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
MANTA was a randomized, parallel-group, phase 2 trial in patients with metastatic or advanced estrogen receptor positive breast cancer. The registry records 333 participants, four arms, no masking, and progression-free survival as the primary endpoint.
| Feature | MANTA |
|---|---|
| Trial acronym | MANTA |
| ClinicalTrials.gov identifier | NCT02216786 |
| Phase | Phase 2 |
| Status | Completed |
| Condition | Estrogen Receptor Positive Breast Cancer |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | None |
| Primary purpose | Treatment |
| Enrollment | 333 |
| Number of arms | 4 |
| Lead sponsor | Queen Mary University of London |
| Sponsor type | Other |
| Start date | 2014-01-16 |
| Primary completion date | 2021-12-31 |
2. Clinical Question
The registered clinical question concerns the use of AZD2014 in combination with fulvestrant in patients with metastatic or advanced estrogen receptor positive breast cancer. The primary efficacy question is framed around progression-free survival, measured from randomisation until documented progression or death.
Population
Patients with metastatic or advanced estrogen receptor positive breast cancer.
Intervention
AZD2014 in combination with fulvestrant.
Comparator structure
The registry records a four-arm randomized parallel design involving AZD2014, everolimus, and fulvestrant. The registry data presented here do not specify the treatment assignment composition of each individual arm.
Primary question
How does progression-free survival, defined from randomisation to first documented progression or death, perform across the randomized study arms?
3. Trial Design
Registered interventions
AZD2014
- Registered as a study intervention.
- Evaluated in combination with fulvestrant according to the trial title.
Everolimus
- Registered as a study intervention.
Fulvestrant
- Registered as a study intervention.
- Specifically identified in the trial title as part of the AZD2014 combination.
Four randomized arms
- The registry records four arms within a parallel randomized design.
- The available registry information does not provide the arm-by-arm enrollment counts.
4. Endpoints
The registered primary endpoint is progression-free survival. Its definition is explicitly tied to randomisation, investigator-assessed tumour progression using RECIST 1.1, death from any cause, and an assessment period of up to 100 weeks.
| Endpoint | Registry definition | Time frame |
|---|---|---|
| Progression-free survival | Defined as the time from the date of randomisation to the date of first documented tumour progression based on investigator assessment using RECIST 1.1 or death from any cause, whichever occurs first. | Date of randomisation to date of first documented progression, assessed up to 100 weeks. |
Why the endpoint is a time-to-event outcome
Progression-free survival is not simply a proportion of patients who progress. Each patient contributes a follow-up time beginning at randomisation, and the event is the first occurrence of either documented tumour progression or death from any cause.
This structure creates two important statistical features. First, patients can have different lengths of observable follow-up. Second, a patient who has not experienced progression or death by the last available assessment may contribute censored follow-up rather than a confirmed event.
The registry specifies that the first event is determined using investigator assessment of tumour progression based on RECIST 1.1, with death from any cause treated as the alternative event when it occurs first.
5. Planned Analysis
The registry lists progression-free survival as the primary endpoint but does not post a formal statistical analysis for that endpoint. The registry therefore provides the endpoint definition and assessment framework without reporting an estimated treatment effect, confidence interval, or p-value.
Primary endpoint status
Primary endpoint: progression-free survival
Assessment window: up to 100 weeks
Registry-posted statistical analysis: none
How progression-free survival is typically analysed
For a randomized trial with PFS as a time-to-event endpoint, the usual statistical framework begins with a Kaplan-Meier estimator. This estimates the probability of remaining progression-free over time while accommodating patients whose event status is not observed for the full follow-up period.
A treatment comparison is commonly expressed with a hazard ratio, usually obtained from a Cox proportional-hazards model. A hazard ratio below 1 would describe a lower estimated instantaneous rate of the PFS event in one group relative to another, whereas a hazard ratio above 1 would describe a higher estimated instantaneous event rate.
A log-rank test is also commonly used for comparing time-to-event distributions between randomized groups. The exact statistical method used for MANTA cannot be established from the registry information presented here because no formal statistical analysis has been posted.
| Statistical component | Typical role for a PFS endpoint | MANTA registry status |
|---|---|---|
| Kaplan-Meier estimator | Estimate progression-free survival over time | Not reported as a posted analysis |
| Hazard ratio | Summarize relative instantaneous event rates between groups | Not reported |
| Confidence interval | Quantify uncertainty around an estimated treatment effect | Not reported |
| Log-rank test | Compare time-to-event distributions | Not reported |
| Cox model | Estimate a hazard ratio under a proportional-hazards framework | Not reported |
| RECIST 1.1 | Define investigator-assessed tumour progression | Explicitly specified by the registry |
6. Statistical Methodology
Kaplan-Meier estimation
Kaplan-Meier estimation is the natural descriptive framework for an endpoint such as MANTA's progression-free survival. Instead of treating every patient as if follow-up were identical, it uses the observed event times and the numbers remaining at risk to estimate the probability of remaining event-free over time.
Here, di is the number of events at time ti and ni is the number of patients at risk immediately before that time.
For MANTA, the event is not merely radiographic progression. The registry explicitly defines the PFS event as the first documented tumour progression based on investigator assessment using RECIST 1.1 or death from any cause, whichever occurs first.
Hazard ratios
For a randomized comparison of time-to-event outcomes, a hazard ratio can provide a compact relative measure of the event rate between treatment groups. If the proportional-hazards assumption is reasonable, a single Cox-model hazard ratio summarizes the relative hazard over the analysis period.
The hazard ratio is a relative time-to-event measure. It is not a probability, an absolute risk difference, or the percentage of patients who benefit.
Confidence intervals
A confidence interval around a hazard ratio or other effect estimate describes statistical uncertainty under the relevant model and sampling framework. It is especially important in a multi-arm phase 2 study because each individual pairwise comparison may contain fewer patients than the total enrollment of 333.
The absence of a posted confidence interval means that the registry record does not provide a numerical assessment of the precision of a treatment-effect estimate.
Log-rank comparison
The log-rank test is commonly used when the scientific question concerns whether two or more time-to-event distributions differ. It uses the ordering of events over follow-up rather than reducing each patient to a simple binary outcome.
For a four-arm randomized study, the statistical question can be framed in several ways: an overall comparison across all arms, prespecified pairwise comparisons, or another multiple-comparison strategy. The available registry information does not specify which comparison framework was used.
RECIST 1.1 and event definition
RECIST 1.1 provides the assessment framework identified by the registry for investigator-determined tumour progression. Statistically, this is important because the classification of progression determines when the PFS event occurs.
Because death from any cause is also part of the registered PFS definition, a patient does not need to have a documented tumour progression before becoming a PFS event through death.
7. Statistical Methods Explained
Why is progression-free survival analysed as a time-to-event outcome?
Patients can experience progression at different times, and some may reach the end of observation without experiencing progression or death. A time-to-event method retains the timing of events and accommodates incomplete follow-up through censoring.
Why does death count as a PFS event?
The registered definition explicitly states that progression or death from any cause is counted, whichever occurs first. This prevents a patient who dies before documented progression from being treated as if they remained progression-free indefinitely.
Why is RECIST 1.1 important statistically?
The PFS event depends on identifying tumour progression. RECIST 1.1 supplies the assessment framework used by investigators to determine whether a progression has occurred. Consequently, the assessment rule is part of the definition of the statistical endpoint, not merely a clinical description.
What would a hazard ratio tell us?
A hazard ratio would compare the estimated instantaneous rates of progression or death between groups. For example, a hazard ratio below 1 would indicate a lower estimated event rate in the numerator group relative to the reference group. It would not mean that the same percentage of patients avoided progression.
Why does a confidence interval matter?
An effect estimate without its uncertainty can give an incomplete picture. A confidence interval shows how precisely the treatment effect has been estimated. Wide intervals indicate greater uncertainty, while narrower intervals indicate greater statistical precision under the model.
Why is a four-arm trial statistically different from a two-arm trial?
With four randomized arms, there can be more than one scientifically relevant comparison. An overall test across all arms and individual pairwise comparisons answer different questions. If multiple comparisons are made, the interpretation of p-values can also depend on how multiplicity was handled. The registry information does not specify the multiplicity strategy for MANTA.
What does an unmasked design mean for progression assessment?
Because the registry classifies the study as having no masking, treatment assignment was not masked in the registered design. For an investigator-assessed endpoint such as RECIST 1.1 progression, knowledge of treatment assignment can be relevant to the potential for assessment-related bias. The registry information does not quantify the effect of such a possibility.
8. Randomization and the Four-Arm Structure
Randomization is central to the design because it establishes treatment assignment through a randomized process rather than allowing investigators or patients to select the study arm. In principle, this supports comparison of outcomes across randomized groups.
| Design feature | Registered information | Statistical implication |
|---|---|---|
| Allocation | Randomized | Supports comparative inference based on randomized treatment assignment. |
| Design model | Parallel | Patients remain associated with their randomized study arm rather than following a crossover design described in the registry. |
| Number of arms | 4 | Creates multiple potential treatment comparisons. |
| Masking | None | Knowledge of assignment can be relevant to subjective or investigator-assessed outcomes. |
| Enrollment | 333 | Total randomized study size is divided across four arms. |
The four-arm structure is particularly relevant to interpretation because a single overall comparison would not necessarily answer the same question as each pairwise comparison. The choice of contrast should follow the scientific objective and prespecified statistical analysis plan.
9. Analysis Population and Follow-Up
The registry reports total enrollment of 333 and describes the study as randomized. It does not provide the arm-specific enrollment counts in the information available here, nor does it identify separate efficacy or safety analysis populations.
| Population feature | Registered information |
|---|---|
| Total enrollment | 333 |
| Randomization | Yes |
| Number of arms | 4 |
| Arm-specific enrollment | Not reported in the available registry information |
| Primary efficacy population | Not specified in the available registry information |
| Safety population | Not specified in the available registry information |
| Primary endpoint follow-up | Assessed up to 100 weeks |
The 100-week assessment window should not be interpreted as a statement that every participant was followed for exactly 100 weeks. It identifies the time frame attached to the registered endpoint.
10. Missing Data and Censoring
Time-to-event analyses distinguish between an observed event and incomplete event information. If a patient has no documented progression or death during the period in which they are observed, their PFS time may be censored rather than treated as an event.
The registry definition establishes the event but does not describe the censoring rules, missing-data strategy, sensitivity analyses, or imputation procedures used in the statistical analysis.
Censoring
Represents follow-up that ends without an observed PFS event during the available observation period.
Event
First documented tumour progression by investigator assessment using RECIST 1.1, or death from any cause, whichever occurs first.
Missing assessment
The registry information does not specify how missing tumour assessments were handled.
Sensitivity analysis
The registry information does not report sensitivity analyses for the primary endpoint.
11. Multiplicity and Multiple Arms
Multiplicity becomes relevant whenever several statistical hypotheses are tested. In a four-arm randomized study, investigators may be interested in an overall treatment comparison, selected pairwise comparisons, or a hierarchy of comparisons.
For example, if four arms were compared pairwise without adjustment, there would be six possible pairwise contrasts. That arithmetic follows from the four-arm structure, but it does not establish that all six comparisons were actually performed in MANTA.
This is a property of a four-arm design. It is not evidence that six hypothesis tests were conducted in the MANTA analysis.
Multiplicity procedures can control the familywise type I error rate when several confirmatory comparisons are planned. Possible approaches include hierarchical testing, Bonferroni-type procedures, or other prespecified adjustment methods. The registry information does not state which approach, if any, was used.
12. Interim Analysis
The available registry information does not report an interim analysis plan, interim efficacy boundary, alpha-spending method, or stopping rule for MANTA.
This distinction matters because an interim analysis can change the statistical interpretation of a p-value and the confidence interval if repeated looks at accumulating data are incorporated into a confirmatory testing framework.
13. Non-Inferiority, Equivalence, and Superiority
The registry information identifies progression-free survival as the primary endpoint but does not specify a non-inferiority or equivalence objective, nor does it provide a non-inferiority margin.
Accordingly, a non-inferiority interpretation cannot be assigned to the registered endpoint from the available information. In a non-inferiority design, the margin is a central component of the hypothesis and determines whether the observed confidence interval is compatible with preservation of an acceptable proportion of the reference treatment effect.
That logic differs from a conventional superiority comparison, where the principal question is whether the evidence supports a difference in a prespecified direction.
14. Bayesian Methods
The available registry information does not report Bayesian statistical methods, prior distributions, posterior probabilities, Bayesian decision thresholds, or credible intervals.
For this reason, the registered MANTA design is best described using the randomized time-to-event framework and the endpoint definition that are explicitly recorded, without assigning a Bayesian analysis that is not documented in the registry.
15. Safety and Adverse Events
The information available for this analysis identifies the study interventions and primary efficacy endpoint but does not provide arm-specific serious adverse event counts or other numerical safety results.
| Safety element | Registry information available here |
|---|---|
| Serious adverse events by arm | Not reported |
| Grade-specific adverse events | Not reported |
| Overall adverse event rates | Not reported |
| Discontinuations due to adverse events | Not reported |
| Arm-specific safety population | Not reported |
Safety and efficacy address different statistical questions. Progression-free survival evaluates the registered primary time-to-event endpoint, whereas adverse-event analyses describe harms and tolerability in relation to treatment exposure. They should not be collapsed into a single numerical measure.
16. Trial Timeline
Study start
The registry records 2014-01-16 as the trial start date.
Randomized parallel study
The study is registered as a randomized, parallel-group phase 2 trial with four arms and no masking.
Primary endpoint assessment
Progression-free survival is assessed from randomisation to first documented tumour progression or death from any cause, whichever occurs first, up to 100 weeks.
Primary completion
The registry records 2021-12-31 as the primary completion date.
Current study status
The registry identifies the MANTA study as completed.
17. What a PFS Hazard Ratio Would — and Would Not — Mean
If a Cox model were used to estimate a PFS hazard ratio, an HR below 1 would indicate a lower estimated instantaneous rate of progression or death in one randomized group relative to its reference group.
It would not mean that the same percentage of patients avoided progression, that the median PFS differed by the same percentage, or that every individual patient experienced the same relative reduction.
A confidence interval would quantify uncertainty around the estimated hazard ratio. An estimate alone cannot show whether the treatment effect has been estimated precisely or imprecisely.
Because MANTA defines PFS using either investigator-assessed progression or death, the hazard ratio would describe the combined event defined by the protocol rather than progression alone.
18. Clinical Interpretation vs Statistical Interpretation
Statistical interpretation
The registered primary endpoint is a time-to-event outcome. Its formal analysis would ordinarily require methods that account for event timing and censoring, with effect estimates accompanied by measures of uncertainty.
Clinical interpretation
The clinical question concerns whether different randomized treatment strategies are associated with differences in the duration of time before tumour progression or death.
The distinction is important. Statistical analysis determines how the observed event times are summarized and compared; clinical interpretation determines what those differences mean in the context of the disease and treatment strategies being studied.
19. Important Limitations and Interpretation Issues
- No posted statistical analysis: the registry records the primary endpoint but does not post a formal statistical analysis with an effect estimate, confidence interval, or p-value.
- Four-arm design: multiple possible comparisons exist, and the interpretation of any collection of hypothesis tests depends on the prespecified comparison strategy.
- Unspecified arm sizes: total enrollment is 333, but the available registry information does not provide enrollment counts for each of the four arms.
- Unmasked design: the study is registered as having no masking. Because progression is based on investigator assessment, knowledge of treatment assignment may be relevant to assessment-related bias.
- Endpoint censoring: progression-free survival is a censored time-to-event endpoint, and the registry information does not specify the detailed censoring rules.
- Missing data: the available registry information does not specify how missing tumour assessments or other missing endpoint information were handled.
- Proportional hazards: if a Cox proportional-hazards model is used, the hazard-ratio interpretation depends on the model's assumptions. The registry does not report whether such a model was used or whether its assumptions were evaluated.
- Assessment framework: progression is based on investigator assessment using RECIST 1.1, making the assessment process part of the primary endpoint definition.
- Generalizability: the study population is specifically defined as patients with metastatic or advanced estrogen receptor positive breast cancer, so applicability to other populations should be considered separately.
20. Why This Trial Matters Statistically
MANTA is a useful teaching example because its registered endpoint combines several central concepts in clinical-trial biostatistics: randomization, a multi-arm parallel design, investigator-assessed tumour progression, censoring, and a composite time-to-event definition involving progression or death.
| Concept | How it appears in MANTA |
|---|---|
| Randomization | The study is registered as randomized. |
| Parallel design | The registry identifies a parallel-group design. |
| Multi-arm trial | Four randomized arms are registered. |
| Time-to-event analysis | Progression-free survival is the primary endpoint. |
| Censoring | PFS requires accounting for patients who do not experience an event during observed follow-up. |
| Composite event definition | PFS ends at first documented progression or death from any cause, whichever occurs first. |
| RECIST 1.1 | Investigator assessment using RECIST 1.1 determines documented tumour progression. |
| Multiplicity | Four arms create multiple potential treatment contrasts. |
| Unmasked assessment | The registry identifies the study as having no masking. |
| Statistical reporting | No formal statistical analyses are posted in the registry information available here. |
21. Statistical Concepts in This Trial
MANTA connects several foundational ideas in survival analysis and randomized clinical-trial design. The most important statistical lesson is that the definition of the endpoint determines the appropriate structure of the analysis.
Time-to-event endpoint
PFS preserves information about when progression or death occurs rather than reducing follow-up to a simple yes/no outcome.
Censoring
Patients without a PFS event during observed follow-up can contribute partial information through their censoring time.
Hazard ratio
A commonly used relative summary for treatment comparisons in time-to-event analysis, when the underlying model is appropriate.
Multi-arm inference
Four randomized arms create multiple potential contrasts, making the choice of comparison and multiplicity strategy important.
22. Sources
- ClinicalTrials.gov: MANTA — NCT02216786.
- PubMed: PubMed record for the MANTA study publication.
Explore Clinical Biostats
Clinical trial results can be studied alongside broader statistical methods, clinical-trial design concepts, and quantitative analysis workflows.
23. Record Summary
MANTA is a completed randomized phase 2 study with 333 enrolled participants, four parallel arms, no masking, and a primary endpoint of progression-free survival. The registered PFS definition is precise: time from randomisation to the first documented tumour progression based on investigator assessment using RECIST 1.1 or death from any cause, whichever occurs first, assessed up to 100 weeks.
From a statistical perspective, the endpoint calls for a time-to-event framework capable of handling differing follow-up times and censoring. Kaplan-Meier estimation, log-rank comparison, and Cox proportional-hazards modeling are standard approaches for this type of endpoint, but the available registry information does not identify which formal statistical procedures were used for MANTA. The four-arm structure also makes the definition of planned comparisons and multiplicity particularly important.