Pembrolizumab in Treating Patients With EGFR Mutant, Tyrosine Kinase Inhibitor Naive Advanced Non-Small Cell Lung Cancer: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Pembrolizumab in Treating Patients With EGFR Mutant, Tyrosine Kinase Inhibitor Naive Advanced Non-Small Cell Lung Cancer (NCT02879994), a na phase 2 study. The primary outcome is Overall Response Rate (ORR) Determined as the Percentage of Patients Achieving Complete Response or Partial Response as Respectively Defined in RECIST 1.1.

2. Clinical Question

3. Trial Design

This was a phase II, single-group study of pembrolizumab for patients with advanced non-small cell lung cancer described as EGFR mutant and tyrosine kinase inhibitor naïve. The record lists the study as completed, with enrollment of 11. It also says that enrollment ceased for lack of efficacy after 11 of 25 planned patients were treated. These figures describe the enrolled and planned populations; the registry narrative does not provide a fuller account of screening, exclusions, or the analysis population.

There was no random allocation, masking, or control group recorded. The study therefore describes outcomes among patients receiving pembrolizumab, rather than estimating a treatment contrast against another group. Without a concurrent comparator, observed outcomes cannot separate treatment effects from other influences on response or follow-up.

Pembrolizumab was administered intravenously over 30 minutes on day 1, with treatment repeating every 3 weeks for up to 35 courses in the absence of progression or unacceptable toxicity. The primary endpoint was overall response rate, with a stated timeframe of up to 14 months. Secondary outcomes included adverse events, progression-free survival, and overall survival. The registry record does not report a detailed statistical analysis plan, the handling of treatment discontinuation, or a fuller description of follow-up for each outcome.

The enrollment stopped early according to the registry narrative, which attributes the decision to lack of efficacy. The record does not report the decision process, any interim monitoring arrangements, or whether the stopping decision followed a prespecified statistical boundary. Those details should not be inferred from the fact that enrollment ended before the planned enrollment was reached.

4. Participant Flow and Accounting

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

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Enrollment was ceased due to lack of efficacy after 11 of 25 planned patients were treated. Only 1 patient had an objective response, but repeat analysis of this patient's tumor definitively showed the original report of an epidermal growth factor receptor (EGFR) mutation to be erroneous.

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: Male and female subjects with EGFR mutant, tyrosine kinase inhibitor naïve, PD-L1+, advanced non-small cell lung cancer

6. Primary Endpoint

The registry lists the primary outcome as Overall Response Rate (ORR) Determined as the Percentage of Patients Achieving Complete Response or Partial Response as Respectively Defined in RECIST 1.1, measured over Up to 14 months.

The primary endpoint was overall response rate, defined as the percentage of participants achieving a complete response or partial response. In general, response endpoints classify participants according to prespecified changes in tumor findings. A complete response means disappearance of all target lesions under the criteria cited in the record; a partial response is defined there as a decrease of at least 30% in the sum of the longest diameters of target lesions. The endpoint combines those response categories.

Response rate is a binary outcome at the participant level: each participant is classified as having or not having a qualifying response during the assessment period. It can provide a direct description of how often a defined tumor response was observed in a treated group. It does not by itself describe how long a response lasts, whether progression is delayed, or how survival compares with another treatment. The registry lists a timeframe of up to 14 months.

The record is not consistent about the response criteria version: the endpoint title refers to RECIST 1.1, while its description refers to RECIST v1.0. The record also states that the sole reported objective response came from a participant whose original EGFR mutation report was later shown by repeat tumor analysis to be erroneous. These details affect how confidently the response result can be interpreted for the intended population.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

What does a single-group result estimate?

In a single-group study, the primary response result describes the observed response experience among participants receiving the study treatment. It is not a treatment difference because there is no concurrent comparator group. General statistical interpretation distinguishes a group’s observed outcome from a causal contrast: a contrast requires a comparison that can account for what would have happened under another condition. Randomization is one way to support such comparisons, but this study is not recorded as randomized.

For a binary endpoint such as response, each participant is classified into an outcome category under the endpoint rules. The reported count is therefore a count of participants classified as having an event, rather than a continuous measurement such as tumor size. A response proportion can describe the frequency of the event in the analyzed group, but its uncertainty depends on the number analyzed and the analysis method. The registry narrative reports one objective response but does not give a formal estimate or confidence interval.

Why does endpoint definition matter?

A response endpoint depends on the criteria used to classify tumor findings. The record’s endpoint title names RECIST 1.1, while the description names RECIST v1.0. The record does not clarify which version was applied. In general, different versions or implementation details can affect classification rules, so readers need a consistent definition to interpret or reproduce an endpoint analysis. The record also does not describe whether response assessments were independently reviewed or how disagreements were handled.

The reported response also has a population-classification issue: repeat tumor analysis found that the sole participant with an objective response did not have the EGFR mutation originally reported. This does not permit a recalculation of the response result for a corrected population because the registry does not report the complete participant-level classifications or analysis details. It does, however, make the single reported response difficult to interpret as an outcome in the intended EGFR-mutant population.

What can a confidence interval tell a reader?

In general, a confidence interval describes the statistical uncertainty around an estimate under a stated method and assumptions. A narrow interval indicates greater precision than a wide interval, but neither the interval nor its interpretation substitutes for a comparator when the question is whether treatment changes outcomes. The registry record does not report a confidence interval for the primary endpoint. It also does not report a hypothesis test, a p-value, or a model for the primary endpoint.

With a small group, estimates of binary outcomes can be particularly sensitive to the classification of individual participants. This is a general property of sparse event data, not a separate calculation for this study. Here, the response narrative makes participant classification especially salient because the only reported objective response was associated with an erroneous original mutation report. The registry’s description should therefore be read as a limited outcome account, not as a precise estimate for the intended biomarker-defined group.

What do the planning assumptions mean?

The record describes sample-size planning illustrations based on the observed effect, a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. It states that the effect to detect is a specified fraction of the observed effect. These are planning assumptions, not a description of this study’s single-group design: the trial itself did not have two randomized arms. The record’s illustrations should not be read as observed power or as evidence that a treatment effect was established. Sample-size illustrations concern hypothetical future study designs under chosen assumptions; they do not change the uncertainty or limitations of the observed trial.

For time-to-event outcomes, the record’s planning note describes total event counts using the Schoenfeld approach; other planning illustrations are expressed as participants per arm. These are general distinctions between event-driven planning for time-to-event endpoints and participant-based planning for other endpoints. The registry record does not report the numerical planning results in the registry record described for this analysis, nor does it report the assumptions needed to interpret any separate sample-size calculation as the trial’s own design justification.

Why can over-dispersion matter?

Over-dispersion is a general concern when observed variability is greater than a statistical model expects. It can make standard errors too small if the model does not allow for the extra variability, leading to confidence intervals that are too narrow and tests that appear more decisive than warranted. This trial’s primary outcome is binary, and the registry record does not describe a model that would allow an over-dispersion assessment. The point is a general modeling consideration, not a reported issue or adjustment in this study.

The record reports no analyses in its analysis listing. It does not provide the statistical methods used for the primary endpoint, the approach to missing outcome assessments, or the analysis population definition. Consequently, readers cannot determine from the registry narrative whether estimates were based on all treated participants, an evaluable subset, or another population. These omissions limit independent reproduction and appraisal of the numerical analysis.

10. Exact Estimates for Each Arm

No two-arm comparison is shown for this trial, but each arm's result can still be summarized with exact interval estimates computed from the posted counts. The Clopper-Pearson interval is exact and conservative; the Wilson interval is a common closer-to-nominal alternative; the Bayesian credible interval uses a uniform Beta(1, 1) prior. For small groups the three can differ noticeably.

R code and output

arm    <- c("Treatment (Pembrolizumab)")
n      <- c(11)
events <- c(1)
for (i in seq_along(n)) {
  ci <- binom.test(events[i], n[i])$conf.int          # Clopper-Pearson exact interval
  cr <- qbeta(c(0.025, 0.975), 1 + events[i], 1 + n[i] - events[i])   # Bayesian, uniform Beta(1,1) prior
  wi <- (function(x, n) { p <- x / n; z <- qnorm(0.975); d <- 1 + z^2 / n
    c((p + z^2 / (2 * n) - z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d, (p + z^2 / (2 * n) + z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d) })(events[i], n[i])
  cat(sprintf("ARM|%s|%d|%d|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f\n", arm[i], n[i], events[i], 100 * events[i] / n[i],
              100 * ci[1], 100 * ci[2], 100 * wi[1], 100 * wi[2], 100 * cr[1], 100 * cr[2]))
}

ARM  Treatment (Pembrolizumab)  11  1  9.1  0.2  41.3  1.6  37.7  2.1  38.5

11. Multiplicity, Missing Data and Analysis Populations

The registry lists three outcome measures, including the primary overall response rate and secondary outcomes covering adverse events, progression-free survival, and overall survival. It does not report a multiplicity strategy for interpreting results across multiple outcomes. In general, testing several outcomes can increase the chance of at least one apparently notable result when each is assessed separately. Whether or how that issue was addressed in this study is not reported in the registry record.

The registry’s analysis listing contains no analyses. It does not report a statistical analysis plan describing endpoint hierarchies, adjustment for multiple testing, prespecified subgroups, or rules for selecting which results to emphasize. This should not be taken to mean that no such plan existed; its contents are not reported in the registry record.

Missing-data methods and analysis populations are also not reported in the registry record. The record does not explain how participants without evaluable tumor assessments, incomplete follow-up, or treatment discontinuation were handled for the primary or secondary outcomes. It does not define whether the primary analysis included all treated participants or only those considered evaluable. Without these details, the reported response narrative cannot be independently reproduced as a complete analysis.

In general, missing outcome information can affect estimates if the probability of being assessed is related to the outcome. The direction and size of such an effect depend on the missingness pattern and the analysis method; neither is described. The early enrollment stop and the erroneous original EGFR mutation report for the participant with an objective response add interpretive context, but the record does not provide enough information to quantify their impact on an analysis.

12. 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 0% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: respiratory failure (9.1%), pneumonitis (9.1%), hypotension (9.1%), hypercalcemia (9.1%), back pain (9.1%).

The serious adverse event figures describe participants affected in the pembrolizumab treatment group. The record reports 6 participants with serious adverse events among 11 at risk. It separately reports 6 participants with other adverse events among 11 at risk, and 2 deaths. These categories should not be added together: the record does not say whether participants or events overlap across categories, and a combined total would not be justified.

The serious terms listed include respiratory failure, pneumonitis, hypotension, hypercalcemia, back pain, and elevated alanine aminotransferase. Each is reported as 9.1%. These percentages are presented as recorded; the registry does not provide further detail about event timing, severity grades for individual terms, recurrence, or attribution to treatment.

In general, adverse-event counts describe events observed during the defined reporting period and population; they do not by themselves establish that treatment caused an event. The registry record does not report causes of death or provide a causal assessment for the listed serious events. Because the study has no control group, the safety figures cannot be used to estimate a comparative excess risk against another treatment.

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

The result is a registry account of a small, single-group phase II study in which enrollment stopped for lack of efficacy before all planned patients had been treated. The record reports one objective response, then notes that repeat tumor analysis showed the original EGFR mutation report for that participant to be erroneous. This means the response cannot straightforwardly be interpreted as a response in a confirmed EGFR-mutant participant under the original classification.

The record does not report a formal response estimate with a confidence interval, a comparative effect, or the results for progression-free survival and overall survival. Consequently, it does not establish the size or precision of an effect, and it does not establish how outcomes compare with another treatment. The lack of a comparator is a design limitation, not something that can be resolved by interpreting the observed response count as a treatment difference.

The record also does not describe the analysis population, missing-data handling, response review procedures, or the statistical basis for stopping enrollment. In general, these details help readers judge whether outcome classifications and estimates are robust. Their absence from the registry record limits independent interpretation and reproduction.

The safety information is likewise descriptive. The reported participant counts and serious adverse event terms do not establish causation or comparative safety. They should be interpreted alongside the record’s lack of a control group and the absence of further event-level details.

Overall, the registry supports a narrow description: a study of pembrolizumab in the intended population stopped early for lack of efficacy, and the only reported objective response was associated with an erroneous original EGFR mutation report. It does not support a precise estimate of efficacy or a causal claim about treatment benefit or harm.

14. Why This Trial Matters Statistically

This trial illustrates how endpoint interpretation depends on both the statistical design and the accuracy of participant classification. The primary outcome was a binary response measure, and the registry reports one objective response. However, repeat analysis showed that the participant’s original EGFR mutation report was erroneous. When a key eligibility or subgroup characteristic is misclassified, even a correctly recorded outcome may not answer the intended clinical question for that subgroup.

The study also illustrates the limits of single-group evidence. A treated group can be described, but without a comparator the observed outcomes cannot distinguish treatment effects from other explanations. This matters especially when enrollment stops early and the record does not report a formal estimate, uncertainty interval, or stopping rule. Statistical interpretation should reflect those limits rather than treating an observed count as a comparative effect.

Finally, the inconsistency between the RECIST version in the endpoint title and description shows why precise outcome definitions matter. Reproducibility depends on knowing which criteria were applied and how assessments were made. The registry does not resolve that discrepancy or describe the analysis methods. The case therefore highlights a practical biostatistical lesson: clarity about population, endpoint definition, analysis population, and uncertainty is central to interpreting a clinical trial result.

15. 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 limitations constrain interpretation. The study was single-group, with no randomized comparator, so the registry record cannot support a comparative treatment effect. Enrollment stopped for lack of efficacy after 11 of 25 planned patients had been treated. The record does not report the statistical rule or process underlying that decision, so early stopping cannot be interpreted as a formal futility analysis.

The primary endpoint definition is internally inconsistent: the title refers to RECIST 1.1, whereas the description refers to RECIST v1.0. The record does not clarify which version was used. It also does not report whether tumor assessments were independently reviewed or how assessment disagreements were resolved. The sole reported objective response was in a participant whose original EGFR mutation report was later shown to be erroneous, complicating interpretation for the intended population.

The registry does not report a formal response estimate, confidence interval, inferential test, analysis population, or handling of missing assessments. Results for the listed secondary efficacy outcomes are not reported in the record. Details about the timing and attribution of serious adverse events and deaths are also not reported. The record does not provide enough information to assess how these issues might affect estimated outcomes.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Pembrolizumab in Treating Patients With EGFR Mutant, Tyrosine Kinase Inhibitor Naive Advanced Non-Small Cell Lung Cancer — NCT02879994.
• Registry record last updated 2021-07-14; status last verified 2019-09.
• Results first posted 2021-07-14.
• 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.

19. Record Summary

NCT02879994 is a completed phase II, single-group study of pembrolizumab in patients with advanced EGFR-mutant, tyrosine kinase inhibitor-naïve non-small cell lung cancer. The registry lists enrollment of 11 and states that enrollment stopped for lack of efficacy after 11 of 25 planned patients had been treated.

The primary endpoint was overall response rate, with a timeframe of up to 14 months. The record reports one objective response, but repeat tumor analysis showed that the participant’s original EGFR mutation report was erroneous. The endpoint title and description cite different RECIST versions. The record also lists adverse events, progression-free survival, and overall survival as secondary outcomes, but does not report their efficacy results in the registry record. No comparator group or detailed primary analysis methods are reported.