HD IL-2 + Ipilimumab in Patients With Metastatic Melanoma: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for HD IL-2 + Ipilimumab in Patients With Metastatic Melanoma (NCT01856023), a randomized phase 4 study. The primary outcome is Estimated One-year OS in the Evaluable Population in Each Treatment Arm Separately.

2. Clinical Question

This was an open-label, randomized, multi-center study. Randomization supports a comparison of treatment sequences, but the primary endpoint is described as an estimate for each arm separately rather than a direct estimate of the difference between arms. The posted log-rank test is the reported inferential comparison of survival over time. The result should therefore be read with both features in view: the arm-specific one-year estimates, and the test comparing the time-to-event data.

3. Trial Design

The registry describes a phase IV, open-label, randomized, multi-center study of treatment sequence in metastatic melanoma. Participants were assigned to one of two arms: high-dose interleukin-2 followed by ipilimumab, or ipilimumab followed by high-dose interleukin-2. The planned drug courses and cycles differed in order, not in the listed components. Masking was none, so the study was unblinded.

Enrollment was 29. The registry reports the primary endpoint in an evaluable population and gives an arm-specific evaluable count of 8 for Treatment Arm 1 and 10 for Treatment Arm 2. These endpoint counts should not be treated as the full randomized group sizes: the record separately defines who was evaluable for the primary analysis. The reason that the endpoint population differs from enrollment is not reported in the registry record.

The primary outcome was estimated one-year overall survival in each treatment arm separately. The registry describes the endpoint as the percentage alive at one year, estimated with the Kaplan-Meier method. Death from any cause was the event; patients alive at their last evaluation were censored up to one year. Progression-free survival is listed as a secondary outcome, but its numerical result is not reported in the registry record.

The study started in 2013-05 and its primary completion was 2015-07. Its status is terminated. The reason for termination, whether recruitment ended as planned, and any consequences of termination for the analysis are not reported in the registry record. The planned sample-size justification, target enrollment, and statistical analysis plan are also not reported in the registry record. The posted analyses should therefore be interpreted as the results that are available in the record, rather than as confirmation that every planned analysis was completed.

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: Patients will be randomized to receive both treatments in different sequence: Arm 1 IL-2 followed by Ipi; Arm 2 IPI followed by IL-2 Evaluable Patients will have received at least 50% of dose of both study drugs and completed assessment; Safety Population are all patients who received 1 dose of either study drug; Intent to treat

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: patients who received at least 50% of both research drugs as planned-and had their disease re-evaluated after baseline; patients who exhibited PD or died before the end of one course of treatment were also considered evaluable

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 Estimated One-year OS in the Evaluable Population in Each Treatment Arm Separately, measured over start of first treatment to date of death from any cause and patients alive at their last evaluation date were censored up to 1 year..

Overall survival measures time from the start of the first treatment to death from any cause. The registry’s primary endpoint focuses on the estimated percentage alive at one year in each arm. Patients who were alive at their last evaluation were censored up to one year, meaning that their exact later survival time was not observed within the stated follow-up window.

The Kaplan-Meier method is used for this kind of time-to-event outcome because it can incorporate both observed deaths and censored follow-up. In general, censoring allows a participant’s observed time to contribute information up to the last point at which their status is known, without treating that time as a death. The resulting estimate is a survival probability over time, expressed as a percentage at the one-year point.

Overall survival is a clinically interpretable endpoint because it concerns death from any cause rather than a disease assessment alone. That general advantage does not remove uncertainty: the estimate depends on the observed follow-up and the assumptions behind the time-to-event analysis. In this study, the endpoint was reported separately for the evaluable population in each arm. The arm-specific percentages therefore describe estimated survival within each arm; the log-rank test is the separately reported analysis comparing survival experience between arms.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The primary endpoint is a time-to-event outcome. For each patient, the relevant time begins at the start of the first treatment and ends at death from any cause, or at the last evaluation for patients who were alive. The registry states that patients alive at their last evaluation were censored up to one year. This structure differs from an ordinary comparison of a single observed yes-or-no outcome because patients may have different lengths of follow-up.

What does Kaplan-Meier estimation do?

In general, the Kaplan-Meier method estimates the probability of remaining event-free over time while retaining information from participants whose follow-up ends before the event is observed. A censored participant contributes information while observed and is not counted as having had the event at the censoring time. At a chosen time point, the estimated survival probability can be expressed as a percentage. The registry uses this method to estimate one-year survival separately for each treatment arm.

The method relies on assumptions about the censoring process. In general, its interpretation is most straightforward when censoring is not related to a participant’s underlying event risk in a way that would systematically distort the estimate. The record specifies the censoring rule for patients alive at their last evaluation but does not report further detail about loss to follow-up or the reasons for censoring.

How should the confidence intervals be read?

The 95% confidence intervals describe uncertainty in each arm-specific estimate. In general, a confidence interval is a range generated by a statistical procedure that would cover the underlying quantity at the stated rate across repeated studies under its assumptions. It is not a guarantee that the true value lies within the particular interval. The intervals reported for this study are broad, so the one-year estimates have limited precision. They are intervals for each arm separately, not a confidence interval for the difference between arms.

What does the log-rank test compare?

The registry reports a Log Rank comparison between Treatment Arm 1 and Treatment Arm 2, with p = .001. In general, the log-rank test compares survival experience across the follow-up period by comparing observed events with the events expected under a null hypothesis of equal survival patterns. It uses information about when events occur, rather than comparing only the survival percentages at one selected time. A p-value describes how unusual data at least as inconsistent with the null would be if the null and the test assumptions held. It is not the probability that the null is true, and it does not measure the magnitude or clinical importance of a difference.

The registry does not report a hazard ratio, a confidence interval for a between-arm contrast, or detailed event counts for the primary endpoint. The posted arm-specific estimates and the log-rank p-value consequently provide different pieces of information: point estimates at one year for each arm, and a test of the broader survival experience. The record does not report details that would permit an independent reconstruction of the log-rank test.

Why does the analysis population matter?

The primary endpoint population is defined by treatment exposure and disease reassessment, with an exception for patients who exhibited progressive disease or died before the end of one course. This definition is not equivalent, on its face, to an analysis of every randomized participant. In general, restricting an analysis to participants who meet post-randomization criteria can affect the comparability created by randomization, because eligibility for the analysis may depend on events occurring after assignment. That is a methodological consideration, not a claim about the direction or size of any effect in this study.

The record does not report the number excluded from this endpoint population by reason, the handling of all missing endpoint information, or a separate intention-to-treat analysis. It also does not report the planned statistical analysis plan. Those omissions limit assessment of how the posted analysis corresponds to the original analysis strategy.

What can planning illustrations establish?

In general, a sample-size calculation describes the number of participants or events that a hypothetical design would require under chosen assumptions. It is a planning exercise, not a measure of the power observed after a study has ended. The registry’s stated planning assumptions use a two-sided significance level of 0.05, equal allocation, a normal approximation, and effects expressed as fractions of the observed effect. For time-to-event outcomes, the described convention is to give the total number of events; for other outcomes, it is participants per arm. These assumptions do not establish the study’s original design rationale, which is not reported in the registry record. Any planning illustration based on an observed effect should not be interpreted as observed power.

The registry does not report multiplicity handling, interim analyses, or additional details about the statistical plan. In general, multiplicity matters when several outcomes, groups, time points, or analyses are tested, because repeated testing can affect the overall chance of a false-positive finding. Without the planned testing hierarchy or adjustment strategy, the role of the reported p-value among possible analyses cannot be fully assessed. The absence of that information is a limitation of the registry record, not evidence that a particular procedure was or was not used.

10. Exact Estimates for Each Arm

No two-arm comparison is shown for this trial, but each arm's result can still be summarized with exact interval estimates computed from the posted counts. The Clopper-Pearson interval is exact and conservative; the Wilson interval is a common closer-to-nominal alternative; the Bayesian credible interval uses a uniform Beta(1, 1) prior. For small groups the three can differ noticeably.

R code and output

arm    <- c("Treatment Arm 1", "Treatment Arm 2")
n      <- c(8, 10)
events <- c(6, 8)
for (i in seq_along(n)) {
  ci <- binom.test(events[i], n[i])$conf.int          # Clopper-Pearson exact interval
  cr <- qbeta(c(0.025, 0.975), 1 + events[i], 1 + n[i] - events[i])   # Bayesian, uniform Beta(1,1) prior
  wi <- (function(x, n) { p <- x / n; z <- qnorm(0.975); d <- 1 + z^2 / n
    c((p + z^2 / (2 * n) - z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d, (p + z^2 / (2 * n) + z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d) })(events[i], n[i])
  cat(sprintf("ARM|%s|%d|%d|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f\n", arm[i], n[i], events[i], 100 * events[i] / n[i],
              100 * ci[1], 100 * ci[2], 100 * wi[1], 100 * wi[2], 100 * cr[1], 100 * cr[2]))
}

ARM  Treatment Arm 1  8  6  75.0  34.9  96.8  40.9  92.9  40.0  92.5
ARM  Treatment Arm 2  10  8  80.0  44.4  97.5  49.0  94.3  48.2  94.0

11. Analyses Reported by the Sponsor

The posted analysis reports a Log Rank comparison between the treatment arms with p = .001. The primary endpoint estimates themselves are reported separately for each arm, with confidence intervals, rather than as a direct between-arm difference. The registry record does not report a recalculated two-arm effect estimate for these binary percentage summaries, so there is no corresponding recalculated comparison to place alongside the sponsor’s test.

The log-rank result uses time-to-event information and is not interchangeable with an unadjusted comparison of the two one-year percentages. In general, a re-calculation from arm-level summaries may not reproduce a test based on individual event times, censoring, and follow-up. The record does not include the event-time data needed to independently recreate the log-rank analysis. It also does not report a hazard ratio or confidence interval for the between-arm contrast.

Accordingly, the sponsor-reported log-rank result is the available inferential comparison in the registry record. The arm-specific percentages and confidence intervals provide complementary descriptive information, but they do not independently verify or replace that test. No discrepancy is identified in the record between a sponsor analysis and a recalculated result.

12. Multiplicity, Missing Data and Analysis Populations

The record reports one primary outcome, estimated one-year overall survival, and lists progression-free survival as a secondary outcome. It does not report a multiplicity adjustment, a hierarchy of hypotheses, or an explanation of how the primary and secondary outcomes were handled together. In general, when several outcomes or analyses are tested, multiplicity procedures can control the chance of false-positive findings across those tests. Without a stated strategy, the reported p-value should be understood as the result of the specific Log Rank analysis recorded, not as evidence that multiplicity across every possible analysis has been addressed.

Progression-free survival is named as a secondary outcome, but its numerical result and analysis details are not reported in the registry record. The record also does not report whether interim analyses were planned or conducted, whether there were other prespecified analyses, or whether the reported comparison was the only survival comparison performed. Those details matter when assessing how selectively the posted result may represent the full analysis plan.

Missing-data handling is described only in part for the primary endpoint: patients alive at their last evaluation were censored up to one year. The reasons for censoring, completeness of follow-up, and handling of missing disease or survival information beyond that description are not reported in the registry record. In general, censoring is not the same as treating a participant as having survived through the end of follow-up; it uses information up to the last known evaluation and relies on assumptions about the relation between censoring and outcome risk.

The endpoint analysis population is evaluable patients who received at least 50% of both research drugs and had disease re-evaluated after baseline, with patients who exhibited progressive disease or died before one course also considered evaluable. The record does not report a complete participant flow by randomized arm or reasons for non-evaluability. It does not report a separate all-randomized analysis. The criteria and absent flow details make it difficult to assess the impact of missing assessments and post-randomization selection on the endpoint results.

13. 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 0% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: supraventricular tachycardia, ventricular tachycardia (15.4%), autoimmune encephalitis, intracranial hemorrhage (15.4%), colitis (12.5%), pneumonia (12.5%), respiratory failure, hypoxia (12.5%).

The registry reports serious adverse event figures separately from the evaluable population used for the primary endpoint. For Treatment Arm 1, 9 of 13 participants at risk were affected by serious adverse events; for Treatment Arm 2, 10 of 16 were affected. The registry also reports 9 of 13 and 10 of 16, respectively, for other adverse events, and deaths of 1 and 2 in the corresponding safety groups. These safety denominators should not be substituted for the primary endpoint counts.

The listed serious adverse event terms and percentages are supraventricular tachycardia, ventricular tachycardia and autoimmune encephalitis, intracranial hemorrhage, each reported at 15.4%; colitis, pneumonia, and respiratory failure, hypoxia, each at 12.5%; and diarrhea at 7.7%. The registry record does not give additional context about event timing, severity grades, attribution to treatment, or whether terms could occur in the same participant. The percentages should therefore be read as posted, without assuming they represent mutually exclusive groups.

These figures describe reported adverse events in the registry. They do not by themselves establish that a drug caused an event or support a comparative safety conclusion. The safety and efficacy populations and their denominators differ, and the record does not report further details needed to reconcile them.

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

The posted results describe one-year overall survival estimates for the evaluable population in each randomized treatment sequence. Treatment Arm 1 had an estimate of 73.0%, with a 95% confidence interval from 38.0% to 91.0%; Treatment Arm 2 had an estimate of 75.0%, with a 95% confidence interval from 40.0% to 92.0%. The registry also reports a Log Rank comparison with p = .001. These are distinct statistical summaries: the percentages estimate survival at one year within each arm, while the test compares survival experience over follow-up.

The data do not establish which sequence is preferable. The registry does not report a between-arm effect size with a confidence interval, and the arm-specific intervals are broad. A p-value alone does not communicate the magnitude or precision of a difference. The log-rank finding should therefore not be converted into a claim about the size, direction, or clinical importance of an effect beyond what the record states.

The results also do not represent a stated analysis of all randomized participants. The primary endpoint population is defined by treatment receipt and disease reassessment, with the additional inclusion provision for early progression or death. The record does not report a full flow of participants into and out of this analysis. The uncertainty in the estimates and the evaluable-population definition limit how broadly the results can be interpreted.

More generally, a randomized design is intended to support comparison of assigned strategies, but the interpretation depends on how participants are analyzed and how missing follow-up is handled. The record does not report the planned analysis strategy, detailed censoring reasons, or a separate analysis of all randomized participants. It also does not report the reason the study was terminated. These gaps prevent a complete assessment of the relationship between the original design and the posted analysis.

The results are best understood as the registry’s reported survival estimates and log-rank test for this study, with their stated population and uncertainty. They do not provide medical advice, prove that either sequence caused a particular outcome, or establish that the findings apply beyond the population and analysis described in the record.

15. Why This Trial Matters Statistically

This trial raises a useful statistical distinction between describing survival at a fixed time and testing survival patterns over time. The one-year Kaplan-Meier estimates summarize each arm at a clinically interpretable point, while the log-rank analysis uses the event-time information across follow-up. Those methods answer related but different questions. Reading them together helps prevent a point estimate at one time from being mistaken for a complete comparison of survival experience.

The record also highlights the importance of analysis populations. Randomization assigns treatment sequences, but the primary endpoint is described for evaluable patients meeting post-assignment criteria. In general, post-randomization exclusions can complicate interpretation because the analyzed groups may not preserve all the balance created by random assignment. Participant flow and an all-randomized analysis would help readers understand this issue; they are not reported in the registry record.

Precision is another central lesson. The arm-specific confidence intervals are broad, so the percentages should not be treated as exact descriptions of survival. A statistically reported p-value and imprecise arm-specific estimates can coexist because they summarize different features of the data. Neither alone supplies the full picture of effect size, uncertainty, and clinical relevance.

Finally, a registry report is most informative when its endpoint definitions, missing-data procedures, analysis population, multiplicity strategy, and reasons for study completion are all clear. Several of these details are not reported in the registry record. This limits how fully the analysis can be reconstructed and how confidently readers can connect the reported test to the original analysis plan. The study therefore provides an instructive case in reading survival results cautiously, distinguishing descriptive estimates from formal tests, and treating missing methodological details as genuine limits on interpretation.

16. 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 study’s primary endpoint was reported for an evaluable population rather than a clearly described analysis of all randomized participants. Eligibility for that population depended on treatment receipt and disease reassessment, with an additional rule for patients who exhibited progressive disease or died before the end of one course. The record does not report the number or reasons for participants who were not included in the primary endpoint analysis by arm. That omission limits assessment of selection after randomization.

The arm-specific one-year estimates have broad 95% confidence intervals. The endpoint counts are 8 in Treatment Arm 1 and 10 in Treatment Arm 2, indicating that the estimates are based on a limited evaluable population. The registry does not report a confidence interval for a direct between-arm effect or a hazard ratio. It reports a log-rank p-value, but not the event-time details needed to reconstruct the analysis independently.

The registry does not report the planned sample-size justification, target enrollment, statistical analysis plan, interim analysis procedures, multiplicity strategy, or detailed missing-data handling. Although the record states how living participants were censored, it does not describe censoring reasons or the completeness of follow-up. The reason the study was terminated is also not reported in the registry record.

Progression-free survival is listed as a secondary outcome, but its numerical result and analysis details are not reported in the registry record. Safety summaries use groups with denominators distinct from the primary endpoint population, and the record does not give details on event attribution, timing, or overlap among terms. These gaps constrain interpretation of both efficacy and safety summaries and prevent a comprehensive appraisal of the full study results.

17. Related Tutorials

18. Related Statistical Calculators

19. Sources

• ClinicalTrials.gov: HD IL-2 + Ipilimumab in Patients With Metastatic Melanoma — NCT01856023.
• Registry record last updated 2023-12-05; status last verified 2023-03.
• Results first posted 2023-12-05.
• 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.

20. Record Summary

NCT01856023 is a randomized, open-label, phase IV, multi-center study of treatment sequence in metastatic melanoma. Treatment Arm 1 received high-dose interleukin-2 followed by ipilimumab; Treatment Arm 2 received ipilimumab followed by high-dose interleukin-2. Enrollment was 29, and the study status is terminated.

The primary outcome was estimated one-year overall survival in the evaluable population, using the Kaplan-Meier method. The registry reports estimates of 73.0% for Treatment Arm 1 and 75.0% for Treatment Arm 2, each with a 95% confidence interval, and a Log Rank comparison with p = .001. Progression-free survival is listed as a secondary outcome, but numerical results are not reported in the registry record.

The record defines an evaluable population but does not report a full participant flow, a separate all-randomized analysis, or several details of the statistical plan. The reason for termination is not reported in the registry record. These limitations should be considered when interpreting the posted estimates and test.