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
MAINTAIN was a randomized, double-masked, parallel phase 2 trial evaluating ribociclib after progression on CDK4/6 inhibition in patients with HR+ HER2- advanced breast cancer. The registry lists progression-free survival as the primary endpoint, with disease status assessed approximately every 12 weeks.
| Feature | MAINTAIN |
|---|---|
| Phase | Phase 2 |
| Condition | Metastatic Breast Cancer; Breast Carcinoma |
| Population | Patients with HR+ HER2- advanced breast cancer after progression on CDK4/6 inhibition |
| Allocation | Randomized |
| Design | Parallel |
| Masking | Double |
| Primary purpose | Treatment |
| Enrollment | 169 |
| Number of arms | 2 |
| Status | Completed |
| Start date | 2016-03 |
| Primary completion date | 2022-01-04 |
| Lead sponsor | Melissa K Accordino |
| Sponsor type | OTHER |
| ClinicalTrials.gov | NCT02632045 |
2. Clinical Question
The central statistical question is whether changing treatment to include ribociclib after progression on CDK4/6 inhibition is associated with a different progression-free survival experience than the corresponding placebo-controlled treatment approach in patients with HR+ HER2- advanced breast cancer.
Population
Patients with HR+ HER2- advanced breast cancer after progression on CDK4/6 inhibition.
Intervention
Ribociclib (LEE-011) with fulvestrant.
Comparator
Placebo with fulvestrant.
Primary question
Does the ribociclib-containing treatment produce a different progression-free survival outcome than placebo/fulvestrant?
3. Trial Design
Ribociclib / Fulvestrant
- Ribociclib (LEE-011)
- Fulvestrant
Placebo / Fulvestrant
- Placebo
- Fulvestrant
The registry classifies the allocation as randomized and the design model as parallel. With two treatment arms, the parallel structure means that participants are compared according to the treatment strategy assigned at randomization rather than being assigned sequentially to both study arms.
4. Endpoints
The registry identifies progression-free survival (PFS) as the primary endpoint. The endpoint is explicitly defined as a time-to-event measure beginning at randomization.
| Endpoint | Registry definition | Time frame |
|---|---|---|
| Progression Free Survival (PFS) | Interval from the time of randomization until objective disease progression (local assessment) or death from any cause. Disease status is assessed with comprehensive radiographic studies every three treatment cycles, approximately every 12 weeks (+/- 1 week). | Up to approximately 30 months |
RECIST 1.1 progression definition
The registry defines disease progression according to RECIST 1.1 as at least a 20% increase in the sum of diameters of target lesions, using the smallest sum on study as the reference, including the baseline sum when it is the smallest on study. The sum must also show an absolute increase of at least 5 mm. The appearance of one or more new lesions is also considered progression.
Because either progression or death can terminate the PFS interval, PFS is a composite time-to-event endpoint. Patients without either event by their last evaluable assessment contribute follow-up until censoring.
5. Statistical Methodology
Why PFS is a time-to-event endpoint
PFS is not simply a proportion of patients who progressed. Each patient contributes a duration from randomization until an event or censoring. This makes the timing of progression clinically and statistically relevant: an event occurring early and an event occurring late are not treated as equivalent observations.
Kaplan-Meier estimation
For a randomized trial with PFS as its primary endpoint, a standard descriptive approach is the Kaplan-Meier estimator. It estimates the probability of remaining progression-free over time while accommodating patients whose PFS experience is censored before an event is observed.
Here, di is the number of events at time ti and ni is the number at risk immediately before that time.
Hazard ratios and Cox regression
A standard comparative analysis for a randomized PFS endpoint is a Cox proportional-hazards model, which expresses the treatment contrast as a hazard ratio. A hazard ratio below 1 indicates a lower estimated instantaneous event rate in the treatment group under the model; a hazard ratio above 1 indicates a higher estimated instantaneous event rate.
The hazard ratio is a relative time-to-event measure. It is not a probability of progression, not a difference in median PFS, and not the proportion of patients who benefit.
Log-rank comparison
A standard hypothesis test for two randomized survival curves is the log-rank test. It compares the observed and expected numbers of events between treatment groups over the follow-up period. The test addresses evidence against equality of the underlying event-time experience; it does not itself quantify the size of the treatment effect.
Censoring
Survival analysis relies on information from patients who have not yet experienced the event. Such patients are generally censored at an appropriate follow-up time. The interpretation of a Kaplan-Meier curve therefore depends not only on observed events but also on the rules used to determine when observations are censored.
6. Planned Analysis
The registry's primary endpoint is PFS through approximately 30 months. For an endpoint of this type, the statistical analysis would ordinarily describe the PFS distribution using Kaplan-Meier methods and compare the randomized treatment groups using a time-to-event test such as the log-rank test. A Cox proportional-hazards model is commonly used to estimate a hazard ratio and its confidence interval.
What would be estimated?
The analysis would typically estimate the PFS experience over time and summarize the relative treatment effect using a hazard ratio with a confidence interval.
What would be tested?
A formal hypothesis test would typically evaluate whether the PFS distributions differ between the randomized treatment groups.
What determines an event?
Objective disease progression or death from any cause, using the registry's RECIST 1.1 progression definition.
What determines follow-up?
Radiographic disease assessment is scheduled every three treatment cycles, approximately every 12 weeks (+/- 1 week).
ClinicalTrials.gov does not report a PFS hazard ratio, confidence interval, p-value, median PFS, or other numerical efficacy estimate in the statistical-analysis information represented here. No formal statistical analyses were posted to ClinicalTrials.gov.
7. Statistical Methods Explained
Why is PFS analyzed differently from a simple response rate?
A response rate records whether a prespecified tumor response occurred. PFS records when progression or death occurred. The additional time dimension means that two patients who both eventually progress can contribute very different amounts of information depending on when progression occurs.
Why is death included in the PFS endpoint?
The registry defines PFS as the interval from randomization to objective disease progression or death from any cause. Including death prevents a patient who dies before documented radiographic progression from being treated as though they remained progression-free indefinitely.
Why use Kaplan-Meier estimation?
Not every participant will necessarily experience progression or death during the observation period. Kaplan-Meier estimation allows those patients to contribute information up to the point at which their follow-up ends, rather than requiring every participant to have an observed event.
What would a hazard ratio mean in MAINTAIN?
A hazard ratio would compare the estimated instantaneous rate of a PFS event between the ribociclib/fulvestrant and placebo/fulvestrant groups. An HR of 0.70, for example, would describe an estimated 30% lower hazard under the fitted model, but it would not mean that exactly 30% fewer patients progressed.
Why does the RECIST 1.1 definition matter?
The endpoint depends on how progression is classified. The registry specifies both a relative increase of at least 20% and an absolute increase of at least 5 mm in the sum of target-lesion diameters, while also defining the appearance of new lesions as progression. These rules determine which observations become PFS events.
Why does the assessment schedule matter?
Radiographic assessment is scheduled every three treatment cycles, approximately every 12 weeks (+/- 1 week). Progression is therefore observed at assessment times rather than continuously. The timing of these assessments is part of how PFS is measured and can affect when progression is detected.
8. Safety Results
The registry reports serious adverse events by treatment arm. These figures describe participants affected by a serious adverse event relative to the number at risk in each reported arm.
| Reported safety measure | Affected | At risk |
|---|---|---|
| Placebo/Fulvestrant | 0 | 59 |
| Ribociclib (LEE-011)/Fulvestrant | 4 | 60 |
The serious-adverse-event comparison should be interpreted separately from the PFS endpoint. Efficacy and safety answer different statistical questions, and a treatment effect on PFS cannot be used as a substitute for an assessment of adverse-event experience.
9. Clinical Biostats Interpretation of the Primary Endpoint
PFS measures the time from randomization until the first qualifying event: objective disease progression by the registry's local-assessment definition or death from any cause. It therefore incorporates both radiographic progression and mortality into one time-to-event outcome.
If the ribociclib/fulvestrant group had a hazard ratio below 1 relative to placebo/fulvestrant, that would indicate a lower estimated instantaneous rate of progression or death under the specified time-to-event model. It would not establish that every patient experienced delayed progression or that the same relative effect applied at every point in time.
A confidence interval around a hazard ratio describes statistical uncertainty surrounding the estimated relative treatment effect. A narrow interval indicates greater precision than a wide interval. The interval does not describe the range of outcomes that individual patients can experience.
A p-value addresses the evidence against a specified null hypothesis under the statistical model. It does not tell the reader how large the treatment effect is. Effect magnitude is better described using measures such as a hazard ratio, confidence interval, median PFS, or time-specific PFS estimates when those quantities are available.
A Cox hazard ratio is easiest to interpret as a single relative effect when the proportional-hazards assumption is reasonable. If hazards change substantially relative to one another over time, a single HR may not fully describe the shape of the treatment-group difference. The registry information represented here does not provide enough statistical-analysis detail to assess that assumption.
10. Limitations and Interpretation Issues
- No posted numerical efficacy analysis: although the registry indicates that results are posted, the statistical-analysis information represented here does not include a PFS estimate, confidence interval, or p-value.
- No median PFS reported here: the registry information represented here does not provide a median PFS or its confidence interval.
- No subgroup estimates: the registry information represented here does not report subgroup-specific PFS estimates or treatment-effect interactions.
- No baseline statistical table: baseline demographic and disease-characteristic estimates are not reported in the information represented here.
- Local progression assessment: the registry defines progression using local assessment, which is relevant when considering how PFS events were classified.
- Assessment timing: radiographic evaluations occur approximately every 12 weeks (+/- 1 week), so progression is identified through scheduled assessments rather than continuously observed.
- Composite endpoint: PFS combines objective progression and death from any cause. The two event types have different clinical meanings even though both terminate the endpoint.
- Safety and efficacy are distinct: the reported serious-adverse-event counts should not be combined with PFS into a single summary measure.
- Analysis details: the registry information represented here does not specify the fitted survival model, stratification factors, multiplicity procedure, interim-analysis procedure, or missing-data/imputation method.
11. Why This Trial Matters Statistically
MAINTAIN is a useful teaching case because its primary endpoint illustrates the central statistical challenge of oncology trials: disease progression is not merely a yes-or-no outcome. The timing of progression matters, patients may remain event-free at the end of observation, and death is incorporated directly into the endpoint definition.
| Statistical concept | How it appears in MAINTAIN |
|---|---|
| Randomization | The trial uses randomized allocation between two treatment arms. |
| Parallel design | The two randomized groups are followed as parallel treatment strategies. |
| Blinding | The registry classifies masking as double. |
| Time-to-event analysis | PFS records time from randomization to progression or death. |
| Kaplan-Meier estimation | A standard method for describing the PFS distribution when censoring occurs. |
| Hazard ratio | A standard model-based way to quantify a relative difference in event rates over time. |
| RECIST 1.1 | Provides the registry's rules for classifying objective disease progression. |
| Censoring | Allows patients without an observed PFS event during available follow-up to contribute partial follow-up. |
| Scheduled assessment | Radiographic disease status is assessed every three treatment cycles, approximately every 12 weeks (+/- 1 week). |
| Safety analysis | Serious adverse events are reported separately by treatment arm. |
12. Trial Timeline
Trial begins
The registry lists March 2016 as the study start date.
Randomized treatment comparison
The study is classified as a phase 2, randomized, double-masked, parallel trial with two arms and enrollment of 169.
Primary endpoint time frame
The registry specifies PFS assessment over a time frame of up to approximately 30 months.
Primary completion
The registry lists January 4, 2022 as the primary completion date.
13. Randomization and Treatment Contrast
Randomization creates the principal framework for comparing the two treatment strategies. The statistical contrast is between the PFS experience associated with the randomized ribociclib/fulvestrant strategy and the PFS experience associated with placebo/fulvestrant.
| Dimension | Ribociclib group | Control group |
|---|---|---|
| Study drug | Ribociclib (LEE-011) | Placebo |
| Background treatment | Fulvestrant | Fulvestrant |
| Randomization | Randomized | Randomized |
| Masking | Double | Double |
| Design structure | Parallel | Parallel |
Because both arms contain fulvestrant, the treatment contrast is focused on the presence versus absence of ribociclib within the randomized treatment strategies described by the registry. The value of randomization is that treatment assignment, rather than investigators' choice of therapy, defines the primary comparison.
14. Understanding RECIST-Based PFS
The progression component of PFS depends on radiographic measurement. The registry specifies a quantitative threshold rather than allowing any visible change to count automatically as progression.
Relative threshold
The sum of diameters of target lesions must increase by at least 20% relative to the smallest sum on study.
Absolute threshold
The increase must also represent an absolute increase of at least 5 mm.
New lesions
The appearance of one or more new lesions is considered progression.
Assessment interval
Comprehensive radiographic studies are performed every three treatment cycles, approximately every 12 weeks (+/- 1 week).
This structure illustrates why endpoint definitions are statistical as well as clinical specifications. Changing the progression rule changes which observations are counted as events, which in turn changes the PFS data set used for survival analysis.
15. What the Registry Reports — and What It Does Not Report
| Item | Registry information |
|---|---|
| Trial status | Completed |
| Enrollment | 169 |
| Primary endpoint | Progression Free Survival (PFS) |
| PFS time frame | Up to approximately 30 months |
| Results posted | Yes |
| Posted statistical analyses | No formal statistical analyses are listed in the statistical-analysis information represented here |
| PFS hazard ratio | Not reported in the statistical-analysis information represented here |
| PFS confidence interval | Not reported in the statistical-analysis information represented here |
| PFS p-value | Not reported in the statistical-analysis information represented here |
| Median PFS | Not reported in the statistical-analysis information represented here |
| Subgroup estimates | Not reported in the statistical-analysis information represented here |
| Serious adverse events | 0/59 for Placebo/Fulvestrant; 4/60 for Ribociclib (LEE-011)/Fulvestrant |
This distinction is important when interpreting a registry. A posted results record can contain outcome information without supplying every statistical estimate needed for a complete inferential analysis. The absence of a reported hazard ratio or p-value is different from evidence that the treatment comparison had no effect.
16. Statistical Interpretation vs Clinical Interpretation
Statistical interpretation
The primary endpoint is a randomized time-to-event comparison. Appropriate analysis focuses on event timing, censoring, the estimated relative treatment effect, and uncertainty around that estimate.
Clinical interpretation
PFS represents the time until objective progression or death. The clinical meaning of the endpoint therefore depends on both how progression is defined and how long patients remain free of either event.
A statistically estimated treatment effect and a clinical judgment about its importance are related but distinct. The magnitude, precision, duration, safety profile, and clinical context all matter when interpreting a time-to-event result.
17. Important Statistical Distinctions
A hazard ratio compares instantaneous event rates under a statistical model. It is not the probability that a patient will progress, and it should not be translated directly into a percentage of patients who benefit.
Median PFS identifies the time at which the estimated progression-free survival probability reaches 50%, whereas a hazard ratio summarizes a relative event-rate contrast. They answer different questions and can provide complementary information.
A p-value evaluates evidence against a null hypothesis. It does not quantify the clinical magnitude of a treatment effect. A complete interpretation pairs hypothesis testing with an effect estimate and its confidence interval.
Both events terminate PFS, but they are not clinically interchangeable. The composite endpoint is useful because it captures either loss of disease control through objective progression or death from any cause.
18. Why the Analysis Population Matters
Randomization establishes the treatment comparison, but interpretation also depends on who is included in each analysis. For a primary randomized efficacy analysis, the usual framework is to preserve randomized treatment assignment rather than redefining groups according to treatment received after randomization.
The registry information represented here does not specify a detailed efficacy analysis population beyond the randomized trial structure. It also does not provide the statistical-analysis conventions needed to determine how every potential protocol deviation, treatment discontinuation, or missing assessment was handled.
19. Overall Statistical Perspective
MAINTAIN illustrates a compact but important clinical-trial analysis framework. The trial is randomized, double-masked, parallel, and phase 2, with PFS designated as the primary endpoint. The endpoint begins at randomization and ends at objective disease progression or death, with progression defined using specific RECIST 1.1 criteria and radiographic assessment scheduled approximately every 12 weeks.
From a statistical perspective, the key analytical structure is therefore a two-group time-to-event comparison. Kaplan-Meier methods provide a natural way to describe PFS over time, while a log-rank test and Cox model are standard tools for comparing groups and estimating relative treatment effects. Interpretation must account for censoring, the composite nature of PFS, the RECIST progression rule, and the assumptions underlying any fitted proportional-hazards model.
The registry does not provide the numerical inferential results needed to characterize the magnitude or precision of the PFS treatment effect. The serious-adverse-event counts provide a separate safety signal, with 0/59 affected participants reported for Placebo/Fulvestrant and 4/60 for Ribociclib (LEE-011)/Fulvestrant.
20. Sources
- ClinicalTrials.gov: NCT02632045 — MAINTAIN.
Continue through the Clinical Biostats knowledge graph
Explore the broader statistical methods used to understand randomized clinical trials, survival endpoints, and treatment-effect estimation.
21. Record Summary
MAINTAIN is a randomized, double-masked phase 2 trial with 169 enrolled participants and two parallel treatment arms evaluating ribociclib (LEE-011) with fulvestrant versus placebo with fulvestrant in HR+ HER2- advanced breast cancer after progression on CDK4/6 inhibition. Its primary endpoint is PFS, defined from randomization to objective disease progression or death from any cause, with progression determined using the registry's RECIST 1.1 criteria and radiographic assessment approximately every 12 weeks.
Statistically, the trial provides a clear example of how a randomized treatment question becomes a time-to-event analysis. The important concepts are the definition of the event, the timing of assessments, censoring, Kaplan-Meier estimation, comparative survival testing, hazard-ratio interpretation, confidence intervals, and the distinction between efficacy and safety. The registry reports serious adverse events of 0/59 for Placebo/Fulvestrant and 4/60 for Ribociclib (LEE-011)/Fulvestrant, while numerical inferential analyses for the primary PFS endpoint are not reported in the statistical-analysis information represented here.