Study of MEDI4736 Monotherapy and in Combination With Tremelimumab Versus Standard of Care Therapy in Patients With Head and Neck Cancer: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Study of MEDI4736 Monotherapy and in Combination With Tremelimumab Versus Standard of Care Therapy in Patients With Head and Neck Cancer (NCT02369874), a randomized phase 3 study. The primary outcome is Overall Survival (OS).

2. Clinical Question

The study was a Phase III randomized, open-label, parallel-group trial. Its stated purpose was to assess efficacy and safety of the two durvalumab strategies against standard of care in the target population. The primary endpoint was time-to-event rather than a simple count of participants alive at a single time point, so the reported analysis includes median survival estimates and hazard ratios with confidence intervals.

These results address the trial’s specified comparisons in the enrolled study population. They do not, by themselves, resolve whether effects differ by PD-L1 status or establish conclusions for populations not described in the registry record. The subgroup results are secondary analyses and should be interpreted with that distinction in mind.

3. Trial Design

The registry record describes a randomized, open-label, parallel-group Phase III study conducted across multiple centers and globally. The listed study period began on 2015-09-09, and primary completion was recorded as 2018-09-10. The trial status is completed. Enrollment is recorded as 736.

Participants were assigned to one of three groups: durvalumab monotherapy, durvalumab plus tremelimumab, or standard of care. The registry record gives endpoint-specific group counts for the primary survival analysis: 247 for the combination group, 240 for durvalumab, and 249 for standard of care. These are the counts attached to the posted primary outcome, not a substitute for a full accounting of all analysis populations or follow-up.

Open-label means that masking was listed as none. That design feature can matter for outcomes influenced by decisions or judgments made during follow-up. Overall survival is based on death from any cause, but the registry record does not describe the details of outcome ascertainment, censoring, follow-up procedures, or protections against bias.

The comparison structure allows each active strategy to be assessed against standard of care. It also supports the listed secondary comparison between the combination and durvalumab in the PD-L1-negative subgroup. The registry record does not report the exact standard-of-care regimen, randomization procedures, allocation ratio, stratification factors, or the trial’s original statistical design assumptions. Those details should not be inferred from the observed group counts or from the planning illustrations associated with the posted 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.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Full analysis set - inclusive of 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.

6. Primary Endpoint

The registry lists the primary outcome as Overall Survival (OS), measured over September 2015 to September 2018 (36 months).

Overall survival is the time from randomization until death from any cause. It is a time-to-event endpoint: the analysis uses both the time at which deaths occur and the information contributed by participants who have not had the event by the time their observation ends. The registry record identifies the primary analysis population as the full analysis set, inclusive of all randomized participants, regardless of PD-L1 status.

As a general statistical matter, overall survival is used in clinical trials because it captures a consequential outcome and does not depend on a judgment that a tumor has progressed or responded. The median is the time by which half of the survival distribution has experienced the event, subject to the estimation method and available follow-up. It does not describe the survival time of every participant, nor does it alone summarize the entire survival curve.

The registry record also reports hazard ratios for comparisons between groups. In general, a hazard ratio compares event rates over follow-up, with its interpretation depending on the time-to-event model and assumptions used. It is not a difference in median survival. The registry record does not provide sufficient detail to evaluate all modeling and censoring assumptions. The reported confidence intervals therefore matter alongside the point estimates when judging how precisely the comparisons are estimated.

7. Results Posted in the Registry

• Durvalumab + Tremelimumab vs Standard of Care (SoC): hazard ratio 1.04 (95% CI 0.85–1.26), reported p = 0.762 (as reported in the registry).
• Durvalumab vs Standard of Care (SoC): hazard ratio 0.88 (95% CI 0.72–1.08), reported p = 0.199 (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.

Overall Survival (OS) in PD-L1 Negative Participants

• 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:
  • Durvalumab + Tremelimumab vs Standard of Care (SoC): hazard ratio 0.92 (95% CI 0.73–1.17), reported p = 0.459 (as reported in the registry).
  • Durvalumab + Tremelimumab vs Durvalumab: hazard ratio 1.08 (95% CI 0.85–1.36) (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 Durvalumab + Tremelimumab vs Standard of Care (SoC) and the control. The interval gives the range of effects that remain compatible with the data.
  • The 95% confidence interval includes no difference (a ratio of 1), so these data do not show a statistically significant difference between Durvalumab + Tremelimumab vs Durvalumab and the control. The interval gives the range of effects that remain compatible with the data.

Overall Survival (OS) in PD-L1 Positive Participants

• Endpoint type: Time to event: compared as a hazard ratio.
• Test: Hazard ratio as reported by the sponsor; it was not recalculated.
• Result:
  • Durvalumab vs Standard of Care (SoC): hazard ratio 0.93 (95% CI 0.63–1.39) (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 Durvalumab vs Standard of Care (SoC) and the control. The interval gives the range of effects that remain compatible with the data.

Progression Free Survival (PFS)

• Endpoint type: Time to event: compared as a hazard ratio.
• Test: Hazard ratio as reported by the sponsor; it was not recalculated.
• Result:
  • Durvalumab + Tremelimumab vs Standard of Care (SoC): hazard ratio 1.09 (95% CI 0.90–1.33) (as reported in the registry).
  • Durvalumab vs Standard of Care (SoC): hazard ratio 1.02 (95% CI 0.84–1.25) (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 Durvalumab + Tremelimumab vs Standard of Care (SoC) and the control. The interval gives the range of effects that remain compatible with the data.
  • The 95% confidence interval includes no difference (a ratio of 1), so these data do not show a statistically significant difference between Durvalumab vs Standard of Care (SoC) and the control. The interval gives the range of effects that remain compatible with the data.

The registry record lists secondary outcomes that extend the primary overall-survival question. They include overall survival within PD-L1 subgroups, progression-free survival, response and disease-control measures, duration of response, and proportions alive or alive without progression. The posted secondary results include median time-to-event summaries and hazard-ratio comparisons for some outcomes and subgroups.

Across the reported time-to-event comparisons, the confidence intervals for hazard ratios include 1. This broad pattern does not establish that the groups are equivalent; it means the reported estimates do not clearly distinguish a relative hazard difference from the null value at the stated confidence level. Some secondary comparisons do not have a p-value or analysis method reported. The registry record also does not report an interaction test to determine whether effects differ by PD-L1 status.

These results relate to the primary outcome by adding perspectives on survival subgroups and disease progression or response. They should not override the primary full-analysis-set overall-survival comparison. In general, secondary outcomes can help describe the range of observed outcomes, but their interpretation depends on whether they were prespecified, how many comparisons were made, and how missing data and censoring were handled. Details about these features are not reported in the registry record.

Unadjusted secondary findings carry less weight as confirmatory evidence because examining many outcomes increases the chance that some results appear notable by chance. Without a reported testing hierarchy or multiplicity adjustment, the secondary analyses are best regarded as supportive or exploratory descriptions. That caution applies even when a particular point estimate looks different from the primary estimate: it does not establish a reliable treatment difference or a subgroup-specific effect.

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

• Overall Survival (OS) in PD-L1 Negative Participants
• Overall Survival (OS) in PD-L1 Positive Participants
• Progression Free Survival (PFS)
• Objective Response Rate (ORR)
• Duration of Response (DoR)
• Disease Control Rate (DCR)
• Percentage of Participants Alive and Progression Free (APF)
• Percentage of Participants Alive

9. Statistical Interpretation

10. Statistical Methods Explained

The primary endpoint is time-to-event overall survival, defined as time from randomization to death from any cause. The registry record reports group medians and sponsor-reported hazard ratios. A median gives a single point on the survival distribution, while a hazard ratio summarizes a relative event-rate comparison over follow-up. These are different summaries and should not be read as interchangeable quantities.

What does the hazard ratio represent?

In general, a hazard is the rate at which an event occurs among people who have not yet experienced it, at a given point in follow-up. A hazard ratio compares that rate between groups. A value of 1 represents equal hazards under the comparison; a value below or above 1 indicates a lower or higher estimated hazard for the first named group relative to the comparator. This interpretation relies on the analysis and its assumptions. The registry record identifies the primary comparisons as Log Rank analyses but does not report enough detail to assess the full statistical model, proportional-hazards assumption, censoring approach, or any covariate adjustment.

How should the confidence interval be read?

A confidence interval expresses uncertainty around an estimate under the statistical procedure used. In repeated use of a method, intervals constructed at a stated confidence level are designed to cover the underlying quantity at that rate, under the method’s assumptions. For an individual trial, the interval gives a range of values compatible with the observed data and analysis. It is not a probability statement that the true value lies within that particular interval. For a hazard ratio, an interval that includes 1 does not demonstrate equality; it indicates that the analysis has not isolated a relative difference from the null value at that confidence level.

Why does the analysis use time-to-event methods?

Time-to-event methods can account for when events occur and can retain information from participants whose event has not been observed by the end of their recorded follow-up. In general, such participants are censored at the last time their event status is known. The registry record does not report the censoring rules or the extent and pattern of missing follow-up. Those omissions limit an independent assessment of how follow-up information was handled. Overall survival is defined as death from any cause, but the registry record does not give further operational details about event ascertainment.

What role does the Log Rank method play?

The primary comparisons are labeled Log Rank. In general, the Log Rank test compares time-to-event experience across groups while accounting for event timing and censoring. The registry’s reported hazard ratios provide an effect estimate, while the p-values provide a test statistic’s compatibility with a null comparison under the analysis. A p-value is not the probability that the null hypothesis is true, nor does it measure the size or importance of an effect. Interpretation should consider the estimate, interval, p-value, and analysis context together.

Why can over-dispersion matter?

General explanation: over-dispersion occurs when observed variation is greater than a statistical model expects. It can make standard errors too small if the extra variation is ignored, giving intervals that are too narrow and tests that appear more certain than warranted. The registry record does not report over-dispersion as a concern or describe an analysis that would require this adjustment. It should therefore be treated as a general statistical issue, not as a finding about this trial.

What does randomization contribute?

In general, random assignment helps balance both measured and unmeasured participant characteristics across groups on average, supporting causal comparison when the trial is conducted and analyzed appropriately. The study is described as randomized, but the registry record does not report the sequence-generation method, allocation concealment, stratification, or detailed adherence and crossover information. Open-label status means masking was none, a feature relevant to decisions and assessments made during follow-up, even though overall survival is defined using death from any cause.

The registry record reports the primary analysis population as the full analysis set, inclusive of all randomized participants. It does not describe how exclusions, missing endpoint data, treatment discontinuation, or protocol deviations were handled. Those details matter because analysis populations and follow-up methods can affect the interpretation of randomized comparisons. The posted statistical summaries support a reading of the reported estimates, but not a complete audit of the trial’s analytic procedures.

11. Confidence Intervals and Precision

The primary survival estimates are accompanied by 95% confidence intervals, which show the range of values compatible with the data under the analysis method and its assumptions. For the combination versus standard-of-care hazard ratio, the interval extends from 0.85 to 1.26 around an estimate of 1.04. For durvalumab versus standard of care, it extends from 0.72 to 1.08 around an estimate of 0.88. Both intervals include 1, so neither comparison isolates a relative hazard difference from that reference value at the stated confidence level.

The median overall survival intervals also indicate uncertainty in each group’s median estimate: 5.5 to 8.2 months for the combination, 6.1 to 9.8 months for durvalumab, and 7.3 to 9.2 months for standard of care. Overlap between separate group intervals is not itself a formal test of a treatment contrast. The reported hazard ratios and their intervals are the direct comparative summaries in the primary analysis.

Precision depends on factors including the number and timing of events, follow-up, and variability in the data. The registry record provides endpoint group counts but does not report the total primary-endpoint event count, follow-up distribution, or detailed censoring information. The intervals should therefore be read as the reported uncertainty, not as a complete account of every source of uncertainty or potential bias.

Subgroup estimates are less precise when they are based on smaller populations, and their intervals in the posted results are broad. The registry record does not report a formal interaction test establishing whether the treatment comparison varies by PD-L1 status. A difference between subgroup point estimates alone would not establish effect modification.

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 sample-size figures associated with the time-to-event outcomes are total event counts, not participants per arm. They are planning illustrations based on the observed effect, rather than the original trial’s design calculations. The registry record does not report the original sample-size justification or the assumptions used to plan this trial.

The planning illustrations assume a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. They use the stated fraction of the observed effect as the effect to detect. Rows for smaller fractions of the observed effect require more total events than rows for the full observed effect. This is a general feature of study planning: smaller effects are harder to distinguish from random variation, so more information is typically needed.

The 80% power column describes the planned event count for a hypothetical design that would have an 80% chance of detecting the specified effect under the stated assumptions, if that effect were the true one and the design assumptions held. The 90% power column describes the corresponding target for a higher chance of detection. These are planning concepts, not guarantees about a completed study’s results. The listed event counts do not mean that the trial observed that many events.

These calculations are not observed power. They are illustrations derived from the reported effect and should not be used as evidence that the completed trial did or did not have adequate power. The trial’s own planning assumptions, including its original target effect and event target, are not reported in the registry record.

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 = 1.040, so ln(HR) = 0.0392.
2. d = 4 × (1.96 + 0.842)² / [ln(1.040)]² = 20,409.7, rounded up to 20,410 events in total for 80% power.

The smaller-effect rows in the table above repeat the same calculation with the effect multiplied by 0.75 and 0.5 (for ratios, applied to the logarithm of the ratio).

R code

alpha <- 0.05
z_a   <- qnorm(1 - alpha / 2)          # two-sided
hr <- 1.04                              # 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(2369874)
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 = 20,410 (80% power), 27,323 (90% power)
 75% of observed effect: events = 36,284 (80% power), 48,574 (90% power)
 50% of observed effect: events = 81,639 (80% power), 109,292 (90% power)
Simulated power with 20,410 events: 0.797 (20000 replicates)

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

14. Analyses Reported by the Sponsor

The registry record reports the sponsor’s hazard-ratio analyses for the primary comparisons, including the method, confidence intervals, and p-values. The displayed primary hazard ratios are 1.04 for durvalumab plus tremelimumab versus standard of care and 0.88 for durvalumab versus standard of care. Their reported p-values are 0.762 and 0.199, respectively.

The record indicates no discrepancy between the posted sponsor analysis and the comparison summaries. The hazard-ratio interpretation therefore follows the sponsor-reported analysis; no separate unadjusted hazard-ratio re-calculation is used in its place. The group medians are descriptive summaries and should not be treated as a substitute for the sponsor’s comparative time-to-event analysis.

In general, a re-calculation from summary statistics may not reproduce a sponsor’s analysis because it can lack participant-level event times, censoring information, stratification, covariate adjustment, or other details. The registry record does not provide those details for an independent reconstruction. The reported hazard ratios, intervals, and p-values should be read as the analysis posted in the registry, with the stated limitations rather than as a complete analytic report.

15. Multiplicity, Missing Data and Analysis Populations

The registry record lists a primary overall-survival endpoint and multiple secondary outcomes, including subgroup survival analyses and measures of progression, response, and disease control. It does not report a multiplicity adjustment, a formal testing hierarchy, or how the family of statistical tests was defined. When many outcomes or subgroup comparisons are tested, the chance of at least one apparently notable result can increase. In general, prespecified procedures such as a hierarchy or multiplicity adjustment can help control that issue. Without reported procedures, secondary p-values and findings should not be treated as if they were isolated confirmatory tests.

The primary analysis population is described as the full analysis set, inclusive of all randomized participants, regardless of PD-L1 status. The registry record does not report detailed rules for inclusion in each secondary analysis, handling of treatment discontinuation, crossover, protocol deviations, or missing outcome data. It also does not describe censoring rules, the completeness of follow-up, or sensitivity analyses for missing data. These are important for time-to-event outcomes because participants may contribute information up to a censoring time without having an observed event.

The registry record identifies the study as randomized and open-label, with masking listed as none. It does not report randomization procedures, allocation concealment, stratification factors, or analysis adjustments. Nor does it report a detailed statistical analysis plan, interim analyses, or the original sample-size justification. None of these unreported details can be inferred from the endpoint counts or the posted estimates.

These gaps do not invalidate the reported numbers, but they limit how completely an external reader can assess bias, precision, and the prespecification of analyses. The primary comparison is more directly interpretable than the collection of secondary results because it is the designated primary endpoint, yet full evaluation still depends on information not reported in the registry record.

16. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

Deaths

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: pneumonia (3.8%), tumour haemorrhage (2.8%), pneumonia aspiration (2.1%), anaemia (2.0%), general physical health deterioration (1.7%).

The registry record reports serious adverse events for 79 participants in the combination group, 69 in the durvalumab group, and 61 in the standard-of-care group. The corresponding numbers at risk for these serious-event summaries were 246, 237, and 240. For other adverse events, the affected counts were 203, 180, and 207, with the same respective at-risk counts listed. These are counts as reported; no additional rates have been calculated.

The registry record also lists serious adverse-event terms and percentages, including pneumonia at 3.8%, tumour haemorrhage at 2.8%, pneumonia aspiration at 2.1%, anaemia at 2.0%, general physical health deterioration at 1.7%, and hypercalcaemia at 1.7%. The record does not provide enough information in these summaries to assess event timing, duration, severity grading, exposure-adjusted rates, or statistical comparisons between groups.

Death counts are also listed in the safety record: 206 for the combination group, 186 for durvalumab, and 199 for standard of care. These safety counts should not be interpreted as treatment-caused deaths or as a stand-alone comparison of survival. Overall survival is reported separately as a time-to-event endpoint. Differences in exposure and follow-up, which are not detailed in the safety summary, can affect how adverse-event counts are interpreted.

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

The results describe the estimated overall-survival experience in the randomized groups and the uncertainty around the reported comparisons. The combination-versus-standard-of-care hazard ratio was 1.04, with a 95% confidence interval from 0.85 to 1.26. The durvalumab-versus-standard-of-care hazard ratio was 0.88, with a 95% confidence interval from 0.72 to 1.08. Both intervals include 1, and the reported primary p-values are 0.762 and 0.199. These results do not provide strong statistical evidence of a difference in the primary analysis.

That interpretation is not the same as proving that the treatments have identical effects. A non-significant result can reflect an effect near the null, uncertainty that spans potentially relevant effects, or both. The confidence intervals show that the estimates are not precise enough to rule out a range of relative effects. The registry record does not report a prespecified equivalence or non-inferiority framework, so the results should not be framed as demonstrating equivalence or non-inferiority.

The medians summarize each group’s survival distribution but do not capture every feature of the survival curves. The hazard ratios are comparative summaries and are not differences in median survival. Neither statistic alone describes an individual participant’s likely course or establishes a treatment effect for people outside the study population.

Secondary outcomes and subgroup analyses add descriptive information, but the registry record does not report a multiplicity procedure or a formal interaction analysis for PD-L1 status. Those results therefore do not establish subgroup-specific effects. Likewise, adverse-event counts describe reported safety events but do not, on their own, establish causation or provide a fully adjusted comparison.

The interpretation is limited to the population, endpoint definitions, and analyses recorded in the registry. It is not medical advice and should not be used to make an individual treatment decision.

18. Why This Trial Matters Statistically

This trial is statistically informative because it compares two treatment strategies with a standard-of-care group in a randomized, parallel-group design and specifies overall survival as its primary outcome. Randomization is an important design feature for comparative inference because, in general, it reduces systematic differences between groups at assignment. The open-label design and the unreported details of allocation and follow-up remain relevant when assessing the total evidence.

The endpoint is time-to-event, so the analysis uses more information than a single status count at one time point. The posted medians and hazard ratios offer complementary summaries: medians describe a location in each group’s survival distribution, while hazard ratios compare event rates over follow-up under the analysis. Their intervals make visible the uncertainty that point estimates alone would conceal.

The primary comparisons have intervals that include the no-difference reference for the hazard ratio. Statistically, this highlights the distinction between an observed estimate and a precise, decisive conclusion. The findings do not establish equality, and they do not support a strong claim of a clear survival advantage in either comparison. The secondary endpoints expand the picture but need cautious interpretation because multiplicity procedures and several analysis details are not reported.

The registry record also illustrates why results need to be read alongside design and analysis information. Group counts, confidence intervals, and p-values are useful, but they cannot replace details about censoring, missing data, analysis populations, original planning assumptions, and prespecified testing. Recognizing what is reported and what remains unspecified is part of sound biostatistical interpretation.

19. Limitations

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

The registry record does not report the specific standard-of-care regimen, the randomization method, allocation concealment, stratification factors, or the detailed statistical analysis plan. It also does not report the trial’s original sample-size justification or design assumptions. Planning illustrations based on observed effects are not a substitute for those original design details.

For overall survival, the record defines the endpoint and reports sponsor analyses, but does not describe censoring rules, follow-up completeness, event adjudication, or sensitivity analyses. These details matter for evaluating time-to-event results. The primary group counts are endpoint-specific, and should not be mistaken for complete information about every analysis population or the amount of follow-up available.

The registry lists multiple secondary outcomes and subgroup analyses but does not report a multiplicity adjustment or hierarchy. Some secondary comparisons lack a reported p-value or method, and no formal interaction test for PD-L1 status is reported. The subgroup results should therefore be interpreted cautiously and cannot establish that treatment effects differ by biomarker status.

The study was open-label, with masking listed as none. The record does not describe safeguards for decisions that may be influenced by knowledge of assigned treatment. In the safety summaries, counts are available, but the record does not provide exposure-adjusted comparisons, detailed event timing, or enough context to assess causal attribution. These limitations constrain interpretation of the posted results; they do not change the numbers recorded for the analyses.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: Study of MEDI4736 Monotherapy and in Combination With Tremelimumab Versus Standard of Care Therapy in Patients With Head and Neck Cancer — NCT02369874.
• Registry record last updated 2021-02-10; status last verified 2021-01.
• Results first posted 2019-10-03.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

23. Record Summary

The registry record describes a completed randomized, open-label, parallel-group Phase III study of durvalumab, durvalumab plus tremelimumab, and standard of care in recurrent or metastatic squamous cell carcinoma of the head and neck. Enrollment is recorded as 736. The primary endpoint was overall survival in the full analysis set, regardless of PD-L1 status.

The sponsor-reported primary hazard ratios were 1.04 for the combination versus standard of care and 0.88 for durvalumab versus standard of care. Both 95% confidence intervals include 1, and the reported p-values are 0.762 and 0.199. The registry also reports secondary survival, progression, response, and safety outcomes. Interpretation is limited by unreported details about the original design assumptions, missing data, censoring, and multiplicity procedures.