← Clinical Trials
ER Positive Breast Cancer Phase 1/2 Randomized NCT01992952

FAKTION: Complete Statistical Analysis of Fulvestrant with AZD5363 in Advanced Aromatase Inhibitor-Resistant Breast Cancer

An independent statistical review of the randomized phase 1/2 FAKTION trial evaluating AZD5363 with fulvestrant in estrogen receptor positive breast cancer, with emphasis on its dose-finding and progression-free survival objectives.

FAKTION  ·  NCT01992952  ·  Randomized, quadruple-masked, parallel design
Registry note

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

FAKTION is a randomized, quadruple-masked, parallel phase 1/2 treatment trial evaluating fulvestrant with or without AZD5363 in estrogen receptor positive breast cancer. The registry identifies a phase 1b dose-finding objective and a phase 2 progression-free survival objective.

149
Enrollment
Total participants
2
Arms
Parallel design
1/2
Phase
Phase 1/2
6 mo
Primary time frame
Registry-defined endpoint frame
FeatureFAKTION
Trial acronymFAKTION
ClinicalTrials.gov identifierNCT01992952
PhasePhase 1/2
StatusActive, not recruiting
ConditionEstrogen Receptor Positive Breast Cancer
Enrollment149
AllocationRandomized
Design modelParallel
MaskingQuadruple
Primary purposeTreatment
Number of arms2
Lead sponsorVelindre NHS Trust
Sponsor typeOther government
Start date2014-05-07
Primary completion date2019-03

2. Clinical Question

The clinical question is structured in two stages. In phase 1b, the registry asks whether the maximum tolerated dose of AZD5363 can be established when AZD5363 is combined with fulvestrant and whether a recommended phase 2 dose can be established. In phase 2, the question shifts from dose selection to anti-tumour activity, measured by progression-free survival.

Population

Patients with estrogen receptor positive breast cancer, in the setting described by the trial title as advanced aromatase inhibitor-resistant breast cancer.

Intervention

AZD5363 administered in combination with fulvestrant.

Comparator

Placebo in combination with fulvestrant.

Primary questions

What is the maximum tolerated dose of AZD5363 with fulvestrant, and does the combination demonstrate anti-tumour activity as measured by PFS?

3. Trial Design

01
Enroll149 participants
02
Phase 1bAZD5363 dose finding
03
Phase 2Randomized comparison
04
AssessPFS and disease progression
05
Follow-upUp to 6 months after last randomization
Allocation
Randomized
Participants in the phase 2 comparison were allocated using a randomized design.
Structure
Parallel
The registry identifies a parallel trial design rather than a crossover or factorial design.
Masking
Quadruple masked
The registry identifies quadruple masking.
Purpose
Treatment
The primary purpose recorded by the registry is treatment.
Two statistical questions are embedded in one trial. The phase 1b endpoint is a dose-finding endpoint, whereas the phase 2 endpoint is a time-to-event efficacy endpoint. These questions require different statistical thinking: dose finding focuses on tolerability and dose selection, while PFS compares the distribution of time until progression or death.

4. Treatment Arms

ARM 1

AZD5363 + Fulvestrant

  • AZD5363
  • Fulvestrant
ARM 2

Placebo + Fulvestrant

  • Placebo
  • Fulvestrant

The registry lists AZD5363, placebo, and fulvestrant as the study interventions. The two-arm structure makes the central phase 2 comparison conceptually straightforward: fulvestrant is present in both groups, while the randomized treatment contrast is the addition of AZD5363 versus placebo.

5. Endpoints

EndpointRegistry definitionTime frame
Phase 1b primary outcome measure Maximum Tolerated Dose of AZD5363 in combination with fulvestrant 6 months
Phase 2 primary outcome Progression free survival (PFS): from date of randomization until the date of first documented progression or date of death from any cause, whichever came first. The registry defines progression according to strict RECIST v1.1 criteria, with lesions compared with baseline measurements to assess progression. From randomization until progression or death, assessed up to 6 months after the last patient is randomised

Phase 1b: Maximum Tolerated Dose

The phase 1b primary endpoint is the maximum tolerated dose of AZD5363 in combination with fulvestrant. The registry states that the objective is to establish the MTD and a recommended phase 2 dose.

Statistically, an MTD endpoint is different from a conventional superiority endpoint. The objective is not to estimate a treatment effect against a control group. Instead, the dose-selection process evaluates tolerability across dose levels and identifies a dose that satisfies the protocol's safety criteria. The registry does not report the dose-escalation rules, dose levels, dose-limiting toxicity definition, or decision boundaries in the information available here.

Phase 2: Progression-Free Survival

The phase 2 primary endpoint is PFS. The registry defines PFS as the time from randomization to the first documented progression or death from any cause, whichever occurs first. Progression is based on RECIST v1.1 criteria, with lesions compared with baseline measurements.

This definition creates a standard time-to-event structure. Patients who have not experienced progression or death by the end of their observed follow-up contribute information until the point at which their event-free observation ends. The resulting data are naturally suited to survival-analysis methods rather than a simple comparison of proportions at a single time point.

6. Statistical Methodology

Time-to-event analysis for PFS

For a randomized trial whose primary endpoint is PFS, the natural statistical framework is time-to-event analysis. The analysis considers both when an event occurs and the fact that some participants may not experience an event during observed follow-up.

Conceptual survival function
S(t) = P(T > t)

Here, T represents the time from randomization to the first qualifying PFS event. The survival function therefore represents the probability of remaining free of the defined event beyond time t.

For PFS, the event is not simply death. It is the earlier of documented disease progression and death from any cause, according to the registry definition. That distinction matters because a patient can have a PFS event without dying.

Kaplan-Meier estimation

A Kaplan-Meier estimator is commonly used to describe the PFS distribution. It produces an estimated event-free probability over time while accounting for right censoring.

Kaplan-Meier estimator
Ŝ(t) = ∏ti ≤ t (1 − di/ni)

At each observed event time, di is the number of events and ni is the number of participants at risk immediately before that time.

The registry does not post a statistical result for PFS, so there is no reported Kaplan-Meier estimate, median PFS, hazard ratio, confidence interval, or p-value to interpret here.

Cox proportional-hazards model

For a randomized PFS comparison, a Cox proportional-hazards model is a standard method for estimating a relative treatment effect. A treatment hazard ratio below 1 would indicate a lower estimated instantaneous event rate in the AZD5363-plus-fulvestrant group under the fitted model.

Hazard-ratio interpretation
HR = hAZD5363 + fulvestrant(t) / hplacebo + fulvestrant(t)

An HR of 1 represents equal modeled hazards. An HR below 1 indicates a lower modeled hazard in the numerator group; an HR above 1 indicates a higher modeled hazard.

A hazard ratio is not the same as a risk ratio or an absolute difference in PFS probability. It is a relative time-to-event measure whose interpretation depends on the fitted model and its assumptions.

Log-rank comparison

A log-rank test is another standard method for comparing two survival distributions. It uses the ordering of event times and compares the observed and expected numbers of events between randomized groups across follow-up.

For this trial, the registry establishes the randomized phase 2 PFS endpoint but does not post the formal statistical test, test statistic, p-value, or treatment-effect estimate.

Censoring

Time-to-event methods must define what happens when a participant has not experienced progression or death by the end of observable follow-up. Such an observation is generally censored rather than treated as an event. The timing and rules for censoring can materially affect a PFS analysis, particularly when follow-up differs among participants.

7. Planned Analysis

The registry identifies PFS as the phase 2 primary outcome and defines the event as the first documented progression or death from any cause, measured from randomization. For an endpoint of this type, the conventional analysis would use a time-to-event framework such as Kaplan-Meier estimation to describe PFS and a comparison such as a log-rank test and/or Cox proportional-hazards model to quantify the treatment contrast.

The phase 1b endpoint would ordinarily be evaluated through the observed tolerability and dose-limiting toxicity experience at successive AZD5363 dose levels, with the protocol's dose-escalation rules determining how the maximum tolerated dose and recommended phase 2 dose are selected. The registry does not report those dose-escalation rules in the information available here.

Registry results status: no formal statistical analyses are posted to ClinicalTrials.gov for the registered primary endpoints. Consequently, the registry provides the endpoint definitions and study design but does not provide a numerical PFS treatment effect, confidence interval, or p-value for interpretation.

8. Statistical Methods Explained

Why is PFS a time-to-event endpoint?

PFS records not only whether progression or death occurs, but also the time until that event. Participants can have different lengths of follow-up, and some may remain event-free when observation ends. A time-to-event framework preserves this timing information and handles censoring explicitly.

Why does the definition use whichever comes first?

The registry defines PFS from randomization to the first documented progression or death from any cause, whichever comes first. This prevents a participant who dies before documented progression from being treated as if they remained progression-free indefinitely.

What would a Kaplan-Meier curve show?

A Kaplan-Meier curve would estimate the probability of remaining free of progression or death over time. A vertical separation between treatment-group curves would describe differences in the estimated PFS distributions, while the timing of events and censoring would determine the shape of those curves.

What would a hazard ratio add?

A hazard ratio would summarize the relative instantaneous event rate between the two randomized groups under a Cox model. It would provide a compact relative-effect measure, but it would not replace the need to examine absolute PFS probabilities and the underlying survival curves.

Why is the confidence interval important?

A confidence interval communicates statistical uncertainty around an estimated treatment effect. A narrow interval indicates greater precision than a wide interval, all else equal. It does not describe the range of outcomes that individual patients will experience.

Why does the p-value not measure treatment-effect size?

A p-value addresses compatibility of the observed data with a specified null hypothesis under the statistical model. It is affected by sample size and variability and therefore should not be interpreted as a measure of how large or clinically important an effect is. Effect estimates and confidence intervals provide the information needed to describe magnitude and precision.

Why is randomization important for the PFS comparison?

Randomization creates the basis for comparing outcomes between groups while reducing systematic differences in measured and unmeasured baseline characteristics in expectation. The resulting treatment comparison is anchored to assignment rather than to the treatment participants happened to receive after randomization.

9. Interpreting the Phase 1b Endpoint

The phase 1b endpoint asks for the maximum tolerated dose of AZD5363 in combination with fulvestrant. This is fundamentally a dose-selection problem rather than a conventional two-group efficacy comparison.

What MTD means

MTD generally refers to the highest dose that satisfies a prespecified tolerability criterion under the protocol's dose-escalation framework.

What MTD does not mean

MTD does not mean that the selected dose is necessarily the dose with the greatest efficacy, nor does it establish superiority over placebo.

Recommended phase 2 dose

The registry states that the phase 1b objective includes establishing a recommended phase 2 dose.

Protocol dependence

The statistical interpretation of MTD depends heavily on the prespecified dose levels, toxicity definitions, observation windows, and escalation rules.

Because those detailed dose-escalation rules are not reported in the registry information available here, the MTD endpoint should be understood conceptually rather than assigned a numerical dose or decision boundary.

10. PFS as a Statistical Endpoint

PFS combines two possible events: disease progression and death. This makes it a composite time-to-event endpoint, but not every composite endpoint has the same clinical interpretation. Here, the registry explicitly defines progression according to RECIST v1.1 and includes death from any cause as the alternative first event.

ComponentRole in the registered PFS definition
Time originDate of randomization
First eventFirst documented progression or death from any cause
Progression frameworkStrict RECIST v1.1 criteria
Radiologic comparisonLesions compared with baseline measurements
Assessment windowUp to 6 months after the last patient is randomised

The time origin is especially important. Because PFS begins at randomization, the randomized treatment comparison is preserved from the point at which participants enter the phase 2 efficacy comparison.

11. What a PFS Hazard Ratio Would Mean

Statistical interpretation

If a Cox analysis produced an HR below 1 for AZD5363 plus fulvestrant versus placebo plus fulvestrant, it would indicate a lower estimated instantaneous rate of progression or death in the AZD5363 group under that model.

It would not mean that the same percentage of patients avoided progression, that every participant experienced that proportional reduction, or that the median PFS differed by the same percentage.

Confidence interval

The confidence interval around a PFS hazard ratio would describe statistical uncertainty around the estimated relative effect. A confidence interval that spans 1 would indicate that the data are compatible with both a lower and a higher hazard under the corresponding confidence framework.

Absolute PFS probabilities

Time-specific PFS probabilities would provide a complementary absolute description. For example, a PFS probability at a specified time answers a different question from a hazard ratio: it describes the estimated proportion remaining free of progression or death at that time rather than the relative event rate over follow-up.

12. Proportional-Hazards Assumption

The Cox proportional-hazards model commonly used for PFS assumes that the hazard ratio is reasonably interpretable as a common relative hazard over the relevant follow-up period. If the treatment effect changes substantially over time, a single hazard ratio can compress a more complicated pattern into one number.

This issue is important whenever a time-to-event treatment effect is not approximately proportional. A complete analysis would therefore benefit from examining the estimated survival curves, the timing of events, and, where appropriate, diagnostics or alternative summaries rather than relying on a hazard ratio alone.

Educational distinction: proportional hazards is an assumption of a particular statistical model. It is not an assumption required for the Kaplan-Meier estimator itself. The two methods therefore provide different layers of information about the same time-to-event endpoint.

13. Censoring and Follow-Up

The PFS time frame extends from randomization until progression or death, with assessment continuing up to 6 months after the last patient is randomised. A participant who has not experienced either event by the relevant end of observation would generally contribute a censored PFS time.

Censoring is informative because it determines how much follow-up information each participant contributes. Valid survival analysis generally relies on assumptions about the relationship between censoring and the event process. Differential or informative loss to follow-up can complicate interpretation.

ConceptStatistical role
EventProgression or death, whichever occurs first
CensoringEnds an observed event-free interval without counting an event
Time originRandomization
Follow-up boundaryUp to 6 months after the last patient is randomised
Analysis consequenceParticipants can contribute different amounts of event-free follow-up

14. Randomization and Treatment Contrast

The phase 2 comparison is particularly interpretable because fulvestrant appears in both treatment descriptions. The randomized contrast is therefore the addition of AZD5363 versus placebo in combination with fulvestrant.

Conceptual treatment contrast
AZD5363 + fulvestrant   vs   placebo + fulvestrant

This structure isolates the incremental randomized treatment comparison associated with AZD5363, subject to the usual assumptions and conduct of a randomized trial.

Randomization is important because it establishes the comparison before outcome information is observed. In an efficacy analysis based on randomized assignment, the treatment groups are compared according to the allocation generated by the study design rather than by selectively choosing participants after outcomes are known.

15. Masking and Bias Control

The registry identifies FAKTION as quadruple masked. Masking can reduce the risk that knowledge of treatment assignment influences aspects of trial conduct or outcome assessment.

For a PFS endpoint, masking can be particularly relevant when progression depends on clinical and radiologic assessment. The registry specifies strict RECIST v1.1 criteria for progression, providing an objective framework for defining the event, while masking can further reduce opportunities for treatment knowledge to influence assessment.

Masking does not eliminate every possible source of bias. Its value depends on how the masking is implemented, maintained, and applied throughout treatment and outcome assessment.

16. Missing Data and Analysis Considerations

Time-to-event analysis does not require every participant to have an observed progression event. Participants can be censored when event-free follow-up ends. This is different from simply deleting participants with incomplete follow-up from a conventional binary analysis.

For PFS, the key statistical questions include whether censoring rules are prespecified, whether assessments occur according to the planned schedule, and whether the reasons for incomplete follow-up could be related to prognosis or treatment. The registry information available here does not report detailed missing-data or imputation procedures.

Censoring is not imputation

A censored observation contributes follow-up information up to its censoring time. It is not equivalent to filling in an unobserved progression date.

Timing matters

Because PFS is measured over time, the date and reason for an incomplete observation can influence the amount of information contributed to the analysis.

17. Multiplicity and Endpoint Structure

FAKTION has primary outcomes in two phases, with different statistical purposes. The phase 1b endpoint concerns dose tolerance and selection, while the phase 2 primary endpoint concerns PFS. These should not be treated as though they were two interchangeable efficacy tests of the same hypothesis.

EndpointPhaseStatistical role
Maximum Tolerated Dose of AZD5363 in combination with fulvestrantPhase 1bDose-finding and tolerability
Progression free survivalPhase 2Primary efficacy endpoint

Multiplicity becomes relevant when a trial formally tests several hypotheses or repeatedly examines accumulating efficacy data. The registry information available here does not report an alpha-allocation strategy, multiplicity adjustment, or interim-efficacy boundary for the phase 2 PFS analysis.

It would therefore be inappropriate to assign an unreported multiplicity procedure to the trial. The correct interpretation is that the registry establishes the endpoint and design, while the detailed statistical analysis plan would be needed to establish the exact confirmatory testing framework.

18. Non-Inferiority, Bayesian, Factorial, and Crossover Considerations

Non-inferiority

The registry does not identify the phase 2 PFS question as a non-inferiority analysis and does not provide a non-inferiority margin.

Bayesian methods

The registry information available here does not report a Bayesian primary analysis or Bayesian decision rule.

Factorial design

The design model is recorded as parallel, not factorial.

Crossover

The registry does not identify a crossover design.

These distinctions matter because each design creates a different statistical structure. A non-inferiority trial requires an explicit margin; a factorial design estimates effects across multiple randomized factors; a crossover design uses within-participant treatment comparisons; and a Bayesian trial requires a prior and a prespecified posterior decision framework.

19. Phase 1b and Phase 2 Require Different Statistical Thinking

One of the most useful statistical features of FAKTION is the transition from dose finding to randomized efficacy evaluation.

FeaturePhase 1bPhase 2
Primary objectiveEstablish MTD and recommended phase 2 doseEstablish anti-tumour activity using PFS
Primary endpointMaximum tolerated dose of AZD5363 with fulvestrantProgression-free survival
Statistical objectDose/tolerability decisionTime-to-event treatment comparison
Natural summaryDose-level toxicity experience and dose-selection ruleSurvival function and treatment-effect estimate
Key uncertaintyRelationship between dose and tolerabilityUncertainty around the PFS treatment effect

A dose-finding result should not be interpreted like a hazard ratio, and a PFS hazard ratio should not be interpreted as a dose-selection criterion. The endpoints answer different questions within the same development program.

20. Registry Status and What Is Reported

The registry lists FAKTION as active, not recruiting. The trial began on 2014-05-07 and has a primary completion date of 2019-03. Enrollment is recorded as 149 participants across 2 arms.

2014-05-07

Trial start

The registry records the study start date as 2014-05-07.

2019-03

Primary completion

The registry records the primary completion date as 2019-03.

Current registry status

Active, not recruiting

The ClinicalTrials.gov record currently identifies the study status as active, not recruiting.

The registry reports the primary endpoint definitions but does not post formal statistical analyses for those endpoints. As a result, the statistical record available on ClinicalTrials.gov supports detailed discussion of design and planned methods, but not a numerical treatment-effect interpretation for PFS.

21. Limitations

These limitations concern the amount of statistical detail reported in the registry, not the conceptual suitability of the registered endpoints. PFS is a well-established time-to-event endpoint, but its interpretation ultimately depends on the prespecified analysis rules and the observed event and censoring data.

22. Why This Trial Matters Statistically

FAKTION provides a useful teaching example because its registered objectives span two distinct stages of clinical development. The same study framework moves from establishing a tolerable dose to evaluating a randomized time-to-event efficacy endpoint.

Statistical conceptHow it appears in FAKTION
RandomizationThe phase 2 comparison uses randomized allocation.
Parallel-group designThe registry identifies a parallel design with 2 arms.
MaskingThe registry identifies quadruple masking.
Dose findingPhase 1b seeks the maximum tolerated dose and a recommended phase 2 dose.
Time-to-event analysisPhase 2 uses PFS as the primary efficacy endpoint.
Composite event definitionPFS ends at progression or death, whichever comes first.
RECIST assessmentProgression is defined using strict RECIST v1.1 criteria.
CensoringPFS analysis requires handling participants without a progression or death during observation.
Kaplan-Meier estimationA natural method for describing the PFS distribution.
Cox proportional hazardsA standard model for estimating a relative PFS treatment effect.
Confidence intervalsWould quantify uncertainty around a treatment-effect estimate.
Phase-specific objectivesDose tolerance and randomized efficacy require different statistical frameworks.

23. Clinical Interpretation vs Statistical Interpretation

Statistical interpretation

The registered phase 2 endpoint is a time-to-event comparison of PFS. Its formal analysis would normally describe the PFS distributions and quantify the relative treatment effect while accounting for censoring.

Clinical interpretation

The clinical question is whether adding AZD5363 to fulvestrant produces anti-tumour activity in the registered population, with PFS serving as the primary phase 2 measure of that activity.

Dose-development interpretation

The phase 1b endpoint addresses whether a tolerable AZD5363 dose can be established with fulvestrant and carried forward as a recommended phase 2 dose.

Evidence interpretation

The registry provides the design and endpoint definitions but does not post the formal statistical analyses needed to quantify the observed PFS treatment effect.

24. A Practical Statistical Reading of FAKTION

A reader approaching this trial should separate four questions.

  1. What was randomized? The phase 2 design compares AZD5363 plus fulvestrant with placebo plus fulvestrant.
  2. What was measured? The primary phase 2 endpoint was PFS, defined from randomization to progression or death.
  3. How should it be analysed? PFS naturally calls for survival-analysis methods that incorporate event timing and censoring.
  4. What does the registry actually quantify? The registry establishes the endpoint and design but does not post a formal statistical analysis or numerical PFS treatment effect.

This separation is important because a clinical trial record can contain a precise endpoint definition without containing the numerical analysis required to answer the corresponding efficacy question. The endpoint definition establishes what would constitute an event; the statistical analysis determines how the observed event times are compared.

25. Sources

Continue through the Clinical Biostats clinical-trial library

Clinical trial records can be read most effectively by separating study design, endpoint definition, statistical method, and interpretation of uncertainty.

26. Record Summary

FAKTION is a randomized, quadruple-masked, parallel phase 1/2 treatment trial with 149 enrolled participants and 2 arms. Its statistical structure is notable because the registered primary objectives operate at two different stages: phase 1b seeks the maximum tolerated dose of AZD5363 in combination with fulvestrant and a recommended phase 2 dose, while phase 2 evaluates progression-free survival as the measure of anti-tumour activity.

The phase 2 PFS definition is explicit: time begins at randomization and ends at the first documented progression or death from any cause, whichever occurs first, with progression defined using strict RECIST v1.1 criteria and assessed up to 6 months after the last patient is randomised. That structure naturally leads to Kaplan-Meier and other time-to-event methods, with hazard ratios, confidence intervals, and survival probabilities providing complementary descriptions of treatment effect and uncertainty.

Statistical takeaway: FAKTION illustrates why trial interpretation begins with the endpoint definition and design before moving to statistical estimation. The registry establishes a randomized PFS question and a phase 1b dose-finding question, but it does not post formal statistical analyses for those primary endpoints.