Dose Escalation and Expansion Study of FLX475 Monotherapy and in Combination With Pembrolizumab: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Dose Escalation and Expansion Study of FLX475 Monotherapy and in Combination With Pembrolizumab (NCT03674567), a non-randomized phase 1, phase 2 study. The primary outcome is Safety and Tolerability of FLX475 as a Single Agent and in Combination With Pembrolizumab Measured by the Incidence of Adverse Events, Including Dose-limiting Toxicities and Maximum Tolerated Dose.

2. Clinical Question

3. Trial Design

This is a Phase 1/2, open-label, non-randomized study with sequential dose escalation and cohort expansion. The registry lists enrollment of 323 and describes two parts. In Part 1, eligible participants entered sequential cohorts receiving successively higher doses of FLX475, either alone or with pembrolizumab. In Part 2, expansion cohorts received FLX475 monotherapy or the combination; the description says participants were initially enrolled in Stage 1, with additional participants in each cohort potentially enrolled in Stage 2.

The posted adverse-event groups reflect these different study components. Monotherapy dose-escalation groups are listed at 25 mg, 50 mg, 75 mg, and 100 mg. Combination dose-escalation groups are listed at 50 mg, 75 mg, and 100 mg. Expansion groups are identified as monotherapy cohorts M1 through M5 and combination cohorts C1 through C4, C8, C9, C10, and C11; C10 and C11 are described as lead-in cohorts. These are cohort labels, not randomized treatment assignments.

The trial is open-label and has no masking. Because allocation was non-randomized and the study followed a sequential model, cohort membership and treatment were not assigned through a randomized comparison. The registry itself cautions that the arms are not a like-for-like control comparison. Any apparent differences across dose levels or cohorts may reflect the way participants entered the study, cohort composition, or other factors; the posted data do not establish that a difference was caused by treatment.

The registry lists the study as completed. The recorded start date is 2018-09-25, and the primary completion date is 2024-12-31. RAPT Therapeutics, Inc., a member of the GSK plc group of companies, is listed as sponsor. The registry record does not report additional detail about how doses were selected or how transitions between escalation and expansion were governed.

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.

Only the first 6 of 20 groups are shown.

Overall Study

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 21 groups are shown.

6. Primary Endpoint

The registry lists the primary outcome as Safety and Tolerability of FLX475 as a Single Agent and in Combination With Pembrolizumab Measured by the Incidence of Adverse Events, Including Dose-limiting Toxicities and Maximum Tolerated Dose, measured over Approximately 18 weeks.

The primary outcome is titled as safety and tolerability of FLX475 as a single agent and in combination with pembrolizumab. It names the incidence of adverse events, including dose-limiting toxicities and maximum tolerated dose. The outcome description specifies treatment-emergent adverse events, the unit is participants, and the stated timeframe is approximately 18 weeks.

Safety outcomes are important in dose-escalation research because treatment-related harms and tolerability inform decisions about whether a dose or regimen merits further evaluation. In general, a dose-limiting toxicity is an adverse event used under a study’s defined rules to limit escalation or guide dose selection. A maximum tolerated dose is a dose boundary determined using those rules. The registry record does not provide the detailed criteria used for either measure, nor does it report a primary endpoint result that identifies a maximum tolerated dose.

The adverse-event listing provides counts of participants at risk and counts with serious or other adverse events by cohort. Those counts offer descriptive safety information, but they do not by themselves give the full endpoint analysis. The registry does not report the event definitions, severity grading rules, attribution rules, or a complete analytic method for the primary outcome. In particular, the posted group counts should not be read as a complete account of every safety assessment conducted during the study.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The design and reported results call for a descriptive interpretation. This study was non-randomized, open-label, and sequential, with dose-escalation cohorts followed by expansion cohorts. In a randomized trial, random allocation can help make treatment groups comparable in expectation. That protection is absent. Cohorts may differ in ways that are not captured by the event counts, so comparing their raw counts does not estimate a randomized treatment effect.

What do the cohort counts represent?

The registry lists participants at risk for each adverse-event category and participants affected. These are counts, not rates adjusted for time at risk. A participant count tells how many people in the listed group had at least one event in that category as recorded; it does not convey the number of recurrent events or how long each participant was observed. The registry record does not describe the counting rules in enough detail to determine whether every event category uses the same event-level conventions.

Group sizes vary, so counts alone are not suitable for direct comparisons across cohorts. Even a percentage based on the counts would not address differences in cohort composition, follow-up, or treatment assignment. This is particularly important in sequential escalation, where participants enter different cohorts by dose and study stage rather than by random allocation. The posted results should therefore be interpreted group by group and in the context of the study design.

Why does over-dispersion matter?

In general, over-dispersion means that observed variation is greater than a simple statistical model expects. For event counts, this can arise when participants have differing underlying risks or when events cluster within participants or groups. A model that ignores over-dispersion can give uncertainty estimates that are too narrow. The registry record does not report a count model, a dispersion estimate, or an analysis addressing over-dispersion; the posted safety data are participant counts rather than an inferential count-model result.

What can a confidence interval add?

In general, a confidence interval expresses the uncertainty around an estimated quantity under a stated statistical method. It can indicate which values are reasonably compatible with the observed data and model assumptions. It does not remove bias from a non-randomized comparison, and it does not turn an association into a causal effect. The registry record reports no confidence intervals for the listed safety counts or serious-event terms, so the precision of those estimates cannot be assessed from the record using intervals.

The registry lists no analyses. It does not report a statistical model for the primary safety outcome, an adjusted analysis, or a hypothesis test for between-cohort differences. There are no posted recalculated comparisons to interpret. The summary counts remain the appropriate basis for describing what was recorded, while formal claims of differences would require methods and assumptions not reported in the registry record.

How should planning illustrations be read?

The planning illustration is based on the observed effect and uses a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. It represents participants per arm for a specified fraction of the observed effect; it is not the trial’s own design justification and is not observed power. Such planning values are conditional on their assumptions and should not be mistaken for a feature of this non-randomized trial. The registry record does not report a sample-size justification, and the planning illustration does not substitute for one.

For this trial, the posted safety counts do not define a controlled effect to detect. The study has no randomized control comparison, and the assumptions in the planning illustration are not the study’s allocation scheme. Any planning use therefore requires a separate, clearly specified future comparison and an effect assumption appropriate to that question. The registry record does not report such a future design.

10. Multiplicity, Missing Data and Analysis Populations

The registry record does not report a multiplicity plan. It does not say whether statistical hypotheses were prespecified, whether any adjustment for multiple outcomes or comparisons was used, or how such adjustments would have been applied. This matters because the study includes several treatment cohorts, doses, and adverse-event categories. Looking across many groups or outcomes can produce apparent differences by chance, especially if those comparisons were not planned in advance. The posted descriptive counts should not be interpreted as confirmatory tests.

Missing-data handling is also not reported in the registry record. There is no description of how participants with incomplete follow-up were handled, how missing safety observations were classified, or whether follow-up differed across cohorts. The at-risk counts identify the group denominator for the listed adverse-event categories, but they do not explain whether any participants were excluded from a particular analysis or why.

Analysis populations are not reported in the registry record. The record does not specify whether the safety analyses included everyone who received treatment, everyone enrolled, or another defined group. Nor does it provide a participant flow accounting for enrollment, treatment exposure, discontinuation, or outcome availability. The registry’s enrollment of 323 should not be assumed to be the analysis denominator for any particular outcome; the adverse-event groups have their own listed at-risk counts.

These omissions limit assessment of how completely the posted counts represent the study experience and whether selective availability of data could affect interpretation. They do not establish that data were missing or that an analysis was inappropriate; they mean the handling and analysis populations cannot be evaluated 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: pneumonitis (10.0%), pneumothorax spontaneous (10.0%), diarrhoea (8.3%), pyrexia (8.3%), electrocardiogram qt prolonged (7.7%).

For each cohort, the registry lists the number of participants at risk and the number affected by serious adverse events, other adverse events, and deaths. Read each count within its own cohort. For example, the monotherapy expansion cohort M3 lists 1 participant with a serious adverse event among 13 at risk, while combination cohort C3 lists 3 among 24 at risk. Those are descriptive counts; the cohort design does not support a randomized comparison between them.

The serious-event terms are listed with percentages, including Pneumonitis at 10.0% and Pneumothorax spontaneous at 10.0%. The registry record does not report the denominator or calculation details for those percentages. They should not be equated with the proportions in any single cohort. In addition, an event term may not correspond to a unique participant count, and the record does not explain whether participants could contribute to multiple terms.

Deaths are separate from the serious adverse-event count in the listing. A death reported for a cohort does not, by itself, show that treatment caused it. The registry record does not provide the attribution or clinical details needed to make that determination. The counts describe what is posted, not a causal safety comparison.

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

The results describe safety events recorded in cohorts receiving FLX475 alone or with pembrolizumab. They show that adverse events, including serious adverse events, were reported in the listed groups, and that the recorded counts differ across cohorts. The seriousness categories, term percentages, and death counts are separate kinds of information and should each be interpreted according to their definitions and denominators.

These findings do not establish whether FLX475 or pembrolizumab caused a particular event. The registry record does not provide enough information about attribution, event definitions, analysis populations, or follow-up completeness to resolve those questions. A death count is not evidence of treatment causation, and a cohort-level count cannot distinguish treatment effects from differences in the participants or observation across cohorts.

The study is not a randomized comparison. It therefore does not establish that one dose, monotherapy, or combination treatment is safer than another. Raw differences in counts across groups are not equivalent to estimates from a controlled comparison. The record also does not report inferential analyses, uncertainty intervals, or the detailed results needed to evaluate a treatment contrast.

The primary outcome description focuses on safety and tolerability over approximately 18 weeks and names dose-limiting toxicities and maximum tolerated dose. However, the registry does not report a primary endpoint result identifying a maximum tolerated dose or provide the detailed criteria used to define dose-limiting toxicity. The study summary also names preliminary anti-tumor activity as an aim, but the posted outcome information described in this record does not establish an anti-tumor effect.

In short, the registry supports a descriptive account of adverse events by cohort. It does not support medical advice, a conclusion that a regimen works or does not work, or a causal claim about efficacy or harm. The limits reflect the non-randomized design and the statistical and clinical details not reported in the registry record.

13. Why This Trial Matters Statistically

This study illustrates why early-phase safety data need to be interpreted alongside design. Sequential dose escalation and cohort expansion can provide a structured way to observe tolerability across regimens and doses. But the resulting cohorts are not randomized arms, so their observed counts do not have the same interpretation as a controlled treatment comparison. The distinction between describing events and estimating a causal effect is central to biostatistical interpretation.

The record also illustrates the importance of denominators and endpoint definitions. A count of participants affected can be informative, but it does not convey event recurrence, time at risk, or the rules used to classify events. Percentages for serious-event terms add a scale for interpretation, yet their denominator and calculation details are not reported in the registry record. Without those details, the percentages cannot be independently assessed from the cohort listings.

Finally, the study record highlights how much statistical interpretation depends on reporting. Information about analysis populations, missing observations, multiplicity, endpoint criteria, and uncertainty would help readers judge completeness and precision. Their absence does not erase the descriptive safety information, but it narrows the conclusions that can be drawn. This is a useful example of separating what a registry count directly records from what would require a prespecified statistical analysis or a randomized design.

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 principal limitation for interpretation is the non-randomized design. Cohorts receiving different doses or regimens are not like-for-like controls, so differences in their posted event counts cannot be attributed to treatment. The sequential design and expansion structure also mean that cohorts enter under different study stages and treatment assignments rather than through random allocation.

The primary safety outcome is not fully operationalized in the registry record. Although it names adverse events, dose-limiting toxicities, and maximum tolerated dose, the record does not report detailed event grading, attribution, dose-limiting criteria, or a result identifying the maximum tolerated dose. The listed treatment-emergent adverse events are therefore not a complete substitute for a full endpoint analysis.

Analysis populations, missing-data procedures, and participant flow are not reported in the registry record. The record does not explain how incomplete follow-up was handled or whether the posted at-risk groups include all participants who received treatment. The percentages for serious-event terms also lack a reported denominator and calculation description.

There are no reported inferential analyses or confidence intervals for the safety counts. The absence of those statistics limits assessment of precision, while the absence of a randomized comparison limits causal interpretation regardless of precision. The registry record also does not provide results for the stated preliminary anti-tumor activity aim in the outcome information. These limitations define what can be concluded from the posted record; they do not establish that unreported analyses or procedures were not conducted.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: Dose Escalation and Expansion Study of FLX475 Monotherapy and in Combination With Pembrolizumab — NCT03674567.
• Registry record last updated 2026-09-03; status last verified 2026-08.
• Results first posted 2026-09-03.
• 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

NCT03674567 is a completed, open-label, non-randomized Phase 1/2 study of FLX475, also called tivumecirnon, as monotherapy and in combination with pembrolizumab in advanced cancer. The study used sequential dose escalation followed by expansion cohorts. The registry lists enrollment of 323, a start date of 2018-09-25, and a primary completion date of 2024-12-31.

The primary outcome concerns safety and tolerability, including treatment-emergent adverse events, dose-limiting toxicities, and maximum tolerated dose, with a timeframe of approximately 18 weeks. The posted safety data provide cohort-level counts for serious and other adverse events and deaths, plus percentages for selected serious-event terms. No statistical analyses are listed. The registry record does not report analysis populations, missing-data handling, multiplicity methods, or a randomized treatment comparison, so the results are best read descriptively.