A Global Study to Assess the Effects of MEDI4736 Following Concurrent Chemoradiation in Patients With Stage III Unresectable Non-Small Cell Lung Cancer: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Global Study to Assess the Effects of MEDI4736 Following Concurrent Chemoradiation in Patients With Stage III Unresectable Non-Small Cell Lung Cancer (NCT02125461), a randomized phase 3 study. The primary outcome is Progression Free Survival Based on Blinded Independent Central Review (BICR) According to Response Evaluation Criteria in Solid Tumors (RECIST 1.1).

2. Clinical Question

3. Trial Design

The registry describes a phase III, randomized, double-blind, placebo-controlled, parallel-group study. Its official title also describes the study as multi-centre and international. Masking is recorded as quadruple. The trial enrolled 713 patients and is marked completed. AstraZeneca is listed as the sponsor.

The study addressed sequential treatment after definitive, platinum-based concurrent chemoradiation in patients with locally advanced, unresectable stage III non-small cell lung cancer who had not progressed. The treatment groups received intravenous MEDI4736 or matching placebo. The registry record does not report the dose, dosing schedule, treatment duration, or detailed eligibility criteria beyond the population described.

The primary outcome was progression-free survival assessed by blinded independent central review according to RECIST 1.1. The full analysis set included all randomized patients, analyzed on an intent-to-treat basis. Tumor scans were scheduled at baseline and approximately every 8 weeks up to 48 weeks, then approximately every 12 weeks until confirmed progression. The primary outcome was assessed through the 13 Feb 2017 data cutoff, up to a maximum of approximately 3 years.

The recorded study start was 2014-05-07, and the primary completion date was 2017-02-13. These dates describe the registry record; they do not by themselves establish when every participant began or completed treatment. The original statistical design assumptions, including the planned treatment duration and sample-size justification, 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: Eligible patients with locally advanced, unresectable Stage III non-small cell lung cancer were randomized in a 2:1 ratio to receive either durvalumab (MEDI4736) 10 milligrams (mg) / kilogram (kg) every 2 weeks (Q2W) or placebo.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: FAS included all randomized patients, analyzed on an intent-to-treat (ITT) basis.

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: Baseline analysis was based on the FAS consisting of all randomized patients.

6. Primary Endpoint

The registry lists the primary outcome as Progression Free Survival Based on Blinded Independent Central Review (BICR) According to Response Evaluation Criteria in Solid Tumors (RECIST 1.1), measured over Tumor scans performed at baseline then every ~8 weeks up to 48 weeks, then every ~12 weeks thereafter until confirmed disease progression. Assessed until 13 Feb 2017 DCO; up to a maximum of approximately 3 years..

Progression-free survival (PFS) measures the time from randomization until objective disease progression or death from any cause when progression has not been documented. In this study, progression was defined using RECIST 1.1: an increase in the sum of the longest diameters of target lesions, a measurable increase in a non-target lesion, or the appearance of new lesions. Tumor status was assessed by blinded independent central review.

PFS is a time-to-event outcome. It incorporates both when an event occurs and whether a participant remains free of a recorded event during follow-up. The registry reports median PFS for each group, estimated using the Kaplan-Meier technique, and a hazard ratio from the sponsor's log-rank analysis. The median describes the estimated time by which half of the group has experienced progression or death; it is not an average survival time and does not describe every participant's course.

Using PFS can capture disease progression before death, which makes it a different outcome from overall survival. Its interpretation depends on the outcome definition, assessment schedule, censoring, and review process. The registry specifies the definition and scan schedule, but does not report detailed handling of missing scans or censoring in the record.

7. Results Posted in the Registry

• Durvalumab (MEDI4736) vs Placebo: hazard ratio 0.52 (95% CI 0.42–0.65), reported p < 0.001 (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.

Objective Response Rate (ORR) Based on BICR Assesments According to RECIST 1.1

• 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:
  • Durvalumab (MEDI4736): 30.0% (133/443) versus 17.8% (38/213) on control; risk difference 12.2 percentage points (95% CI 5.5 to 18.9), risk ratio 1.68 (1.22–2.32); unadjusted two-proportion z-test p = <0.001.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the proportion with the event is statistically higher for Durvalumab (MEDI4736) than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Proportion of Patients Alive and Progression Free at 12 Months From (APF12) Based on BICR Assessments According to RECIST 1.1

• 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:
  • Durvalumab (MEDI4736): 55.9% (266/476) versus 35.4% (84/237) on control; risk difference 20.4 percentage points (95% CI 12.9 to 28.0), risk ratio 1.58 (1.30–1.91); unadjusted two-proportion z-test p = <0.001.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the proportion with the event is statistically higher for Durvalumab (MEDI4736) than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Proportion of Patients Alive and Progression Free at 18 Months From (APF18) Based on BICR Assessments According to RECIST 1.1

• 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:
  • Durvalumab (MEDI4736): 44.1% (210/476) versus 27.0% (64/237) on control; risk difference 17.1 percentage points (95% CI 9.9 to 24.3), risk ratio 1.63 (1.29–2.06); unadjusted two-proportion z-test p = <0.001.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the proportion with the event is statistically higher for Durvalumab (MEDI4736) than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

The posted secondary outcomes cover several related but distinct aspects of disease control: objective response, and the proportions alive and progression free at specified follow-up times. The reported group estimates generally favor the MEDI4736 group in direction, in keeping with the primary PFS comparison. These endpoints complement rather than replicate PFS: response is based on tumor assessment, while a fixed-time alive-and-progression-free proportion combines survival and progression status at a stated time.

The secondary outcomes also include duration of response, time to distant metastasis or death, survival at a specified time, time to second progression or death, and health-related quality of life. The registry lists these outcomes, but the record does not provide numerical results for all of them. Their status should not be inferred from the available results for other outcomes.

Secondary analyses generally carry less confirmatory weight when multiple outcomes are assessed. Without a reported multiplicity procedure, separate unadjusted p-values do not control the chance of false-positive findings across the family of analyses. The registry record does not report whether a prespecified hierarchy or adjustment was used, so the strength of any confirmatory interpretation for secondary outcomes is uncertain.

The posted secondary estimates are useful for describing the pattern of measured outcomes, but they should be interpreted alongside their endpoint definitions, follow-up windows, analysis populations, and uncertainty intervals. They do not independently establish a broad effect across all outcomes, nor do they replace the sponsor-reported primary time-to-event result.

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

• Objective Response Rate (ORR) Based on BICR Assesments According to RECIST 1.1
• Duration of Response (DoR) Based on BICR Assessments According to RECIST 1.1
• Proportion of Patients Alive and Progression Free at 12 Months From (APF12) Based on BICR Assessments According to RECIST 1.1
• Proportion of Patients Alive and Progression Free at 18 Months From (APF18) Based on BICR Assessments According to RECIST 1.1
• Time to Death or Distant Metastasis (TTDM) Based on BICR Assessments According to RECIST 1.1
• Percentage of Patients Alive at 24 Months (OS24)
• Time to Second Progression or Death (PFS2)
• Time to Deterioration of Global Health Status / Health-Related Quality of Life (HRQoL), Assessed Using European Organization for Research and Treatment of Cancer 30-Item Core Quality of Life Questionnaire (EORTC QLQ-C30)

9. Statistical Interpretation

10. Statistical Methods Explained

What does a time-to-event outcome capture?

PFS records the time from randomization to progression or death, rather than simply counting whether an event ever occurred. Participants without a recorded event by the analysis cutoff contribute information up to the point they were last assessed under the analysis rules. This allows the analysis to use follow-up of different lengths, although the result depends on how censoring and missing assessments are handled. Those detailed rules are not reported in the registry record.

How is the Kaplan-Meier median obtained?

The Kaplan-Meier method estimates the proportion remaining free of the event over time while accounting for censoring. The median PFS is the time at which this estimated proportion reaches one half. It is a summary of the estimated event-time distribution, not the arithmetic average of individual follow-up times. The registry reports median estimates and 95% confidence intervals for the two groups.

What does the hazard ratio describe?

A hazard is the instantaneous event rate among participants who have not yet had the event, considered at a point in follow-up. The hazard ratio compares that rate between groups over the analysis. A value below 1 means the estimated rate is lower for the first-named group under the comparison as defined. It does not mean that the probability of progression or death is reduced by the same proportion, and it is not a ratio of median times. The sponsor-reported estimate is 0.52, with a 95% confidence interval from 0.42 to 0.65.

The registry identifies the sponsor's comparison method as log-rank. In general, a log-rank test compares event-time patterns across groups while accounting for the timing of events and censored observations. The registry reports p < 0.001 for the primary comparison. The precise implementation details and any additional model adjustments are not reported in the registry record.

What does a confidence interval add?

A confidence interval conveys uncertainty around an estimate. In repeated use of the same method under its assumptions, intervals constructed this way would cover the target quantity at the stated rate. A 95% interval is not a statement that there is a 95% probability that the particular fixed effect lies inside the calculated interval. The interval's width reflects statistical precision, which is influenced by the amount and distribution of information available and by the analysis method.

How should binary outcomes be understood?

For outcomes such as objective response and the alive-and-progression-free proportions, the registry reports percentages and confidence intervals. The posted recalculations for these outcomes include unadjusted risk differences and p-values. A risk difference compares the two group proportions on an absolute scale; it is not a time-to-event measure. For objective response, the sponsor's listed analysis uses Fisher's exact test and reports p < 0.001. The registry does not list a sponsor-reported test for the posted fixed-time alive-and-progression-free outcomes.

Unadjusted analyses do not account for other variables unless the method specifically includes them. The registry does not report adjusted secondary estimates in the record. The recalculated binary comparisons are therefore descriptive, unadjusted comparisons based on the posted group summaries; the sponsor's own analysis takes precedence when a discrepancy is identified.

Why does multiplicity matter?

When many outcomes or comparisons are examined, the chance of at least one apparently unusual result can rise if each is assessed separately without adjustment. This is the general reason that secondary analyses are usually interpreted with less confirmatory weight than a prespecified primary outcome. The registry lists multiple secondary outcomes and reports no multiplicity procedure in the record. The detailed statistical plan, including any prespecified hierarchy or adjustment, is not reported in the registry record.

The primary time-to-event comparison is reported by the sponsor, and the record marks no discrepancy. The secondary binary recalculations should not be treated as replacements for the sponsor's primary inferential analysis. A small p-value is not a substitute for examining the effect estimate, its interval, the endpoint definition, and the analysis population.

11. Confidence Intervals and Precision

The primary PFS estimates are reported with 95% confidence intervals. For MEDI4736, the median was 16.8 months, with an interval from 13.0 to 18.1 months. For placebo, the median was 5.6 months, with an interval from 4.6 to 7.8 months. These intervals show the uncertainty around each group's median estimate; they do not give the range of individual participant outcomes.

The sponsor-reported hazard ratio was 0.52, with a 95% confidence interval from 0.42 to 0.65. The interval lies below 1, consistent with the direction of the reported comparison. Its limits describe a range of effect estimates compatible with the data under the analysis assumptions and method. The interval does not capture every possible source of uncertainty, such as limitations in outcome measurement or unreported analysis decisions.

Precision is distinct from statistical significance. A p-value addresses compatibility with a null hypothesis, whereas a confidence interval communicates the scale and uncertainty of an estimate. The primary p-value is reported as p < 0.001. The interval provides more detail about the plausible range of the hazard ratio than the p-value alone.

The intervals for the group medians should not be used as a substitute for a confidence interval for the difference between medians. The registry does not report such a contrast. Likewise, the hazard ratio interval pertains to the sponsor-reported time-to-event comparison and should not be interpreted as an interval for a difference in months. For the secondary outcomes, the posted group-level intervals and unadjusted comparisons have their own estimands and should be read separately.

Confidence intervals depend on the method and assumptions used to construct them. The registry identifies the primary outcome's Kaplan-Meier method and the sponsor's log-rank comparison, but does not report all implementation details. The limits are therefore best read as precision measures for the stated analysis, not as protection against every limitation of the record.

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 figures are illustrations based on the observed effect, not a report of the trial's original design calculation. For the primary time-to-event outcome, the figures are total events required, not participants per arm. The listed assumptions are a two-sided significance level of 0.05, equal allocation between groups, a normal approximation, and an effect set to the stated fraction of the observed effect; the time-to-event illustration uses the Schoenfeld method.

For the binary outcomes, the planning figures are participants per arm. The 80% power column describes the approximate number per arm under the stated assumptions for a design with an 80% chance of detecting the assumed effect if that effect is real. The 90% power column describes the corresponding planning target for a 90% chance. These are design concepts, not guarantees about a future study's result.

The rows for 100% of the observed effect use the full observed effect as the planning target. Rows using 75% or 50% of the observed effect assume a smaller effect to detect. Smaller effects are harder to distinguish from random variation, so they require more participants or events under the same planning assumptions. The figures are not observed power. They do not tell how likely the completed trial was to detect its actual effect, and they should not be treated as a retrospective measure of trial quality.

The original trial's own sample-size justification and design assumptions are not reported in the registry record. These planning illustrations use a simplified framework and should not be read as evidence that the original protocol used these assumptions or targets.

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 = 0.520, so ln(HR) = -0.6539.
2. d = 4 × (1.96 + 0.842)² / [ln(0.520)]² = 73.4, rounded up to 74 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 <- 0.52                              # 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(2125461)
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 = 74 (80% power), 99 (90% power)
 75% of observed effect: events = 131 (80% power), 175 (90% power)
 50% of observed effect: events = 294 (80% power), 394 (90% power)
Simulated power with 74 events: 0.784 (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 112 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 <- 5.6          # control-arm median, months (converted from the posted unit)
hr          <- 0.52          # 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: 74         p (1-sided)   0.0001
 Events: 33        ~HR at bound   0.2739
  Month: 16    P(Cross) if HR=1   0.0001
            P(Cross) if HR=0.52   0.0338
  IA 2: 67%                   Z   2.5114
     N: 112         p (1-sided)   0.0060
 Events: 66        ~HR at bound   0.5380
  Month: 25    P(Cross) if HR=1   0.0060
            P(Cross) if HR=0.52   0.5603
      Final                   Z   1.9930
     N: 112         p (1-sided)   0.0231
 Events: 99        ~HR at bound   0.6692
  Month: 60    P(Cross) if HR=1   0.0250
            P(Cross) if HR=0.52   0.9000



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

15. Analyses Reported by the Sponsor

For the primary outcome, the registry reports the sponsor's log-rank hazard ratio of 0.52, its 95% confidence interval from 0.42 to 0.65, and p < 0.001. The posted record indicates no discrepancy between the sponsor's primary result and the analysis represented for that endpoint. The sponsor's own primary analysis is therefore the relevant reported inferential result; no separate unadjusted re-calculated hazard ratio is presented.

For objective response rate, the sponsor's listed analysis is a Fisher exact test with p < 0.001. The record also contains unadjusted recalculations for binary outcomes, including risk differences and corresponding intervals. These recalculations are derived from posted summary statistics and are not a substitute for the sponsor's full analysis. For the 12- and 18-month alive-and-progression-free outcomes, no sponsor-reported analysis is listed in the record.

Where a sponsor analysis and a simple recalculation differ, the sponsor's own analysis takes precedence. The registry discrepancy indicator is false for this record, but that does not establish that every analysis detail is available. The registry does not report all statistical-plan specifications or implementation details needed to reproduce the sponsor's analyses in full.

16. Multiplicity, Missing Data and Analysis Populations

The registry record lists multiple secondary outcomes, including response, fixed-time progression-free status, time-to-event outcomes, and health-related quality of life. Examining several outcomes creates a multiplicity issue: when each is tested separately, the chance of at least one statistically significant result by chance can increase. This is a general statistical concern, not a claim that any particular posted result is false.

The registry record does not report a multiplicity adjustment, a hierarchical testing strategy, or the complete prespecified testing plan. It is therefore not possible from this record to determine whether secondary p-values were intended as confirmatory, how the family of outcomes was defined, or whether any adjustment was applied outside the listed results. Unadjusted secondary comparisons should be read as exploratory or descriptive unless further analysis-plan information establishes otherwise.

The primary analysis population is described as the full analysis set, including all randomized patients and analyzed on an intent-to-treat basis. This indicates that the primary endpoint was analyzed according to randomized group, as stated in the registry. The record does not provide comparable population details for every secondary endpoint in the available results.

Missing-data handling, censoring rules beyond the outcome description, and sensitivity analyses are not reported in the registry record. For time-to-event outcomes, these details matter because participants may have differing follow-up and some may not have a recorded event by the cutoff. For binary outcomes, the treatment of participants with unavailable assessments can affect denominators and estimates. The registry's posted counts and estimates should be read in light of these unreported details, rather than assuming a particular method.

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: pneumonia (6.9%), pneumonitis (3.6%), radiation pneumonitis (3.6%), atrial fibrillation (1.1%), chronic obstructive pulmonary disease (1.1%).

The registry reports serious adverse events among 138 of 475 participants at risk in the MEDI4736 group and 54 of 234 participants at risk in the placebo group. It also reports other adverse events affecting 436 of 475 participants at risk in the MEDI4736 group and 212 of 234 at risk in the placebo group. The record separately lists 262 deaths in the MEDI4736 group and 154 in the placebo group.

These are descriptive counts with the stated at-risk denominators. They should not be interpreted as proof that treatment caused or prevented an event. The registry summary does not supply the context needed to attribute each event, assess timing or exposure, or compare risk while accounting for follow-up. The efficacy and adverse-event denominators differ, and the record does not explain every denominator difference.

The most frequently listed serious adverse-event terms include pneumonia, pneumonitis, radiation pneumonitis, atrial fibrillation, chronic obstructive pulmonary disease, and diarrhoea. The listed percentages describe the registry's event reporting; they do not establish comparative causation. A complete clinical safety assessment would require the full safety reporting context, which is not reported in this record.

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

The primary result is an estimate from a randomized comparison of MEDI4736 with placebo after concurrent chemoradiation in the population described by the registry. The sponsor-reported hazard ratio and median PFS estimates describe the observed time-to-event pattern under the stated definition and follow-up. They support a statistical interpretation of the comparison in this trial, not a guarantee about an individual participant's outcome.

The hazard ratio is a relative measure of event rates over follow-up, and the median is a group-level summary. Neither gives the probability that a particular person will benefit, the length of time any individual will remain progression free, or a direct estimate of overall survival. The registry includes secondary outcomes related to response, survival, progression, and quality of life, but these are separate questions and not all have results reported in the record.

The results apply most directly to patients meeting the trial's described population criteria: locally advanced, unresectable stage III non-small cell lung cancer without progression after definitive platinum-based concurrent chemoradiation. The registry does not provide enough information to assess how results may differ across patient subgroups or settings outside that population.

Statistical significance does not by itself establish clinical importance, explain the balance of benefits and harms, or remove uncertainty. The confidence intervals convey uncertainty for the reported estimates, while unreported details—such as full analysis-plan specifications, missing-data handling, and the broader safety context—limit what can be concluded from the registry record alone. No medical recommendation follows from these statistical results.

19. Why This Trial Matters Statistically

This trial is statistically notable for combining a randomized, placebo-controlled design with a time-to-event primary outcome assessed by blinded independent central review. The outcome definition links progression and death, and the registry describes a scheduled imaging approach and an intent-to-treat primary population. Those features help make the comparison and its target population explicit.

The primary result illustrates why time-to-event analyses use more than a comparison of group medians. The sponsor's hazard ratio summarizes relative event rates over follow-up, while Kaplan-Meier medians provide a separate, readily interpretable summary of the event-time distributions. Their confidence intervals communicate uncertainty on different scales. Interpreting both together avoids treating a single summary as a complete account of the data.

The record also illustrates an important distinction between a sponsor-reported analysis and a simple recalculation from summary statistics. A recalculation can help readers understand an unadjusted comparison, but it may not reproduce the full sponsor analysis or its protocol decisions. For the primary time-to-event outcome, the sponsor's reported result is the relevant analysis in the record.

Finally, the trial demonstrates why secondary outcomes and planning illustrations need careful framing. Related endpoints can add context, but multiple unadjusted tests are less persuasive as confirmatory evidence. Planning figures based on an observed effect illustrate how required information changes when a target effect is smaller; they are not observed power and do not reveal the original design assumptions.

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 fuller statistical appraisal are not reported in the registry record. These include the original sample-size justification, detailed statistical analysis plan, any interim analyses, full censoring and missing-data procedures, and whether secondary outcomes followed a prespecified hierarchy or multiplicity adjustment.

The primary endpoint used blinded independent central review under RECIST 1.1, but the record does not provide the operational details for adjudication, handling of disputed assessments, or how missed or unevaluable scans entered the analysis. The scheduled scans and data cutoff provide useful context, but do not by themselves establish the amount or pattern of missing follow-up.

The registry provides selected secondary outcome results rather than numerical results for every listed secondary endpoint. It also does not provide detailed subgroup findings in the record. As a result, the available results cannot establish whether the primary pattern was consistent across participant subgroups or across all secondary measures.

The safety summary presents event counts and selected serious adverse-event terms, but does not give enough context to assess exposure-adjusted rates, event timing, attribution, or the full balance of harms. The registry record also does not report treatment dose or duration. These omissions constrain interpretation of both the efficacy and safety summaries; they do not establish that the information is absent from all study materials.

21. Related Tutorials

22. Related Statistical Calculators

23. Sources

• ClinicalTrials.gov: A Global Study to Assess the Effects of MEDI4736 Following Concurrent Chemoradiation in Patients With Stage III Unresectable Non-Small Cell Lung Cancer — NCT02125461.
• Registry record last updated 2023-10-10; status last verified 2023-09.
• Results first posted 2019-01-30.
• 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

NCT02125461 is a completed phase III randomized, double-blind, placebo-controlled, parallel-group study sponsored by AstraZeneca. It enrolled 713 patients with stage III unresectable non-small cell lung cancer who had not progressed after definitive platinum-based concurrent chemoradiation. Participants received intravenous MEDI4736 or matching placebo.

The primary endpoint was PFS by blinded independent central review using RECIST 1.1, analyzed in the full analysis set on an intent-to-treat basis. The sponsor reported a hazard ratio of 0.52 (95% confidence interval 0.42 to 0.65; p < 0.001). Median PFS was 16.8 months in the MEDI4736 group and 5.6 months in the placebo group. The record lists multiple secondary outcomes and selected safety counts, while detailed design assumptions, multiplicity procedures, and several analysis details are not reported in the registry record.