This page separates the registered trial design from statistical interpretation. The registry identifies progression-free survival as the primary endpoint and describes its radiologic assessment framework. No formal statistical analyses are posted on ClinicalTrials.gov.
1. Trial at a Glance
eXalt3 is a phase 3 randomized, open-label, parallel-group treatment trial comparing X-396 (ensartinib) with crizotinib in patients with non-small cell lung cancer. The registry identifies progression-free survival as the primary endpoint, assessed by independent radiology review based on RECIST v. 1.1 criteria.
| Feature | eXalt3 |
|---|---|
| Trial acronym | eXalt3 |
| Phase | Phase 3 |
| Condition | Non-small Cell Lung Cancer |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | None |
| Primary purpose | Treatment |
| Enrollment | 290 |
| Number of arms | 2 |
| Interventions | X-396 (ensartinib) and crizotinib |
| Lead sponsor | Xcovery Holdings, Inc. |
| Sponsor type | Industry |
| ClinicalTrials.gov | NCT02767804 |
| Current registry status | Active, not recruiting |
2. Clinical Question
The clinical question is whether treatment with X-396 (ensartinib) should be compared with crizotinib with respect to progression-free survival in patients with non-small cell lung cancer. The registry defines PFS as the primary endpoint and specifies independent radiology review using RECIST v. 1.1 criteria.
Population
Patients with non-small cell lung cancer enrolled in the eXalt3 phase 3 trial.
Intervention
X-396 (ensartinib), a drug intervention.
Comparator
Crizotinib, a drug intervention.
Primary question
How does progression-free survival compare between the randomized ensartinib and crizotinib groups over the registered 36-month time frame?
3. Trial Design
X-396 (ensartinib)
- Drug intervention
- Randomized treatment arm
- Primary endpoint: progression-free survival
Crizotinib
- Drug intervention
- Randomized comparator arm
- Primary endpoint: progression-free survival
The parallel randomized structure is important statistically because the treatment comparison is anchored to randomized assignment. With no masking, radiologic assessment has a potential for assessment bias if knowledge of treatment assignment affects interpretation. The registry addresses this concern for the primary endpoint by specifying independent radiology review based on RECIST v. 1.1 criteria.
4. Trial Timeline and Registry Status
Study begins
The registry lists June 2016 as the study start date.
Parallel treatment comparison
The study compares X-396 (ensartinib) with crizotinib in a randomized parallel-group design.
Primary PFS assessment
Progression-free survival is registered with a 36-month time frame and independent radiology review using RECIST v. 1.1 criteria.
Registered primary completion date
The registry lists December 30, 2025 as the primary completion date.
The registry currently identifies the study as ACTIVE_NOT_RECRUITING. That status indicates that active enrollment is not occurring under the registry status, while the registered study itself remains an active record.
5. Primary Endpoint
| Endpoint | Time frame | Registry definition |
|---|---|---|
| Progression-free survival (PFS) | 36 months | As assessed by independent radiology review based on RECIST v. 1.1 criteria |
What PFS measures
Progression-free survival is a time-to-event endpoint. In a clinical-trial setting, the event is generally defined by protocol-specified disease progression or death, with patients who have not experienced the event by their last evaluable assessment contributing censored follow-up. For eXalt3, the registry specifically identifies PFS as the primary endpoint and specifies independent radiology review according to RECIST v. 1.1.
The distinction between an endpoint definition and its statistical estimator is important. RECIST v. 1.1 provides the framework for determining radiologic progression, while survival-analysis methods are used to summarize the resulting time-to-event data.
6. Planned Analysis
The registry identifies progression-free survival as the primary endpoint, with a 36-month time frame and assessment by independent radiology review based on RECIST v. 1.1 criteria. No formal statistical analyses are posted on ClinicalTrials.gov, and the registry does not provide an effect estimate, confidence interval, p-value, event count, median PFS, or treatment-group PFS estimate for this endpoint.
For a randomized two-arm trial with a time-to-event primary endpoint such as PFS, a conventional statistical analysis would generally estimate the PFS distribution with the Kaplan-Meier method, compare treatment groups with a log-rank test, and estimate the relative treatment effect with a Cox proportional-hazards model. If the statistical analysis plan specifies stratification factors, the corresponding log-rank test and Cox model would normally incorporate those prespecified strata.
Here, di represents the number of observed events at event time ti, while ni is the number of patients at risk immediately before that time. Censored patients contribute information while they remain under observation.
A Cox model would typically express the relative treatment effect as a hazard ratio. A hazard ratio below 1 would correspond to a lower estimated instantaneous rate of the PFS event in the treatment group under the fitted model; a hazard ratio above 1 would correspond to a higher estimated instantaneous event rate.
7. Statistical Methodology
Independent radiology review
The primary endpoint is specified as being assessed by independent radiology review. This is a methodological feature rather than an effect estimate. In an unmasked trial, independent review can help separate radiologic endpoint determination from the treatment team's knowledge of treatment assignment.
The distinction matters because progression is not simply a laboratory value or a patient-reported measurement. It depends on interpreting imaging according to prespecified criteria. An independent review process provides a standardized assessment framework for the primary efficacy endpoint.
RECIST v. 1.1
The registry specifies RECIST v. 1.1 as the criteria underlying the radiologic assessment of PFS. RECIST is used to classify changes in measurable disease and to determine whether a protocol-defined progression event has occurred.
Statistically, the important consequence is that radiologic assessments become the basis for constructing each patient's progression history. Those individual event times, together with censoring information, then feed the time-to-event analysis.
Kaplan-Meier estimation
Kaplan-Meier estimation is the standard descriptive framework for a PFS curve. Rather than requiring every patient to experience progression, it uses each patient's available follow-up and accounts for censoring.
The Kaplan-Meier curve is a time-specific estimate of the proportion remaining event-free. It should not be interpreted as a simple proportion of patients who have completed treatment or as a direct measure of tumor response.
Log-rank comparison
A log-rank test is commonly used to compare two Kaplan-Meier survival distributions. It evaluates whether the observed pattern of events over follow-up differs between treatment groups under the test's assumptions.
The log-rank test is not a measure of clinical magnitude. A p-value can indicate evidence against a null hypothesis while saying little by itself about the size, precision, or clinical meaning of the treatment effect.
Cox proportional-hazards model
A Cox proportional-hazards model is commonly used to estimate a hazard ratio for a two-group PFS comparison. The model compares the estimated instantaneous event rates between treatment groups over follow-up.
A hazard ratio of 1 represents equal modeled hazards at a given time. Values below 1 indicate a lower modeled hazard in the numerator group, while values above 1 indicate a higher modeled hazard.
The proportional-hazards assumption is an important consideration. A single Cox hazard ratio is most naturally interpreted when the relative hazards are reasonably stable over time. If the treatment effect changes substantially during follow-up, a single hazard ratio can compress a more complicated time-varying pattern into one summary number.
8. Statistical Methods Explained
Why is PFS a time-to-event endpoint?
PFS incorporates both whether progression or death occurs and when it occurs. Two patients can both experience progression but contribute very different amounts of information if their events occur at different times. Time-to-event analysis preserves that timing information.
Why use independent radiology review?
The trial is unmasked, so treatment assignment is known. Independent radiology review provides a separate assessment of the imaging endpoint and can reduce the influence of treatment knowledge on progression classification. It does not make the overall trial blinded; it specifically addresses assessment of the radiologic endpoint.
Why does RECIST v. 1.1 matter statistically?
Statistical analysis depends on a consistent definition of the event. RECIST v. 1.1 provides the framework for determining radiologic progression, while the survival-analysis method determines how those progression events and censored observations are summarized over time.
Why is Kaplan-Meier preferable to simply reporting the percentage progressed?
A simple percentage does not adequately account for different follow-up durations or censoring. Kaplan-Meier estimation uses the available observation time from patients who have not yet experienced an event, allowing the analysis to describe the probability of remaining progression-free at different points in time.
What would a PFS hazard ratio mean?
A PFS hazard ratio would describe the relative instantaneous rate of progression or death between the two randomized treatment groups under a Cox model. It would not mean that the same percentage of every patient's risk was reduced, nor would it directly equal a difference in median PFS or a difference in the percentage of patients progression-free at a particular time.
Why is the confidence interval important?
A confidence interval communicates statistical precision around an estimated effect. A narrow interval indicates greater precision than a wide interval, all else being equal. It does not describe the range of outcomes for individual patients and does not replace examination of the underlying PFS curves and absolute event-free probabilities.
9. Censoring and Follow-Up
Time-to-event analysis depends on distinguishing an observed event from censoring. A patient who remains progression-free at the last evaluable assessment may contribute follow-up information up to that point without being counted as having experienced progression.
Observed event
The patient experiences the protocol-defined PFS event during observed follow-up.
Censored observation
The patient's event status is not observed after the last relevant follow-up time, so the available follow-up contributes information up to that point.
Why censoring matters
Incorrect treatment of censored patients can bias the estimated PFS distribution and distort comparisons between treatment groups.
Administrative time frame
The registered PFS time frame is 36 months, providing the stated temporal framework for the primary endpoint.
For Kaplan-Meier and Cox methods to be appropriate, censoring is generally assumed to be non-informative conditional on the analysis framework: broadly, the reason a patient becomes censored should not itself be strongly related to the unobserved event process after accounting for relevant information. The ClinicalTrials.gov record does not describe a specific missing-data or censoring strategy beyond the endpoint definition.
10. Randomization and Causal Interpretation
Randomization is central to the causal interpretation of a two-arm treatment comparison. By assigning patients randomly to the ensartinib or crizotinib groups, the trial design aims to balance measured and unmeasured prognostic factors in expectation.
The value of randomization is distinct from the statistical test. A p-value does not create randomization, and a statistically significant result would not by itself establish that a treatment comparison is free of all sources of bias. Conversely, the randomized design provides an important foundation for interpreting a prespecified treatment comparison when the analysis follows the randomized groups appropriately.
| Design feature | Statistical role |
|---|---|
| Randomized allocation | Provides the basis for comparing treatment groups under a common causal framework. |
| Parallel design | Maintains the randomized treatment comparison across the study period. |
| No masking | Treatment assignment is not masked, making independent radiology review particularly relevant to the radiologic endpoint. |
| Two treatment arms | Creates a direct comparison between ensartinib and crizotinib. |
| Primary PFS endpoint | Focuses the confirmatory question on a time-to-event efficacy outcome. |
11. Analysis Population Considerations
The registry information identifies an enrollment of 290 patients but does not provide a detailed statistical analysis-population definition in the information reported here.
For a randomized efficacy endpoint, an intention-to-treat analysis would normally preserve patients according to their randomized treatment assignment. This approach maintains the benefits of randomization and avoids selectively excluding patients because of treatment discontinuation or other post-randomization events.
A safety analysis, by contrast, is often based on patients who actually received study treatment because adverse-event exposure is tied to treatment received. The exact safety analysis population and treatment-exposure rules should be taken from the trial's statistical analysis plan when available.
12. What a PFS Effect Estimate Would and Would Not Mean
If a Cox analysis produced a PFS hazard ratio below 1 for ensartinib versus crizotinib, that would indicate a lower estimated instantaneous rate of the progression-or-death event in the ensartinib group under the fitted model.
It would not mean that the same percentage of every patient experienced a reduction in risk, nor would it directly state the difference in the probability of being progression-free at 36 months.
Time-specific PFS probabilities and median PFS would provide absolute descriptions of the treatment groups. These measures answer different questions from a hazard ratio and are most informative when interpreted together.
A p-value addresses evidence against a specified null hypothesis. It does not quantify the magnitude of a treatment effect, the probability that a treatment is beneficial, or the clinical importance of a result.
A confidence interval around a hazard ratio would show the uncertainty associated with the estimated relative effect under the statistical model. Its width depends on the information available from the observed events and follow-up, not simply on the nominal enrollment number.
13. What Is Known and What Remains Unreported
| Item | Registry record |
|---|---|
| Primary endpoint | Progression-free survival (PFS) |
| PFS time frame | 36 months |
| PFS assessment | Independent radiology review based on RECIST v. 1.1 criteria |
| Enrollment | 290 |
| Treatment groups | X-396 (ensartinib) and crizotinib |
| Formal statistical analyses posted | None |
| PFS hazard ratio | Not reported in the registry information |
| PFS confidence interval | Not reported in the registry information |
| PFS p-value | Not reported in the registry information |
| Median PFS | Not reported in the registry information |
| PFS event counts by arm | Not reported in the registry information |
This distinction is statistically important. A registry can define an endpoint and its assessment framework without posting the numerical results or formal inferential analysis for that endpoint. Endpoint definition is therefore not equivalent to evidence about the observed treatment effect.
14. Important Limitations and Interpretation Issues
- No formal registry analysis: the ClinicalTrials.gov record does not post formal statistical analyses for the registered primary endpoint.
- No numerical PFS estimate in the registry record: the available registry information does not provide a hazard ratio, confidence interval, p-value, median PFS, or event count by treatment group.
- Unmasked design: treatment assignment is not masked, so independent radiology review is an important feature of the primary endpoint assessment.
- RECIST dependence: PFS classification depends on the radiologic assessment framework specified by the registry, so endpoint determination and statistical analysis are separate but linked components of the evidence.
- Censoring: PFS analysis requires appropriate handling of patients whose event status is not observed through the end of follow-up.
- Proportional-hazards assumption: if a Cox model is used, the interpretation of a single hazard ratio depends on the adequacy of the proportional-hazards framework.
- Analysis population: the registry information does not provide a detailed efficacy analysis-population definition in the record summarized here.
- Generalizability: the enrolled population and eligibility criteria determine which patients with non-small cell lung cancer are represented by the randomized comparison.
15. Why This Trial Matters Statistically
eXalt3 is a useful teaching case for understanding how a randomized oncology trial converts radiologic disease assessments into a formal time-to-event endpoint. The key statistical chain is straightforward but important: patients are randomized to two treatments, disease is assessed using a prespecified radiologic framework, progression-free survival is constructed as a time-to-event variable, and the resulting observations can be analyzed with survival-analysis methods.
| Concept | How it appears in eXalt3 |
|---|---|
| Randomization | Patients are randomized between two treatment groups. |
| Parallel-group design | The trial uses a parallel design rather than a crossover or factorial structure. |
| Time-to-event analysis | The primary endpoint is progression-free survival over a 36-month time frame. |
| Independent assessment | PFS is assessed by independent radiology review. |
| RECIST v. 1.1 | Provides the radiologic criteria underlying the primary endpoint assessment. |
| Kaplan-Meier estimation | A standard method for describing PFS over time. |
| Log-rank testing | A standard method for comparing two time-to-event distributions. |
| Cox modeling | A standard framework for estimating a relative hazard of progression or death. |
| Censoring | Allows patients without an observed event to contribute available follow-up. |
| Confidence intervals | Would quantify statistical precision around an estimated treatment effect. |
| Unmasked treatment | Makes independent radiologic assessment relevant to endpoint objectivity. |
16. Statistical Methods Explained: A Deeper View
Why does randomization matter more than simply adjusting for covariates?
Randomization creates treatment groups through an allocation mechanism that is independent of patients' prognostic characteristics in expectation. Statistical adjustment can account for measured variables, but it cannot generally reproduce the protection against unmeasured confounding created by proper randomization.
Why is PFS different from objective response?
A response endpoint generally classifies whether a prespecified amount of tumor shrinkage occurs. PFS incorporates the timing of progression or death and therefore captures duration without progression as well as the occurrence of the event.
Why is independent review not the same as blinding?
Blinding conceals treatment assignment from specified trial participants or investigators. Independent radiology review instead establishes a separate assessment process for imaging. A trial can therefore be unmasked while still using independent review of a radiologic endpoint.
Why can a hazard ratio not be converted directly into a difference in median PFS?
A hazard ratio is a relative comparison of instantaneous event rates under a model. Median PFS is the time at which the estimated survival function reaches 0.5. The two measures summarize different aspects of the event-time distribution and are not algebraically interchangeable without additional assumptions.
Why does the 36-month time frame not mean every patient must be followed for 36 months?
A registered time frame identifies the period associated with the endpoint assessment. In a time-to-event analysis, patients can contribute different amounts of follow-up. Patients who have an event earlier contribute an observed event time, while patients without an observed event may be censored at their last evaluable follow-up.
Why does a p-value not tell us how large the treatment effect is?
The p-value measures compatibility of the observed data with a specified null hypothesis under the statistical testing framework. Effect size is described by measures such as a hazard ratio or an absolute PFS difference, while a confidence interval communicates precision around an estimate.
17. A Practical Statistical Workflow for eXalt3
The most important conceptual point is that the statistical analysis begins after the clinical endpoint has been operationalized. Imaging review determines when progression has occurred according to the trial's criteria. The statistical analysis then uses those event times and censoring information to estimate the PFS distribution and compare randomized treatment groups.
This separation is useful when evaluating any clinical trial: endpoint definition, endpoint measurement, analysis method, and interpretation of the effect estimate are related but distinct steps.
18. Clinical Interpretation vs Statistical Interpretation
Statistical interpretation
The registered primary endpoint is a randomized comparison of progression-free survival, assessed through independent radiology review using RECIST v. 1.1 over a 36-month time frame. Standard analysis would treat PFS as a censored time-to-event outcome.
Clinical interpretation
The trial is designed to compare two drug interventions in patients with non-small cell lung cancer. The clinical meaning of the comparison depends on the observed PFS results together with the precision of those estimates and the trial's design characteristics.
The two perspectives should not be collapsed into a single number. A statistically estimated treatment effect describes the observed data under a model; clinical interpretation considers what that effect means in the context of disease, treatment, follow-up, and patient outcomes.
19. Data Interpretation Checklist
| Question | Why it matters |
|---|---|
| What is the endpoint? | Defines exactly what treatment comparison is being tested. |
| How is progression determined? | Determines which observations become PFS events. |
| Who performs the assessment? | Independent review can reduce the influence of treatment assignment on radiologic classification. |
| How are censored observations handled? | Determines how incomplete follow-up contributes to the PFS estimate. |
| What analysis population is used? | Determines which randomized patients contribute to the efficacy comparison. |
| What is the effect measure? | A hazard ratio, median PFS, or time-specific PFS probability answers different questions. |
| How precise is the estimate? | The confidence interval describes statistical uncertainty around an estimated effect. |
| Are model assumptions reasonable? | For example, a Cox hazard ratio relies on assumptions concerning the relationship of hazards over time. |
20. Record Summary
eXalt3 is a phase 3 randomized parallel-group trial comparing X-396 (ensartinib) with crizotinib in non-small cell lung cancer. The registry identifies progression-free survival as the primary endpoint, with a 36-month time frame and assessment by independent radiology review based on RECIST v. 1.1 criteria.
Statistically, the trial is a classic time-to-event framework: randomization establishes the treatment comparison, radiologic assessment determines progression, and survival-analysis methods can account for both event timing and censoring. Kaplan-Meier estimation, log-rank comparison, and Cox proportional-hazards modeling are standard approaches for this endpoint type, although formal statistical analyses are not posted in the ClinicalTrials.gov record summarized here.
The registry records an enrollment of 290, two treatment arms, a phase 3 randomized design, no masking, a start date of 2016-06, and a primary completion date of 2025-12-30. The registry status is ACTIVE_NOT_RECRUITING.
21. Sources
- ClinicalTrials.gov: NCT02767804 — eXalt3.
- PubMed: PubMed record associated with eXalt3.
Continue through the Clinical Biostats clinical-trial library
Explore additional clinical-trial analyses and statistical methods used to interpret randomized evidence, survival endpoints, and clinical-study designs.