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
VERONICA was a randomized, open-label phase 2 study comparing venetoclax plus fulvestrant with fulvestrant in women with estrogen receptor-positive, HER2-negative locally advanced or metastatic breast cancer after recurrence or progression during or after CDK4/6 inhibitor therapy.
| Feature | VERONICA |
|---|---|
| Phase | Phase 2 |
| Population | Women with estrogen receptor-positive (ER+)/HER2-negative locally advanced or metastatic breast cancer who experienced disease recurrence or progression during or after CDK4/6 inhibitor therapy |
| Design | Randomized, parallel-group, unmasked |
| Primary purpose | Treatment |
| Enrollment | 103 |
| Interventions | Venetoclax + fulvestrant versus fulvestrant |
| Status | Terminated |
| Start date | 2018-09-06 |
| Primary completion date | 2020-08-05 |
| Lead sponsor | Hoffmann-La Roche |
2. Clinical Question
The central question was whether adding venetoclax to fulvestrant improved clinical benefit compared with fulvestrant alone in women with ER-positive, HER2-negative locally advanced or metastatic breast cancer whose disease had recurred or progressed during or after CDK4/6 inhibitor therapy.
Population
Women with ER-positive, HER2-negative locally advanced or metastatic breast cancer who experienced disease recurrence or progression during or after CDK4/6 inhibitor therapy.
Intervention
Venetoclax plus fulvestrant.
Comparator
Fulvestrant.
Primary question
Does venetoclax plus fulvestrant produce a higher clinical-benefit rate than fulvestrant alone?
3. Trial Design
Venetoclax + Fulvestrant
- Venetoclax
- Fulvestrant
Fulvestrant
- Fulvestrant
The registry describes the study as randomized and parallel with no masking. That design establishes a randomized comparison between the two treatment strategies, while the open-label structure means treatment assignment was not masked.
4. Trial Timeline and Status
Study start
The VERONICA study began.
Primary completion
The registry lists 2020-08-05 as the primary completion date.
Study status
The study was terminated due to the sponsor's decision. The registry states that there were no safety concerns.
5. Endpoints
Primary Endpoint: Clinical Benefit
Clinical Benefit was defined as Complete Response (CR), Partial Response (PR) or Stable Disease (SD) lasting ≥ 24 weeks, as determined by the investigator according to RECIST v1.1.
| Endpoint | Registry definition | Time frame |
|---|---|---|
| Clinical Benefit | CR, PR or SD lasting ≥ 24 weeks from randomization in participants with measurable disease at baseline, as determined by the investigator according to RECIST v1.1. | Randomization through till 6 months after the last participant is enrolled into the study (up to approximately 23 months) |
| Progression-Free Survival | Time-to-event endpoint evaluated in the ITT population. | Randomization through till 6 months after the last participant is enrolled into the study (up to approximately 23 months) |
| Objective Response | Objective Response in participants with measurable disease at baseline. | Randomization through till 6 months after the last participant is enrolled into the study (up to approximately 23 months) |
| Overall Survival | Time from randomization to death from any cause. | Randomization to death from any cause, through till the end of the study (up to approximately 32 months) |
RECIST v1.1 Response Definitions
For target lesions assessed by CT or MRI, the registry defines complete response as disappearance of all target lesions. Partial response requires a ≥ 30% decrease in the sum of diameters of target lesions, taking the baseline sum as reference. Stable disease is neither sufficient shrinkage to qualify for partial response nor sufficient increase to qualify for disease progression. Disease progression requires a ≥ 20% increase in the sum of diameters of target lesions, taking the smallest sum on study (nadir) as reference, together with an absolute increase of at least 5 mm.
6. Analysis Populations
| Population | Registry definition and role |
|---|---|
| Baseline Measurable Disease Population | All randomized participants with measurable disease at baseline. Used for the Clinical Benefit and Objective Response analyses. |
| ITT population | All randomized participants whether or not they were assigned to the arm where the study treatment was administered. Used for PFS and OS analyses. |
This distinction is statistically important. The clinical-benefit and objective-response analyses were restricted to randomized participants with measurable disease at baseline, whereas the time-to-event analyses used the ITT population. Consequently, the denominators underlying the different endpoint analyses need not be identical.
7. Statistical Methodology
Cochran-Mantel-Haenszel Analysis
The registry reports the Cochran-Mantel-Haenszel method for the primary Clinical Benefit endpoint and for Objective Response. The corresponding effect measure is the risk difference between the venetoclax-plus-fulvestrant group and the fulvestrant group.
A negative risk difference means that the estimated percentage of participants meeting the endpoint was lower in the venetoclax-plus-fulvestrant group than in the fulvestrant group. A positive value would indicate a higher percentage in the venetoclax-plus-fulvestrant group.
Cox Proportional-Hazards Regression
The registry reports Cox regression for PFS, using the hazard ratio as the effect measure. The registry also states for OS that the hazard ratios were estimated by Cox regression, although the posted statistical method for that endpoint is listed as the log-rank test.
An HR below 1 indicates a lower estimated instantaneous event rate in the venetoclax-plus-fulvestrant group under the fitted model. An HR above 1 indicates a higher estimated instantaneous event rate.
Log-Rank Test
The posted statistical method for Overall Survival is the log-rank test. The log-rank framework compares the observed pattern of events over follow-up between randomized groups and is designed for time-to-event data in which some participants may be censored.
Confidence Intervals
The registry reports two-sided 95% confidence intervals for the primary Clinical Benefit risk difference, PFS hazard ratio, Objective Response risk difference, and OS hazard ratio. The record notes that the primary and secondary efficacy endpoints were updated to report a 95% confidence interval for the Clinical Benefit estimate and a 95% confidence interval for the Cox proportional-hazards model for PFS, following reporting conventions.
8. Primary Result: Clinical Benefit
The primary endpoint compared the percentage of participants achieving Clinical Benefit, defined as CR, PR, or SD lasting ≥ 24 weeks. The analysis population was the Baseline Measurable Disease Population, and the statistical method was the Cochran-Mantel-Haenszel method.
Risk difference for Clinical Benefit
95% CI: −16.86 to 12.94 · P = 0.7286
Hypothesis: Superiority
| Feature | Primary Clinical Benefit analysis |
|---|---|
| Comparison | Venetoclax + Fulvestrant vs Fulvestrant |
| Endpoint | CR, PR or SD lasting ≥ 24 weeks |
| Analysis population | Baseline Measurable Disease Population |
| Statistical method | Cochran-Mantel-Haenszel |
| Effect measure | Risk Difference |
| Estimate | −1.96 |
| 95% CI | −16.86 to 12.94 |
| P-value | 0.7286 |
| Hypothesis | Superiority |
The estimated risk difference of −1.96 percentage points describes the difference in the percentage of participants achieving the defined clinical-benefit endpoint between the two randomized groups. In this comparison, the point estimate is slightly below zero, meaning the estimated Clinical Benefit percentage was lower in the venetoclax-plus-fulvestrant group.
The 95% confidence interval from −16.86 to 12.94 is relatively broad and crosses zero. It represents uncertainty around the estimated between-group difference; it does not describe the range of individual patient responses.
The P-value of 0.7286 is evidence against the null hypothesis only in the context of the prespecified superiority test. It is not a measure of the size or clinical importance of the observed effect. A P-value should therefore be read alongside the risk difference and its confidence interval rather than as a standalone measure of treatment effect.
Because this endpoint was defined using response and durable stable disease, its interpretation also depends on investigator assessment according to RECIST v1.1 and on the Baseline Measurable Disease Population used for analysis.
9. Secondary Result: Progression-Free Survival
Progression-Free Survival was analyzed in the ITT population. The registry reports Cox regression with hazard ratio as the effect measure.
Hazard ratio for PFS
95% CI: 0.61–1.45 · P = 0.7853
Hypothesis: Superiority
| Feature | PFS analysis |
|---|---|
| Comparison | Venetoclax + Fulvestrant vs Fulvestrant |
| Analysis population | ITT population |
| Statistical method | Regression, Cox |
| Effect measure | Hazard Ratio |
| Estimate | 0.94 |
| 95% CI | 0.61 to 1.45 |
| P-value | 0.7853 |
| Time frame | Randomization through till 6 months after the last participant is enrolled into the study (up to approximately 23 months) |
| Hypothesis | Superiority |
The PFS hazard ratio of 0.94 is close to 1. Under the Cox model, this corresponds to an estimated instantaneous event rate that is approximately 6% lower in the venetoclax-plus-fulvestrant group, although the confidence interval shows substantial uncertainty around that point estimate.
The 95% CI of 0.61 to 1.45 includes 1.00. Thus, the interval is compatible with both a lower and a higher hazard under the model. The width of the interval is also important: the point estimate alone does not capture the uncertainty of the comparison.
The P-value of 0.7853 does not measure the magnitude of the PFS effect. It addresses the statistical evidence for the superiority hypothesis under the specified analysis. The hazard ratio, confidence interval, and P-value therefore provide complementary information.
As with any Cox analysis, interpretation of a single hazard ratio is most straightforward when the proportional-hazards assumption is reasonable. The registry does not provide a proportional-hazards diagnostic in the posted analysis.
10. Secondary Result: Objective Response
Objective Response was evaluated in the Baseline Measurable Disease Population using the Cochran-Mantel-Haenszel method and reported as a risk difference.
Risk difference for Objective Response
95% CI: −12.29 to 8.37 · P = 0.5978
Hypothesis: Superiority
| Feature | Objective Response analysis |
|---|---|
| Comparison | Venetoclax + Fulvestrant vs Fulvestrant |
| Analysis population | Baseline Measurable Disease Population |
| Statistical method | Cochran-Mantel-Haenszel |
| Effect measure | Risk Difference |
| Estimate | −1.96 |
| 95% CI | −12.29 to 8.37 |
| P-value | 0.5978 |
| Hypothesis | Superiority |
The estimated risk difference is −1.96 percentage points, indicating a slightly lower estimated Objective Response percentage in the venetoclax-plus-fulvestrant group.
The 95% CI of −12.29 to 8.37 crosses zero. The interval therefore encompasses both a lower and a higher Objective Response percentage in the venetoclax-plus-fulvestrant group relative to fulvestrant alone.
The P-value of 0.5978 is not an effect-size measure. Its interpretation belongs to the superiority hypothesis tested for this endpoint. The confidence interval provides the more direct description of the statistical uncertainty surrounding the risk-difference estimate.
11. Secondary Result: Overall Survival
Overall Survival was defined as time from randomization to death from any cause, through the end of the study. The analysis used the ITT population. The posted statistical method is the log-rank test, with hazard ratios estimated by Cox regression.
Hazard ratio for Overall Survival
95% CI: 1.02–3.43 · P = 0.0403
Hypothesis: Superiority
| Feature | OS analysis |
|---|---|
| Comparison | Venetoclax + Fulvestrant vs Fulvestrant |
| Analysis population | ITT population |
| Statistical method | Log Rank |
| Effect measure | Hazard Ratio |
| Estimate | 1.87 |
| 95% CI | 1.02 to 3.43 |
| P-value | 0.0403 |
| Hypothesis | Superiority |
| Time frame | Randomization to death from any cause, through till the end of the study (up to approximately 32 months) |
The OS hazard ratio of 1.87 is above 1. Under the Cox model, the estimated instantaneous rate of death was higher in the venetoclax-plus-fulvestrant group than in the fulvestrant group over the analyzed follow-up. Equivalently, the point estimate is approximately an 87% higher estimated hazard in the venetoclax-plus-fulvestrant group under the model.
The 95% CI of 1.02 to 3.43 lies above 1.00, although its lower bound is close to 1 and the interval is fairly wide. This indicates uncertainty about the magnitude of the relative difference even though the point estimate is above 1.
The P-value of 0.0403 addresses the superiority hypothesis under the specified statistical test. It does not measure the magnitude, clinical importance, or certainty of the estimated hazard ratio. Nor does it establish a causal mechanism for the observed difference.
The analysis is based on the ITT population, preserving randomized treatment assignment for the efficacy comparison. As a time-to-event analysis, OS also depends on censoring and follow-up. The proportional-hazards assumption is relevant to interpretation of the Cox-derived hazard ratio, and the registry does not provide a posted diagnostic for that assumption.
12. Safety Results
The registry reports serious adverse events by treatment arm. These figures describe participants affected by serious adverse events relative to the participants at risk in each arm.
| Safety measure | Venetoclax + Fulvestrant | Fulvestrant |
|---|---|---|
| Serious adverse events | 4/50 | 2/51 |
The registry states that the study was terminated due to the sponsor's decision and that there were no safety concerns. The serious-adverse-event counts should nevertheless be interpreted as a safety description rather than as an efficacy comparison or a formal test of treatment-related risk.
13. Statistical Methods Explained
Why was the Cochran-Mantel-Haenszel method used for Clinical Benefit?
The posted analysis used the Cochran-Mantel-Haenszel method for the Clinical Benefit endpoint, which is a categorical outcome expressed as a percentage of participants. This method provides a framework for comparing treatment groups while allowing stratified categorical analyses when the relevant stratification structure is specified.
What does a risk difference of −1.96 mean?
A risk difference of −1.96 means that the estimated percentage of participants meeting the endpoint was 1.96 percentage points lower in the venetoclax-plus-fulvestrant group than in the fulvestrant group. The sign matters: a positive value would indicate a higher estimated percentage in the venetoclax-plus-fulvestrant group, whereas a negative value indicates a lower estimated percentage.
Why does the Clinical Benefit confidence interval matter?
The 95% confidence interval of −16.86 to 12.94 shows the statistical uncertainty around the estimated risk difference. Because it includes zero, the observed point estimate should not be interpreted in isolation as evidence of a definite between-group difference.
What does a PFS hazard ratio of 0.94 mean?
A hazard ratio of 0.94 means that the estimated instantaneous rate of progression or death was approximately 6% lower in the venetoclax-plus-fulvestrant group under the fitted Cox model. It does not mean that 6% fewer participants experienced progression or death, and it is not an absolute risk reduction.
Why can the OS hazard ratio be greater than 1?
The hazard ratio is directional. A value above 1 means the estimated instantaneous event rate is higher in the first-named group than in the comparator. Here, the reported OS hazard ratio of 1.87 therefore describes a higher estimated hazard of death for venetoclax plus fulvestrant relative to fulvestrant alone under the Cox model.
Why are PFS and OS analyzed differently from response endpoints?
Clinical Benefit and Objective Response are categorical endpoints summarized as percentages and analyzed using the Cochran-Mantel-Haenszel method. PFS and OS are time-to-event endpoints, so their analyses account for the timing of events and censoring. The registry therefore reports Cox regression for PFS and a log-rank test with Cox-derived hazard ratios for OS.
14. Interpreting Confidence Intervals and P-values Together
| Endpoint | Effect | 95% CI | P-value | Statistical reading |
|---|---|---|---|---|
| Clinical Benefit | RD −1.96 | −16.86 to 12.94 | 0.7286 | Interval crosses zero |
| PFS | HR 0.94 | 0.61 to 1.45 | 0.7853 | Interval crosses one |
| Objective Response | RD −1.96 | −12.29 to 8.37 | 0.5978 | Interval crosses zero |
| Overall Survival | HR 1.87 | 1.02 to 3.43 | 0.0403 | Interval is above one |
For a risk difference, zero represents no difference between groups. For a hazard ratio, one represents equal estimated hazards. These reference values provide a useful way to read the confidence intervals, but the confidence interval remains an uncertainty statement rather than a range of individual treatment effects.
A P-value and a confidence interval answer related but different questions. The P-value describes the compatibility of the observed data with a specified null hypothesis under the statistical model. The confidence interval describes uncertainty around the estimated effect. Neither quantity by itself describes the clinical importance of the treatment difference.
15. Superiority Testing
The posted analyses identify superiority as the hypothesis for the Clinical Benefit, PFS, Objective Response, and OS comparisons. This matters because the direction of the effect is part of the interpretation.
Risk difference endpoints
For Clinical Benefit and Objective Response, superiority corresponds to seeking a higher percentage of participants with the favorable endpoint in the venetoclax-plus-fulvestrant group.
Time-to-event endpoints
For PFS and OS, a hazard ratio below 1 would represent a lower estimated event hazard for the venetoclax-plus-fulvestrant group, while a hazard ratio above 1 represents a higher estimated event hazard.
This directional distinction prevents an important interpretive error: the same numerical direction does not have the same clinical meaning for every endpoint. A higher response percentage and a higher hazard of death are not analogous outcomes.
16. Registry Reporting Update
The registry states that the primary and secondary efficacy endpoints were updated to report a 95% confidence interval for the Clinical Benefit estimate and a 95% confidence interval for the Cox proportional-hazards model for PFS, following reporting conventions.
17. Study Termination and Interpretation
The ClinicalTrials.gov record lists the study as TERMINATED. The registry states that termination was due to the sponsor's decision and that there were no safety concerns.
Termination status is a design and interpretation consideration rather than an efficacy result by itself. The statistical interpretation should therefore remain anchored to the analyses actually posted in the registry: Clinical Benefit, PFS, Objective Response, and OS, together with their specified analysis populations and statistical methods.
18. Limitations
- Termination: the registry reports that the study was terminated due to the sponsor's decision. The record states that there were no safety concerns.
- Population differences across endpoints: Clinical Benefit and Objective Response use the Baseline Measurable Disease Population, whereas PFS and OS use the ITT population.
- Open-label design: masking was none. Investigator determination of Clinical Benefit according to RECIST v1.1 therefore occurred in an unmasked study.
- Confidence-interval width: the reported intervals, particularly for the time-to-event endpoints, demonstrate uncertainty around the point estimates and should be considered when interpreting the results.
- Hazard-ratio interpretation: Cox hazard ratios are model-based relative measures and are most naturally interpreted when the proportional-hazards assumption is reasonable.
- Censoring: PFS and OS analyses depend on follow-up and censoring patterns. A hazard ratio is not an absolute probability of experiencing the event.
- Multiplicity: multiple efficacy endpoints are reported. A collection of nominal P-values should not automatically be interpreted as though every endpoint were an independent confirmatory test with its own error budget.
- Endpoint definitions: Clinical Benefit incorporates stable disease lasting ≥ 24 weeks as well as complete and partial responses, so it should not be equated with objective tumor response alone.
19. Why This Trial Matters Statistically
VERONICA is a useful teaching example because the registry contains several distinct statistical structures within a relatively compact randomized phase 2 study.
| Concept | How it appears in VERONICA |
|---|---|
| Randomization | Randomized comparison of venetoclax + fulvestrant versus fulvestrant |
| Parallel design | Two parallel treatment groups |
| Open-label design | Masking was none |
| Categorical endpoint | Clinical Benefit and Objective Response expressed as percentages |
| Risk difference | Primary Clinical Benefit and secondary Objective Response effect measure |
| Cochran-Mantel-Haenszel method | Used for Clinical Benefit and Objective Response |
| Time-to-event analysis | PFS and OS analyzed over follow-up |
| Cox regression | Hazard ratio estimation for PFS and OS |
| Log-rank test | Posted statistical method for OS |
| ITT principle | PFS and OS analyzed in all randomized participants |
| Measurable-disease population | Clinical Benefit and Objective Response analyzed among randomized participants with measurable disease at baseline |
| Confidence intervals | 95% two-sided intervals reported for all posted efficacy effect estimates |
| Superiority hypothesis | Specified for the primary and posted secondary efficacy analyses |
20. Statistical Concepts in This Trial
Learn more about the methods used in this trial:
21. Related Statistical Calculators
22. Sources
- ClinicalTrials.gov: NCT03584009 — VERONICA.
Continue through the Clinical Biostats knowledge graph
Explore statistical tutorials and calculators for the methods used to analyze randomized clinical trials and time-to-event endpoints.
23. Record Summary
VERONICA provides a useful example of how a randomized phase 2 trial can combine categorical response endpoints with time-to-event outcomes. The primary Clinical Benefit analysis used the Cochran-Mantel-Haenszel method and reported a risk difference of −1.96 with a two-sided 95% confidence interval of −16.86 to 12.94 and a P-value of 0.7286. Secondary analyses reported a PFS hazard ratio of 0.94, an Objective Response risk difference of −1.96, and an OS hazard ratio of 1.87, each with its corresponding two-sided 95% confidence interval and P-value.
The most informative statistical reading is therefore not a single number. It combines the endpoint definition, analysis population, effect measure, confidence interval, P-value, and time-to-event methodology. The registry's report that the study was terminated due to the sponsor's decision, with no safety concerns, is also important context for interpreting the completed study record.