A Study of BGB324 (Bemcentinib) in Combination With Erlotinib in Patients With Non-Small Cell Lung Cancer: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study of BGB324 (Bemcentinib) in Combination With Erlotinib in Patients With Non-Small Cell Lung Cancer (NCT02424617), a non-randomized phase 1, phase 2 study. The primary outcome is Number of Participants With Treatment-emergent Adverse Events (TEAE).

2. Clinical Question

The study also listed pharmacokinetic measures for bemcentinib and erlotinib, dose-limiting toxicity assessment, and time to progression among its secondary outcomes. Numerical results for those outcomes are not reported in the registry record. In particular, no estimate of a treatment effect on tumor response or progression is included in the posted results.

The clinical question is consequently descriptive: what adverse events were recorded among participants exposed to the study drugs, and what can the reported counts tell readers about safety in this study? Because treatment assignment was not randomized and the arms differed in purpose and participant characteristics, their event counts should not be interpreted as a controlled estimate of the comparative effects of bemcentinib, erlotinib, or their combination.

3. Trial Design

This was a multicenter, open-label Phase 1/2 study of bemcentinib in participants with Stage IIIb or Stage IV non-small cell lung cancer. The registry record gives an enrollment of 40 and lists the study status as completed. Allocation was nonrandomized, masking was absent, and the study model was sequential. The record lists a study start of 2015-04-19 and primary completion of 2021-08-25.

The study included a Phase 1 run-in arm in which bemcentinib was given as monotherapy, with safety and tolerability as its purpose. Phase 1 Arm A combined bemcentinib with erlotinib in participants previously treated with erlotinib. The arm description states that a standard 3+3 design was used to determine a bemcentinib dose that could be administered safely with erlotinib. The detailed dose decision rules and the resulting selected dose are not reported in the registry record.

Phase 2 Arms B and C also used bemcentinib with erlotinib. The Arm B description concerns participants with an activating EGFR mutation who were T790M negative and had progressed after an approved EGFR tyrosine kinase inhibitor. Arm C concerns participants with an activating EGFR mutation who had received at least 12 weeks of erlotinib. The descriptions identify different treatment histories and eligibility settings; they do not establish that these arms were intended as direct comparative groups.

The absence of randomization and masking matters when reading arm-level results. Participants were not assigned by chance to a common treatment comparison, and the groups were not described as exchangeable. Differences in event counts across arms could reflect group size, treatment history, follow-up, or other differences, rather than a treatment effect. The registry record does not report a randomized efficacy comparison, a protocol-level statistical analysis plan, or the detailed conduct of any interim analyses.

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: The safety analysis population included all participants who received at least 1 dose of bemcentinib or erlotinib.

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 Number of Participants With Treatment-emergent Adverse Events (TEAE), measured over First dose of study drug to 28 days post last dose (maximum study treatment exposure was 1554 days; maximum follow-up = 1582 days).

The primary endpoint was the number of participants with treatment-emergent adverse events. An adverse event was defined as an untoward medical occurrence temporally associated with study treatment, whether or not considered related to treatment. For this endpoint, events counted as treatment-emergent if they occurred from the first dose through 28 days after the last dose of study drug.

The endpoint is a participant count: it records whether participants experienced events during the specified observation window, rather than measuring the number of separate episodes or the severity of every event. The registry record defines the safety analysis population as all participants who received at least one dose of bemcentinib or erlotinib. The maximum study treatment exposure was 1554 days, and maximum follow-up was 1582 days; these maxima do not mean that every participant was observed for those durations.

Counting participants with adverse events is used in safety reporting because it gives a straightforward description of how many treated participants experienced events during follow-up. It does not, by itself, determine whether an event was caused by a study drug. Nor does a participant count alone account for differences in time exposed to treatment. The primary endpoint is therefore a safety description, not a measure of efficacy or a randomized estimate of causation.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

Statistical interpretation depends on the outcome being measured, how participants entered each arm, and what analysis the registry reports. In this study, the primary endpoint is a count of participants with treatment-emergent adverse events. The registry also reports serious-event counts by arm, but it does not report a statistical comparison between arms or confidence intervals for those counts. As a result, the available information supports descriptive reading rather than a formal comparative conclusion.

What does a participant count represent?

A participant count identifies how many people in an analysis group met an event definition at least once during the specified period. It is distinct from a count of event episodes: one participant may have multiple adverse events, while still contributing once to a count of participants affected. The registry gives the numbers at risk and affected for serious and other adverse events, but the posted counts do not describe how often each person experienced events or how long events lasted.

For the primary endpoint, treatment-emergent status is based on timing: an event must occur from the first dose until 28 days after the last dose. The event definition does not require that investigators judge the event to be caused by treatment. The registry explicitly includes events whether or not considered related to the study treatment. Therefore, the endpoint is a measure of events observed in temporal association with treatment, not proof of drug causation.

Why is an arm comparison difficult?

General statistical context: a randomized comparison aims to make treatment groups comparable, on average, with respect to factors that might influence outcomes. Randomization was not used in this study. Its arms also differed in phase, purpose, and participant treatment history. A crude difference between arm counts would therefore mix possible treatment effects with differences in who entered each arm, when participants were followed, and the circumstances of treatment. The registry record itself cautions that these arms are not a like-for-like control comparison.

Arm-level counts also have different denominators: the reported numbers at risk are 8, 8, 11, and 13. Small group sizes make estimates sensitive to the experience of individual participants. General statistical context: uncertainty is usually greater when a quantity is estimated from fewer observations. Without confidence intervals or an appropriate model, the counts alone do not quantify that uncertainty or test whether apparent differences are more than random variation.

What does a confidence interval add?

General statistical context: a confidence interval gives a range of values compatible with the data and the method used to estimate a parameter. It helps distinguish the observed estimate from the uncertainty around it. No confidence interval for the adverse-event counts is reported in the registry record. The reader should therefore treat the counts as descriptions of these participants, not as precise estimates of event frequency in all similar patients.

How should event percentages be read?

The registry lists six serious adverse-event terms, each with 12.5%. These values are reproduced as reported. Their denominator and arm distribution are not stated alongside the terms, so they cannot be interpreted as arm-specific rates or used to compare treatment groups. General statistical context: a percentage is only interpretable when its underlying population and event definition are clear. A percentage without a clear denominator can create a misleading impression of precision or comparability.

What do the planning assumptions mean?

The planning illustration is described as using the observed effect, two-sided significance level 0.05, equal allocation between two arms, and a normal approximation. It is an illustration, not the trial's own design or a report of observed power. The effect to detect is described as a stated fraction of the observed effect; the posted information does not give a numerical effect estimate for the primary endpoint or a numerical planning result in the record.

General statistical context: a sample-size calculation asks how many participants might be needed under specified assumptions to detect a chosen effect with a chosen operating characteristic. It does not show that a study had that power after the fact, and it does not guarantee that an effect will be detected. Planning illustrations are especially difficult to apply when the observed effect estimate is unavailable or unstable. They should not be mistaken for the enrollment target used in this study; the registry reports enrollment of 40, while the study's own sample-size justification is not reported in the registry record.

What analyses are not documented?

The registry record lists 13 outcome measures and zero analyses. The detailed statistical analysis plan, handling of missing outcomes, any covariate adjustment, and any interim analysis procedures are not reported in the registry record. The listed secondary outcomes include pharmacokinetic measures, dose-limiting toxicity assessment, and time to progression, but numerical results for these measures are not reported in the record. Consequently, the posted adverse-event counts should not be treated as a complete statistical account of every planned outcome.

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("Phase 1- Run in Arm (Bemcentinib Monotherapy)", "Phase 1- Arm A (Bemcentinib + Erlotinib)", "Phase 2- Arm B (Bemcentinib + Erlotinib)", "Phase 2- Arm C (Bemcentinib + Erlotinib)")
n      <- c(8, 8, 11, 13)
events <- c(8, 8, 11, 13)
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  Phase 1- Run in Arm (Bemcentinib Monotherapy)  8  8  100.0  63.1  100.0  67.6  100.0  66.4  99.7
ARM  Phase 1- Arm A (Bemcentinib + Erlotinib)  8  8  100.0  63.1  100.0  67.6  100.0  66.4  99.7
ARM  Phase 2- Arm B (Bemcentinib + Erlotinib)  11  11  100.0  71.5  100.0  74.1  100.0  73.5  99.8
ARM  Phase 2- Arm C (Bemcentinib + Erlotinib)  13  13  100.0  75.3  100.0  77.2  100.0  76.8  99.8

11. Multiplicity, Missing Data and Analysis Populations

The registry lists 13 outcome measures, including the primary treatment-emergent adverse-event count and secondary pharmacokinetic measures, dose-limiting toxicity assessment, and time to progression. It lists zero analyses. The statistical analysis plan, including any approach to multiple outcomes or multiple comparisons, is not reported in the registry record.

Multiplicity arises when several outcomes or comparisons are examined: the more analyses conducted, the greater the chance that some results may appear notable by chance. General statistical context: methods for handling multiplicity can control the overall error rate or prioritize a prespecified primary question. The record does not state whether such methods were used, so no adjustment—or lack of adjustment—should be inferred.

How missing outcome data were handled is not reported in the registry record. The record also does not describe imputation, censoring rules for outcomes, or sensitivity analyses. These details matter because the interpretation of an outcome can depend on why observations are unavailable and how the analysis treats them. No assumptions about these procedures can be made from the posted safety counts alone.

The stated safety analysis population included participants who received at least one dose of bemcentinib or erlotinib. Other analysis populations, such as populations defined for efficacy or pharmacokinetic analyses, are not reported in the registry record. Nor does the record give a participant flow accounting for enrollment, treatment receipt, follow-up, and inclusion in each outcome analysis. The arm-specific numbers at risk are reported for adverse events, but they do not supply a full description of missing data or analysis-set membership for every outcome.

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: dyspnoea (12.5%), dysphagia (12.5%), cerebellar ischaemia (12.5%), nervous system disorder (12.5%), transient ischaemic attack (12.5%).

The serious adverse-event counts are 2 of 8 in the Phase 1 run-in arm, 3 of 8 in Phase 1 Arm A, 2 of 11 in Phase 2 Arm B, and 5 of 13 in Phase 2 Arm C. The registry records zero deaths in each arm. These values describe the participants at risk and affected in their respective arms; they are not results from a randomized comparison.

The registry separately lists 8 of 8, 8 of 8, 11 of 11, and 13 of 13 participants with other adverse events across those arms. A serious event is not interchangeable with any adverse event, and these group counts do not reveal how many separate episodes occurred or whether an event was judged related to treatment. The serious-event terms listed are dyspnoea, dysphagia, cerebellar ischaemia, nervous system disorder, transient ischaemic attack, and mental status changes, each shown at 12.5%. Their denominator and arm-specific distribution are not reported in the registry record.

General statistical context: sparse counts can be imprecise, and a difference between counts in separate nonrandomized arms is not, by itself, evidence of a treatment effect. Follow-up duration can also affect the opportunity to observe an event. The registry reports maximum treatment exposure and maximum follow-up, not a common observation time for every participant.

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

The posted safety data describe adverse events observed among participants who received at least one dose of bemcentinib or erlotinib. They include arm-specific counts for serious and other adverse events, zero recorded deaths in each arm, and a list of serious-event terms. The primary endpoint definition covers treatment-emergent events from the first dose through 28 days after the last dose.

These findings do not establish that a study drug caused a particular event. The endpoint includes events whether or not considered related to treatment, and temporal association is not the same as causal attribution. Nor do the counts show how the combination compares with erlotinib alone, bemcentinib alone, or another treatment in a controlled setting. The study was nonrandomized, open-label, and sequential, with arms serving different purposes and involving different treatment histories.

The counts also do not establish the frequency of events in a larger population with high precision. Arm sizes were limited, and the registry record does not report confidence intervals for these safety results. General statistical context: estimates from small groups are uncertain, and a single participant can have a substantial influence on a group-level summary. The reported percentages for serious-event terms lack a stated denominator and arm-specific breakdown in the registry record.

The study's primary endpoint is safety-focused. No numerical efficacy results are reported in the registry record for time to progression or other measures that would allow an efficacy conclusion. The study record therefore supports a description of events recorded in the treated participants, but not a conclusion about comparative benefit, treatment effectiveness, or the balance of benefits and harms.

Finally, the registry record does not report the detailed statistical analysis plan, missing-data procedures, or methods for handling multiple outcomes. Those omissions limit assessment of how completely the posted results reflect the intended analyses. Interpretations should remain tied to the descriptive information that is reported.

14. Why This Trial Matters Statistically

This study illustrates why trial structure is central to statistical interpretation. It combined a Phase 1 monotherapy run-in, a Phase 1 combination dose-finding arm, and two Phase 2 combination arms. These parts addressed different purposes and enrolled participants with different treatment histories. Treating the arms as if they were parallel randomized groups would ignore those design differences and could produce an unjustified comparative interpretation.

The study also shows the value of defining an outcome precisely. The primary endpoint was a count of participants with treatment-emergent adverse events, and the registry specified the event window and safety population. That definition clarifies what the count represents. At the same time, it does not turn an event count into proof of causation or a measure of efficacy.

Small arm-level counts call for careful language. Counts can be informative as a record of what occurred in the study, but the absence of reported confidence intervals, comparative analyses, and detailed analysis methods limits conclusions about precision and differences. The serious-event terms are listed with percentages, yet their denominator and arm-level distribution are not reported in the registry record; this demonstrates why percentages need context to be interpretable.

More broadly, the registry record distinguishes outcome definitions from outcome analyses. It lists 13 outcome measures but zero analyses, and does not report numerical results for the listed time-to-progression or pharmacokinetic outcomes. A clinical question may be named in a registry without the corresponding result being available in that record. Statistical reporting should therefore separate the question a trial set out to examine from the results that are actually reported.

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.

The principal limitation for interpretation is the nonrandomized, sequential design. The arms are not like-for-like control groups, and their participant histories and purposes differ. Arm-level event counts cannot isolate the effect of a treatment from differences between the groups. Open-label conduct is reported, but the registry record does not detail how this may have affected event identification or reporting.

The registry gives counts for serious and other adverse events, but does not report confidence intervals or an adjusted comparison. It lists serious-event terms with percentages without reporting their denominator or their distribution by arm. The primary endpoint counts participants with treatment-emergent events, but these counts do not describe event episodes, duration, severity grading, or attribution to each study drug.

Maximum study treatment exposure was 1554 days and maximum follow-up was 1582 days, but participant-level follow-up durations are not reported in the registry record. Thus, the opportunity to observe events may not have been equal across participants. The record does not report how missing data were handled, the detailed analysis populations beyond the safety population, or the statistical analysis plan.

The registry lists secondary outcomes, including time to progression, dose-limiting toxicity assessment, and pharmacokinetic measures, but does not report their numerical results in the record. It lists 13 outcome measures and zero analyses. The trial's own sample-size justification, any interim analysis procedures, and methods for handling multiple outcomes are not reported in the registry record. These gaps prevent fuller appraisal of the planned and completed statistical analyses.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: A Study of BGB324 (Bemcentinib) in Combination With Erlotinib in Patients With Non-Small Cell Lung Cancer — NCT02424617.
• Registry record last updated 2025-02-26; status last verified 2025-02.
• Results first posted 2025-02-26.
• 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

The registry record describes a completed, multicenter, open-label Phase 1/2 study of bemcentinib as monotherapy and in combination with erlotinib in participants with Stage IIIb or Stage IV non-small cell lung cancer. Allocation was nonrandomized and the study model was sequential. Enrollment was 40.

The primary endpoint was the number of participants with treatment-emergent adverse events, from first dose through 28 days after the last dose. The safety population included participants who received at least one dose of bemcentinib or erlotinib. Arm-level serious and other adverse-event counts are reported, as are zero deaths in each arm. The serious-event terms are listed with percentages, but their denominator and arm distribution are not reported.

The record lists 13 outcome measures and zero analyses. Numerical results for time to progression and the listed pharmacokinetic outcomes are not reported in the registry record. The nonrandomized arm structure and the absence of detailed statistical methods limit comparative interpretation of the posted safety data.