Pembrolizumab With Talimogene Laherparepvec or Placebo in Unresected Melanoma: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Pembrolizumab With Talimogene Laherparepvec or Placebo in Unresected Melanoma (NCT02263508), a randomized phase 3 study. The primary outcome is Phase 3: Progression Free Survival (PFS) by Blinded Independent Central Review (BICR) Assessed Using Modified RECIST 1.1.

2. Clinical Question

The registry describes a randomized, double-masked, parallel-group phase 3 study. Its primary objectives were to evaluate progression-free survival and overall survival. The posted primary result concerns progression-free survival, not the overall survival objective. The registry record also lists a phase 1b component, whose primary objective was safety as assessed by dose-limiting toxicity; it is a separate part of the study and should not be conflated with the randomized phase 3 comparison.

3. Trial Design

This was a phase 1b/3, multicenter trial with randomized allocation, double masking, and a parallel-group design. The phase 3 comparison was talimogene laherparepvec plus pembrolizumab versus placebo plus pembrolizumab. The registry lists an enrollment of 713 and reports 346 participants in each phase 3 arm for the primary endpoint. The primary analysis population is described as all participants randomized in phase 3.

The study began on 2014-12-08, and the recorded primary completion date is 2021-03-11. The registry status is terminated. The reason for termination is not reported in the registry record, so the status alone does not explain why the trial ended or whether it was related to efficacy, safety, feasibility, or another consideration.

The primary progression-free survival endpoint was measured from randomization until the data cut-off date of 02 March 2020. The registry reports median follow-up of 25.5 months in the placebo plus pembrolizumab arm and 25.6 months in the talimogene laherparepvec plus pembrolizumab arm, with the ranges recorded for each group. The primary analysis was specified to be conducted when 407 progression-free survival events had occurred. This describes the analysis timing reported in the registry, but the broader interim-monitoring plan is not reported in the registry record.

The registry lists additional outcomes, including response measures, modified irRC-RECIST progression-free survival, and overall survival. Their populations, follow-up definitions, and analysis results differ from the primary endpoint and should be interpreted separately. The trial's original sample-size justification and other design assumptions are not reported in the registry record.

4. Participant Flow and Accounting

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

Before assignment: In Phase 3 participants were randomized equally to 1 of 2 arms. Randomization was stratified by stage of disease: less advanced stages (IIIB, IIIC, and IVM1a) versus more advanced stages (IVM1b and IVM1c) and by prior serine/threonine protein kinase B-Raf (BRAF) inhibitor therapy (no prior BRAF inhibitor versus BRAF inhibitor with or without mitogen-activated extracellular signal-regulated kinase \[MEK\] inhibitor).

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: The DLT analysis set included DLT-evaluable participants enrolled in Phase 1b who had the opportunity to be on treatment for at least 6 weeks from the initial dose of pembrolizumab and received at least 2 doses of talimogene laherparepvec and 2 doses of pembrolizumab in combination (ie, on the same day), or those who otherwise experienced a DLT within 6 weeks after starting the combination therapy.

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 enrolled participants

6. Primary Endpoint

The registry lists the primary outcome as Phase 3: Progression Free Survival (PFS) by Blinded Independent Central Review (BICR) Assessed Using Modified RECIST 1.1, measured over From randomization until the data-cut-off date of 02 March 2020; median (range) time on follow-up was 25.5 (0.6, 44.7) months in the Placebo + Pembrolizumab arm and 25.6 (0.3, 45.8) months in the Talimogene Laherparepvec + Pembrolizumab arm..

Progression-free survival measures the time from randomization until the earlier of documented progressive disease or death from any cause. In this trial, progression was assessed by blinded independent central review using modified RECIST 1.1. The registry defines progression as a specified increase in target-lesion size from nadir or the appearance of a new lesion.

As a time-to-event endpoint, progression-free survival incorporates both whether an event occurred and when it occurred. Participants who had not experienced an event by the analysis cut-off did not simply disappear from the analysis: the record says they were censored at their last evaluable tumor assessment, if available, or otherwise at randomization. General statistical explanation: censoring allows a time-to-event analysis to use participants' observed follow-up without treating an event-free observation as though it were an event. It relies on assumptions about how censoring relates to the event process; the registry does not report diagnostics of those assumptions.

Progression-free survival is used to summarize the time until worsening disease or death, combining those event types in a single endpoint. It does not directly describe overall survival, symptoms, or quality of life. Its interpretation also depends on how progression is assessed and on the follow-up and censoring process. The registry's independent central review approach specifies who assessed progression for the primary endpoint; it does not remove all uncertainty inherent in a time-to-event estimate.

7. Results Posted in the Registry

• Phase 3: Placebo + Pembrolizumab vs Phase 3: Talimogene Laherparepvec + Pembrolizumab: hazard ratio 0.86 (95% CI 0.71–1.04), reported p = 0.130 (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.

Phase 3: Complete Response Rate Assessed Using Modified irRC-RECIST (iCRR)

• 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:
  • Phase 3: Talimogene Laherparepvec + Pembrolizumab: 14.5% (50/346) versus 8.1% (28/346) on control; risk difference 6.4 percentage points (95% CI 1.7 to 11.0), risk ratio 1.79 (1.15–2.77); unadjusted two-proportion z-test p = 0.008.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the proportion with the event is statistically higher for Phase 3: Talimogene Laherparepvec + Pembrolizumab than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Phase 3: Progression Free Survival Assessed Using Modified irRC-RECIST (iPFS)

• Endpoint type: Time to event: compared as a hazard ratio.
• Test: Hazard ratio as reported by the sponsor (Stratified log-rank test); it was not recalculated.
• Result:
  • Phase 3 : Placebo + Pembrolizumab vs Phase 3: Talimogene Laherparepvec + Pembrolizumab: hazard ratio 1.05 (95% CI 0.82–1.34), reported p = 0.140 (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 Phase 3 : Placebo + Pembrolizumab vs Phase 3: Talimogene Laherparepvec + Pembrolizumab and the control. The interval gives the range of effects that remain compatible with the data.

Phase 3: Overall Survival Excluding Stage IVM1c Participants

• Endpoint type: Time to event: compared as a hazard ratio.
• Test: Hazard ratio as reported by the sponsor (Stratified log-rank test); it was not recalculated.
• Result:
  • Phase 3: Placebo + Pembrolizumab vs Phase 3: Talimogene Laherparepvec + Pembrolizumab: hazard ratio 0.88 (95% CI 0.63–1.24), reported p = 0.470 (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 Phase 3: Placebo + Pembrolizumab vs Phase 3: Talimogene Laherparepvec + Pembrolizumab and the control. The interval gives the range of effects that remain compatible with the data.

The secondary results do not form a uniform pattern. The modified irRC-RECIST complete-response outcome has a sponsor-reported logistic-regression estimate whose confidence interval excludes 1, while the reported modified irRC-RECIST progression-free survival and overall-survival comparisons have confidence intervals that include 1. The posted response result therefore differs in inferential appearance from the primary progression-free survival result, but it addresses a different endpoint and uses a different analysis method.

These outcomes do not overturn or substitute for the primary endpoint. The complete-response rate is a binary response measure; modified irRC-RECIST progression-free survival is a time-to-event outcome with a different response framework from the primary RECIST 1.1 endpoint; overall survival concerns death rather than progression or death. Endpoint definitions, assessment schedules, follow-up, and analysis populations can affect what each result means. The registry reports the follow-up timing for these secondary outcomes, but does not report every analysis detail or a complete statistical analysis plan.

General statistical explanation: when several outcomes are examined, the chance of finding at least one apparently notable result can increase, even if no true effect exists across the set. Adjusting for multiplicity can control that risk, but the registry record does not report whether or how multiplicity was handled for the secondary outcomes. Consequently, unadjusted secondary p-values carry less confirmatory weight than a prespecified primary analysis. They are useful for describing the pattern of results and for motivating further questions, but should not be treated as independent proof of efficacy.

The secondary findings are also not all equally precise. Some estimates have reported confidence intervals, while some outcome values or limits are absent in the registry record. The overall reading is therefore one of a more favorable-looking response result alongside secondary time-to-event results that do not establish a clear group difference, set against a primary progression-free survival result that itself is not statistically clear.

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

• Phase 1b: Objective Response Rate (ORR)
• Phase 1b: Best Overall Response (BOR)
• Phase 1b: Durable Response Rate (DRR)
• Phase 1b: Duration of Response (DOR)
• Phase 1b: Disease Control Rate (DCR)
• Phase 1b: Progression-free Survival (PFS)
• Phase 1b: Overall Survival (OS)
• Phase 3: Complete Response Rate Assessed Using Modified irRC-RECIST (iCRR)

9. Statistical Interpretation

10. Statistical Methods Explained

The primary endpoint was analyzed using Kaplan-Meier methods to estimate progression-free survival over time. General statistical explanation: the Kaplan-Meier estimator updates the estimated probability of remaining event-free as events occur, while retaining censored participants in the risk set up to their censoring time. This is useful when follow-up durations vary and not every participant has an observed event by the analysis cut-off. The registry gives the censoring rules used for participants without an event, but does not report further checks of censoring assumptions.

What does the reported hazard ratio describe?

The primary comparison used a stratified log-rank test and reported a hazard ratio of 0.86, with a 95% confidence interval from 0.71 to 1.04. A hazard ratio compares event rates over time between groups under the time-to-event analysis. It is not a ratio of median survival times, nor is it a direct difference in months. The comparison is labeled placebo plus pembrolizumab versus talimogene laherparepvec plus pembrolizumab in the registry. The estimate is close to 1, and its confidence interval includes 1; those features indicate uncertainty about the direction and size of a relative difference.

General statistical explanation: the proportional-hazards interpretation of a single hazard ratio is most straightforward when the relative hazards are reasonably stable over time. The registry record does not report a proportional-hazards assessment or enough detail to evaluate whether that assumption held. A hazard ratio is also not an individual-level prediction and should not be interpreted as the probability that a participant will experience progression or death.

Why use a stratified log-rank test?

The registry identifies the stratified log-rank test as the method for the primary comparison. General statistical explanation: a log-rank test compares the observed and expected event experience across groups over follow-up. Stratification can account for prespecified grouping factors in the comparison, helping preserve the trial's planned structure. The specific stratification factors are not reported in the registry record. The reported p-value is 0.13. It summarizes how compatible the observed test statistic is with the null hypothesis under the test's assumptions; it is not the probability that the null hypothesis is true.

For the primary endpoint, the registry says the analysis was specified when 407 progression-free survival events had occurred, and gives 02 March 2020 as the data cut-off date. This information documents the event-based timing of the primary analysis. The registry record does not report a full statistical analysis plan, an interim efficacy-monitoring scheme, stopping boundaries, or the original sample-size justification. None should be inferred from the event target alone.

How should the Kaplan-Meier medians be read?

The reported medians are 8.5 months for placebo plus pembrolizumab and 14.3 months for talimogene laherparepvec plus pembrolizumab. Each median is accompanied by a 95% confidence interval. General statistical explanation: a Kaplan-Meier median is the time at which the estimated event-free proportion reaches the median point, if the curve reaches that point. It is a summary of the estimated curve, not a description of every participant's experience. A median comparison does not use all the timing information in the same way as the hazard-ratio analysis, and medians alone do not establish a treatment effect.

The record reports median follow-up of 25.5 months in the placebo arm and 25.6 months in the combination arm, with ranges. Follow-up duration is not the same as progression-free survival: it describes observation time, whereas progression-free survival is defined by time to the specified event. The record's event and censoring definitions are important to interpreting both the Kaplan-Meier estimates and the hazard ratio.

How do the secondary methods differ?

The complete-response rate was analyzed with logistic regression for the sponsor-reported odds ratio. General statistical explanation: logistic regression models the odds of a binary outcome and can produce an odds ratio that accounts for the model terms included. The registry does not state all model covariates. Its reported odds ratio is not interchangeable with a risk difference or a risk ratio. The modified irRC-RECIST progression-free survival and the overall-survival analyses used stratified log-rank tests and reported hazard ratios, so they share a time-to-event framework but concern different endpoints and follow-up definitions.

General statistical explanation: over-dispersion matters for count outcomes when variability exceeds what a simple count model assumes, because standard errors can then be too small and uncertainty understated. The posted primary outcome is time-to-event, not an over-dispersed count outcome; the registry does not describe an over-dispersion model for these analyses. This illustrates why methods should be matched to endpoint type rather than transferred automatically from one kind of outcome to another.

The registry reports that primary progression status was assessed by blinded independent central review. Double masking and independent review are design features relevant to reducing opportunities for assessment to depend on treatment knowledge. They do not guarantee the absence of bias, and the registry record does not report an assessment of masking success. The analysis population for the primary endpoint is explicitly all participants randomized in phase 3; analysis-population details for every secondary endpoint are not reported in the registry record.

11. Confidence Intervals and Precision

The primary hazard ratio is 0.86, with a 95% confidence interval from 0.71 to 1.04. The interval describes the uncertainty around the estimated relative hazard under the analysis method. It includes 1, the value corresponding to no relative difference in hazards. The reported p-value is 0.13, consistent with the result not meeting a conventional significance threshold of 0.05.

The interval is more informative than the point estimate alone because it shows a range of effects reasonably compatible with the data and the model. It does not mean there is a 95% probability that the true effect lies within this particular interval. General statistical explanation: a confidence interval procedure is designed to cover the true parameter at its stated rate over repeated applications under its assumptions. The interval's practical meaning depends on the quality of the endpoint measurements, the analysis model, and the validity of the trial's conduct and follow-up.

The medians also have reported 95% confidence intervals: 5.72 to 13.54 months for placebo plus pembrolizumab, and 10.25 to 22.11 months for talimogene laherparepvec plus pembrolizumab. These intervals overlap. Overlap of separate intervals is not itself a formal test of the difference between groups; the trial's reported between-group time-to-event analysis is the more direct inferential comparison.

The secondary outcomes have their own confidence intervals and should not inherit the primary endpoint's precision. For example, the complete-response rate estimates have uncertainty intervals, while some secondary time-to-event estimates are not fully reported in the registry record. When a confidence limit or median is absent, it should not be reconstructed from other numbers. Precision is therefore outcome-specific, and missing interval information limits how closely some secondary estimates can be assessed.

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 use the observed effect as a starting point and then consider planning for 100%, 75%, or 50% of that effect. For the binary complete-response outcome, the reported figures are participants per arm for 80% power and participants per arm for 90% power. For the time-to-event comparisons, the figures are total events required, not participants per arm. The event totals are not directly comparable to participant counts because participants may be censored before an event or may have different follow-up.

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. For time-to-event planning, the registry identifies the Schoenfeld approach; the other planning illustrations report participant counts per arm. These are planning assumptions for an illustration based on the observed effect, not the trial's original design assumptions. The trial's own sample-size justification is not reported in the registry record.

In plain terms, the 80% power column describes the planned sample size or event total that would give an 80% chance of detecting the stated effect if the assumptions and effect were correct. The 90% power column represents a higher chance of detecting that effect, and typically requires more information. Planning for a smaller effect requires more participants or events, which is why the rows for smaller fractions of the observed effect have larger requirements.

These figures are not observed power. They are hypothetical planning illustrations and should not be read as a retrospective measure of the trial's success or as evidence that a result is true. Because the illustration is anchored to an observed effect, it is especially important not to treat it as an independent confirmation of that effect.

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 = 0.860, so ln(HR) = -0.1508.
2. d = 4 × (1.96 + 0.842)² / [ln(0.860)]² = 1,380.2, rounded up to 1,381 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 <- 0.86                              # 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(2263508)
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 = 1,381 (80% power), 1,848 (90% power)
 75% of observed effect: events = 2,454 (80% power), 3,285 (90% power)
 50% of observed effect: events = 5,521 (80% power), 7,391 (90% power)
Simulated power with 1,381 events: 0.808 (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 2,159 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 713 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 <- 8.5          # control-arm median, months (converted from the posted unit)
hr          <- 0.86          # 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: 1506         p (1-sided)   0.0001
  Events: 623        ~HR at bound   0.7428
    Month: 17    P(Cross) if HR=1   0.0001
              P(Cross) if HR=0.86   0.0338
    IA 2: 67%                   Z   2.5114
      N: 2160         p (1-sided)   0.0060
 Events: 1246        ~HR at bound   0.8673
    Month: 26    P(Cross) if HR=1   0.0060
              P(Cross) if HR=0.86   0.5603
        Final                   Z   1.9930
      N: 2160         p (1-sided)   0.0231
 Events: 1869        ~HR at bound   0.9119
    Month: 60    P(Cross) if HR=1   0.0250
              P(Cross) if HR=0.86   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

The registry reports the sponsor's primary analysis as a stratified log-rank test with a hazard ratio of 0.86, a 95% confidence interval of 0.71 to 1.04, and a p-value of 0.13. That is the trial's reported inferential result for the primary endpoint. Any simple re-calculation from posted summary statistics is unadjusted and cannot reproduce the full time-to-event analysis, its stratification, censoring, or participant-level event timing.

The registry also reports sponsor analyses for secondary outcomes, including logistic regression for complete response and stratified log-rank tests for secondary time-to-event endpoints. Their model-based results are the appropriate record of the sponsor's stated analyses. An unadjusted re-calculation from group summaries may differ because it does not necessarily use the same model or covariates. It should not replace the sponsor's analysis when describing the trial's reported conclusions.

The record does not flag a discrepancy for the primary result. In any case, a recalculation is a check based on limited summary data, not an alternative full analysis. The original analysis plan and further model details are not reported in the registry record, which limits independent reconstruction.

16. Multiplicity, Missing Data and Analysis Populations

The registry identifies progression-free survival and overall survival as primary phase 3 objectives, and the posted primary endpoint is progression-free survival. It does not report a full statistical analysis plan that explains how the primary objectives were ordered, whether testing was hierarchical, or how type I error was controlled across primary and secondary endpoints. The registry record also does not report a multiplicity adjustment for the secondary analyses.

Multiplicity matters because testing many outcomes can raise the chance of at least one result appearing statistically notable by chance. The secondary complete-response result has a reported p-value below 0.05, but without a stated multiplicity strategy it should not be interpreted as having the same confirmatory status as a prespecified primary test. The registry's listing of many outcomes does not itself establish which were prespecified for formal hypothesis testing or how their tests were related.

For the primary progression-free survival analysis, the record specifies all participants randomized in phase 3 as the population and describes censoring for those without an event. The registry gives the primary data cut-off date and the event count at which the primary analysis was specified. It does not report details about missing tumor assessments beyond the censoring rule, the amount of missing endpoint data, sensitivity analyses for informative censoring, or how protocol deviations were handled.

Analysis-population details for every secondary endpoint are not reported in the registry record. In particular, the record does not give enough information to determine whether all secondary analyses used the same randomized population or different evaluable subsets. That uncertainty affects how directly secondary estimates can be compared with the primary result. The registry also does not report interim-analysis boundaries or the full procedure for handling repeated looks at accumulating data. These details should not be inferred from the event-driven timing statement.

17. 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: pyrexia (9.5%), malignant melanoma (7.8%), general physical health deterioration (4.8%), autoimmune hepatitis (4.8%), cholecystitis (4.8%).

The registry lists serious adverse event figures separately for the phase 1b combination group and the two phase 3 groups. Serious adverse events are reported for 8 of 21 participants in the phase 1b talimogene laherparepvec plus pembrolizumab group, 141 of 344 in the placebo plus pembrolizumab group, and 154 of 344 in the phase 3 combination group. The record lists deaths as 6, 156, and 146 for those groups, respectively. These are descriptive counts; the registry does not give a causal attribution for the deaths in the registry record recorded.

The listed serious-event terms include pyrexia at 9.5%, malignant melanoma at 7.8%, and general physical health deterioration, autoimmune hepatitis, cholecystitis, and cytokine release syndrome at 4.8% each. These terms are not a complete safety profile, and the registry's figures do not by themselves establish that a particular event was caused by treatment. Interpretation also depends on event definitions, observation time, and exposure, which are not fully specified in the safety summary.

Safety counts should be read alongside the study population and follow-up rather than treated as a simple measure of comparative risk. The phase 1b group is a different study component from the randomized phase 3 groups. The registry record does not report adjusted between-group safety comparisons in the registry record for this summary.

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

The primary data describe a randomized comparison of adding talimogene laherparepvec to pembrolizumab against placebo plus pembrolizumab in unresectable melanoma. The sponsor-reported primary analysis estimates a relative progression-free survival hazard ratio of 0.86, with a 95% confidence interval from 0.71 to 1.04 and a p-value of 0.13. The interval includes the no-difference value, so the primary result does not establish a statistically clear difference under the reported test.

This does not show that the groups are equivalent. A non-significant test can reflect a small effect, uncertainty, or both; it does not demonstrate that meaningful differences are absent. Nor should the point estimate alone be taken as proof of benefit or harm. The result is a group-level estimate, not a prediction for any individual participant.

The registry also reports different results for secondary outcomes. A response measure has a reported model-based result with a confidence interval excluding 1, while secondary time-to-event comparisons do not show similarly clear evidence. Because endpoints differ and the registry does not report how multiplicity was handled, the secondary findings should be interpreted with restraint and should not be used to replace the primary analysis.

The findings do not answer every question about safety, durability, quality of life, or outcomes in populations beyond those represented in the trial. The registry status is terminated, but the reason is not reported in the registry record. That status should not be interpreted as a clinical conclusion. Overall, the posted results define what was estimated and how uncertain it is; they do not establish a definitive treatment recommendation.

19. Why This Trial Matters Statistically

This trial illustrates why a clinical result cannot be reduced to its point estimate or to a single p-value. The reported primary hazard ratio is accompanied by a confidence interval that spans the no-difference value, making uncertainty central to interpretation. The primary comparison is therefore best understood through its estimate, interval, and test result together.

It also shows why endpoint definition matters. The primary outcome combines progression and death and uses blinded independent central review under modified RECIST 1.1. Other outcomes use a response definition, a different progression framework, or overall survival. Results for these endpoints answer related but distinct questions; they should not be treated as interchangeable measures of the same effect.

The registry includes an event-based timing statement for the primary analysis and describes censoring rules. These details help readers understand how follow-up contributes to a time-to-event analysis. At the same time, the absence of a full statistical analysis plan, detailed stratification factors, multiplicity procedures, and the original sample-size justification limits how fully the analysis can be reconstructed from the registry alone.

Finally, the record's planning illustrations distinguish participant counts from total event requirements and show how required information grows when planning for smaller effects. That distinction is important for interpreting time-to-event trials: event accrual, not enrollment alone, drives the information available for many survival analyses. The trial provides a useful example of reading an estimate alongside its uncertainty and recognizing the boundaries of registry-level reporting.

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.

Several trial-specific details needed for a full appraisal are not reported in the registry record. These include the reason for termination, the original sample-size justification, the stratification factors for the primary test, the full statistical analysis plan, and any interim-monitoring boundaries. Without these details, the rationale for the event target and the complete inferential framework cannot be assessed from the record alone.

The primary result is limited by statistical uncertainty: its confidence interval includes 1, and the reported p-value is 0.13. The registry provides median progression-free survival estimates, but those medians do not replace the primary hazard-ratio analysis. The proportional-hazards assessment and other model diagnostics are not reported, so the extent to which a single hazard ratio summarizes the pattern over time cannot be judged from the registry record.

Secondary outcomes involve different endpoints and analysis methods, and multiplicity handling is not reported. Some secondary values or confidence limits are absent in the record. These omissions make it harder to compare the secondary findings consistently and increase the importance of distinguishing confirmatory from exploratory evidence.

The safety summary supplies counts and selected serious-event terms, but the record does not provide a detailed adjusted safety comparison in the registry record. It also does not give enough information to assess event exposure time or attribution. Finally, trial results apply most directly to the enrolled study population and procedures; the registry record does not establish how well the findings extend to other populations or care settings.

21. Related Tutorials

22. Related Statistical Calculators

23. Sources

• ClinicalTrials.gov: Pembrolizumab With Talimogene Laherparepvec or Placebo in Unresected Melanoma — NCT02263508.
• Registry record last updated 2022-11-14; status last verified 2022-11.
• Results first posted 2022-05-16.
• 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

NCT02263508 is a randomized, double-masked, parallel-group phase 1b/3 trial of talimogene laherparepvec with pembrolizumab versus placebo with pembrolizumab in unresectable melanoma. The primary phase 3 endpoint was progression-free survival by blinded independent central review using modified RECIST 1.1, analyzed with Kaplan-Meier methods and a stratified log-rank test.

The sponsor-reported primary hazard ratio was 0.86, with a 95% confidence interval of 0.71 to 1.04 and p = 0.13. The interval includes 1, so the primary analysis does not establish a statistically clear difference. Secondary outcomes show a mixed pattern, and multiplicity handling is not reported in the registry record. The record lists the study status as terminated, but does not report the reason.