AZD9291 Versus Placebo in Patients With Stage IB-IIIA Non-small Cell Lung Carcinoma, Following Complete Tumour Resection With or Without Adjuvant Chemotherapy.: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for AZD9291 Versus Placebo in Patients With Stage IB-IIIA Non-small Cell Lung Carcinoma, Following Complete Tumour Resection With or Without Adjuvant Chemotherapy. (NCT02511106), a randomized phase 3 study. The primary outcome is Assess the Efficacy of AZD9291 Compared to Placebo as Measured by Disease Free Survival (DFS)..

2. Clinical Question

The study is a phase III, randomized, double-blind, placebo-controlled comparison with a parallel-group design. Its stated aim is to assess efficacy and safety after complete tumour resection. The endpoint definition counts death from any cause in the absence of documented recurrence as an event, so it captures more than recurrence alone. The registry record describes the primary analysis population as stage IIA-IIIA patients, while the broader trial description includes stage IB-IIIA disease. Results for one population should not automatically be generalized to every stage included in the trial description.

3. Trial Design

This phase III trial compared AZD9291 with a matching placebo in patients with epidermal growth factor receptor mutation-positive stage IB-IIIA non-small cell lung carcinoma after complete tumour resection, with or without adjuvant chemotherapy. The study used randomized allocation, a parallel-group model, and triple masking. The registry describes the study as double-blind in its official title; its masking field records triple masking. The active arm received AZD9291 80 mg or 40 mg orally once daily, and the comparator arm received matching placebo on the same schedule.

The registry records enrollment of 682 and lists AstraZeneca as sponsor. The recorded start date is 2015-10-21, and the primary completion date is 2022-04-11. The study status is active, not recruiting. These registry-level details describe the trial as a whole; the primary endpoint results are reported for the Full Analysis Set of stage IIA-IIIA patients. The arm sizes reported with those results are 233.0 for AZD9291 and 237.0 for placebo. They should not be treated as a replacement for the overall enrollment figure, since they refer to the primary analysis groups.

The primary outcome is disease-free survival, defined from randomization until recurrence or death in the absence of recurrence. The stated assessment period is up to approximately 5 years after the first patient is randomized, with maximum follow-up of 70 months. Overall survival is also listed as a secondary outcome, with follow-up described as up to approximately 7 years after the first patient is randomized, with maximum follow-up of 86 months. The registry record does not report the original design assumptions in enough detail to reconstruct the trial’s planned sample-size rationale or its full statistical plan.

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: During the 28 day screening period, participants were enrolled based on the presence of at least 1 of the 2 most frequent Epidermal growth factor receptor (EGFR) mutations in their tumor. At the time of enrolment, all participant were required to have central confirmation of EGFR mutation status per eligibility criteria (Exon 19 deletion or L858R mutation).

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Full Analysis Set (FAS) stage IIA-IIIA patients

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 Assess the Efficacy of AZD9291 Compared to Placebo as Measured by Disease Free Survival (DFS)., measured over Up to approximately 5 years after the first patient is randomized (maximum follow up of 70 months).

The primary endpoint is disease-free survival (DFS), measured in months. For each patient, the registry defines DFS as the time from randomization to disease recurrence or to death from any cause when recurrence has not occurred. This definition combines recurrence and death into a time-to-event outcome. It therefore asks how long patients remain alive without recorded recurrence, rather than measuring recurrence alone.

The registry reports median DFS for each group and a hazard ratio comparing AZD9291 with placebo. A median is a summary of the event-time distribution: it marks the time by which the estimated event experience has reached the halfway point. A hazard ratio instead compares the event rates over follow-up, taking account of the timing of events and the time participants remain under observation. These summaries answer related but different questions. Neither should be read as a direct estimate of an individual patient’s outcome.

General: Time-to-event endpoints are useful when the timing of an event matters and participants may have different lengths of follow-up. They can incorporate information from participants who have not yet experienced the event by the end of their observed follow-up, provided the analysis handles censoring appropriately. The registry record does not report the detailed censoring rules or missing-data procedures. The endpoint’s stated timeframe is up to approximately 5 years after the first patient is randomized, with maximum follow-up of 70 months; this is a follow-up description, not a guarantee that every participant was observed for the same duration.

7. Results Posted in the Registry

• AZD9291 80mg Tablet vs Placebo Tablet: hazard ratio 0.23 (95% CI 0.18–0.30), 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.

Overall Survival (OS)

• 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:
  • AZD9291 80mg Tablet vs Placebo Tablet: hazard ratio 0.49 (95% CI 0.33–0.73), reported p < 0.001 (as reported in the registry).
• Interpretation:
  • The 95% confidence interval excludes no difference (a ratio of 1), so the hazard of the event is statistically lower for AZD9291 80mg Tablet vs Placebo Tablet than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

The secondary outcomes cover several kinds of evidence: overall survival, DFS rates at specified follow-up times, health-related quality of life and symptoms measured by the SF-36v2 Health Survey, and plasma concentrations of AZD9291 and its listed metabolites. These outcomes do not all address the same question. Survival endpoints assess time to events, patient-reported measures concern health status and symptoms, and plasma concentration measures describe drug or metabolite exposure.

The posted overall-survival comparison points in the same general direction as the primary DFS comparison. That directional alignment adds context, but it does not make the secondary endpoint equivalent to the primary endpoint. Overall survival has its own event definition, follow-up period, analysis, and uncertainty. Median overall-survival values are not reported in the registry record, so the secondary result cannot be expressed as a median time comparison.

Secondary results generally carry less confirmatory weight when many outcomes or analyses are considered and no multiplicity adjustment is reported. Testing more outcomes creates more opportunities for apparently strong results by chance. The registry record does not report the multiplicity strategy, so it is not possible to determine from the record whether the secondary tests were adjusted or how their inferential status was defined. Their results are best read as additional evidence, with attention to outcome type, analysis, and uncertainty, rather than as independent confirmation of every claim.

The registry lists 12 outcome measures and 10 posted secondary outcomes. A listed outcome does not necessarily have a complete numerical result in the record. In particular, missing endpoint values and unreported analysis details constrain how comprehensively the secondary evidence can be assessed. No conclusions about outcomes absent from the posted record should be inferred from the results that are available.

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

• Disease Free Survival (DFS) Rate at 2, 3 and 5 Years
• Overall Survival (OS)
• Overall Survival Rate at 2, 3 and 5 Years
• Patient Health-related Quality of Life and Symptoms (HRQoL) by SF-36v2 Health Survey.
• Plasma Concentrations of AZD9291
• Plasma Concentrations of AZ5104 Metabolites
• Plasma Concentrations of AZ7550 Metabolites

9. Statistical Interpretation

10. Statistical Methods Explained

The primary outcome is a time-to-event measure, and the registry reports both group medians and a hazard ratio. The median summarizes the estimated event-time distribution separately in each arm. The hazard ratio compares the event hazard between arms over the observed follow-up. These statistics are complementary: the medians give time units that can be read for each group, while the hazard ratio expresses a relative comparison across time. They do not provide the same information, and neither alone describes the entire pattern of event times.

What does a hazard ratio describe?

General: A hazard is the instantaneous event rate among people who have not yet had the event and remain under observation. A hazard ratio compares that rate between groups. A value below 1 indicates a lower estimated hazard in the first group named relative to the comparator; a value above 1 indicates a higher estimated hazard. The interpretation depends on the endpoint, group ordering, and analysis. A hazard ratio is not a risk ratio, an odds ratio, or a direct comparison of the proportion of participants who had an event by a particular date.

The reported primary hazard ratio is 0.23, with AZD9291 compared with placebo. The registry identifies the comparison method as Log Rank and reports a 95% confidence interval. The record labels the reported analysis as a superiority analysis. No additional adjusted hazard-ratio estimate is reported in the registry record. The estimate should be understood as the sponsor-reported time-to-event comparison, not as a newly calculated effect from the median values.

How does the log-rank method fit?

General: The log-rank method compares event-time experience between groups while accounting for the ordering and timing of events and the number of participants still at risk over follow-up. It is commonly used for time-to-event comparisons. The registry identifies Log Rank as the method for the reported primary and secondary hazard-ratio analyses. The record does not give sufficient detail to reconstruct the full analysis procedure, such as the exact handling of censoring, stratification, or any model-based adjustments. Those details should not be inferred from the method name alone.

In a time-to-event analysis, participants who have not had the event by the end of their observed follow-up may contribute information up to that point. Their follow-up is generally treated as censored rather than as an event. General: Censoring allows analyses to use incomplete follow-up, but interpretation relies on assumptions about the relationship between censoring and event experience. The registry record does not report the censoring rules or the checks used to evaluate those assumptions. It also does not report detailed procedures for missing data.

What does the confidence interval add?

General: A confidence interval gives a range of values compatible with the data and the statistical method under its assumptions. It describes uncertainty in an estimated quantity, not a range containing a fixed proportion of individual outcomes. For the hazard ratio, an interval that excludes 1 is conventionally consistent with evidence of a difference under the corresponding test framework. The interval’s width also matters: a narrow interval indicates greater precision than a wide interval, all else equal. The confidence interval does not address unmeasured sources of bias or establish that the analysis assumptions are correct.

The reported p-value addresses how compatible the observed test result is with the null hypothesis under the specified statistical framework. It is not the probability that the null hypothesis is true, nor does it measure the magnitude or clinical importance of an effect. The registry reports the primary p-value as below < 0.001 and the overall-survival p-value as < 0.001. Their interpretation is tied to the reported analyses; the registry record does not report whether or how the tests were adjusted for multiplicity.

The primary analysis population is described as the Full Analysis Set of stage IIA-IIIA patients. The registry does not provide a fuller definition of that set in the endpoint record, so its precise inclusion rules should not be assumed. The broader trial population is described as stage IB-IIIA, which means the endpoint analysis population and the trial’s overall eligibility description are not identical. The available summary does not provide enough information to evaluate all population-specific exclusions or the handling of participants without endpoint follow-up.

Overall survival is a secondary time-to-event endpoint. Its reported hazard ratio is sponsor-reported, and the registry provides a confidence interval and p-value, but the group median values are absent. Other listed secondary outcomes include DFS rates at specified follow-up times, health-related quality of life and symptoms measured by the SF-36v2 Health Survey, and plasma concentrations of AZD9291 and metabolites. The registry lists 12 outcome measures, 10 of which are posted as secondary outcomes. The presence of an outcome in the registry does not mean that every numerical result or analysis detail is available in the record.

The registry record does not report the complete statistical analysis plan, the original sample-size justification, whether interim analyses were planned or conducted, or the detailed multiplicity strategy. It also does not state the censoring and missing-data procedures in the endpoint information. These absences limit independent assessment of the analyses, but they do not alter the estimates that the registry reports. General statistical concepts can clarify what the reported measures mean; they cannot fill in unreported design or analysis decisions.

11. Confidence Intervals and Precision

The primary DFS medians are reported with 95% confidence intervals. For AZD9291, the median is 65.8 months, with a lower confidence limit of 54.4 months; an upper limit is not reported in the registry record. For placebo, the median is 21.9 months, with limits of 16.6 and 27.5 months. The intervals communicate estimation uncertainty around the medians. The missing upper limit for the AZD9291 median means that the record does not provide a complete two-sided interval for that estimate.

The hazard-ratio estimate for the primary comparison is 0.23, with a 95% confidence interval from 0.18 to 0.30. This interval is entirely below 1, so the reported data are compatible with a lower event hazard in the AZD9291 group under the analysis method. The interval’s limits also show that the effect is not estimated as a single exact value. It does not capture every possible source of uncertainty, including potential bias or unreported analysis choices.

General: A confidence interval is a measure of statistical precision under the assumptions of the analysis. It should not be read as a range containing 95% of participants’ individual outcomes, or as a guarantee that the true value lies inside the observed interval. Precision and the size of an estimate are different: an estimate can be far from the null while still having uncertainty, and an interval does not determine clinical importance.

For overall survival, the registry reports a confidence interval for the hazard ratio, but the group medians are not reported in the registry record. The registry provides a more precise numerical hazard-ratio estimate than the median results allow for that endpoint, but the absence of medians limits interpretation in time units. The reported maximum follow-up is 86 months for the overall-survival timeframe; it does not establish that all participants had that length of follow-up.

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 for these time-to-event comparisons are totals of events, not participants per arm. The planning illustration uses the observed effect as a reference and also considers planning for 75% and 50% of that observed effect. For the primary endpoint, the total-event targets are 15, 26, and 59 for 80% power across those effect assumptions, and 20, 35, and 78 for 90% power. The corresponding overall-survival targets are 63, 111, and 249 for 80% power, and 84, 148, and 333 for 90% power.

The stated planning assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, a normal approximation, and an effect to detect set to the stated fraction of the observed effect. The time-to-event event totals use the Schoenfeld method. These are planning illustrations based on the observed effect, not a statement of the trial’s original design assumptions; those original assumptions are not reported in the registry record.

The 80% power column describes a planning target that would give the test an 80% chance of detecting the assumed effect if that effect and the planning assumptions were correct. The 90% power column uses a higher probability target under the same general planning framework. Smaller assumed effects require more events to reach either target. The event targets are not participant counts, and the number of participants needed to accrue them depends on event occurrence and follow-up. These figures are not observed power.

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 = 0.230, so ln(HR) = -1.4697.
2. d = 4 × (1.96 + 0.842)² / [ln(0.230)]² = 14.5, rounded up to 15 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 <- 0.23                              # 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(2511106)
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 = 15 (80% power), 20 (90% power)
 75% of observed effect: events = 26 (80% power), 35 (90% power)
 50% of observed effect: events = 59 (80% power), 78 (90% power)
Simulated power with 15 events: 0.695 (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 47 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 682 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 <- 21.9          # control-arm median, months (converted from the posted unit)
hr          <- 0.23          # 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: 42         p (1-sided)   0.0001
  Events: 7        ~HR at bound   0.0584
  Month: 22    P(Cross) if HR=1   0.0001
            P(Cross) if HR=0.23   0.0338
  IA 2: 67%                   Z   2.5114
      N: 48         p (1-sided)   0.0060
 Events: 14        ~HR at bound   0.2567
  Month: 35    P(Cross) if HR=1   0.0060
            P(Cross) if HR=0.23   0.5603
      Final                   Z   1.9930
      N: 48         p (1-sided)   0.0231
 Events: 21        ~HR at bound   0.4143
  Month: 60    P(Cross) if HR=1   0.0250
            P(Cross) if HR=0.23   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 primary hazard ratio and its confidence interval are the sponsor-reported analysis results recorded for the trial. The registry identifies Log Rank as the comparison method and reports the result as a superiority analysis. The corresponding primary median estimates are also registry results, but the hazard ratio and medians summarize different aspects of the time-to-event data.

No discrepancy is flagged between the registry’s reported comparison and the posted interpretation. The sponsor’s own analysis is the relevant reported result; a simple unadjusted re-calculation from summary statistics would not replace it. In particular, a hazard ratio cannot be recreated simply by contrasting the two reported medians. The record does not provide the event-level data needed to reproduce the time-to-event analysis independently.

The overall-survival comparison is also reported as a sponsor analysis. Its median values are not reported in the registry record, so the summary supports interpretation of the reported relative hazard estimate but not a comparison of median survival times. The registry record does not provide enough detail to independently reproduce every part of either endpoint analysis.

16. Multiplicity, Missing Data and Analysis Populations

Multiplicity arises when a trial evaluates several outcomes, time points, comparisons, or analyses. Each statistical test carries a chance of producing a result that appears unusual under its null hypothesis. As the number of tests increases, the chance of at least one apparently notable result can increase unless the analysis plan addresses that issue. General: Methods such as adjusting significance thresholds or controlling an error rate can be used for this purpose, but the choice depends on the trial’s objectives and prespecified testing structure.

The registry record lists a primary DFS outcome and multiple secondary outcomes, including overall survival, DFS rates at specified times, health-related quality of life and symptoms, and plasma concentration measures. It reports a sponsor analysis for overall survival, but does not report the multiplicity strategy or establish how the secondary outcomes were ordered for confirmatory testing. The record therefore does not support treating every reported secondary p-value as if it had the same evidential status as a single prespecified primary test.

Important details about missing data and censoring are also not reported in the registry record. For a time-to-event endpoint, participants may have follow-up that ends before an event is observed. The analysis then needs rules for censoring and assumptions about whether those who leave observation differ in relevant ways from those who remain under observation. The endpoint record does not specify those rules or describe sensitivity analyses. It also does not report the handling of missing endpoint information in the Full Analysis Set.

The primary population is identified as the Full Analysis Set of stage IIA-IIIA patients, but the registry endpoint record does not give its detailed operational definition. The overall trial description includes stage IB-IIIA disease, so the relationship between eligibility and the primary analysis population matters when judging how broadly the result applies. The record does not report the original sample-size justification, the full statistical analysis plan, or whether interim analyses were planned or conducted. These omissions are limits on independent appraisal, not evidence that a particular analysis choice was or was not made.

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 (1.5%), acute kidney injury (0.6%), ureterolithiasis (0.6%), cataract (0.6%), diarrhoea (0.6%).

The registry reports serious adverse events as counts alongside the number at risk in each group. For AZD9291, 68 serious adverse events are reported among 337 at risk; for placebo, 47 are reported among 343 at risk. The record also reports other adverse events as 312 among 337 at risk for AZD9291 and 262 among 343 at risk for placebo. These are the registry’s counts and denominators; they should be read as posted and not converted into rates without the relevant analysis details.

Deaths are reported as 42 for AZD9291 and 82 for placebo. The registry’s safety summary does not, in the registry record for interpretation, establish the cause, timing, relationship to treatment, or adjudication of each death. Serious adverse events are not necessarily events caused by the assigned treatment. The counts alone do not explain severity patterns, duration, recurrence, exposure time, or how events were collected and coded.

The listed serious-event terms include pneumonia at 1.5%, and acute kidney injury, ureterolithiasis, cataract, diarrhoea, and large intestine polyp at 0.6% each. The registry record does not provide further context about whether these percentages are based on participants, events, or a particular safety analysis convention. These safety figures should therefore be interpreted as registry-reported summaries, not as a complete comparative safety assessment.

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

The registry results show a sponsor-reported primary DFS hazard ratio of 0.23 for AZD9291 compared with placebo in the stated Full Analysis Set of stage IIA-IIIA patients. Its 95% confidence interval is 0.18 to 0.30, and the registry reports a p-value below < 0.001. The reported medians are 65.8 months for AZD9291 and 21.9 months for placebo, although the AZD9291 median’s upper confidence limit is not reported. Taken together, these estimates describe a difference in the trial’s reported time-to-event results; they do not define what will happen to any individual participant.

The result does not establish that every patient would experience the same duration without recurrence, nor does the hazard ratio specify an absolute probability of recurrence or death at a chosen time. It is not a percentage reduction in risk. A relative hazard estimate also cannot be directly translated into an individual prognosis without additional information about baseline event risk, follow-up, and the assumptions of the analysis.

The overall-survival analysis is secondary. Its reported hazard-ratio comparison points in the same broad direction as the primary result, but median overall-survival values are not reported in the registry record. The record also does not report the multiplicity strategy, detailed missing-data procedures, censoring rules, or the complete statistical analysis plan. Those gaps limit independent appraisal of the analyses and the strength of claims that can be made from secondary results.

The trial description covers stage IB-IIIA disease, whereas the primary analysis population is specified as stage IIA-IIIA. The posted primary estimate should be interpreted for the stated analysis population rather than assumed to apply equally to all stages in the broader description. Statistical results do not by themselves resolve questions about clinical importance, effects in other populations, or the balance of benefits and harms. The registry’s safety counts add information about reported adverse events, but counts alone are not a complete assessment of causality or comparative safety.

19. Why This Trial Matters Statistically

This trial illustrates why statistical interpretation of a clinical study should distinguish the endpoint, analysis population, and effect measure. The primary outcome is time to recurrence or death, not a simple proportion with recurrence at one fixed time. Its reported medians and hazard ratio describe different features of the event-time experience. Treating either one as a complete summary would lose information about the other.

The record also illustrates the importance of matching claims to the population actually analysed. The trial description includes stage IB-IIIA disease, while the primary analysis population is listed as stage IIA-IIIA. That distinction affects the scope of an interpretation. A result is not automatically generalizable across every group included in the broad eligibility description just because the overall trial title names those groups.

Precision matters alongside the estimated effect. The hazard-ratio interval provides a range of values compatible with the reported analysis, while the primary median for AZD9291 lacks a reported upper confidence limit. A statistically significant result is not, on its own, a complete account of uncertainty, clinical meaning, or the assumptions underlying an analysis. The registry’s unreported details on missing data, censoring, multiplicity, and the original design rationale are relevant to independent appraisal.

Finally, the planning illustrations show that event targets depend on the effect assumed for planning. They are not observed power and do not reconstruct the trial’s original design. Keeping planning calculations separate from trial results avoids confusing hypothetical design scenarios with the evidence observed in the study. Together, these distinctions make the registry findings easier to interpret without overstating what the reported statistics establish.

20. Limitations

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

The primary endpoint results are reported for the Full Analysis Set of stage IIA-IIIA patients, while the trial description covers stage IB-IIIA disease. The registry record does not explain in detail how the analysis set was defined or how the result should be generalized across the broader stage range. This difference in scope is important when interpreting the primary result.

The primary median DFS confidence interval for AZD9291 has a reported lower limit but no reported upper limit. The overall-survival median estimates are not reported in the registry record, despite the presence of a secondary overall-survival hazard-ratio analysis. These omissions limit interpretation of the endpoints in median-time terms.

For the time-to-event analyses, the registry record does not report detailed censoring rules, missing-data procedures, or sensitivity analyses. It also does not report the full statistical analysis plan, multiplicity strategy, interim-analysis details, or the original sample-size justification. These details cannot be reconstructed from the reported summary statistics. The registry status is active, not recruiting, and the stated maximum follow-up differs between the primary and overall-survival endpoints; those timeframe descriptions do not show how much follow-up each participant contributed.

The safety summary provides counts and selected serious-event terms, but the registry record does not provide enough detail to establish causality, event timing, exposure-adjusted comparisons, or the full context for deaths. The available counts should not be treated as a complete safety profile. Finally, summary-level results do not permit independent reconstruction of the log-rank analyses or assessment of all underlying assumptions.

21. Related Tutorials

22. Related Statistical Calculators

23. Sources

• ClinicalTrials.gov: AZD9291 Versus Placebo in Patients With Stage IB-IIIA Non-small Cell Lung Carcinoma, Following Complete Tumour Resection With or Without Adjuvant Chemotherapy. — NCT02511106.
• Registry record last updated 2026-09-11; status last verified 2026-09.
• Results first posted 2021-07-27.
• 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

NCT02511106 is a phase III randomized, triple-masked, parallel-group comparison of AZD9291 and matching placebo after complete tumour resection in patients with epidermal growth factor receptor mutation-positive stage IB-IIIA non-small cell lung carcinoma. The registry records enrollment of 682, a start date of 2015-10-21, a primary completion date of 2022-04-11, and a status of active, not recruiting.

The primary outcome is DFS in the Full Analysis Set of stage IIA-IIIA patients. The sponsor-reported hazard ratio is 0.23, with a 95% confidence interval of 0.18 to 0.30. Median DFS is 65.8 months for AZD9291 and 21.9 months for placebo; the upper confidence limit for the AZD9291 median is not reported in the registry record. Overall survival is listed as a secondary outcome, with a reported hazard-ratio analysis but no reported median values. Registry-level information on multiplicity, missing-data handling, censoring, and the complete statistical plan is not reported.