This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record. Numerical trial results on this page are restricted to the ClinicalTrials.gov record.
1. Trial at a Glance
BEACON CRC was a randomized, unmasked, phase 3 trial with a sequential design in BRAF V600E-mutant metastatic colorectal cancer. The study enrolled 702 participants across three arms and included a safety lead-in followed by the phase 3 portion of the study.
| Feature | BEACON CRC |
|---|---|
| Trial name | BEACON CRC |
| NCT ID | NCT02928224 |
| Phase | Phase 3 |
| Status | Completed |
| Condition | BRAF V600E-mutant metastatic colorectal cancer |
| Allocation | Randomized |
| Design model | Sequential |
| Masking | None |
| Primary purpose | Treatment |
| Enrollment | 702 |
| Arms | 3 |
| Primary endpoints registered | 6 |
| Outcome measures posted | 70 |
| Statistical analyses posted | 16 |
| Lead sponsor | Pfizer |
| Sponsor type | Industry |
2. Clinical Question
The clinical question was whether treatment strategies containing encorafenib, cetuximab, and, in the triplet strategy, binimetinib could improve prespecified outcomes compared with the phase 3 control regimen in patients with BRAF V600E-mutant metastatic colorectal cancer.
Population
Participants with BRAF V600E-mutant metastatic colorectal cancer.
Interventions
Encorafenib, binimetinib, and cetuximab were the study drugs used in the treatment strategies. The ClinicalTrials.gov record also identify irinotecan, folinic acid, and 5-fluorouracil among the interventions.
Comparator
The phase 3 control arm was the comparator for the primary triplet-versus-control analyses. The ClinicalTrials.gov record identifies the control strategy as irinotecan/cetuximab or infusional 5-fluorouracil/folinic acid/irinotecan/cetuximab.
Primary questions
The registered phase 3 primary analyses compared overall survival and objective response rate for the triplet arm versus the control arm.
3. Trial Design
The sequential structure is important statistically because the study contained a safety lead-in before the phase 3 randomized portion. The registry separately defines safety-lead-in endpoints and phase 3 efficacy endpoints. The safety lead-in therefore should not be treated as though it were simply another randomized treatment comparison.
Study interventions
Triplet safety lead-in
- Encorafenib
- Binimetinib
- Cetuximab
- Safety endpoints included dose-limiting toxicities, adverse events, and dose interruptions, modifications, and discontinuations due to adverse events.
Three randomized arms
- Triplet arm
- Doublet arm
- Control arm
- Phase 3 analyses compared these groups for OS, PFS, and ORR according to the posted analysis records.
4. Registered Primary Endpoints
The registry lists six primary endpoints. They span safety-lead-in binary outcomes and phase 3 efficacy outcomes involving both time-to-event and binary endpoints.
| Primary endpoint | Time frame | Type | Results posted |
|---|---|---|---|
| (Safety Lead-in) Number of Participants With Dose-Limiting Toxicities (DLTs) | Cycle 1 (up to 28 days) | Binary | Yes |
| (Safety Lead-in) Number of Participants With Adverse Events (AEs) | Duration of safety lead-in, approximately 6 months (up to 28 days per cycle) | Binary | Yes |
| (Safety Lead-in) Incidence of Dose Interruptions, Dose Modifications and Discontinuations Due to Adverse Events (AEs) - Interim Analysis | Duration of safety lead-in, approximately 6 months (up to 28 days per cycle) | Binary | Yes |
| (Safety Lead-in) Incidence of Dose Interruptions, Dose Modifications and Discontinuations Due to Adverse Events (AEs) - Final Analysis | From start of study treatment until 30 days post last dose of study treatment (maximum treatment exposure of 280 weeks) | Binary | Yes |
| (Phase 3) Overall Survival (OS) of Triplet Arm vs. Control Arm - Interim Analysis | From randomization to death due to any cause until 204 deaths were observed (maximum treatment exposure of 89.1 weeks for triplet arm and 52.4 weeks for control arm) | Time-to-event | Yes |
| (Phase 3) Objective Response Rate (ORR) by Blinded Independent Central Review (BICR) Per Response Evaluation Criteria in Solid Tumors (RECIST), v1.1 of Triplet Arm vs. Control Arm | Duration of Phase 3, approximately 6 months (up to 28 days per cycle) | Binary | Yes |
Registered endpoint definitions
Overall survival: OS was defined as the time from randomization to death due to any cause.
Objective response rate: ORR per RECIST, v1.1, was defined as the percentage of participants achieving an overall best response of complete response (CR) or partial response (PR), where CR was defined as disappearance of all target and non-target lesions and normalization of tumor marker level, with all lymph nodes non-pathological in size (<10 millimeter [mm] short axis), and PR as at least 30% decrease in sum of diameters of target lesions, taking as reference the baseline sum diameters persistence of one or more non-target lesions and/or maintenance of tumor marker level above the normal limits.
5. Analysis Populations
The registry uses distinct analysis populations for the principal phase 3 endpoints. That distinction is important because the OS analysis and the ORR analysis do not use exactly the same analysis-set definition.
| Endpoint | Analysis population | Why it matters |
|---|---|---|
| Phase 3 OS | The Full Analysis Set (FAS) for the Phase 3 portion of the study consisted of all randomized Phase 3 participants. | The analysis is anchored to randomized treatment assignment. |
| Phase 3 ORR | The Phase 3 Response Efficacy Set consisted of the first 330 participants randomized into the Phase 3 portion of the study and any additional participants randomized on the same day as the 330th randomized participant. Here, "Overall Number of Participants Analyzed" signifies participants evaluable for this outcome measure. | The response analysis uses a prespecified response-efficacy set rather than simply all randomized phase 3 participants. |
| Phase 3 secondary OS and PFS | The FAS for the Phase 3 portion of the study consisted of all randomized Phase 3 participants. | The same randomized full analysis framework was used for these time-to-event comparisons. |
6. Results: Primary Overall Survival Analysis
The registry reports a formal interim analysis of overall survival comparing the phase 3 triplet arm with the phase 3 control arm. The reported method was a stratified log-rank test, applied to the Full Analysis Set.
Triplet arm vs control arm
Stratified log-rank test · Interim analysis
The ClinicalTrials.gov record does not report a hazard ratio, confidence interval, median OS, or survival probability.
The reported P < 0.0001 indicates that the observed time-to-event comparison was statistically inconsistent with the null framework tested by the posted stratified log-rank analysis at the stated threshold. It does not quantify the magnitude of the treatment effect.
Because the registry analysis does not provide a hazard ratio or confidence interval, the strength of the association cannot be summarized here as a percentage reduction in estimated hazard. A P-value is not an effect-size measure: it depends on the data and analysis framework and does not tell us how large the difference in survival was.
The analysis is also an interim analysis. Interim testing has to be interpreted within the trial's prespecified monitoring framework. The ClinicalTrials.gov record identifies the endpoint as an interim analysis but does not provide an alpha-spending rule, efficacy boundary, or multiplicity procedure, so this page does not infer one.
7. Results: Primary Objective Response Rate Analysis
The second formal primary phase 3 analysis compared objective response rate between the triplet and control arms. The registry reports a chi-squared test and a P-value below 0.0001.
Triplet arm vs control arm
Chi-squared test · Phase 3 Response Efficacy Set
The ClinicalTrials.gov record does not report the response percentages, response counts, confidence interval, or effect estimate.
The reported P < 0.0001 indicates a statistically detectable difference in the categorical response outcome under the posted chi-squared analysis. It does not state how large the difference in response rate was.
For an ORR comparison, the clinically interpretable quantities would normally include the response proportion in each arm and an effect measure such as an absolute difference or ratio, accompanied by a confidence interval. Those values are not present in the ClinicalTrials.gov record, so they are not reconstructed here.
The confidence interval is particularly important because it describes precision around the estimated treatment contrast. Since no confidence interval is reported, this page cannot characterize the precision of the ORR effect beyond the reported P-value.
8. Secondary Efficacy Results
The registry contains additional formal statistical analyses for OS, PFS, and ORR. These results show how the trial's statistical framework extended beyond the two primary phase 3 comparisons.
| Endpoint | Comparison | Method | P-value |
|---|---|---|---|
| Overall Survival | Doublet vs Control | Stratified log-rank test | <0.0001 |
| Overall Survival | Triplet vs Doublet | Stratified log-rank test | 0.5958 |
| Progression-Free Survival, BICR | Triplet vs Control | Stratified log-rank test | <0.0001 |
| Progression-Free Survival, Investigator | Triplet vs Control | Stratified log-rank test | <0.0001 |
| Progression-Free Survival, BICR | Doublet vs Control | Stratified log-rank test | <0.0001 |
| Progression-Free Survival, Investigator | Doublet vs Control | Stratified log-rank test | <0.0001 |
| Progression-Free Survival, BICR | Triplet vs Doublet | Stratified log-rank test | 0.1004 |
| Progression-Free Survival, Investigator | Triplet vs Doublet | Stratified log-rank test | 0.3724 |
| Objective Response Rate, Investigator | Triplet vs Control | Chi-squared test | <0.0001 |
| Objective Response Rate, BICR | Doublet vs Control | Chi-squared test | <0.0001 |
| Objective Response Rate, Investigator | Doublet vs Control | Chi-squared test | <0.0001 |
| Objective Response Rate, BICR | Triplet vs Doublet | Cochran-Mantel-Haenszel test | 0.1928 |
| Objective Response Rate, Investigator | Triplet vs Doublet | Cochran-Mantel-Haenszel test | 0.0357 |
Final analysis of overall survival
The registry also posts a prespecified final analysis of overall survival for the triplet arm versus the control arm.
Triplet arm vs control arm · Final OS analysis
Stratified log-rank test
Time frame: From randomization to death due to any cause (maximum treatment exposure of 277.4 weeks for triplet arm and 108 weeks for control arm)
The final OS analysis reports P < 0.0001 using a stratified log-rank test. The ClinicalTrials.gov record does not provide the final hazard ratio, confidence interval, median survival, or absolute survival estimates. Consequently, the final result can be described as a statistically significant time-to-event comparison under the posted analysis, but its magnitude cannot be quantified from the ClinicalTrials.gov record.
The final analysis should also not be confused with the earlier interim analysis. They use different analysis time frames, and the registry separately labels them as interim and final analyses.
9. Time-to-Event Endpoints
Overall survival and progression-free survival are time-to-event outcomes. Unlike a simple binary endpoint measured at one fixed time, a time-to-event analysis uses the timing of events and accommodates participants whose event has not occurred by the time their available follow-up ends.
| Endpoint | Registry definition / time frame | Method |
|---|---|---|
| Overall Survival | OS was defined as the time from randomization to death due to any cause. | Stratified log-rank test |
| PFS per BICR | From first dose to the earliest documented PD or death due to any cause (maximum treatment exposure of 277.4 weeks for triplet arm and 268 weeks for doublet arm) | Stratified log-rank test |
| PFS per Investigator | From first dose to the earliest documented PD or death due to any cause (maximum treatment exposure of 277.4 weeks for triplet arm and 268 weeks for doublet arm) | Stratified log-rank test |
Here, di represents events at time ti and ni represents participants at risk immediately before that time. The estimator allows censored participants to contribute information up to their censoring time.
The stratified log-rank test then compares the observed and expected event patterns across treatment groups while accounting for the prespecified stratification structure when such strata are available in the analysis. A statistically significant log-rank result does not by itself tell us the size of the treatment effect; an effect estimate and confidence interval are needed for that.
10. Objective Response Rate as a Binary Endpoint
ORR is a binary endpoint at the participant level: the participant either achieves the prespecified response category or does not. The BEACON CRC registry analyses use chi-squared testing for several ORR comparisons and the Cochran-Mantel-Haenszel test for the triplet-versus-doublet comparisons.
Chi-squared analysis
The chi-squared test assesses whether the observed distribution of a categorical outcome differs between the compared groups. In this trial it was used for the primary triplet-versus-control ORR analysis and several secondary ORR comparisons.
Cochran-Mantel-Haenszel analysis
The Cochran-Mantel-Haenszel framework can compare categorical outcomes while accounting for stratification. The registry reports it for triplet-versus-doublet ORR analyses.
A binary test result should not be confused with a measure of clinical magnitude. If response rates were available, an absolute response-rate difference could answer a different question from a P-value: how much higher or lower was the observed response proportion? The ClinicalTrials.gov record does not provide those response proportions.
11. Statistical Methodology
Stratified log-rank test
The registry identifies the stratified log-rank test as the principal method for the posted phase 3 OS and PFS analyses. The method is designed for comparing time-to-event distributions between treatment groups while accounting for stratification factors used in the analysis.
At each relevant event time, the log-rank framework compares the observed number of events in a treatment group with the number expected under the null hypothesis of comparable event experience, with the stratified version combining information across strata.
Chi-squared test
The chi-squared test was the reported method for the primary ORR comparison and several secondary ORR analyses. For a binary response endpoint, the underlying data can be represented as a two-by-two treatment-by-response table.
Response / No response
The chi-squared test evaluates whether treatment assignment and response category are statistically associated under the test's assumptions.
Cochran-Mantel-Haenszel test
The Cochran-Mantel-Haenszel method was used for the triplet-versus-doublet ORR comparisons. In general, this method provides a way to compare categorical outcomes while accounting for a stratifying variable or set of strata rather than collapsing all observations into one unadjusted table.
Intention-to-treat principles and the Full Analysis Set
The registry states that the phase 3 Full Analysis Set consisted of all randomized phase 3 participants. This is consistent with the central statistical principle that randomized treatment comparisons should preserve the treatment assignment created by randomization whenever possible.
Binary versus time-to-event endpoints
BEACON CRC illustrates why endpoint type determines the appropriate statistical method. ORR is a binary outcome and was analyzed with categorical-data methods. OS and PFS incorporate the timing of events and were analyzed with stratified log-rank methods. Applying the same test to both endpoint types would discard important information.
BICR versus investigator assessment
The registry separately identifies PFS and ORR assessments performed by blinded independent central review and by investigators. This creates an important methodological distinction: independent central review and investigator assessment are different measurement processes, so their analyses should remain separately labeled rather than silently combined.
12. Statistical Methods Explained
Why use a stratified log-rank test for overall survival?
OS is a time-to-event endpoint, so the analysis should use both whether an event occurred and when it occurred. The stratified log-rank test compares the survival experience between groups while accounting for the analysis's stratification structure. It is therefore more informative for this endpoint than reducing every participant to a single binary alive/dead status at an arbitrary time.
What does P < 0.0001 mean?
For the posted primary OS analysis, P < 0.0001 means the statistical test produced a result at or below that reported threshold under the null comparison represented by the stratified log-rank test. It does not mean there is a probability below 0.0001 that the treatment is effective, and it does not tell us the size of the treatment effect.
Why can a very small P-value coexist with an unknown effect size?
A P-value measures evidence against a specified null hypothesis, not magnitude. Effect size requires an estimate such as a hazard ratio or response-rate difference. Confidence intervals then provide information about the precision of that estimate.
Why is ORR analyzed differently from OS?
ORR classifies participants according to whether they achieved a prespecified tumor response. OS records the time until death and can include censoring. The two endpoint structures therefore require different statistical methods. BEACON CRC used chi-squared or Cochran-Mantel-Haenszel methods for the posted ORR analyses and stratified log-rank tests for OS.
What does the Full Analysis Set contribute to interpretation?
The FAS for phase 3 consisted of all randomized phase 3 participants. This preserves the randomized comparison for the phase 3 time-to-event analyses. It also means that the analysis is not limited to participants who completed treatment or remained on therapy long enough to experience an event.
Why are BICR and investigator results reported separately?
They represent different assessment processes. The registry separately posts PFS and ORR analyses based on blinded independent central review and investigator assessment. Keeping them separate allows readers to see whether conclusions are consistent across assessment methods without treating two measurement processes as though they were the same endpoint.
What does a non-significant P-value in a triplet-versus-doublet comparison mean?
For example, the registry reports P = 0.5958 for OS comparing the triplet with the doublet and P = 0.1004 for BICR PFS. A non-significant result does not prove that the two strategies are identical. It means the posted statistical test did not provide sufficient evidence against its null comparison at conventional significance thresholds. Without the corresponding effect estimate and confidence interval, the magnitude and precision of the difference cannot be determined from the ClinicalTrials.gov record.
13. Secondary Endpoint Time Frames and Methods
| Secondary endpoint | Time frame | Analysis method | P-value |
|---|---|---|---|
| OS, Doublet vs Control | From randomization to death due to any cause until 204 deaths were observed (maximum treatment exposure of 89.7 weeks for doublet arm and 89.1 weeks for triplet arm) | Stratified log-rank test | <0.0001 |
| OS, Triplet vs Doublet | From randomization to death due to any cause until 204 deaths were observed (maximum treatment exposure of 89.7 weeks for doublet arm and 89.1 weeks for triplet arm) | Stratified log-rank test | 0.5958 |
| PFS, Triplet vs Control, BICR | From first dose to the earliest documented PD or death due to any cause (maximum treatment exposure of 277.4 weeks for triplet arm and 268 weeks for doublet arm) | Stratified log-rank test | <0.0001 |
| PFS, Triplet vs Control, Investigator | From first dose to the earliest documented PD or death due to any cause (maximum treatment exposure of 277.4 weeks for triplet arm and 268 weeks for doublet arm) | Stratified log-rank test | <0.0001 |
| PFS, Doublet vs Control, BICR | From first dose to the earliest documented PD or death due to any cause (maximum treatment exposure of 268 weeks for doublet arm and 108 weeks for control arm) | Stratified log-rank test | <0.0001 |
| PFS, Doublet vs Control, Investigator | From first dose to the earliest documented PD or death due to any cause (maximum treatment exposure of 268 weeks for doublet arm and 108 weeks for control arm) | Stratified log-rank test | <0.0001 |
| PFS, Triplet vs Doublet, BICR | From first dose to the earliest documented PD or death due to any cause (maximum treatment exposure of 277.4 weeks for triplet arm and 268 weeks for doublet arm) | Stratified log-rank test | 0.1004 |
| PFS, Triplet vs Doublet, Investigator | From first dose to the earliest documented PD or death due to any cause (maximum treatment exposure of 277.4 weeks for triplet arm and 268 weeks for doublet arm) | Stratified log-rank test | 0.3724 |
| ORR, Triplet vs Control, Investigator | Duration of Phase 3, approximately 6 months (up to 28 days per cycle) | Chi-squared test | <0.0001 |
| ORR, Doublet vs Control, BICR | Duration of Phase 3, approximately 6 months (up to 28 days per cycle) | Chi-squared test | <0.0001 |
| ORR, Doublet vs Control, Investigator | Duration of Phase 3, approximately 6 months (up to 28 days per cycle) | Chi-squared test | <0.0001 |
| ORR, Triplet vs Doublet, BICR | Duration of Phase 3, approximately 6 months (up to 28 days per cycle) | Cochran-Mantel-Haenszel test | 0.1928 |
| ORR, Triplet vs Doublet, Investigator | Duration of Phase 3, approximately 6 months (up to 28 days per cycle) | Cochran-Mantel-Haenszel test | 0.0357 |
14. Safety Results
The ClinicalTrials.gov record provides serious adverse-event counts by arm. These are reported as affected participants divided by participants at risk and should not be confused with the broader adverse-event endpoints registered for the safety lead-in.
| Study group | Serious adverse events | Affected / at risk |
|---|---|---|
| Combined Safety Lead-in | Participants with serious adverse events | 22/37 |
| Phase 3: Triplet Arm | Participants with serious adverse events | 118/222 |
| Phase 3: Doublet Arm | Participants with serious adverse events | 91/216 |
| Phase 3: Control Arm | Participants with serious adverse events | 78/193 |
The safety-lead-in denominator and the phase 3 denominators represent different study components and should not be pooled into one overall safety percentage. The ClinicalTrials.gov record also do not provide a formal between-arm statistical test for these serious-adverse-event counts, so no inferential comparison is added here.
Registered safety-lead-in endpoints
The safety lead-in had four registered primary endpoints: dose-limiting toxicities during Cycle 1; adverse events during the safety lead-in; incidence of dose interruptions, dose modifications, and discontinuations due to adverse events at an interim analysis; and the corresponding final analysis from start of study treatment until 30 days after the last dose, with a maximum treatment exposure of 280 weeks.
15. Safety Lead-In and Interim Analysis
The safety lead-in is an important design feature because it precedes the randomized phase 3 efficacy comparison. Its purpose in the registered endpoint structure was to characterize dose-limiting toxicities and adverse-event-related treatment modifications and discontinuations.
Dose-limiting toxicities
The registered DLT endpoint was measured during Cycle 1, up to 28 days. It is a binary participant-level safety outcome.
Adverse events
The registry defines an AE as any untoward medical occurrence temporally associated with use of study drug, whether or not considered related to study drug.
Dose interruptions
The safety-lead-in endpoints separately considered dose interruptions, dose modifications, and discontinuations due to adverse events.
Interim versus final
The registry posted both an interim and final analysis for the incidence of dose interruptions, modifications, and discontinuations due to adverse events.
16. Multiplicity and Multiple Comparisons
BEACON CRC has multiple registered primary endpoints and multiple posted secondary comparisons. That creates an important statistical distinction between an individual reported P-value and a trial-level confirmatory conclusion.
| Feature | What the ClinicalTrials.gov record shows | Interpretive implication |
|---|---|---|
| Primary endpoints | 6 registered primary endpoints, including safety-lead-in endpoints plus phase 3 OS and ORR | These endpoints do not all represent the same statistical question. |
| Primary phase 3 analyses | 2 formal primary analyses: OS and ORR | The two endpoints should be identified separately rather than treated as one composite outcome. |
| Secondary analyses | Multiple OS, PFS, and ORR comparisons across triplet, doublet, and control arms | Repeated testing raises multiplicity considerations. |
| Hypothesis type | Other / not stated | The registry-reported extract does not state a conventional null/alternative hypothesis label. |
The ClinicalTrials.gov record does not specify an alpha-spending strategy, hierarchical testing sequence, multiplicity adjustment, or familywise error-control procedure for all of the posted comparisons. It would therefore be inappropriate to infer one from the P-values alone.
17. What the P-values Do — and Do Not — Mean
A P-value is calculated under a specified null hypothesis and statistical model. It measures how unusual the observed data, or more extreme data, would be under that null framework. It is not the probability that the null hypothesis is true.
The registry-reported BEACON CRC statistical-analysis records provide P-values for the posted comparisons but do not provide the corresponding effect estimates or confidence intervals. Therefore, this page does not translate P < 0.0001 into a percentage reduction in hazard, a response-rate difference, or any other numerical effect.
A confidence interval would describe the statistical precision around an estimated effect. Without a confidence interval, a P-value alone cannot show whether the effect estimate is tightly or loosely estimated.
18. Understanding the Triplet-versus-Doublet Comparisons
The registry includes direct triplet-versus-doublet analyses in addition to triplet-versus-control and doublet-versus-control comparisons. These direct comparisons are particularly useful for understanding what the addition of the third study component contributes statistically.
| Endpoint | Triplet vs Doublet method | P-value |
|---|---|---|
| Overall Survival | Stratified log-rank test | 0.5958 |
| Progression-Free Survival, BICR | Stratified log-rank test | 0.1004 |
| Progression-Free Survival, Investigator | Stratified log-rank test | 0.3724 |
| Objective Response Rate, BICR | Cochran-Mantel-Haenszel test | 0.1928 |
| Objective Response Rate, Investigator | Cochran-Mantel-Haenszel test | 0.0357 |
These P-values should be read as results of their respective statistical tests, not as direct measures of clinical importance. In particular, a non-significant P-value does not establish equivalence, and a statistically significant P-value does not quantify the magnitude of the observed difference.
19. Stratification and the Cochran-Mantel-Haenszel Framework
The registry explicitly lists the Cochran-Mantel-Haenszel test among the statistical methods for the trial. This method is especially relevant to categorical outcomes because it can preserve information across strata instead of treating all participants as though they came from one unstratified population.
The key statistical idea is to compare treatment and response while accounting for stratification. The registry extract does not identify the individual stratification variables used for the posted Cochran-Mantel-Haenszel analyses, so none are inferred here.
This illustrates an important distinction between a simple chi-squared test and a stratified categorical analysis. When stratification is part of the design or analysis plan, ignoring it can discard relevant structure created by the trial.
20. Limitations
- Effect estimates are not reported in the posted statistical-analysis records provided here: the registry data contain P-values for the formal analyses but not corresponding hazard ratios, response-rate differences, or confidence intervals.
- Median survival is not reported in the ClinicalTrials.gov record: the page therefore does not state median OS or PFS values.
- Absolute response rates are not reported in the ClinicalTrials.gov record: the primary ORR P-value is reported, but the response percentages and counts are not.
- Stratification variables are not specified in the registry-reported extract: although stratified log-rank and Cochran-Mantel-Haenszel methods are reported, the individual stratification factors are not provided.
- Multiplicity procedures are not specified: the ClinicalTrials.gov record does not state an alpha allocation, hierarchical testing procedure, or complete familywise error-control strategy.
- Interim monitoring details are incomplete: the primary OS endpoint is explicitly an interim analysis and is tied to 204 deaths, but the ClinicalTrials.gov record does not specify an interim boundary or alpha-spending rule.
- Analysis populations differ: OS uses the phase 3 Full Analysis Set, whereas ORR uses a Response Efficacy Set based on the first 330 randomized phase 3 participants and additional participants randomized on the same day as the 330th participant.
- Serious adverse-event counts are descriptive: the ClinicalTrials.gov record does not provide formal statistical comparisons of serious adverse-event rates across the phase 3 arms.
- No proportional-hazards assessment is reported: the registry identifies stratified log-rank analyses but does not provide a proportional-hazards diagnostic or a hazard-ratio estimate in the ClinicalTrials.gov record.
21. Why This Trial Matters Statistically
BEACON CRC is a useful statistical teaching example because the ClinicalTrials.gov record brings together several core clinical-trial concepts within one sequential phase 3 program.
| Concept | How it appears in BEACON CRC |
|---|---|
| Randomization | The phase 3 study used randomized allocation. |
| Sequential design | A safety lead-in preceded the phase 3 portion. |
| Binary endpoints | DLTs, adverse events, treatment modifications, and ORR were registered as binary outcomes. |
| Time-to-event endpoints | OS and PFS were analyzed using stratified log-rank tests. |
| Stratified analysis | The registry reports stratified log-rank and Cochran-Mantel-Haenszel methods. |
| Intention-to-treat principle | The phase 3 FAS consisted of all randomized phase 3 participants. |
| Different analysis populations | OS used the FAS; ORR used the Response Efficacy Set. |
| Independent assessment | ORR and PFS were analyzed using both BICR and investigator assessment in secondary comparisons. |
| Interim analysis | The phase 3 primary OS comparison included an interim analysis until 204 deaths were observed. |
| Multiple comparisons | The registry contains multiple primary and secondary comparisons across three phase 3 arms. |
The statistical lesson is broader than any single P-value. A rigorous trial analysis requires identifying the endpoint type, analysis population, comparison, statistical method, time frame, and inferential context before interpreting a numerical result.
22. Related Tutorials
Learn more about the methods used in this trial:
23. Related Calculators
24. Sources
- ClinicalTrials.gov: BEACON CRC, NCT02928224.
- PubMed: PMID 39313594.
- PubMed: PMID 36653241.
- PubMed: PMID 35654691.
- PubMed: PMID 35653981.
- PubMed: PMID 33503393.
Continue through the Clinical Biostats statistical pathway
Use the related tutorials and calculators to move from the trial's endpoints to the underlying statistical methods.
25. Record Summary
BEACON CRC illustrates a modern randomized phase 3 analysis in which a sequential safety lead-in is followed by randomized comparisons across three treatment arms. The ClinicalTrials.gov record reports formal stratified log-rank analyses for overall survival and progression-free survival, chi-squared analyses for several objective response comparisons, and Cochran-Mantel-Haenszel analyses for direct triplet-versus-doublet response comparisons.
The central statistical lesson is that a reported P-value should never be interpreted in isolation. For each result, the reader should identify what endpoint was measured, which participants were analyzed, which groups were compared, what statistical method was used, what time frame applied, and whether an effect estimate and confidence interval were reported. In the registry-reported BEACON CRC data, several formal analyses have P-values but no effect estimates or confidence intervals; the appropriate interpretation is therefore deliberately limited to what those statistical records support.