A Study to Compare How Effective is Encorafenib Plus Binimetinib in Real-world and Clinical Trial Settings: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study to Compare How Effective is Encorafenib Plus Binimetinib in Real-world and Clinical Trial Settings (NCT05954546), a registered phase not stated study. The primary outcome is Overall Survival (OS).

2. Clinical Question

The study’s stated purpose is to compare effectiveness in clinical-trial and real-world settings and to assess whether data from those settings can be combined in future research. Its population is therefore defined partly by the available data and eligibility rules, rather than being a single cohort enrolled under one shared setting. The COLUMBUS and Flatiron records have different starting points for survival follow-up and different follow-up descriptions. Those distinctions matter when interpreting a difference between the settings: an observed contrast could reflect differences in patient selection, data capture, follow-up, or other features, as well as differences in outcomes. The registry record does not report a randomized allocation or a like-for-like control comparison. It also does not report numerical comparative survival results, so the direction and precision of any difference cannot be assessed from the recorded information.

3. Trial Design

This is a nonrandomized comparison of overall survival for patients receiving encorafenib plus binimetinib in two settings: the COLUMBUS clinical trial and Flatiron real-world practice data. The record identifies the study as completed, with enrollment of 275. It describes one primary outcome measure and one analysis. The study is sponsored by Pfizer.

The real-world group was identified from a database. The description requires at least one order or administration of the combination after confirmation of metastatic melanoma after 27 June 2018, age 18 years or older at first treatment initiation, and ECOG status of 0 or 1 at that date. It also describes treatment-history requirements and availability of data on deaths. Follow-up begins on the treatment initiation date and ends at the last date of data availability, which may reflect loss to clinic follow-up, death, or the end of the database. The database timeframe is June 2018 to January 2020.

The clinical-trial and real-world groups do not have identical time origins as described in the registry. COLUMBUS follow-up begins at randomization, while Flatiron follow-up begins at treatment initiation. The registry lists approximate observation periods of 80.5 months for COLUMBUS and 42.2 months for Flatiron. Differences in data source, eligibility, follow-up, and outcome ascertainment are relevant to interpretation. The record explicitly notes that this is not a randomized comparison and that the groups are not a like-for-like control comparison. It does not report how patients were matched or adjusted across settings, nor does it report a detailed statistical analysis plan.

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: Data sources: COLUMBUS trial (NCT01909453) and Flatiron Health Electronic Health Records (EHR) real-world database (RWD). Planned cohorts in this study: 1) ENCO+BINI, clinical trial data (CTD) from COLUMBUS; 2) ENCO+BINI (RWD) from Flatiron EHR. Available retrospective data was evaluated in this observational study from 21-Jul-2023 to 31-Dec-2023 \[approximately 5.35 months\].

Overall Study

Analysis population for the primary outcome, as defined in the registry: Analysis population included all eligible participants whose data were retrieved and observed in this 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.

Population: Eligible participants whose data was retrieved and observed in the study.

6. Primary Endpoint

The registry lists the primary outcome as Overall Survival (OS), measured over COLUMBUS: From date of randomization until death or censoring (approx 80.5 months); Flatiron: From index date until death due or censoring or end of follow-up period (approx 42.2 months).

The primary endpoint is overall survival, measured in months. In COLUMBUS, overall survival is defined from randomization to death from any cause. If a death was not observed, follow-up was censored at the last contact or the data analysis cut-off date, whichever occurred first. In Flatiron, the measure runs from the treatment-initiation index date to death; patients without a recorded death date were censored at their last known activity date or the end of the follow-up period, whichever occurred first.

Overall survival is a time-to-event measure: it incorporates both whether an event was observed and how long the participant was followed before death or censoring. It is commonly used in clinical research because death is an outcome with a clear definition, although the completeness and timing of death records remain important. Censoring allows people without an observed death during their available follow-up to contribute information up to their last known observation. It does not establish what happened after that point.

The registry states that Kaplan-Meier analyses were used. It identifies the primary measure as the median and the dispersion as a 95% confidence interval, but does not report numerical medians or confidence intervals. The differing time origins and data sources mean that the endpoint’s operational definition is not identical in the two settings. Those distinctions should be considered before treating the results as directly comparable.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

Overall survival is analyzed as a time-to-event outcome. For each person, the analysis uses a start date, an event date if death is observed, or a censoring date if death is not observed during the available follow-up. The COLUMBUS definition starts at randomization and counts death from any cause as the event. The Flatiron definition starts at treatment initiation and uses the recorded date of death. For people without an observed death, the record describes censoring at the last contact or activity date, or at the relevant cut-off or end of follow-up, as applicable to the source.

What does Kaplan-Meier analysis estimate?

In general, the Kaplan-Meier method estimates the probability of remaining alive over time while accommodating participants whose follow-up ends before a death is recorded. Rather than treating every censored person as if they had died, the method uses the information available up to censoring. A survival curve can therefore represent changing survival estimates across follow-up, and the median is the time at which the estimated survival reaches the halfway point, if the curve reaches that point. A confidence interval conveys uncertainty around an estimate; it is not a guarantee that the true value lies within the interval.

The registry identifies the median and a 95% confidence interval as the reported outcome format, but it does not give their numerical values. It also does not report Kaplan-Meier curves, numbers at risk over time, or the method used to calculate confidence intervals. Those omissions mean readers cannot assess how the estimates change over follow-up or how much information remains at later times.

Why does censoring matter?

As a general statistical concept, censoring is useful when a person’s event time is not observed in full. It preserves information from the time that person was followed, but it does not reveal their later outcome. Kaplan-Meier estimates rely on assumptions about censoring, including that the censoring process does not systematically remove people with different underlying survival prospects in a way that distorts the estimate. If follow-up ends for reasons associated with health status or access to care, the resulting estimates may be affected. The record lists possible end points of follow-up but does not provide details on the frequency or reasons for censoring in either group.

What changes when the data sources differ?

The registry uses COLUMBUS trial data for one group and Flatiron electronic health record data for the other. These settings can differ in how patients are selected, how treatment starts are documented, and how deaths and follow-up are captured. The endpoint definitions are related but not identical: the COLUMBUS clock begins at randomization, while the Flatiron clock begins at treatment initiation. In general, a comparison is easier to interpret when time origins, eligibility, outcome definitions, and follow-up are aligned. The registry does not report a harmonization procedure or a method to account for these differences.

The record specifies approximate follow-up periods of 80.5 months in COLUMBUS and 42.2 months in Flatiron. These descriptions indicate that the observation windows differ, but they do not by themselves show the distribution of follow-up or the number of people observed at each point. A median can be estimated from censored data when the available follow-up supports it; without the estimates and risk information, the adequacy and precision of each median cannot be evaluated.

How should a hazard ratio be understood?

In general, a hazard ratio compares the event rates between groups over time, under the assumptions of the model used to estimate it. It is a relative measure and is not the same as a difference in median survival or a direct statement about the probability that an individual will live for a particular period. If the relative event rate changes over follow-up, a single hazard ratio may not summarize the contrast well. The registry entries do not state a numerical hazard ratio or describe a hazard-ratio model. They also do not report whether model assumptions were checked. Consequently, no comparative hazard-ratio interpretation can be made from the record details presented.

What can the listed death counts establish?

The record lists 131 deaths for the clinical-trial group and 28 for the real-world group. Event counts are relevant to time-to-event analyses, but they do not provide the full information needed to compare survival. Interpretation also depends on the number of people in each group, duration of observation, censoring, and the timing of events. The record reports enrollment of 275 but does not give setting-specific group sizes in the details. It would therefore be inappropriate to convert the death counts into rates or percentages, or to infer a comparative effect from the counts alone.

Overall, the registry identifies an appropriate family of methods for a censored survival endpoint, but does not report the numerical output or enough analysis detail to independently evaluate the comparison. The median and confidence interval format, endpoint definitions, and Kaplan-Meier method are described; the estimates, risk sets, censoring summaries, and model details are not reported in the registry record.

10. Multiplicity, Missing Data and Analysis Populations

The record identifies one primary outcome measure and one analysis: overall survival. It does not report a multiplicity adjustment, a hierarchy of outcomes, or whether other analyses were planned. Because the listed record contains only one primary outcome, there is no reported basis for describing how the study handled testing across multiple primary endpoints or multiple comparisons. Any broader multiplicity strategy is not reported in the registry record.

The outcome definition explains how participants without an observed death were censored. COLUMBUS participants were censored at last contact or the analysis cut-off, whichever came first; Flatiron participants without a death date were censored at last known activity or the end of follow-up, whichever came first. The registry does not report how many participants were censored, the timing of censoring, missingness in key variables, or whether sensitivity analyses assessed the effect of missing or incomplete data. It also does not say how missing baseline or outcome information was handled beyond the stated eligibility requirement for available death information.

The analysis population is described as eligible participants whose data were retrieved and observed in the study. The real-world eligibility criteria include age 18 years or older, ECOG status of 0 or 1 at treatment initiation, specified treatment-history conditions, and availability of death data. The record does not provide a detailed flow of screened, excluded, or analyzed participants by setting, nor does it report setting-specific sample sizes. Enrollment is listed as 275, but the relationship between that figure and the analysis groups is not detailed.

These gaps limit assessment of whether the analysis population represents the intended populations and whether incomplete follow-up could influence estimates. The registry record does not report a statistical plan, interim analyses, or a sample-size justification. No assumptions about those features can be made from the outcome description alone.

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

The study addresses a relevant statistical challenge: comparing survival outcomes across a clinical trial and routine-care data source. Such comparisons can inform research on how evidence from different settings might be brought together, but they require careful attention to who enters each dataset, when follow-up begins, and how outcomes are recorded. The registry describes those settings and the endpoint definitions, but does not report the numerical survival estimates needed to quantify a result.

The record’s stated purpose is to compare effectiveness across settings and produce results relevant to whether data from the settings can be combined in future research. That purpose should not be mistaken for a conclusion that the datasets are already comparable or suitable for pooling. Establishing comparability would require evidence about population alignment, outcome measurement, follow-up, and the analysis used to handle differences. Those details and conclusions are not reported in the registry record.

The record reports death counts for each group, but counts alone do not show survival duration or the relative timing of deaths. Nor do they establish that a treatment caused any observed difference. This is not a randomized comparison, and the record warns that the groups are not like-for-like controls. Differences between groups could reflect patient or data-source characteristics rather than treatment effects. The absence of reported survival estimates and confidence intervals prevents assessment of the size and precision of any observed contrast.

Therefore, the statistical meaning of the available record is primarily methodological and descriptive. It defines overall survival, states that Kaplan-Meier analysis was used, and describes distinct follow-up origins and censoring rules. It does not provide enough numerical evidence to conclude whether survival differed, by how much, or with what uncertainty. It also does not establish that a result from one setting can be generalized to the other. Those conclusions would require fuller comparative results and analytic detail.

12. Why This Trial Matters Statistically

Clinical trials and real-world data answer related but not identical questions. A clinical trial follows a defined protocol and population, while routine-care records reflect treatment and follow-up as documented in practice. Statistical comparisons across these sources may help researchers understand the reach of trial findings and explore whether data can be combined. This study is framed around that problem for encorafenib plus binimetinib in metastatic BRAFV600-mutant melanoma.

The central statistical issue is comparability. The record describes different eligibility pathways, different sources of information, distinct time origins for overall survival, and different approximate follow-up periods. These are not minor labels: they affect what population is represented and what elapsed time the endpoint summarizes. Explicitly stating those definitions is useful because it gives readers a basis for assessing whether a cross-setting contrast is interpretable.

The comparison also illustrates why event counts are not enough. Although the record lists deaths in both groups, a survival analysis uses event timing and censoring as well as counts. Numerical estimates and uncertainty would be needed to judge the magnitude and precision of a contrast. The registry entries do not report those results, so the record cannot establish a quantitative survival finding.

Finally, the stated goal of assessing whether clinical-trial and real-world data can be combined is itself a methodological question. Evidence of compatibility would need to address population differences, outcome ascertainment, and analysis choices. The record identifies the motivation and basic endpoint framework, while leaving the comparative estimates and key analytic details unreported. That distinction is important for readers who may otherwise treat a cross-setting comparison as equivalent to a randomized trial result.

13. 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 main limitation is that the registry record does not report the numerical primary outcome results. It identifies the median and 95% confidence interval as the outcome format, but gives no median, confidence interval, or hazard ratio. The death counts alone cannot establish the direction or magnitude of a survival contrast, and the precision of any estimate cannot be assessed.

The groups come from different settings and are explicitly not described as like-for-like controls. The record does not report setting-specific group sizes, baseline characteristics, matching, adjustment, or other methods used to address differences between the groups. The enrollment figure of 275 does not clarify how participants were distributed across the clinical-trial and real-world groups.

Follow-up is not defined identically: COLUMBUS time begins at randomization, while Flatiron time begins at treatment initiation. The registry also reports approximate follow-up periods of 80.5 months and 42.2 months, respectively. It does not give risk sets, censoring summaries, or details that would allow readers to evaluate how unequal observation and data availability affect the estimates.

The real-world data cover June 2018 to January 2020, and the eligibility description requires available death information. The record does not describe the completeness of death ascertainment, missing data handling, or sensitivity analyses. It also does not report a statistical analysis plan, interim analyses, or sample-size justification. These limitations leave important questions about representativeness, measurement, and uncertainty unresolved.

14. Related Tutorials

15. Related Statistical Calculators

16. Sources

• ClinicalTrials.gov: A Study to Compare How Effective is Encorafenib Plus Binimetinib in Real-world and Clinical Trial Settings — NCT05954546.
• Registry record last updated 2025-02-05; status last verified 2025-02.
• Results first posted 2025-02-05.
• 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.

17. Record Summary

NCT05954546 is a completed study sponsored by Pfizer comparing overall survival for encorafenib plus binimetinib in COLUMBUS clinical-trial data and Flatiron real-world practice data. The registry lists enrollment of 275 and one primary outcome measure: median overall survival in months with a 95% confidence interval. Kaplan-Meier analysis is specified.

Overall survival begins at randomization in COLUMBUS and at treatment initiation in Flatiron. The record describes approximate follow-up periods of 80.5 months and 42.2 months, respectively, and lists 131 deaths in the clinical-trial group and 28 in the real-world group. It does not report numerical survival estimates, confidence intervals, or a hazard ratio. The comparison is not randomized, and the record cautions that the groups are not like-for-like controls.