This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record.
1. Trial at a Glance
BEATcc is a completed, randomized phase 3 clinical trial in stage IVB carcinoma of the cervix. The registry describes a two-arm, parallel-group treatment study with no masking and an enrollment of 410 participants.
| Feature | BEATcc |
|---|---|
| Trial acronym | BEATcc |
| ClinicalTrials.gov identifier | NCT03556839 |
| Phase | Phase 3 |
| Status | Completed |
| Condition | Carcinoma of the Cervix, Stage IVB |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | None |
| Primary purpose | Treatment |
| Enrollment | 410 |
| Number of arms | 2 |
| Lead sponsor | Grupo Español de Investigación en Cáncer de Ovario |
| Sponsor type | Other |
2. Clinical Question
The registry title describes a treatment strategy consisting of platinum chemotherapy plus paclitaxel with bevacizumab and atezolizumab in metastatic carcinoma of the cervix. Because the registry record identifies a randomized two-arm comparison but does not provide the arm-specific treatment assignments in the available record, the statistical question is most precisely framed as whether the randomized treatment groups differ in the two prespecified time-to-event outcomes.
Population
Patients with carcinoma of the cervix, stage IVB.
Interventions listed
Atezolizumab, bevacizumab, cisplatin/carboplatin, and paclitaxel.
Randomization
The trial uses randomized allocation with two treatment arms.
Primary question
Does the randomized treatment comparison affect progression-free survival and overall survival over the 48-month endpoint time frame?
3. Trial Design
BEATcc is a phase 3, randomized, parallel-group treatment trial with no masking. The registry reports enrollment of 410 participants and two arms. The study began on 2018-09-25 and has a primary completion date of 2025-08-31.
Treatment assignment
- The registry identifies two randomized treatment arms.
- The available record does not provide an arm-specific sample size.
- The registry intervention list includes atezolizumab, bevacizumab, cisplatin/carboplatin, and paclitaxel.
Comparator assignment
- The registry identifies a second randomized treatment arm.
- The available record does not provide an arm-specific sample size.
- The two groups form the parallel randomized comparison for the primary endpoints.
4. Trial Timeline
Study start
The registry lists 2018-09-25 as the study start date.
Randomized treatment comparison
The trial uses randomized allocation, a parallel design, no masking, and a treatment-focused primary purpose.
Primary endpoint time frame
Both progression-free survival and overall survival are registered with a 48-month time frame.
Primary completion
The registry lists 2025-08-31 as the primary completion date.
5. Endpoints
The registry identifies two primary endpoints. Both are time-to-event outcomes, but they define the event differently. That distinction is central to understanding how the trial would be analyzed statistically.
| Primary endpoint | Time frame | Registry definition |
|---|---|---|
| Progression-free survival | 48 months | Time from the date of randomization to the date of first documentation of disease progression or death due to any cause, whichever occurs first based on investigator assessment using the Response Evaluation Criteria in Solid Tumors version 1.1 (RECIST v 1.1). |
| Overall survival | 48 months | Time from the date of randomization to the date of death due to any cause. |
Why the endpoint definitions matter
Progression-free survival combines two possible events: documented disease progression and death. The first of those events determines the PFS event time. Overall survival has a single event definition: death from any cause. Both endpoints begin at the date of randomization, which gives the randomized comparison a common time origin.
PFS
Captures the time until either the first documented progression or death, whichever occurs first.
OS
Captures the time from randomization until death from any cause.
6. Planned Analysis
The registry identifies progression-free survival and overall survival as the primary endpoints, each assessed over 48 months. The appropriate statistical framework for both endpoints is therefore a time-to-event analysis rather than a simple comparison of proportions at a single fixed time.
Progression-free survival
For PFS, the analysis would typically estimate the distribution of time from randomization to the first qualifying event. Kaplan-Meier estimation is commonly used to describe the survival function over time, while a log-rank test can compare the time-to-event distributions between randomized groups. A Cox proportional-hazards model can provide a hazard ratio with a confidence interval as a relative measure of treatment effect.
Patients without a qualifying event by their last evaluable follow-up are generally right-censored according to the prespecified analysis rules.
Overall survival
For OS, the event is death from any cause. Kaplan-Meier estimation can describe the probability of remaining alive over time, while a log-rank comparison and Cox proportional-hazards model are standard approaches for a randomized two-group survival analysis.
A participant who remains alive at the end of available follow-up contributes censored survival time rather than an observed death time.
What would constitute a complete survival analysis
A full analysis would ordinarily report the number of events, the number censored, Kaplan-Meier estimates over time, a measure of treatment effect such as a hazard ratio with its confidence interval, and the statistical comparison between randomized groups. The ClinicalTrials.gov record does not post those numerical statistical results.
7. Statistical Methodology
Kaplan-Meier estimation
Kaplan-Meier estimation is designed for time-to-event outcomes in which some participants may not have experienced the event by the time their follow-up ends. Instead of assigning those participants an artificial event time, the method retains their information until censoring.
Here, di is the number of events at time ti and ni is the number at risk immediately before that time.
For BEATcc, the Kaplan-Meier framework is particularly natural because both primary endpoints are explicitly defined as elapsed time from randomization to an event. It also allows the analysis to display how event-free or overall survival changes during follow-up rather than reducing the outcome to a single binary endpoint.
Log-rank comparison
A log-rank test compares survival experience between randomized groups across the observed follow-up period. Unlike a simple comparison of event percentages, it incorporates the timing of events and the information contributed by censored observations.
Cox proportional-hazards model
A Cox model estimates a hazard ratio comparing the instantaneous event rates between treatment groups. For a randomized trial, this provides a compact relative measure of the treatment effect on a time-to-event endpoint.
The hazard ratio is not a risk ratio, not an absolute risk difference, and not the percentage of participants who benefit.
Right censoring
Censoring is unavoidable in many survival studies because participants may remain alive or free of progression when their available follow-up ends. A censored observation tells the analysis that the participant was event-free through a particular point in time; it does not establish what happened afterward.
RECIST v1.1 and PFS
The registry specifically defines PFS using first documentation of disease progression or death, with progression based on investigator assessment using RECIST v1.1. Thus, the statistical event definition depends not only on elapsed follow-up but also on the protocol's disease-assessment process.
8. Statistical Methods Explained
Why is PFS a time-to-event endpoint?
PFS does not simply ask whether progression occurred. It asks when progression or death first occurred after randomization. Two participants can both experience progression while contributing very different amounts of follow-up, so the timing of the event is statistically important.
Why is death included in the PFS definition?
The registry defines PFS as the time to first documentation of disease progression or death from any cause, whichever occurs first. Including death prevents a participant who dies before documented progression from being treated as if the participant remained progression-free indefinitely.
Why is overall survival statistically different from PFS?
OS has death as its sole event, whereas PFS has progression and death as competing ways to reach the endpoint definition. Consequently, the two endpoints measure different clinical processes even though both begin at randomization.
What does a hazard ratio actually measure?
A hazard ratio compares modeled instantaneous event rates between groups. A value below 1 corresponds to a lower modeled event rate in the numerator group relative to the reference group. It does not directly tell us the difference in median survival, the absolute probability of an event, or the fraction of patients who benefit.
Why does the confidence interval matter?
A point estimate alone does not describe statistical precision. A confidence interval shows the range of effect estimates compatible with the specified statistical model and sampling framework at the stated confidence level. Wider intervals indicate greater uncertainty than narrower intervals.
Why can a Kaplan-Meier curve be more informative than a single hazard ratio?
A hazard ratio compresses the treatment comparison into one relative measure. A Kaplan-Meier curve can show when events occur, how survival changes over time, and whether the separation between groups appears consistent across follow-up. If the proportional-hazards assumption is questionable, the single Cox hazard ratio may not adequately describe the full time pattern.
Why does randomization matter for the interpretation?
Randomization creates the basis for comparing outcomes between treatment assignments while reducing systematic differences in measured and unmeasured baseline characteristics in expectation. That is why the primary comparison should be anchored to randomized assignment rather than selectively comparing participants according to treatment received after randomization.
9. Interpreting the 48-Month Endpoint Window
Both registered primary endpoints use a 48-month time frame. This means that the registry specifies a follow-up window for the primary endpoint definitions rather than defining the endpoints as simple outcomes measured at a single visit.
Time origin
Both endpoints begin at the date of randomization.
Time horizon
The registered time frame for both PFS and OS is 48 months.
PFS event
First documented progression or death, whichever occurs first.
OS event
Death due to any cause.
A time frame of 48 months should not be interpreted as saying that every participant is observed for exactly 48 months. In a time-to-event analysis, follow-up can end earlier for reasons such as censoring, while some participants can contribute substantially more information depending on the study's follow-up procedures.
10. Why the Two Primary Endpoints Complement Each Other
PFS and OS provide different perspectives on treatment effects. PFS focuses on disease progression or death and can therefore capture treatment effects before enough deaths have accumulated to make OS mature. OS, by contrast, is based solely on death and has an unambiguous event definition, but it can be influenced by the subsequent treatment pathway after the randomized intervention.
| Feature | Progression-free survival | Overall survival |
|---|---|---|
| Starting point | Date of randomization | Date of randomization |
| Primary time frame | 48 months | 48 months |
| Event definition | First documented disease progression or death | Death due to any cause |
| Progression assessment | Investigator assessment using RECIST v1.1 | Not required for the event definition |
| Typical statistical framework | Kaplan-Meier, log-rank, Cox model | Kaplan-Meier, log-rank, Cox model |
The distinction is important when interpreting any future results. A treatment effect on PFS does not automatically establish the same magnitude of effect on OS, and an OS comparison cannot be reduced to a statement about tumor progression alone.
11. Analysis Population and Randomization
The registry identifies BEATcc as randomized, but the available record does not specify the statistical analysis populations used for each endpoint. In a randomized phase 3 efficacy analysis, the intention-to-treat principle is commonly used because it preserves the original treatment assignment as the basis of comparison.
For survival endpoints, the analysis population also interacts with censoring. A participant who is included in the analysis but has no observed event contributes follow-up information until the relevant censoring time. This is one reason that survival analysis cannot be reduced to a simple count of participants with events.
12. Non-Masking and Statistical Interpretation
The registry specifies masking as "None." This is a design characteristic rather than a statistical result, but it matters particularly for an endpoint involving investigator assessment of disease progression.
Because PFS progression is based on investigator assessment using RECIST v1.1, knowledge of treatment assignment can potentially affect clinical assessment or decisions surrounding assessment. That does not establish that bias occurred; it identifies a design feature that should be considered when interpreting the endpoint.
What randomization addresses
Randomization establishes the intended basis for comparing treatment assignments.
What masking can address
Masking can reduce certain forms of knowledge-related assessment or treatment-management bias.
What RECIST contributes
RECIST v1.1 provides the framework specified by the registry for defining progression.
What remains statistical
The resulting event times still require appropriate time-to-event methods and censoring rules.
13. Statistical Interpretation of Future Hazard-Ratio Results
Because numerical primary endpoint results have not been posted in the registry record, there is no trial-specific hazard ratio to interpret here. The correct framework for reading a future reported hazard ratio is nevertheless straightforward.
A hazard ratio below 1 would indicate a lower estimated instantaneous event rate in the numerator treatment group relative to the reference group. The magnitude should be interpreted together with its confidence interval and the definition of the endpoint.
An interval that includes 1 would indicate that the estimated treatment effect is compatible with no relative hazard difference under the specified confidence-interval framework. That is distinct from proving that the treatments are identical.
An interval that excludes 1 would indicate evidence against a hazard ratio of exactly 1 under the specified model and confidence level. It would still be necessary to examine the magnitude, absolute outcomes, censoring, and assumptions before interpreting clinical importance.
A p-value addresses compatibility of the observed data with a specified null hypothesis under the analysis framework. It does not quantify the magnitude of treatment benefit, the probability that a treatment works, or the probability that the null hypothesis is true.
14. Proportional-Hazards Assumption
The Cox proportional-hazards model is commonly used for randomized time-to-event comparisons, but its interpretation depends on the proportional-hazards assumption when a single hazard ratio is used as the primary summary.
Proportional hazards means that the relative hazard between the treatment groups is approximately stable over time. If hazards change materially over follow-up, a single hazard ratio may conceal important features of the survival curves.
These quantities answer different statistical questions and should not be substituted for one another.
For BEATcc, the registered 48-month time horizon makes the time dimension particularly important. Any eventual statistical report should therefore be interpreted using both the relative treatment estimate and the underlying time-to-event pattern.
15. Missing Data and Censoring
The primary endpoints are time-to-event outcomes, so missingness and censoring require explicit statistical rules. The registry definitions establish the event times but do not specify the full censoring and imputation procedures in the available record.
For PFS, disease assessments may determine whether progression has occurred, while for OS the event is death from any cause. If an event is not observed before the participant's follow-up ends, survival methods generally treat the observation as censored at the relevant time rather than assigning an unobserved event.
16. Multiplicity and the Two Primary Endpoints
BEATcc has two registered primary endpoints: progression-free survival and overall survival. When a trial has more than one primary endpoint, the statistical analysis must distinguish between estimating treatment effects and controlling the overall probability of a false-positive confirmatory conclusion.
| Endpoint | Role | Time frame | Statistical issue |
|---|---|---|---|
| Progression-free survival | Primary | 48 months | Time-to-event comparison with progression or death as the event |
| Overall survival | Primary | 48 months | Time-to-event comparison with death as the event |
The appropriate multiplicity strategy depends on the prespecified statistical analysis plan. Possibilities in clinical-trial design include requiring both endpoints to meet a criterion, allocating alpha across endpoints, or specifying another logical relationship between the hypotheses. The ClinicalTrials.gov record does not provide that multiplicity strategy.
17. Interim Analysis and Sequential Monitoring
An interim analysis can change the statistical interpretation of a trial because repeatedly examining accumulating data without adjustment can increase the probability of rejecting a true null hypothesis. A prespecified group-sequential or alpha-spending framework can control the overall type I error while permitting interim evaluation.
The available registry information identifies the two primary endpoints and their 48-month time frames but does not provide an interim-analysis schedule, alpha-spending function, efficacy boundary, futility boundary, or information fractions.
What an interim analysis can do
Allow prespecified evaluation of accumulating evidence before the planned endpoint horizon is complete.
Why boundaries matter
Stopping rules must be incorporated into the statistical design to preserve the intended error rate.
What alpha spending does
It allocates the overall type I error across planned analyses according to a prespecified rule.
What is documented here
The available registry record does not specify an interim-monitoring procedure.
18. Subgroup and Stratified Analysis
The available registry information does not report prespecified stratification factors, subgroup analyses, interaction tests, or subgroup-specific treatment-effect estimates. Consequently, no subgroup effect should be inferred from the overall randomized design.
When subgroup analyses are eventually reported, the key statistical distinction is between a treatment effect estimated within a subgroup and evidence that treatment effects differ between subgroups.
A formal interaction test or another prespecified heterogeneity analysis is needed to assess whether treatment effect differs between subgroups.
19. Bayesian Methods
No Bayesian statistical method is identified in the available registry information. The primary endpoints are registered as conventional randomized time-to-event outcomes, for which frequentist survival-analysis methods such as Kaplan-Meier estimation, log-rank testing, and Cox modeling are commonly used.
If a future statistical report uses Bayesian analyses, those analyses would need to specify the prior distributions, likelihood, posterior estimand, decision threshold, and sensitivity of conclusions to the prior assumptions.
20. What Can and Cannot Be Learned From the Registry Record
The registry provides enough design information to understand the basic inferential structure of BEATcc: it is a completed phase 3 randomized, parallel-group treatment trial with two primary time-to-event endpoints and a 48-month time frame.
| Question | What the registry establishes |
|---|---|
| What population was studied? | Carcinoma of the cervix, stage IVB. |
| How was treatment assigned? | Randomized allocation. |
| How many participants were enrolled? | 410. |
| How many arms were there? | Two. |
| Was the trial masked? | No; masking is listed as none. |
| What were the primary endpoints? | Progression-free survival and overall survival. |
| What was the endpoint time frame? | 48 months for both primary endpoints. |
| Are numerical primary endpoint analyses posted? | No formal statistical analyses are posted to ClinicalTrials.gov. |
This distinction is important because trial design information and trial results answer different questions. Knowing that a randomized phase 3 trial was conducted establishes the framework for inference, but it does not by itself establish the magnitude, direction, precision, or statistical significance of the treatment effect.
21. Limitations
- No posted numerical efficacy analysis: the registry record does not contain formal statistical analyses for the two primary endpoints, so no trial-specific hazard ratio, confidence interval, p-value, median survival estimate, or event count can be interpreted from the registry.
- Arm-specific enrollment is not reported: the available record states total enrollment of 410 and two arms but does not provide the number assigned to each arm.
- Comparator treatment details are not fully specified in the available record: the registry lists the interventions but does not provide an arm-by-arm treatment description in the ClinicalTrials.gov record.
- Masking: the registry specifies no masking, which can be relevant when an endpoint depends on investigator assessment of disease progression.
- RECIST-based progression: PFS depends on investigator assessment using RECIST v1.1, making the endpoint partly dependent on disease-assessment procedures.
- Censoring: time-to-event analysis requires assumptions about the censoring mechanism and prespecified rules for handling follow-up that ends before an event.
- Multiplicity: two primary endpoints create a need for a prespecified strategy controlling the interpretation of multiple primary hypotheses; that strategy is not identified in the available registry information.
- Interim monitoring: the available record does not specify whether interim efficacy or futility analyses were performed or how repeated testing was controlled.
- Analysis populations: the available record does not identify the precise efficacy and safety analysis populations used in the statistical analysis.
- Generalizability: the trial's population is defined by the registry condition of stage IVB carcinoma of the cervix, so applicability to populations outside that clinical setting requires separate consideration.
22. Why This Trial Matters Statistically
BEATcc is a useful statistical teaching case because it combines randomized treatment allocation with two clinically distinct time-to-event primary endpoints. The design illustrates why clinical-trial interpretation requires more than identifying whether a p-value is below a conventional threshold.
| Statistical concept | How it appears in BEATcc |
|---|---|
| Randomization | Participants are allocated randomly to two treatment arms. |
| Parallel-group design | The two randomized arms are evaluated concurrently. |
| Time-to-event analysis | Both primary endpoints measure elapsed time from randomization to an event. |
| Kaplan-Meier estimation | A standard framework for estimating survival functions with censoring. |
| Hazard ratio | A standard model-based relative measure for randomized survival comparisons. |
| RECIST v1.1 | The registered framework used for investigator assessment of disease progression for PFS. |
| Multiple primary endpoints | PFS and OS are both registered as primary endpoints. |
| Censoring | Participants may contribute follow-up without experiencing the event during observed follow-up. |
| Non-masked design | The registry specifies no masking. |
| Clinical interpretation | Relative time-to-event measures should be interpreted alongside endpoint definitions and absolute survival experience when those results become available. |
23. Statistical Questions Readers Should Ask
What was the event?
For PFS, first documented progression or death. For OS, death from any cause.
When did follow-up begin?
Both primary endpoint definitions begin at randomization.
How was progression defined?
By investigator assessment using RECIST v1.1.
What was the analysis time frame?
The registered time frame was 48 months for both primary endpoints.
Was the comparison randomized?
Yes. The registry specifies randomized allocation.
Was masking used?
No. The registry specifies no masking.
24. Interpreting Future Results Without Overreading Them
When formal numerical results become available, interpretation should begin with the endpoint definition rather than the p-value. For PFS, the result would describe time to first documented progression or death. For OS, it would describe time to death from any cause.
The next layer is the effect estimate. A hazard ratio communicates relative event-rate information, while absolute survival probabilities at clinically meaningful times communicate a different dimension of effect. Confidence intervals then indicate the statistical precision of those estimates.
Finally, the result should be evaluated in the context of the randomized design, censoring, analysis population, multiplicity strategy, and any interim monitoring. These features determine how strongly a numerical result supports the corresponding statistical conclusion.
Endpoint definition tells us what was measured → time-to-event analysis tells us how event timing and censoring were handled → effect estimate describes the treatment contrast → confidence interval describes precision → p-value addresses compatibility with a specified null hypothesis → design features determine how the result should be interpreted.
25. Sources
- ClinicalTrials.gov: BEATcc, NCT03556839.
- PubMed: PubMed record 40968753.
- PubMed: PubMed record 38048793.
- PubMed: PubMed record 31645423.
Continue through the Clinical Biostats knowledge graph
Clinical trial analysis connects study design, endpoint definitions, survival methods, effect measures, uncertainty, and interpretation.
26. Record Summary
BEATcc is a completed phase 3 randomized, parallel-group treatment trial enrolling 410 participants with stage IVB carcinoma of the cervix. The registry identifies two primary endpoints—progression-free survival and overall survival—both measured over a 48-month time frame from randomization. PFS is defined as time to first documentation of disease progression or death, based on investigator assessment using RECIST v1.1, while OS is defined as time to death from any cause.
Statistically, the trial is centered on a randomized time-to-event comparison. Kaplan-Meier estimation provides a natural way to describe the survival functions, log-rank methods can compare event-time distributions, and Cox modeling can summarize relative treatment effects through hazard ratios and confidence intervals. The interpretation of any eventual numerical results will depend not only on the effect estimate but also on censoring, endpoint definition, multiplicity, analysis populations, and the non-masked design.