details on the circumstances or criteria for re-induction.

The listed enrollment is 831. For the posted overall-survival outcome, the registry reports groups of 365 at 10 mg/kg and 362 at 3 mg/kg. The primary outcome population is described as all randomized participants. The record does not explain why the enrollment figure and the outcome-group counts differ, so those figures should be read as separately reported quantities rather than reconciled by inference.

Overall survival was the primary outcome, with the timeframe described as approximately 48 months, assessed up to February 2016. The primary outcome definition starts follow-up at randomization and ends it at death from any cause; participants not known to have died were censored at the last known alive date. Listed secondary outcomes include progression-free survival, response and disease-control measures, duration measures, overall-survival rates, and survival among participants with brain metastases at baseline. Only some secondary outcome results are included in the posted results described in this analysis.

The record lists a study start of 2012-02-17, primary completion of 2016-02-06, and status as completed. It does not report the trial's original sample-size justification, the planned number of events, the detailed statistical analysis plan, or whether interim analyses were conducted. Those design details cannot be inferred from the reported results.

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: 831 participants were enrolled; 727 were randomized to a treatment group; 726 received at least one dose of study treatment. Of the 105 participants not treated, 81 no longer met study criteria, 11 withdrew consent, 4 suffered an Adverse Event, 4 died, and 5 were not treated due to investigator decision or other reasons.

Induction

Reasons for not completing this period:

First Re-Induction Phase

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: All randomized participants

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: All randomized participants

6. Primary Endpoint

The registry lists the primary outcome as Overall Survival (OS), measured over Approximately 48 months (assessed up to February 2016).

Overall survival is the time from randomization until death from any cause. A participant who had not died was censored at the last date they were known to be alive. Censoring allows the analysis to use available follow-up without treating a participant's last observation as if it were a death. The registry record describes the primary population as all randomized participants and gives an assessment timeframe of approximately 48 months, up to February 2016.

The median overall survival is a time-to-event summary. It identifies the point on the estimated survival distribution at which half of the modeled survival experience lies on either side. It is not the average survival time, nor does it say when every participant died. In this trial the median is reported separately for each dose group, with a 95% confidence interval.

Overall survival is used because it captures death from any cause and does not require assigning the cause of death to the melanoma or to treatment. That makes the endpoint clinically direct as a measure of survival, while also meaning that deaths from all causes contribute. The registry record states that the median and its two-sided 95% confidence interval were calculated using the method of Brookmeyer and Crowley.

The endpoint is distinct from progression-free survival, which is separately reported using mWHO criteria. Overall survival addresses how long participants lived during follow-up; progression-free survival and response outcomes describe different aspects of disease status. The registry record does not provide enough detail to infer how those secondary measures affected the primary survival analysis.

7. Results Posted in the Registry

• Ipilimumab (10 mg/kg) vs Ipilimumab (3 mg/kg): hazard ratio 0.84 (95% CI 0.70–0.99), reported p = 0.040 (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 mWHO Criteria

• Endpoint type: Time to event: compared as a hazard ratio.
• Test: Hazard ratio as reported by the sponsor (Log Rank); it was not recalculated.
• Result:
  • Ipilimumab (10 mg/kg) vs Ipilimumab (3 mg/kg): hazard ratio 0.89 (95% CI 0.76–1.04), reported p = 0.155 (as reported in the registry).
• Interpretation:
  • The 95% confidence interval includes no difference (a ratio of 1), so these data do not show a statistically significant difference between Ipilimumab (10 mg/kg) vs Ipilimumab (3 mg/kg) and the control. The interval gives the range of effects that remain compatible with the data.

Best Overall Response Rate (BORR) by mWHO Criteria

• 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:
  • Ipilimumab (3 mg/kg): 12.2% (44/362) versus 15.3% (56/365) on control; risk difference -3.2 percentage points (95% CI -8.2 to 1.8), risk ratio 0.79 (0.55–1.14); unadjusted two-proportion z-test p = 0.212.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Ipilimumab (3 mg/kg) and the control. The interval gives the range of effects that remain compatible with the data.

Disease Control Rate (DCR) by mWHO Criteria

• 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:
  • Ipilimumab (3 mg/kg): 27.9% (101/362) versus 31.5% (115/365) on control; risk difference -3.6 percentage points (95% CI -10.2 to 3.0), risk ratio 0.89 (0.71–1.11); unadjusted two-proportion z-test p = 0.287.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Ipilimumab (3 mg/kg) and the control. The interval gives the range of effects that remain compatible with the data.

The posted secondary outcomes span time to progression or death, response, and disease control. Taken together, their point estimates generally lean toward the 10 mg/kg arm, but the amount and type of evidence vary. The progression-free survival comparison is less conclusive than the primary overall-survival result: its hazard-ratio confidence interval includes 1 and its reported p-value is 0.155. The response and disease-control percentages are higher in the 10 mg/kg group, but their registry-record comparisons are unadjusted and no sponsor-reported inferential analyses are posted for them.

These secondary outcomes are related to the primary question because they describe disease status and event experience during follow-up, but they are not interchangeable with overall survival. A response measure concerns meeting a response criterion; disease control concerns a separate criterion; progression-free survival is a time-to-event measure. Neither a favorable point estimate for a secondary endpoint nor a less conclusive result should be used to replace the primary survival analysis.

The registry lists additional secondary outcomes, including duration of response, duration of stable disease, overall-survival rate, and overall survival among participants with brain metastases at baseline. The results described in the posted data do not include estimates for all of these listed outcomes. Their absence from the reported results cannot be treated as evidence for or against a difference.

Multiplicity matters because examining several outcomes creates more opportunities for a result to appear noteworthy by chance. The registry record does not report whether a hierarchy, adjustment, or other multiplicity strategy was planned. Without that information, secondary p-values and unadjusted calculations carry less weight than the prespecified primary analysis. They are useful for describing the broader pattern and identifying questions for further study, but should not be treated as independent confirmation of the primary result.

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

• Progression Free Survival (PFS) by mWHO Criteria
• Best Overall Response Rate (BORR) by mWHO Criteria
• Disease Control Rate (DCR) by mWHO Criteria
• Duration of Response (DOR) by mWHO Criteria
• Duration of Stable Disease by mWHO Criteria
• Rate of Overall Survival
• Overall Survival of Participants With Brain Metastases at Baseline

9. Statistical Interpretation

10. Statistical Methods Explained

The primary outcome is time-to-event data. Each participant contributes information from randomization until death or, if death was not observed, until the last known alive date. A censored observation still contributes follow-up information up to that date; it does not mean that the participant was known to survive beyond it. The registry record reports medians and 95% confidence intervals and identifies Brookmeyer and Crowley as the method used for those median intervals.

What does the median describe?

A median survival time is the time at which the estimated survival probability reaches its midpoint. It is a summary of the survival distribution, not a mean and not a statement about the outcome of each participant. Because follow-up and censoring can differ among participants, the median is estimated from the observed time-to-event information rather than by simply averaging recorded times. The posted results give a median and an interval for each dose group.

What does the hazard ratio compare?

The hazard ratio summarizes the relative event hazard between groups over follow-up. In the primary analysis, death is the event, and the comparison is 10 mg/kg versus 3 mg/kg. A hazard ratio below 1 indicates that the estimated hazard is lower in the first-named group under that comparison; a value of 1 represents no relative difference in hazard. This is a relative, time-to-event measure, not a difference in median months and not a direct probability that an individual will survive.

The registry reports a log-rank method for the hazard-ratio comparison. The log-rank test compares the event experience across groups over follow-up while taking the timing of events and censoring into account. The posted primary result includes the hazard ratio, its 95% confidence interval, and a p-value. The registry does not report additional details about model specification, covariate adjustment, or assumptions used in estimating the hazard ratio; those details should not be presumed.

How should a confidence interval be read?

A confidence interval conveys the statistical uncertainty around an estimate. In general, a 95% confidence interval is generated by a procedure that, under its assumptions, would cover the underlying parameter in 95% of repeated applications. It is not a range that guarantees the true value lies inside for this particular study, nor is it a range of individual participant outcomes. For the hazard ratio, the interval communicates uncertainty about the relative hazard; for the median, it communicates uncertainty about the estimated median time.

The primary hazard-ratio interval is 0.70 to 0.99. Because it lies below 1, it is compatible with a lower estimated hazard for the 10 mg/kg group in the reported comparison, while its closeness to 1 indicates that a small relative difference remains compatible with the data. The p-value of 0.040 is a measure of how unusual data at least as incompatible with the no-difference hypothesis would be under that hypothesis and the test assumptions. It is not the probability that the null hypothesis is true, and it is not a measure of clinical importance.

How do the secondary measures differ?

Progression-free survival is another time-to-event outcome, reported using mWHO criteria. It concerns the time until an event defined for that outcome, rather than death alone. The registry gives group medians and a log-rank hazard-ratio comparison. The record does not provide expanded detail on the criteria's implementation or on adjudication procedures.

Best overall response rate and disease control rate are binary summaries: they report the percentage meeting the respective outcome definition during the stated assessment period. Such measures reduce the outcome to whether the criterion was met, rather than describing how long it took to occur. The registry presents confidence intervals for the percentages. It also contains unadjusted risk-difference calculations for these endpoints, but does not post sponsor-reported inferential analyses for them. A risk difference expresses the contrast in proportions between groups; its confidence interval describes uncertainty in that contrast.

Time-to-event outcomes can be affected by censoring and follow-up duration, whereas binary outcomes depend on the defined assessment window and how participants with incomplete assessment are handled. The registry record does not report the detailed missing-data procedures for the binary outcomes. It also does not report a multiplicity adjustment for the several secondary endpoints. These limitations matter when interpreting results across multiple outcomes, particularly when a secondary calculation is not identified as a sponsor-reported analysis.

General statistical context: if event rates vary substantially over time, a single hazard ratio may not describe every part of follow-up equally well. The registry record does not report whether proportional-hazards diagnostics or other checks were performed. No conclusion about those checks can be drawn from the posted results alone.

11. Confidence Intervals and Precision

The primary median estimates are 15.7 months for 10 mg/kg and 11.53 months for 3 mg/kg. Their respective 95% confidence intervals are 11.63 to 17.84 months and 9.86 to 13.27 months. Each interval reflects uncertainty in its own group's estimated median. The intervals overlap, but overlap or non-overlap of separate intervals is not by itself the reported test of the between-group comparison. The registry's reported comparison is the hazard ratio from the log-rank analysis.

For that comparison, the hazard ratio is 0.84 and its 95% confidence interval is 0.70 to 0.99. The interval is relatively close to the null value of 1 at its upper end and excludes 1 according to the reported rounded limits. This supports a direction favoring the 10 mg/kg group in the primary analysis, while leaving meaningful uncertainty about the magnitude of the relative hazard difference. The reported p-value is 0.040.

Precision concerns how tightly an estimate is determined by the data, not whether the estimate is important or whether it applies to every patient. A narrower interval generally indicates more precise estimation than a wider interval under comparable conditions. The interval for the primary hazard ratio still spans a range of possible relative effects, so it should not be treated as an exact effect size. The p-value does not replace the interval: it does not show the range of effect sizes compatible with the data.

The progression-free survival hazard ratio is 0.89, with a 95% confidence interval of 0.76 to 1.04. That interval includes 1, and its reported p-value is 0.155. For the response outcomes, confidence intervals around the group percentages also convey uncertainty; the registry's unadjusted risk-difference intervals are wide enough to include no difference. The registry record does not report sponsor inferential tests for those binary secondary outcomes.

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 illustrations are based on the observed effect, not on the original design assumptions for this trial. The registry record does not report the trial's original sample-size justification. These illustrations use a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. They consider detecting the stated fraction of the observed effect: 100%, 75%, or 50%.

For overall survival and progression-free survival, the planning figures are total event counts, using the Schoenfeld method. They are not participants per arm. For the binary response outcomes, the figures are participants per arm. The unit differs because the time-to-event calculations are expressed in events, while the binary-outcome calculations are expressed in participants assigned to each arm.

The 80% power column describes the planning number of events or participants under which the specified effect would be detected by the stated test in 80% of repeated studies, assuming the planning conditions hold. The 90% power column describes the corresponding target for a greater probability of detection under those same conditions. These are design illustrations, not guarantees about a future study. They are also not observed power.

The figures increase when the target effect is reduced from 100% to 75% or 50% of the observed effect. This follows the general principle that smaller differences are harder to distinguish from random variation and therefore require more information. For time-to-event outcomes, the needed information is represented by event counts; for binary outcomes, it is represented by participants per arm. The calculations should not be interpreted as the sample size this trial was designed to use, nor as evidence that its achieved enrollment was adequate or inadequate.

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 = 0.840, so ln(HR) = -0.1744.
2. d = 4 × (1.96 + 0.842)² / [ln(0.840)]² = 1,032.8, rounded up to 1,033 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 <- 0.84                              # 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(1515189)
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 = 1,033 (80% power), 1,383 (90% power)
 75% of observed effect: events = 1,837 (80% power), 2,458 (90% power)
 50% of observed effect: events = 4,132 (80% power), 5,531 (90% power)
Simulated power with 1,033 events: 0.809 (20000 replicates)

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

14. Group-Sequential Design Analysis

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

Under these assumptions the design needs about 1,738 participants in total, with the boundaries below.

Fixed design versus group-sequential alternatives

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

Sensitivity to a smaller true effect

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

For context, this trial enrolled 831 participants.

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

R code and output

suppressWarnings(suppressMessages({ library(gsDesign); library(rpact) }))
med_control <- 11.53          # control-arm median, months (converted from the posted unit)
hr          <- 0.84          # hazard ratio used for planning
lambdaC     <- log(2) / med_control
eta         <- 0.01               # dropout hazard per month (assumption)
acc_months  <- 24                 # accrual duration (assumption)
study_months<- 60                 # total study duration (assumption)

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

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

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

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

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

Method: LachinFoulkes 
     Analysis               Value Efficacy
    IA 1: 33%                   Z   3.7103
      N: 1324         p (1-sided)   0.0001
  Events: 466        ~HR at bound   0.7091
    Month: 18    P(Cross) if HR=1   0.0001
              P(Cross) if HR=0.84   0.0338
    IA 2: 67%                   Z   2.5114
      N: 1738         p (1-sided)   0.0060
  Events: 932        ~HR at bound   0.8483
    Month: 29    P(Cross) if HR=1   0.0060
              P(Cross) if HR=0.84   0.5603
        Final                   Z   1.9930
      N: 1738         p (1-sided)   0.0231
 Events: 1398        ~HR at bound   0.8989
    Month: 60    P(Cross) if HR=1   0.0250
              P(Cross) if HR=0.84   0.9000



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

15. Analyses Reported by the Sponsor

For overall survival, the registry reports a sponsor analysis using the log-rank method: hazard ratio 0.84, 95% confidence interval 0.70 to 0.99, and p = 0.040. The posted summary statistics give the same hazard-ratio result and interval at their stated precision. The registry record indicates no discrepancy between the sponsor-reported result and the posted comparison. The sponsor's own analysis therefore remains the primary inferential result to interpret.

The median survival values and their intervals provide complementary descriptive summaries, but they are not a substitute for the sponsor's between-group test. A recalculated comparison from summary statistics, when available, is an unadjusted re-calculation and may not reproduce the sponsor's complete analysis. It should not supersede a reported analysis. For the primary result in this record, the reported hazard ratio and log-rank p-value are the appropriate inferential figures.

The sponsor also reports a log-rank analysis for progression-free survival, with hazard ratio 0.89, 95% confidence interval 0.76 to 1.04, and p = 0.155. The binary secondary outcomes do not have sponsor-reported inferential analyses in the registry record. Any unadjusted secondary calculations should therefore be kept distinct from the sponsor's reported analyses and interpreted more cautiously.

16. Multiplicity, Missing Data and Analysis Populations

The registry identifies overall survival as the primary outcome and lists several secondary outcomes. It reports results for progression-free survival, best overall response rate, and disease control rate, among other listed outcomes. The record does not report a multiplicity strategy, such as an adjustment of significance thresholds or a prespecified testing hierarchy. With multiple outcomes, interpreting each result as though it were the sole planned test can overstate the strength of evidence. The primary result should therefore be distinguished from exploratory or unadjusted secondary comparisons.

The primary endpoint definition states that participants who had not died were censored at their last known alive date. This describes how survival follow-up was treated for those participants, but it does not give the amount of censoring, reasons for loss to follow-up, or the completeness of follow-up by arm. The registry record does not report those details. It also does not provide detailed missing-data rules for response and disease-control assessments, so the effect of unobserved assessments on those outcomes cannot be evaluated from the record.

The primary population is described as all randomized participants. The record does not give further detail on analysis-population conventions, such as whether all randomized participants were included in every secondary analysis or how any protocol deviations were handled. The outcome-group counts should not be assumed to resolve those questions.

The registry record does not report interim analyses, stopping rules, the trial's original statistical analysis plan, or the original sample-size justification. It also does not report covariate adjustment details or diagnostic checks for the time-to-event analyses. These omissions do not invalidate the posted estimates, but they limit independent assessment of the planned analysis, missing-data handling, and the context for interpreting secondary results.

17. 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: malignant neoplasm progression (16.6%), diarrhoea (11.5%), colitis (9.1%), hypophysitis (4.4%), general physical health deterioration (3.3%).

The registry reports serious adverse events affecting 245 participants at risk of 364 in the 10 mg/kg group and 194 participants at risk of 362 in the 3 mg/kg group. It separately reports other adverse events affecting 319 participants at risk of 364 and 302 participants at risk of 362, respectively. These are counts of participants with events and counts at risk as recorded in the safety results; they should not be interpreted as the total number of events experienced.

The registry lists malignant neoplasm progression at 16.6%, diarrhoea at 11.5%, colitis at 9.1%, hypophysitis at 4.4%, general physical health deterioration at 3.3%, and pyrexia at 2.5% among the top serious-event terms. The posted list does not assign these percentages to an individual arm, so an arm-specific interpretation should not be inferred from it. Death counts are not reported in the safety record.

These figures describe reported safety events and do not by themselves establish that a particular event was caused by treatment. The registry record does not provide detailed event definitions, exposure-adjusted rates, severity grades, or arm-specific breakdowns for the listed terms. The safety counts also do not summarize the full balance of risks and benefits for an individual participant.

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

The primary result is an estimated difference in overall survival between two ipilimumab dose groups. The 10 mg/kg group had a reported median overall survival of 15.7 months, compared with 11.53 months for the 3 mg/kg group. The sponsor-reported hazard ratio was 0.84, with a 95% confidence interval of 0.70 to 0.99 and p = 0.040. These figures describe the observed trial comparison and its statistical uncertainty.

The result does not mean that every participant assigned to 10 mg/kg lived longer than every participant assigned to 3 mg/kg. Group medians do not predict an individual's outcome, and a hazard ratio is not a direct measure of an individual's probability of survival. The result also does not establish the exact size of an effect, since the confidence interval includes a range of relative hazard estimates.

The comparison is between two doses of ipilimumab in the study population. It does not establish how either dose compares with no treatment, another medicine, or a different treatment strategy. Nor does the primary survival result alone settle the balance of survival outcomes and adverse events. Safety results include serious and other adverse-event counts, but the registry does not provide all details needed to compare harms fully or to attribute particular events to treatment.

The secondary outcomes should be interpreted in context. The progression-free survival interval includes no relative difference, while response and disease-control results are reported without sponsor inferential analyses. The record does not describe a multiplicity strategy, so those secondary results are less definitive than the primary analysis. The posted evidence does not answer every question about durability, subgroup effects, or outcomes for participants with particular disease features.

These are population-level results from a randomized, masked study, not personal medical advice. Decisions about treatment require clinical context and assessment beyond the registry statistics.

19. Why This Trial Matters Statistically

This trial is statistically informative because it compares two doses of the same intervention in a randomized, masked, parallel-group design. That structure focuses the central contrast on dose while reducing some sources of systematic imbalance that can affect nonrandomized comparisons. The registry describes the primary population as all randomized participants and uses overall survival, a time-to-event outcome based on death from any cause.

The results also illustrate why a single trial outcome should be read through more than one statistic. The median survival values describe time in each group, while the hazard ratio summarizes the comparative event experience over follow-up. The confidence interval communicates uncertainty around the comparative estimate, and the p-value addresses compatibility with a no-difference hypothesis under the reported test. These quantities answer related but distinct questions.

The secondary results show another important statistical principle: point estimates alone do not determine the strength of evidence. The pattern of response and disease-control percentages leans in the same direction as the primary result, but those outcomes lack sponsor-reported inferential analyses. The progression-free survival interval includes 1. Without a reported multiplicity strategy, the secondary findings should be considered alongside the primary result rather than counted as multiple independent confirmations.

Finally, the registry's planning illustrations distinguish event targets from participant targets and show how smaller targeted effects require more information. They are planning scenarios based on the observed effect, not a reconstruction of the trial's original design. Keeping this distinction clear prevents planning calculations from being mistaken for evidence about the trial's achieved power or design rationale.

20. Limitations

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

The registry record does not report the original sample-size justification or the detailed statistical analysis plan. Although it gives planning illustrations based on the observed effect, those calculations do not show what assumptions informed the trial's original design. The record also does not report interim analyses or stopping rules.

The record identifies the primary population as all randomized participants, but it does not explain the difference between the listed enrollment of 831 and the outcome-group counts of 365 and 362. Those figures should be understood as reported, without inferring a reason for the difference. Further participant-flow details are not reported in the registry record.

For overall survival, censoring at the last known alive date is described, but the registry does not give the amount or pattern of censoring, follow-up completeness, or the reasons participants were last known alive at different dates. The record also does not report model checks or enough detail to assess whether the hazard ratio adequately summarizes the changing event experience over time.

The secondary outcomes are limited by the absence of a reported multiplicity strategy and, for response and disease control, the absence of sponsor-reported inferential analyses. Detailed handling of missing response assessments is not reported. Although safety counts are posted, arm-specific percentages for the listed serious-event terms, exposure-adjusted rates, and deaths are not reported in the safety results. These limitations constrain detailed comparisons of safety and secondary outcomes.

The results apply to the study's described population and comparison. The registry record does not report subgroup estimates for the overall primary result or enough detail to assess how the findings vary across patient characteristics. Generalization beyond the studied population should therefore be cautious.

21. Related Tutorials

22. Related Statistical Calculators

23. Sources

• ClinicalTrials.gov: Phase 3 Trial in Subjects With Metastatic Melanoma Comparing 3 mg/kg Ipilimumab Versus 10 mg/kg Ipilimumab — NCT01515189.
• Registry record last updated 2019-07-31; status last verified 2019-07.
• Results first posted 2017-03-24.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

24. Record Summary

The registry record describes a completed randomized, quadruple-masked, parallel-group Phase 3 study comparing ipilimumab 10 mg/kg with 3 mg/kg in people with previously treated or untreated unresectable or metastatic melanoma. The primary outcome was overall survival, defined from randomization to death from any cause, with censoring at the last known alive date for participants who had not died.

The sponsor-reported primary comparison used the log-rank method and gave a hazard ratio of 0.84, a 95% confidence interval of 0.70 to 0.99, and p = 0.040. Median overall survival was 15.7 months in the 10 mg/kg group and 11.53 months in the 3 mg/kg group. Secondary results include progression-free survival, response, and disease control; the registry record does not report sponsor inferential analyses for the binary secondary outcomes or a multiplicity strategy.

The record lists enrollment of 831 and reports separate outcome-group counts. Its sample-size illustrations are based on the observed effect and are not the original design justification. Safety results include counts of participants with serious and other adverse events, but detailed arm-specific breakdowns for the listed serious-event terms are not reported.