Platform Trial of Novel Regimens Versus Standard of Care (SoC) in Participants With Non-small Cell Lung Cancer (NSCLC) - Sub-study 3: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Platform Trial of Novel Regimens Versus Standard of Care (SoC) in Participants With Non-small Cell Lung Cancer (NSCLC) - Sub-study 3 (NCT06926673), a randomized phase 2 study. The primary outcome is Part 1: Number of Participants With Any Treatment Emergent Adverse Events (TEAEs) and Serious Adverse Events (SAEs) (Arm 4).

2. Clinical Question

3. Trial Design

This is a Phase II, randomized, open-label, parallel-group platform trial sub-study in participants with previously treated NSCLC. It is part of master protocol 205801 (NCT03739710). The registry record identifies GlaxoSmithKline as sponsor and lists the study as completed. It records enrollment of 62 and dates of 2021-12-23 for study start and 2024-05-02 for primary completion.

Randomization means assignment to study groups is intended to use a chance-based process. The registry record identifies the study as randomized, but the randomization procedure, allocation ratio, and any restrictions on assignment are not reported in the registry record. Open-label means masking is listed as none; the record does not describe additional steps to limit awareness of treatment assignment.

The posted safety groups include three Arm 4 groups receiving dostarlimab plus belrestotug at different belrestotug dose levels. Arm 5 groups combine dostarlimab and belrestotug with nelistotug, with dose combinations and study parts identified in the group labels. These group descriptions show that safety results are presented across dose-specific cohorts and parts. They do not, by themselves, establish that every group was randomized against every other group or against SoC.

The registry record does not report the allocation details for these specific groups, the planned enrollment by group, the protocol’s sample-size justification, or whether enrollment changed during conduct. It also does not report the statistical analysis plan or analysis populations for each secondary outcome. The study summary describes safety and pharmacokinetic and pharmacodynamic assessment; response and concentration outcomes appear in the listed secondary outcomes. These design features should be kept in view when interpreting the posted counts: they are descriptive safety results across the listed treatment groups, not a complete account of comparative treatment effects.

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: This is a sub-study of the master study NCT03739710.The master protocol included two parts -Part 1 (non-randomized, safety and PK/PD evaluation) and Part 2 (randomized part comparing safety and efficacy to SoC). Criteria on whether to advance a study regimen from Part 1 to Part 2 has been built into the study design. Having completed Part 1 with the allowable number of patients per protocol, the decision was taken to not enroll participants in Part 2. Part 1 was therefore considered completed.

Only the first 6 of 7 groups are shown.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Safety Population included all participants who received at least 1 dose of standard of care (SoC) or experimental regimen based on actual treatment received.

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.

Only the first 6 of 8 groups are shown.

6. Primary Endpoint

The registry lists the primary outcome as Part 1: Number of Participants With Any Treatment Emergent Adverse Events (TEAEs) and Serious Adverse Events (SAEs) (Arm 4), measured over Up to approximately 97 weeks.

The listed primary outcome is the number of participants in Arm 4 with any TEAEs and SAEs. A TEAE is an event that began after study treatment started, or an event already present that worsened in intensity or frequency after exposure. An AE is an untoward medical occurrence temporally associated with use of a medicinal product, whether or not the event is considered related to that product. An SAE is defined by specified serious consequences, including death, life-threatening events, hospitalization or prolonged hospitalization, disability or incapacity, congenital anomaly or birth defect, and other situations associated with liver injury or impaired liver function. SAEs are a subset of AEs.

This endpoint is used to describe safety events occurring during treatment and follow-up, rather than to measure tumor response. The timeframe is up to approximately 97 weeks, and events were coded using the Medical Dictionary for Regulatory Activities (MedDRA). The posted results distinguish participants with serious events from participants with other reported events, with the at-risk count shown for each group. A count indicates how many participants had an event in the relevant category; it is not a measure of event severity beyond the SAE classification, nor does it establish that treatment caused the event.

The registry record does not report an adjusted treatment effect or a formal comparison for this primary outcome. The listed primary outcome is specific to Arm 4, while the adverse-event results also include Arm 5 groups. The counts therefore support a descriptive reading of the recorded safety experience, not a claim of comparative safety between regimens.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

What does a participant count tell us? The primary outcome is a count of participants who experienced TEAEs and SAEs, rather than a count of every event occurrence. A participant with multiple events may still contribute only once to a participant-level count for a particular outcome, depending on the outcome definition. The posted safety results give both the number at risk and the number affected for each group. Reporting both makes the observed count interpretable in relation to the group under observation. It does not, by itself, account for different lengths of follow-up or show when events occurred.

The at-risk figures for the displayed groups range from 6 to 11. Because these group sizes are limited, one participant represents a substantial part of a group’s observed experience. That is a general property of small samples: estimates can change noticeably with a small change in the observed data. For this reason, the counts are useful descriptive information, but they are not precise evidence of differences between regimens. The registry record does not report confidence intervals for the posted adverse-event counts.

What is the distinction between an AE, a TEAE, and an SAE? In the registry definition, an AE is a medical occurrence temporally associated with use of a medicinal product, regardless of whether it is considered related to the product. A TEAE is an event that was not present before treatment or that worsened after treatment began. An SAE is an event meeting the stated seriousness criteria. The record specifies that SAEs are a subset of AEs. These definitions concern event timing and seriousness; they do not establish that treatment caused an event. Causal attribution would require separate assessment, and the posted counts do not provide that assessment.

MedDRA coding is used to classify adverse events into standardized terms. Standardized coding helps organize events reported in clinical studies, but the listed terms do not convey every clinical detail, severity grade, duration, or causal assessment. The record lists selected serious-event terms and percentages, but does not report their denominators in the relevant listing. Those percentages therefore cannot be safely reinterpreted as risks for a particular group or compared across groups.

Why does randomization matter, and what does it not resolve? In general, random assignment can reduce systematic differences between groups at the point of assignment, supporting fairer comparisons when the groups, outcomes, and analysis are defined appropriately. It does not guarantee that each displayed group is directly comparable with every other group. The registry record identifies the overall sub-study as randomized and parallel, while the safety results include groups identified by dose and study part. It does not report the assignment process or establish which pairwise comparisons were planned. A comparison of the listed counts across groups should not be treated as a randomized contrast unless the record identifies it as one.

Open-label conduct means masking is listed as none. In general, knowledge of treatment assignment can influence some aspects of care, reporting, and assessment. This is particularly relevant to outcomes that involve judgment, although adverse-event definitions and coding can add structure to reporting. The registry record does not describe any additional procedures used to limit awareness or manage assessment bias.

How should confidence intervals and significance tests be understood? In general, a confidence interval describes a range of values compatible with the data under a specified statistical method and its assumptions. It communicates precision as well as the direction and size of an estimated effect. A significance test evaluates compatibility with a specified null hypothesis under a model; a p-value is not the probability that the null hypothesis is true and does not measure clinical importance. The registry record reports no confidence intervals or hypothesis tests for these safety comparisons, so no such inferential interpretation can be made from the counts alone.

What is over-dispersion? In general, over-dispersion occurs when observed variability is greater than a statistical model expects. It can make standard errors too small and confidence intervals too narrow if the model does not allow for the extra variability. It is often discussed for count outcomes, but the primary safety outcome is presented as participant counts with events, and the registry record does not report a model-based analysis or an over-dispersion assessment. It would not be appropriate to claim that over-dispersion was tested or corrected.

The registry record lists a planning illustration based on the observed effect, using a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. Such illustrations are conditional calculations: they describe the group size that would be needed under stated assumptions to detect a specified fraction of the observed effect. They are not the trial’s original sample-size justification, and they are not observed power. The registry record does not report the trial’s sample-size rationale, and planning illustrations should not be used as evidence that the trial was designed to detect a particular effect.

For a safety endpoint, interpretation also depends on how long participants were observed, how events were collected, whether events were graded, how missing follow-up was handled, and whether exposure differed among groups. The timeframe is listed as up to approximately 97 weeks, but the distribution of individual follow-up times and time-to-event methods are not reported in the registry record. The posted participant counts therefore describe the recorded experience without adjusting for exposure time or other differences among participants.

10. Multiplicity, Missing Data and Analysis Populations

The registry record lists 50 outcome measures, including safety, response, disease control, and pharmacokinetic measures. It does not report a statistical analysis plan describing which outcomes were confirmatory, how comparisons were ordered, or how multiplicity was controlled. Multiplicity arises when many outcomes or comparisons are examined: even when each test uses a conventional threshold, the chance of at least one apparently unusual result can increase across a family of tests. Without a stated testing strategy, any individual result should be interpreted as descriptive unless the record identifies it as a prespecified confirmatory analysis.

The record lists 0 analyses and does not post inferential results for the outcomes. It therefore does not provide test statistics, p-values, confidence intervals, or adjusted estimates for the safety groups. This absence does not show that no analyses were conducted; it means that analysis results and details are not reported in the registry record.

The safety population is defined as participants who received at least one dose of SoC or an experimental regimen, based on actual treatment received. The record does not describe the analysis populations used for response, disease control, or pharmacokinetic outcomes, nor does it report how participants with incomplete outcome information were handled. Methods for missing data, treatment discontinuation, censoring, and sensitivity analyses are not reported in the registry record.

These omissions matter because missingness can affect estimates if participants without complete follow-up differ systematically from those with recorded outcomes. General statistical methods can address some missing-data problems, but their suitability depends on why information is missing and on explicit assumptions. No particular method or assumption can be attributed to this sub-study from the registry record.

11. 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: febrile neutropenia (14.3%), dysphagia (14.3%), covid-19 pneumonia (14.3%), pneumonia (14.3%), epilepsy (14.3%).

Each safety group lists the number at risk, the number with serious adverse events, the number with other adverse events, and a death count. For example, the Arm 4 low-dose group records 0 serious events among 6 participants at risk and other adverse events among 6 of 6. These are participant counts, not counts of event episodes. A participant may have experienced multiple events, and the categories should not be summed as though they represented non-overlapping sets of participants.

The serious-event counts are 0, 3, and 2 in the Arm 4 low-, medium-, and high-dose groups, respectively; the corresponding counts in the Arm 5 groups are 3, 3, 3, and 3. The other-event counts are 6, 9, and 8 in Arm 4 and 10, 6, 11, and 9 in Arm 5, in the listed group order. The group sizes differ, so counts alone do not provide a like-for-like comparison. Even when the at-risk count is displayed, these small groups do not support strong conclusions about relative safety without an appropriate analysis and uncertainty estimates.

SAEs are a subset of AEs under the registry definition. The label “other” in the posted results should not be taken as proof that those events are unrelated, mild, or clinically unimportant. The death counts are listed separately, but the record does not report their causes or relationship to treatment. The serious-event term percentages likewise lack a clear denominator in the listing. Accordingly, these figures describe what was recorded, not why events occurred or whether one regimen caused more harm than another.

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

The posted results document adverse-event experience across dose-specific Arm 4 and Arm 5 groups in a Phase II NSCLC sub-study. They show that serious adverse events were recorded in several groups and that other adverse events were recorded in most participants in the displayed groups. They also list deaths and selected serious-event terms. These observations are relevant to understanding the safety information collected during the study, but they should be interpreted in the context of the group sizes, the open-label design, and the absence of inferential comparisons in the registry record.

The counts do not establish that a regimen caused a particular event. The registry definition explicitly includes events whether or not they are considered related to the medicinal product. Nor do the counts establish that a regimen is safer or less safe than another: the posted record does not provide a formal comparison, uncertainty estimates, or details about which group contrasts were planned. Differences in the displayed counts may reflect chance, group composition, time under observation, or other factors; the registry record does not provide enough information to distinguish among these possibilities.

The primary outcome is a safety count for Arm 4, while the posted adverse-event data include Arm 4 and Arm 5. The study summary also describes pharmacokinetic and pharmacodynamic assessment, and the listed secondary outcomes include objective response rate, disease control rate, and drug concentration measures. The information summarized in the registry record does not provide results for those outcomes. It therefore cannot answer whether the regimens changed tumor response, disease control, or drug concentrations, or how these outcomes compare with SoC.

In general, a trial result is most informative when the outcome definition, analysis population, follow-up, uncertainty, and intended comparison are all clear. Here, the safety counts are useful descriptive evidence, but they do not supply all of those elements. The record does not report missing-data methods, individual follow-up distributions, or a statistical analysis plan. Conclusions should remain limited to the reported counts and their stated definitions, without extending them to unreported outcomes or causal claims.

13. Why This Trial Matters Statistically

Platform trials can evaluate multiple novel regimens within a shared research structure. This sub-study illustrates why it is important to distinguish the overall platform description from the specific treatment groups and outcomes reported for an individual sub-study. The title refers to novel regimens versus SoC, but the adverse-event results are organized by Arm 4 and Arm 5 dose groups. The registry record does not specify a direct SoC comparison in those safety results. Clear identification of the actual comparison is essential before interpreting any difference as a treatment effect.

Safety data are central to early-phase evaluation because they describe the types and frequency of events observed during exposure and follow-up. Participant-level counts can be transparent and clinically legible, especially when paired with clear event definitions and at-risk counts. Yet counts from small groups are inherently imprecise: a small number of participants can materially influence the pattern. A reader should distinguish descriptive reporting from a comparative estimate supported by an analysis.

This record also highlights the value of complete registry reporting. An analysis plan, definitions of analysis populations, missing-data handling, follow-up detail, and uncertainty estimates would help readers assess how robustly observed results address the study questions. The record identifies a range of secondary outcomes, but does not post analyses for them. Reporting outcome-specific methods and results would make it easier to understand how safety, response, and pharmacokinetic findings fit together.

The statistical importance is not that the counts settle a question of comparative benefit or risk. Rather, they show how dose-specific safety observations must be read alongside study design, outcome definitions, and the limits of available analysis. For clinical researchers, this is a reminder that the presence of randomization at the trial level does not make every descriptive group contrast a randomized comparison. The target estimand, treatment assignment, analysis set, and uncertainty all matter.

14. 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 posted safety data are descriptive counts from small dose-specific groups. The registry record does not report confidence intervals or formal tests for comparisons between these groups, and group counts alone do not quantify the precision of an effect estimate. The treatment groups also differ in their listed dose combinations and study parts, so apparent differences should not be interpreted as dose-response evidence without a defined analysis.

Although the overall study is identified as randomized, the specific allocation process for the displayed groups and the planned pairwise comparisons are not reported in the registry record. The record does not identify a SoC group among the displayed adverse-event results. Consequently, the results do not provide a direct comparative safety estimate against SoC in the information presented.

Follow-up is described as up to approximately 97 weeks, but the record does not report the actual follow-up distribution, exposure duration by group, event timing, or time-to-event analyses. Unequal time at risk can affect the number of events observed. The posted counts do not adjust for such differences.

The definitions establish that SAEs are a subset of AEs, but the registry record does not report event grades, detailed attribution, participant-level overlap, or the full criteria applied to individual events beyond the stated definition. Death counts are presented without cause, timing, or treatment relationship. Selected serious-event term percentages are listed without a denominator in that listing, limiting interpretation.

The record does not report missing-data methods, analysis populations for the secondary outcomes, interim analyses, multiplicity procedures, or the study’s sample-size justification. It also lists no analyses. These limitations prevent assessment of the precision, robustness, or confirmatory status of the results and restrict interpretation to the posted descriptive safety information.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: Platform Trial of Novel Regimens Versus Standard of Care (SoC) in Participants With Non-small Cell Lung Cancer (NSCLC) - Sub-study 3 — NCT06926673.
• Registry record last updated 2025-07-03; status last verified 2025-06.
• Results first posted 2025-07-03.
• 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.

18. Record Summary

NCT06926673 is a completed, randomized, open-label Phase II platform-trial sub-study in participants with previously treated NSCLC. The study summary describes assessment of safety and pharmacokinetics and pharmacodynamics for regimens containing dostarlimab and belrestotug, with nelistotug included in some groups. The registry lists enrollment of 62, a start date of 2021-12-23, and primary completion on 2024-05-02.

The primary outcome is the number of Arm 4 participants with TEAEs and SAEs, over up to approximately 97 weeks. Posted safety counts cover three Arm 4 dose groups and four Arm 5 groups. The registry also lists response, disease-control, and drug-concentration outcomes, but does not report analysis results for them. No formal comparative safety estimates, confidence intervals, or analysis methods are reported in the registry record. The available counts therefore describe recorded events within the listed groups and do not establish comparative or causal effects.