LUX-Lung 8: A Phase III Trial of Afatinib (BIBW 2992) Versus Erlotinib for the Treatment of Squamous Cell Lung Cancer After at Least One Prior Platinum Based Chemotherapy: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for LUX-Lung 8: A Phase III Trial of Afatinib (BIBW 2992) Versus Erlotinib for the Treatment of Squamous Cell Lung Cancer After at Least One Prior Platinum Based Chemotherapy (NCT01523587), a randomized phase 3 study. The primary outcome is Progression-free Survival, Based on Central Independent Review as Determined by Response Evaluation Criteria in Solid Tumours 1.1.

2. Clinical Question

The registry describes the objective as comparing the efficacy of afatinib, also named BIBW 2992 in the trial title, with erlotinib as second-line treatment in this population. The trial was randomized and open-label, so treatment assignment was randomized but patients and treating teams were not masked to the assigned drug. The primary endpoint was time from randomization to disease progression or death before progression, as determined by central independent review.

The question is specific to patients with advanced squamous cell carcinoma after prior platinum-based chemotherapy. It is not a comparison of first-line treatment, nor does the registry record describe other treatment settings as the target population. The posted primary result concerns progression-free survival; the registry also names overall survival, response, disease control, tumour shrinkage, and patient-reported symptoms among secondary outcomes. Those outcomes address different aspects of disease course and experience and should not be treated as interchangeable with the primary endpoint.

3. Trial Design

The registry describes LUX-Lung 8 as a Phase III, randomized, open-label, parallel-group trial comparing afatinib with erlotinib. The recorded enrollment is 795. The trial status is completed. The registry lists Boehringer Ingelheim as sponsor. The recorded start date is 2012-03-05, and primary completion is recorded as 2013-10-21.

Patients had advanced squamous cell carcinoma of the lung and were being treated after first-line platinum-based chemotherapy. The afatinib group received tablets once daily, and the erlotinib group received tablets once daily. Randomization assigns treatment by chance, which can help balance measured and unmeasured prognostic factors between groups on average. It does not guarantee that every baseline characteristic is balanced in a particular trial. Masking was recorded as none, so the comparison was open-label.

The primary outcome was progression-free survival based on central independent review according to RECIST version 1.1. The primary analysis population was the Randomized Set, defined in the record as all patients randomized regardless of whether they received investigational treatment. For this outcome, the assessment period ran from first treatment administration to the cutoff date of 02 March 2015, up to 1058 days. The registry separately gives a later timeframe for overall survival, from first drug administration from 9 April 2012 until study closure on 27 Dec 2017, approximately 2089 days.

The registry record does not report the original sample-size justification, the detailed allocation procedure, or a fuller statistical analysis plan. The planning illustrations associated with the posted results are not evidence of the trial's original design assumptions. The record also does not report the masking of those assessing other endpoints beyond the stated central independent review for progression-free survival.

4. Participant Flow and Accounting

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

Before assignment: Patients screened to ensure that they met all inclusion/exclusion criteria. Patients were not to be entered to trial treatment if any one of the specific entry criteria were not met. Tumor assessments at screening were completed within 21 days and other screening assessments were completed within 28 days, of randomization.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Randomized Set (RS): All patients who were randomized, regardless of whether they received investigational treatment.

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: Randomized Set (RS): All patients who were randomized, regardless of whether they received investigational treatment.

6. Primary Endpoint

The registry lists the primary outcome as Progression-free Survival, Based on Central Independent Review as Determined by Response Evaluation Criteria in Solid Tumours 1.1, measured over First treatment administration up until cut off date of 02 March 2015 (up to 1058 days)..

Progression-free survival measures the time from randomization until disease progression or death if death occurs before progression. In this trial, progression was determined by central independent review using RECIST version 1.1. The endpoint therefore combines two events into a time-to-event outcome, rather than counting only tumour shrinkage or only survival.

RECIST provides rules for classifying changes in target lesions. The registry description defines complete response as disappearance of all target lesions and partial response as at least 30% decrease in the sum of the longest diameters of target lesions. Overall response is defined as complete response plus partial response. The primary endpoint is not response itself: it is the time until progression or death, with progression assessed under those rules.

Time-to-event measures can use information about when events occur and can account analytically for participants whose event has not been observed by the analysis cutoff, subject to the analysis methods. In general, progression-free survival can describe the timing of disease worsening or death without requiring death to be the event for every participant. It is not a direct measure of symptom relief, quality of life, or length of life by itself. The registry record does not report the detailed censoring rules or missing-data procedures used for this analysis.

7. Results Posted in the Registry

• Afatinib vs Erlotinib: hazard ratio 0.81 (95% CI 0.69–0.96), reported p = 0.010 (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

• 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:
  • Afatinib vs Erlotinib: hazard ratio 0.84 (95% CI 0.73–0.97), reported p = 0.019 (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 Afatinib vs Erlotinib than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Number of Participants With Objective Response According to RECIST 1.1

• Endpoint type: Binary: the number of participants with the event, compared as proportions.
• Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
• Result:
  • Erlotinib: 2.8% (11/397) versus 5.5% (22/398) on control; risk difference -2.8 percentage points (95% CI -5.5 to 0.0), risk ratio 0.50 (0.25–1.02); unadjusted two-proportion z-test p = 0.051.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Erlotinib and the control. The interval gives the range of effects that remain compatible with the data.

Number of Participants With Disease Control According to RECIST 1.1

• Endpoint type: Binary: the number of participants with the event, compared as proportions.
• Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
• Result:
  • Erlotinib: 39.5% (157/397) versus 50.5% (201/398) on control; risk difference -11.0 percentage points (95% CI -17.8 to -4.1), risk ratio 0.78 (0.67–0.92); unadjusted two-proportion z-test p = 0.002.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the proportion with the event is statistically lower for Erlotinib than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

The secondary outcomes cover several domains rather than one single confirmatory question. They include overall survival, RECIST-based objective response and disease control, tumour shrinkage, and patient-reported cough, dyspnoea, pain, and health-related quality-of-life measures. Such outcomes can add context to progression-free survival by addressing survival, tumour response, or symptoms, but each has its own definition and analysis.

The detailed posted results include overall survival and two RECIST-based participant counts. Across these outcomes, the reported estimates do not all have the same degree of precision or the same statistical strength. The objective-response analysis is less conclusive statistically than the reported disease-control analysis, while the overall-survival result is a separate time-to-event comparison. These broad patterns should not be read as a single combined result or as proof that every secondary outcome supports the primary finding.

Secondary outcomes commonly involve multiple comparisons. When many outcomes or analyses are considered, the chance of at least one apparently notable result arising by chance can increase if no adjustment or hierarchy is used. The registry record does not report how multiplicity was addressed, whether the secondary outcomes were prespecified in an inferential hierarchy, or which analyses were intended to be confirmatory. Consequently, unadjusted secondary findings carry less weight than a result supported by a clearly specified confirmatory plan.

Secondary results may still be informative as descriptive evidence and as context for the primary endpoint. Their interpretation should account for their separate estimands, the uncertainty in each estimate, the availability of detailed outcome results, and the unreported multiplicity strategy. They should not be combined informally into a claim that every aspect of disease course or patient experience moved in the same direction.

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

• Overall Survival
• Number of Participants With Objective Response According to RECIST 1.1
• Number of Participants With Disease Control According to RECIST 1.1
• Tumour Shrinkage
• Number of Participants With Status Change in Cough, Dyspnoea and Pain Related Items Over Time in Health Related Quality of Life Questionnaire
• Summary of Time to Deterioration in Coughing, Dyspnoea and Pain.
• Change in Score Over Time in Coughing,Dyspnoea and Pain

9. Statistical Interpretation

10. Statistical Methods Explained

The primary endpoint is a time-to-event outcome, so its analysis uses both event status and the timing of events. A median time is a summary of the estimated event-time distribution. The reported median for afatinib is 2.63 months, with a 95% confidence interval from 2.0 to 2.86; the reported median for erlotinib is 1.94 months, with a 95% confidence interval from 1.87 to 2.1. These summaries do not show the full shape of the distributions, and comparing medians alone does not use all the timing information available to a time-to-event analysis.

What does the hazard ratio describe?

The reported comparison uses a hazard ratio from a Log Rank analysis. A hazard is the rate at which an event occurs among participants who have not yet had the event at a given point in follow-up. A hazard ratio compares those rates between groups over the analyzed period. The reported estimate of 0.814 is for afatinib versus erlotinib. A value below the null indicates a lower estimated event hazard for the first group under that comparison. It is not a direct estimate of the difference in median progression-free survival, and it does not state how many individuals avoid progression or death.

The registry reports a 95% confidence interval from 0.693 to 0.956 and p = 0.010 for this comparison. A confidence interval conveys uncertainty around the estimated hazard ratio under the analysis; a p-value describes how compatible the observed data are with the null hypothesis under the specified test. Neither quantity measures the probability that a treatment is effective or ineffective. The record identifies the method as Log Rank but does not report further details such as proportional-hazards diagnostics or stratification factors.

Why use central review and RECIST?

RECIST version 1.1 supplies standardized criteria for classifying tumour response and progression. Standardized criteria help make assessments more consistent across participants. Central independent review means progression was assessed independently of the treating team, as stated in the primary outcome definition. This distinction matters in an open-label trial, where treatment assignment is known to participants and care teams. The registry does not describe the review workflow or whether central reviewers were masked to treatment assignment.

The primary endpoint definition includes death before documented progression as an event. This makes the outcome broader than radiographic progression alone. It also means that the reported estimate combines progression and death; it cannot isolate the effect on either component from the information given in the primary summary.

How do the secondary analyses differ?

Overall survival is also a time-to-event outcome, reported as median months and analyzed using a Log Rank hazard ratio. Response and disease control are counts of participants meeting RECIST-based outcome definitions. The registry reports logistic regression odds ratios for these binary outcomes. An odds ratio compares the odds of an outcome between groups; it is not the same quantity as a risk difference, which compares the proportions with the outcome. The outcome definitions and estimands matter when reading the reported estimates.

The posted recalculations for the binary outcomes are unadjusted summaries based on the posted group counts. They do not replace the sponsor's reported logistic-regression analyses. Different metrics can produce estimates on different scales, so apparent differences between an odds ratio and an unadjusted risk difference are not, by themselves, evidence of a contradiction. The sponsor's own reported analysis remains the relevant trial analysis for its specified model.

What remains unspecified?

The registry record does not report the detailed statistical analysis plan, the original design assumptions, or the method used to address multiplicity across outcomes. It also does not report the missing-data procedures, the detailed censoring rules, or whether any interim analyses were planned or conducted. These omissions limit how fully a reader can reconstruct the analysis decisions from the registry record alone. General statistical interpretation should therefore remain tied to the endpoint definitions and methods that are explicitly recorded.

Planning illustrations based on observed effects are distinct from a trial's original design justification. They describe hypothetical sample requirements under stated assumptions, not the trial's actual power after results are known. Over-dispersion is a concern in some count-data settings when variability exceeds what a simple model expects; it is not a stated feature or reported adjustment in this trial's primary time-to-event analysis.

11. Confidence Intervals and Precision

The 95% confidence interval for the primary hazard ratio extends from 0.693 to 0.956 around the estimate of 0.814. It lies below the null value, which is consistent with the reported p-value of 0.010 under the Log Rank analysis. The interval is the principal indication of statistical precision in the reported relative comparison: it shows a range of hazard-ratio values compatible with the data and analysis at the stated confidence level. It does not mean there is a 95% probability that the true value lies in this particular interval.

The median progression-free survival estimates also have confidence intervals. For afatinib, the estimate is 2.63 months, with an interval from 2.0 to 2.86; for erlotinib, the estimate is 1.94 months, with an interval from 1.87 to 2.1. These intervals describe uncertainty in the medians, not the hazard ratio. They should not be combined by subtracting their endpoints or treated as a direct confidence interval for a difference between medians.

Confidence intervals for secondary outcomes vary in what they convey because the outcomes and effect measures differ. For example, the overall-survival hazard ratio has a 95% confidence interval from 0.727 to 0.973, while the reported objective-response odds ratio has a 95% confidence interval from 0.98 to 4.32. These intervals are on different scales and refer to different outcomes. Their widths reflect statistical uncertainty under their respective analyses, but do not account for all sources of uncertainty, such as outcome measurement, missing data, or selective emphasis.

Precision is not the same as clinical importance. A relatively narrow interval can still describe an effect whose practical importance depends on context, and a wider interval can leave a broad range of plausible estimates. The registry record does not provide a prespecified threshold for a clinically important difference in progression-free survival. The estimates should therefore be read as statistical summaries of this trial, not as a complete assessment of the value of either treatment.

12. Planning a Similar Trial: Sample Size

This is a planning illustration, not a re-evaluation of the original trial. It asks: if a new study wanted to detect a difference like the one observed, or a smaller one, how large would it need to be? It is deliberately not “observed power” of the finished trial, which is not a meaningful quantity.

The planning illustrations use the observed effect as a starting point and show hypothetical requirements for detecting a stated fraction of that effect. For progression-free survival and overall survival, the figures are total event counts, not participants per arm. For the binary outcomes, the figures are participants per arm. The two kinds of figures are not interchangeable: time-to-event planning is expressed in events, while the binary-outcome illustrations are expressed as group sizes.

The stated planning assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The effect to detect is the stated fraction of the observed effect. For the time-to-event illustrations, the event calculation uses the Schoenfeld method. These are planning calculations based on the observed effects, not the trial's original design justification; the original design assumptions are not reported in the registry record.

The 80% power column describes the hypothetical number of events or participants required, under those assumptions, for an 80% chance of detecting the specified effect if that effect were the true one. The 90% power column expresses the corresponding requirement for a 90% chance. Higher planned power generally requires more information under otherwise unchanged assumptions. When the target is a smaller effect, it is harder to distinguish that effect from random variation, so the planning requirement increases.

For instance, the primary progression-free survival illustration gives total event requirements of 742 for 80% power and 993 for 90% power when planning for 100% of the observed effect. For 75% of the observed effect, the corresponding total event figures are 1,318 and 1,765; for 50%, they are 2,966 and 3,970. These planning 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 = 0.814, so ln(HR) = -0.2058.
2. d = 4 × (1.96 + 0.842)² / [ln(0.814)]² = 741.3, rounded up to 742 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 <- 0.814                              # 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(1523587)
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 = 742 (80% power), 993 (90% power)
 75% of observed effect: events = 1,318 (80% power), 1,765 (90% power)
 50% of observed effect: events = 2,966 (80% power), 3,970 (90% power)
Simulated power with 742 events: 0.798 (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 1,036 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 795 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 <- 1.94          # control-arm median, months (converted from the posted unit)
hr          <- 0.814          # 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: 472         p (1-sided)   0.0001
  Events: 335        ~HR at bound   0.6665
    Month: 11    P(Cross) if HR=1   0.0001
              P(Cross) if HR=0.81   0.0338
    IA 2: 67%                   Z   2.5114
       N: 822         p (1-sided)   0.0060
  Events: 670        ~HR at bound   0.8235
    Month: 19    P(Cross) if HR=1   0.0060
              P(Cross) if HR=0.81   0.5603
        Final                   Z   1.9930
      N: 1036         p (1-sided)   0.0231
 Events: 1004        ~HR at bound   0.8818
    Month: 60    P(Cross) if HR=1   0.0250
              P(Cross) if HR=0.81   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 endpoint, the sponsor-reported analysis is a Log Rank hazard ratio of 0.814, with a 95% confidence interval from 0.693 to 0.956 and p = 0.010. The posted primary comparison uses the same reported estimate and interval. The record does not flag a discrepancy for this result, so the reported analysis can be read directly as the sponsor's primary comparison rather than as a replacement estimate from a recalculation.

The binary secondary outcomes use a different comparison in the sponsor analysis: logistic regression odds ratios. The posted simple unadjusted recalculations use risk differences from the group counts. These measures answer related but distinct questions and are expressed on different scales. A difference between their values does not, on its own, show that either calculation is erroneous. The sponsor's reported model is the trial's stated analysis for those outcomes; the unadjusted recalculations are supplementary descriptions, not substitutes for that model.

For the primary time-to-event endpoint, the posted interpretation uses the sponsor-reported hazard ratio rather than attempting to derive a new hazard ratio from the medians. A hazard ratio cannot be recovered by simply comparing the two reported medians. The record identifies the reported test method but does not include enough detail to independently reconstruct every model or analysis decision.

16. Multiplicity, Missing Data and Analysis Populations

Multiplicity arises when a trial evaluates multiple outcomes, comparisons, or analysis choices. If each is interpreted using a nominal significance threshold without adjustment or a prespecified hierarchy, the chance of at least one result appearing statistically notable by chance can be higher than for a single planned test. This concern is relevant to a record that names outcomes spanning survival, response, disease control, tumour shrinkage, and symptom measures.

The registry record identifies progression-free survival as the primary outcome and lists several secondary outcomes, but it does not report a multiplicity adjustment, a testing hierarchy, or how confirmatory claims were to be controlled across outcomes. The status of the secondary analyses as confirmatory or exploratory is not reported in the registry record. The reported p-values should therefore be interpreted as results for their specified analyses, without assuming a family-wise adjustment or a particular sequence of testing.

The primary analysis population is defined as the Randomized Set: all randomized patients, regardless of whether they received investigational treatment. That definition is useful for understanding who was included in the stated analysis population. It does not, by itself, explain how missing endpoint assessments, treatment discontinuations, or incomplete follow-up were handled.

For progression-free survival, the registry describes the event definition and assessment period but does not report detailed censoring rules or procedures for missing assessments. Missing-data handling for secondary outcomes is also not reported in the registry record. The statistical analysis plan, including interim-analysis methods and any associated decision rules, is not reported in the registry record. These gaps matter because analysis choices can affect estimates and their uncertainty, especially when follow-up or outcome data are incomplete.

It is not possible from the registry record to determine whether all planned outcomes were reported in the same way, whether any analyses were added after results became available, or how unreported outcomes were treated. Those uncertainties do not negate the posted estimates, but they limit the certainty with which the full analysis process can be reconstructed.

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: dyspnoea (7.6%), pneumonia (6.6%), malignant neoplasm progression (5.9%), diarrhoea (4.6%), dehydration (3.1%).

The adverse-event record gives serious-event and other-event counts for each group, along with deaths. Serious adverse events were recorded for 174 of 392 participants at risk in the afatinib group and 175 of 395 at risk in the erlotinib group. Other adverse events were recorded for 383 of 392 at risk in the afatinib group and 371 of 395 at risk in the erlotinib group. The recorded death counts are 77 and 71, respectively.

The most frequently listed serious-event terms include dyspnoea at 7.6%, pneumonia at 6.6%, malignant neoplasm progression at 5.9%, diarrhoea at 4.6%, dehydration at 3.1%, and respiratory failure at 3.0%. These are registry-reported terms and percentages. The record does not provide enough information in this summary to establish whether each event was considered treatment-related, how events overlapped within participants, or how severity was graded beyond the serious-event categorization.

These counts describe reported safety events in the groups; they should not be interpreted as a causal comparison by themselves. Exposure time, event definitions, attribution, and details of follow-up affect safety interpretation. The registry record does not report those details in the adverse-event summary quoted.

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

The primary result is an estimate of progression-free survival in the randomized comparison of afatinib with erlotinib among patients with advanced squamous cell lung cancer after platinum-based chemotherapy. The reported median estimates and hazard ratio describe the observed time-to-event comparison under central independent review and RECIST version 1.1. The hazard-ratio confidence interval lies below the null value, and the reported p-value is 0.010. These statistics indicate that the observed comparison is not readily explained by the null hypothesis under the reported test, but they do not eliminate uncertainty.

The finding does not establish that every participant had a better outcome with afatinib, nor does it predict an individual patient's course. Group summaries do not show the full range of outcomes or identify which individuals benefited, experienced progression, or had adverse events. The result is also not a direct measure of quality of life or symptom relief. Those questions require their own outcome definitions and analyses.

The primary result does not, by itself, settle the interpretation of overall survival, response, disease control, or safety. Those outcomes are distinct and have their own estimates, uncertainty, and analysis considerations. Nor does a statistical comparison answer whether an effect is clinically important; the registry record does not report a threshold for a clinically meaningful difference in the primary outcome.

Generalizability is bounded by the trial population and treatment setting described in the record. The results concern second-line therapy after first-line platinum-based chemotherapy in advanced squamous cell carcinoma of the lung. They should not be automatically extended to other lung cancer types, earlier treatment settings, or populations not described in the record. The open-label design and missing detail about analysis procedures also constrain what can be inferred from the registry summary.

The appropriate conclusion is therefore limited: the posted primary analysis reports a more favorable progression-free survival comparison for afatinib than erlotinib under the specified endpoint and analysis. It is not a guarantee of individual benefit, a complete comparison of all outcomes, or a substitute for evidence about other populations and treatment decisions.

19. Why This Trial Matters Statistically

This trial illustrates how the choice of endpoint shapes the interpretation of a clinical comparison. Progression-free survival combines progression and death into a time-to-event outcome, while overall survival measures a different event and RECIST response counts participants meeting response criteria. These outcomes address related but nonidentical questions. Keeping their definitions separate prevents a median, hazard ratio, response count, and symptom measure from being treated as interchangeable evidence.

The result also demonstrates why an effect estimate should be read with its confidence interval and analysis method. The reported hazard ratio summarizes the relative event hazard over follow-up, whereas the median estimates summarize positions in the time distributions. The Log Rank result and its confidence interval provide more information than a statement based only on whether a p-value crosses a threshold. Precision and clinical importance remain distinct considerations.

The open-label design and central independent review make assessment procedures relevant to interpretation. Randomization supports a group comparison, while central review provides an independent process for the primary progression assessment. These features do not remove every potential source of bias, and the registry record does not report all details needed to assess the analysis process fully.

Finally, the trial is a useful example of the difference between an original design justification and retrospective planning illustrations based on observed effects. The latter can show how sample requirements change under stated assumptions, but they do not reveal the original planned sample-size rationale or the achieved power. Recognizing that distinction helps prevent planning calculations from being mistaken for evidence about the strength or certainty of the observed result.

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 limitations specific to this trial affect how the registry findings can be interpreted. The trial was open-label, with masking recorded as none. The primary progression assessment was based on central independent review, but the registry record does not state whether central reviewers were masked to assigned treatment or describe the review process. The independent assessment is relevant, but it does not make the overall study masked.

The primary endpoint was assessed through the cutoff date of 02 March 2015, while the overall-survival timeframe extends to study closure on 27 Dec 2017. These different observation periods mean the endpoint summaries are not based on the same stated timeframe. The registry record does not provide detailed follow-up completeness or censoring rules for the primary analysis.

The reported primary population is the Randomized Set, including all randomized patients whether or not they received investigational treatment. However, the registry summary does not provide detailed information about missing outcome data, treatment exposure, adherence, or how incomplete assessments were handled. It also does not report subgroup analyses or how any such analyses were prespecified.

The record does not report the original sample-size assumptions, a full statistical analysis plan, multiplicity procedures, or interim-analysis details. These omissions limit reconstruction of the planned inferential framework. The planning illustrations based on observed effects cannot fill that gap. Secondary outcomes span different measures and the record does not explain their inferential hierarchy, so their results should be interpreted cautiously rather than treated as a coordinated set of confirmatory tests.

Finally, the findings are specific to advanced squamous cell carcinoma of the lung after first-line platinum-based chemotherapy and the two study drugs described. The registry information does not establish how the estimates apply beyond that population and treatment context.

21. Related Tutorials

22. Related Statistical Calculators

23. Sources

• ClinicalTrials.gov: LUX-Lung 8: A Phase III Trial of Afatinib (BIBW 2992) Versus Erlotinib for the Treatment of Squamous Cell Lung Cancer After at Least One Prior Platinum Based Chemotherapy — NCT01523587.
• Registry record last updated 2019-02-15; status last verified 2019-02.
• Results first posted 2014-11-21.
• 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

LUX-Lung 8 is recorded as a completed, randomized, open-label Phase III parallel-group trial of afatinib versus erlotinib for advanced squamous cell carcinoma of the lung after platinum-based chemotherapy. Recorded enrollment is 795. Both interventions were tablets taken once daily.

The primary outcome was progression-free survival by central independent review under RECIST version 1.1, analyzed in the Randomized Set. Median progression-free survival was 2.63 months for afatinib and 1.94 months for erlotinib. The reported hazard ratio was 0.814, with a 95% confidence interval from 0.693 to 0.956 and p = 0.010.

The registry also reports secondary outcomes and adverse-event counts. Interpretation is limited by missing detail on the original design assumptions, multiplicity, missing-data procedures, and other analysis-plan decisions. The primary result describes this randomized comparison and population; it does not establish individual outcomes or resolve every secondary or safety question.