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
RAINBO is a randomized, parallel-design phase 2/3 trial in endometrial cancer. The registry describes eight study arms and four primary endpoint frameworks corresponding to p53abn-RED, MMRd-GREEN, NSMP-ORANGE, and POLEmut-BLUE trials.
| Feature | RAINBO |
|---|---|
| Official title | Refining Adjuvant Treatment IN Endometrial Cancer Based On Molecular Features |
| Phase | Phase 2/3 |
| Condition | Endometrial Cancer |
| Design | Randomized, parallel |
| Masking | None |
| Primary purpose | Treatment |
| Enrollment | 1615 |
| Number of arms | 8 |
| Allocation | Randomized |
| Status | Recruiting |
| Lead sponsor | Leiden University Medical Center |
| Sponsor type | Other |
| Start date | 2021-11-11 |
| Primary completion date | 2030-01-01 |
| ClinicalTrials.gov | NCT05255653 |
2. Clinical Question
The central statistical question is whether adjuvant treatment strategies assigned within the RAINBO program produce different recurrence outcomes within molecularly defined trial components of endometrial cancer.
Population
Patients enrolled in a phase 2/3 randomized clinical trial for endometrial cancer.
Interventions
The registry lists olaparib, pelvic external beam radiotherapy, chemotherapy, durvalumab, medroxyprogesterone acetate, megestrol acetate, vaginal brachytherapy, and observation among the interventions.
Comparator framework
The registry describes eight randomized study arms. The available record does not provide an arm-by-arm mapping of each listed intervention in the information summarized here.
Primary question
How do randomized adjuvant treatment strategies affect recurrence-free survival or pelvic recurrence-free survival over the registered 3-year assessment period?
3. Trial Design
Patients are allocated by randomization within the trial design.
The registry classifies the overall study design as a parallel assignment model.
The registry identifies the study as unmasked.
The primary purpose is treatment.
Intervention framework
Drug and radiation strategies
- Olaparib
- Pelvic external beam radiotherapy
- Chemotherapy
- Durvalumab
- Vaginal brachytherapy
Hormonal and observation strategies
- Medroxyprogesterone Acetate
- Megestrol Acetate
- Observation
The eight interventions listed by the registry should not be interpreted as eight simple one-to-one randomized comparisons. The study is organized around multiple named trial components, and the available registry information identifies the interventions and primary endpoints without providing a complete arm-level treatment map.
4. Molecularly Defined Trial Components
The four registered primary endpoints are explicitly labeled p53abn-RED, MMRd-GREEN, NSMP-ORANGE, and POLEmut-BLUE. These labels define four distinct primary endpoint frameworks within the RAINBO program.
| Trial component | Primary endpoint | Time frame |
|---|---|---|
| p53abn-RED trial | Recurrence-free survival | 3 years |
| MMRd-GREEN trial | Recurrence-free survival | 3 years |
| NSMP-ORANGE trial | Recurrence-free survival | 3 years |
| POLEmut-BLUE trial | Pelvic recurrence-free survival | 3 years |
Statistically, this structure is important because the registry does not describe one single primary endpoint applied uniformly across every participant. Instead, three registered components use recurrence-free survival, while the POLEmut-BLUE component uses pelvic recurrence-free survival.
5. Primary Endpoints
| Endpoint | Registry definition | Time frame |
|---|---|---|
| p53abn-RED trial | Recurrence-free survival | 3 years |
| MMRd-GREEN trial | Recurrence-free survival | 3 years |
| NSMP-ORANGE trial | Recurrence-free survival | 3 years |
| POLEmut-BLUE trial | Pelvic recurrence-free survival | 3 years |
Recurrence-free survival
Recurrence-free survival is a time-to-event endpoint. Conceptually, it measures the time from a defined starting point until recurrence or another prespecified event, with patients who have not experienced the event by the end of observation contributing censored follow-up.
The registry identifies recurrence-free survival as the primary endpoint for the p53abn-RED, MMRd-GREEN, and NSMP-ORANGE components, with a 3-year time frame.
Pelvic recurrence-free survival
Pelvic recurrence-free survival is also a time-to-event endpoint, but it focuses specifically on recurrence within the pelvic region rather than recurrence in any location. The registry assigns this endpoint to the POLEmut-BLUE trial at 3 years.
6. Planned Analysis
The ClinicalTrials.gov record currently has no posted statistical analyses or numerical efficacy results for the registered primary endpoints. The trial remains listed as recruiting, with a primary completion date of 2030-01-01.
How recurrence-free survival is typically analysed
A time-to-event endpoint such as recurrence-free survival is typically summarized with the Kaplan-Meier method. The Kaplan-Meier curve estimates the probability of remaining event-free over time while allowing participants to be right-censored when follow-up ends before an event is observed.
Here, di represents the number of events at time ti, while ni is the number of participants at risk immediately before that time.
Typical comparison of randomized groups
For randomized groups with a time-to-event primary endpoint, a log-rank test is commonly used to compare event-time distributions. A Cox proportional-hazards model is also commonly used to estimate a hazard ratio and its confidence interval.
An HR below 1 indicates a lower estimated instantaneous event rate in the numerator group under the fitted model. It is not the same as a relative risk, an absolute risk difference, or a probability that an individual patient will remain recurrence-free.
Three-year assessment
The registry specifies 3 years for all four primary endpoint entries. A time-specific recurrence-free survival estimate at 3 years provides an absolute measure of the proportion estimated to remain event-free at that time point. A hazard ratio, if reported in a future analysis, would instead summarize a relative treatment effect over the analyzed follow-up under the chosen model.
Pelvic recurrence-free survival analysis
The POLEmut-BLUE endpoint would ordinarily require the same general time-to-event framework, but with pelvic recurrence defining the relevant event according to the study's endpoint specification. Interpretation should therefore remain tied to the precise event definition and censoring rules used in the eventual statistical analysis.
7. Statistical Methodology
Randomization
RAINBO uses randomized allocation. Randomization is fundamental to causal comparison because, in expectation, it balances both measured and unmeasured prognostic characteristics across treatment groups within the randomized comparison.
That balance is a property of the randomization process, not a guarantee that every baseline characteristic will be numerically identical between groups. Statistical analysis therefore remains necessary to quantify uncertainty around treatment-effect estimates.
Time-to-event analysis
All four registered primary endpoints are time-based recurrence outcomes. Time-to-event methods are appropriate because participants may have different lengths of follow-up and because some participants may complete observation without experiencing the event of interest.
Right censoring
Censoring occurs when the event has not been observed by the time a participant's event assessment ends. A Kaplan-Meier analysis uses the information available up to that censoring time rather than treating the participant as if an event had occurred.
Why censoring matters
A participant who remains recurrence-free through the last observation contributes evidence of event-free survival up to that point, even though the eventual recurrence status may be unknown.
Key assumption
Standard survival analysis relies on assumptions about the relationship between censoring and the underlying event process. Informative loss to follow-up can complicate interpretation.
Kaplan-Meier estimation
The Kaplan-Meier estimator constructs an event-free survival curve from the sequence of observed events and the numbers at risk. It allows the analysis to incorporate participants with different follow-up durations without assigning an artificial event time to those who are censored.
Hazard ratios and Cox models
A Cox proportional-hazards model is a common method for estimating a relative time-to-event effect. The model expresses the hazard for an individual or treatment group relative to another while allowing the baseline hazard to remain unspecified.
For a binary treatment indicator, the hazard ratio is related to the exponentiated treatment coefficient, exp(β). The proportional-hazards assumption means that the treatment hazard ratio is interpreted as constant over time on the model's hazard scale.
Log-rank testing
The log-rank test compares observed and expected numbers of events across randomized groups over follow-up. It is particularly suited to time-to-event outcomes when the question concerns whether the event-time distributions differ between groups.
8. Statistical Methods Explained
Why is recurrence-free survival a time-to-event endpoint?
Recurrence does not necessarily occur at the same time for every participant. Some participants recur early, some recur later, and some remain recurrence-free during the available follow-up. A time-to-event framework preserves that timing information instead of reducing the outcome to a simple yes-or-no indicator at enrollment.
Why use Kaplan-Meier estimation?
Kaplan-Meier estimation allows investigators to describe the probability of remaining event-free over time while incorporating right-censored observations. This is useful when participants have unequal follow-up durations.
What would a hazard ratio mean?
If a future analysis reports an HR below 1 for an intervention relative to its comparator, it would indicate a lower estimated instantaneous recurrence-related event rate under the fitted Cox model. It would not mean that the same percentage of patients avoided recurrence or that every patient experienced the same relative benefit.
Why report a confidence interval with the hazard ratio?
A point estimate alone does not communicate statistical precision. A confidence interval provides a range of values compatible with the observed data and the statistical model at the stated confidence level. Narrower intervals generally indicate greater precision than wider intervals, although precision and clinical importance are different concepts.
Why is the 3-year time frame important?
A 3-year endpoint specifies a clinically meaningful time horizon for the registered primary analysis. It also makes clear that an eventual 3-year event-free estimate is a time-specific quantity rather than a statement about lifetime recurrence risk.
Why must the four primary endpoints be analysed separately?
The registry identifies four separately named primary endpoint frameworks. Three use recurrence-free survival and one uses pelvic recurrence-free survival. Their populations, treatment comparisons, event definitions, and statistical error-control procedures determine how each result should be interpreted.
9. Multiplicity and Multiple Primary Endpoint Frameworks
The RAINBO registry contains four primary endpoint entries: p53abn-RED, MMRd-GREEN, NSMP-ORANGE, and POLEmut-BLUE. That structure creates an important statistical distinction between the existence of several primary endpoint frameworks and the interpretation of a particular hypothesis test.
| Registered component | Primary endpoint | Statistical issue |
|---|---|---|
| p53abn-RED | Recurrence-free survival at 3 years | Requires a prespecified treatment comparison and analysis population |
| MMRd-GREEN | Recurrence-free survival at 3 years | Requires its own endpoint and comparison definition |
| NSMP-ORANGE | Recurrence-free survival at 3 years | Requires its own endpoint and comparison definition |
| POLEmut-BLUE | Pelvic recurrence-free survival at 3 years | Uses a different event definition from the other three primary endpoints |
When multiple primary hypotheses are tested within one overall development program, the statistical interpretation depends on the prespecified multiplicity strategy. Possible approaches include hierarchical testing, alpha allocation, or another familywise-error-control procedure. The registry information summarized here does not specify such a procedure.
10. Analysis Populations and Interpretation
The registry identifies the study as randomized and parallel but does not provide the detailed analysis-population definitions in the available record summarized here.
For a randomized trial, an intention-to-treat analysis commonly retains participants according to their randomized assignment, preserving the principal benefit of randomization. Per-protocol or as-treated analyses can answer different questions but may be more susceptible to post-randomization differences between groups.
Intention-to-treat principle
Participants are generally analysed according to randomized assignment. This preserves the original treatment allocation as the basis for comparison.
Per-protocol interpretation
A per-protocol analysis can focus on participants who followed specified treatment procedures, but post-randomization exclusions can weaken the protection provided by randomization.
The exact definitions used for RAINBO's eventual efficacy analyses should be taken from the final statistical analysis documentation when those analyses are reported.
11. Interim Analysis and Adaptive Design Considerations
The study is registered as phase 2/3, but the available registry information does not describe an interim-analysis schedule, alpha-spending function, stopping boundary, or adaptive decision rule.
These features are particularly important in a multi-component development program because repeated looks at accumulating efficacy data can affect the probability of a false-positive conclusion unless appropriately controlled.
Interim efficacy analysis
If an interim efficacy analysis is planned, its statistical boundary and information time should be prespecified so that repeated testing does not silently inflate type I error.
Adaptive decisions
If treatment allocation, continuation, or stopping rules are changed using accumulating data, the statistical consequences depend on the exact adaptation rule and how inference is adjusted for it.
The registry does not report these details, so they cannot be assigned to the RAINBO design from the registry record alone.
12. Missing Data, Censoring, and Follow-Up
Time-to-event analysis naturally accommodates participants who have not experienced the event by the end of observed follow-up through censoring. However, censoring is not equivalent to missingness in the ordinary cross-sectional sense.
Censoring versus missing outcome data
A censored participant contributes follow-up information up to a known time. A missing assessment can create uncertainty about whether an event occurred between observed visits. The statistical handling of such situations depends on the protocol-defined endpoint and censoring rules.
The available registry information does not specify a missing-data or imputation strategy for the RAINBO primary endpoints.
13. Stratification and Covariate Adjustment
The registry identifies randomized allocation and multiple molecularly defined trial components but does not provide a complete statistical stratification scheme in the information summarized here.
Stratification can be incorporated at randomization, in the primary statistical test, or in regression modeling. When stratification factors are used, the analysis should preserve the relationship between the randomization scheme and the inferential model.
Covariate adjustment can also improve precision when prognostic baseline variables are prespecified. It should not be confused with post hoc adjustment selected because a particular variable appears imbalanced after randomization.
14. Bayesian Methods
The registry information does not identify a Bayesian analysis for the primary RAINBO endpoints.
In a Bayesian time-to-event analysis, prior information would be combined with the observed data to produce a posterior distribution for a treatment effect. A posterior probability could then quantify evidence for a specified treatment-effect region. That framework is conceptually different from a frequentist confidence interval and P-value.
Neither framework should be assumed merely from the fact that a trial is adaptive, multi-stage, or phase 2/3. The actual statistical analysis plan determines the inferential framework.
15. Safety and Adverse Events
The registry information available here identifies the interventions but does not provide serious adverse-event counts or arm-specific safety results.
Safety is nevertheless a distinct statistical domain from recurrence efficacy. A treatment comparison can have different effects on recurrence and adverse events, and those endpoints should be analysed using their appropriate estimands and denominators.
| Safety information | Registry information available here |
|---|---|
| Serious adverse events by arm | Not reported |
| Grade-specific adverse events | Not reported |
| Treatment discontinuations due to adverse events | Not reported |
| Deaths due to adverse events | Not reported |
When safety results become available, they should be reported separately by treatment exposure and according to the prespecified safety population and event definitions.
16. Trial Timeline
Study start
The registry lists 2021-11-11 as the RAINBO study start date.
Recruiting
The ClinicalTrials.gov record currently lists the study status as recruiting.
Primary completion
The registry lists 2030-01-01 as the primary completion date.
17. What a Future RAINBO Result Would Need to Show
Because no formal statistical analyses are posted to ClinicalTrials.gov, the eventual results should be interpreted through several distinct statistical layers rather than through a single number.
Absolute outcome
A 3-year recurrence-free survival estimate would describe the estimated probability of remaining recurrence-free at the prespecified time point.
Relative effect
A hazard ratio, if used, would quantify a relative time-to-event effect under the fitted model.
Precision
A confidence interval would show statistical uncertainty around an estimated treatment effect.
Multiplicity
The interpretation of each P-value would depend on how the four primary endpoint frameworks are handled within the prespecified testing strategy.
18. Interpreting a Future Hazard Ratio
If an analysis reported an HR of 0.70, for example, the model would estimate approximately a 30% lower instantaneous event rate in the treatment group relative to the comparison group, because 1 − 0.70 = 0.30.
That would not mean that 30% of patients avoided recurrence, that every patient had a 30% reduction in recurrence risk, or that the absolute 3-year recurrence-free survival difference was 30 percentage points.
A Cox hazard ratio is most straightforward to interpret when the proportional-hazards assumption is reasonable. If relative hazards change materially over time, a single HR can summarize the data imperfectly, and time-specific survival estimates or other estimands may be more informative.
19. Why This Trial Matters Statistically
RAINBO provides a useful teaching case because it combines randomized treatment allocation with multiple molecularly defined trial components and time-to-event primary endpoints.
| Statistical concept | How it appears in RAINBO |
|---|---|
| Randomization | Patients are randomly allocated within a parallel trial design. |
| Multiple trial components | Four named primary endpoint frameworks are registered. |
| Time-to-event analysis | All four primary endpoints are recurrence-related time-to-event outcomes. |
| Kaplan-Meier estimation | A standard method for estimating event-free survival over time. |
| Hazard ratio | A common relative effect measure for randomized time-to-event comparisons. |
| Censoring | Allows participants with incomplete event observation to contribute follow-up information. |
| Multiplicity | Four primary endpoint frameworks require attention to the overall hypothesis-testing structure. |
| Endpoint definition | Three components use recurrence-free survival, while POLEmut-BLUE uses pelvic recurrence-free survival. |
| Analysis populations | The eventual statistical analysis must define which randomized participants contribute to each endpoint. |
20. Important Limitations and Interpretation Issues
- No posted statistical analyses: the ClinicalTrials.gov record currently does not contain numerical primary endpoint results or formal statistical analyses.
- Recruiting status: the registry currently lists the study as recruiting, so the registered primary completion date remains in the future.
- Multiple primary frameworks: four named primary endpoints are registered, making the prespecified testing hierarchy or multiplicity strategy important for eventual confirmatory interpretation.
- Different event definitions: recurrence-free survival and pelvic recurrence-free survival should not be treated as interchangeable outcomes.
- Hazard-ratio assumptions: if Cox models are used, proportional hazards should be assessed as part of model interpretation rather than assumed automatically.
- Censoring: time-to-event estimates depend on assumptions about the relationship between censoring and the underlying event process.
- Analysis population: conclusions can differ if intention-to-treat, per-protocol, or other populations are used, so the eventual analysis population should be clearly stated.
- Safety evidence: arm-specific serious adverse-event results are not reported in the registry information summarized here.
- Arm-level treatment mapping: the registry information identifies eight arms and the intervention set but does not provide a complete arm-by-arm mapping in the information available here.
21. Clinical Interpretation vs Statistical Interpretation
Statistical interpretation
The study is structured to compare randomized adjuvant treatment strategies using recurrence-related time-to-event endpoints assessed at 3 years. The eventual evidence will depend on effect estimates, uncertainty intervals, event counts, analysis populations, and the prespecified multiplicity framework.
Clinical interpretation
The clinical meaning of the eventual findings will depend on the magnitude and durability of recurrence outcomes, the specific molecularly defined trial component, and the corresponding safety profile. Statistical significance alone would not describe the full clinical effect.
22. Sources
- ClinicalTrials.gov: RAINBO, NCT05255653.
- PubMed: PMID 36600534.
- PubMed: PMID 40966692.
Continue through the Clinical Biostats clinical trial library
Explore additional clinical trial analyses and statistical explanations across randomized studies, endpoints, and study designs.
23. Record Summary
RAINBO is a randomized, parallel phase 2/3 trial in endometrial cancer with 1615 planned participants and eight study arms. Its four registered primary endpoint frameworks are p53abn-RED, MMRd-GREEN, NSMP-ORANGE, and POLEmut-BLUE. The first three use recurrence-free survival at 3 years, while POLEmut-BLUE uses pelvic recurrence-free survival at 3 years.
From a statistical perspective, the study illustrates the importance of matching the analysis method to the endpoint definition. Recurrence outcomes are naturally treated as time-to-event data, making Kaplan-Meier estimation, log-rank comparisons, and Cox proportional-hazards modeling common analytical tools. At the same time, the presence of four primary endpoint frameworks makes multiplicity and prespecified hypothesis structure central to interpretation.
The ClinicalTrials.gov record currently lists RAINBO as recruiting and does not contain posted statistical analyses or numerical efficacy results. Consequently, the statistical evidence available from the registry at this stage is primarily the trial's design and prespecified endpoint structure rather than observed treatment effects.