This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record. Numerical trial results on this page are restricted to the information reported in the ClinicalTrials.gov record for FIGHT-302.
1. Trial at a Glance
FIGHT-302 was a randomized, open-label, parallel-group phase 3 trial evaluating pemigatinib versus gemcitabine plus cisplatin in participants with unresectable or metastatic cholangiocarcinoma. The registry reports 167 enrolled participants and two treatment arms. The study was terminated because enrollment became difficult after a change in the standard of care for first-line treatment of cholangiocarcinoma; the registry states that safety concerns did not contribute to the decision.
| Feature | FIGHT-302 |
|---|---|
| Phase | Phase 3 |
| Population | Participants with unresectable or metastatic cholangiocarcinoma |
| Design | Randomized, parallel-group, open-label |
| Allocation | Randomized |
| Primary purpose | Treatment |
| Enrollment | 167 |
| Arms | 2 |
| Lead sponsor | Incyte Corporation |
| Sponsor type | Industry |
| ClinicalTrials.gov | NCT03656536 |
| Status | Terminated |
2. Clinical Question
The central question was whether pemigatinib produced better clinical outcomes than gemcitabine plus cisplatin in participants with unresectable or metastatic cholangiocarcinoma. The registry identifies superiority as the hypothesis type for the posted statistical analyses.
Population
Participants with unresectable or metastatic cholangiocarcinoma.
Intervention
Pemigatinib 13.5 mg QD.
Comparator
Gemcitabine 1000 mg/m2 plus cisplatin 25 mg/m2, followed by pemigatinib 13.5 mg QD as described in the registry's treatment-group comparison.
Primary question
Does randomized assignment to pemigatinib improve progression-free survival compared with the gemcitabine-plus-cisplatin treatment sequence?
3. Trial Design
Pemigatinib 13.5 mg QD
- Pemigatinib 13.5 mg QD.
- The registry's randomized-treatment comparison identifies this as the pemigatinib treatment group.
- Statistical efficacy analyses were based on treatment assignment at randomization.
Gemcitabine plus cisplatin
- Gemcitabine 1000 mg/m2.
- Cisplatin 25 mg/m2.
- The registry's statistical comparison describes the treatment sequence as gemcitabine plus cisplatin to pemigatinib 13.5 mg QD.
4. Randomization and Analysis Population
The primary PFS analysis was conducted in the Intent-to-Treat (ITT) Population, defined in the registry as all randomized participants. Treatment groups were defined according to the treatment assignment at the time of randomization.
| Analysis population | Registry definition / role |
|---|---|
| Intent-to-Treat | All randomized participants; treatment groups were defined according to treatment assignment at randomization. |
| ITT for ORR | ITT Population; confidence intervals were calculated using the exact method for the binomial distribution. |
| ITT for overall survival | ITT Population; median survival was estimated using the Kaplan-Meier method and confidence intervals using the Brookmeyer and Crowley method with log-log transformation. |
| ITT for duration of response | ITT Population; median survival was estimated using the Kaplan-Meier method and confidence intervals using the Brookmeyer and Crowley method with log-log transformation. The registry further restricts the analysis to participants meeting the response-related analysis criteria. |
5. Primary Endpoints
The registry lists four primary endpoints. They represent two related PFS outcomes in the randomized treatment period and two corresponding PFS outcomes in the transition period.
| Primary endpoint | Time frame | Registry definition |
|---|---|---|
| Randomized Treatment Period: Progression-free Survival (PFS) | up to 1422 days | PFS was defined as the time from the date of randomization until the date of disease progression according to RECIST v1.1 and assessed by an ICR, or death, whichever occurs first. |
| Randomized Treatment Period: Kaplan-Meier Estimates of PFS of the Indicated Lengths of Time | up to 1422 days | PFS was defined as the time from the date of randomization until the date of disease progression according to RECIST v1.1 and assessed by an ICR, or death, whichever occurs first. Kaplan-Meier estimates indicate the percent probability of a participant being alive at the indicated time after treatment start. |
| Transition Period: PFS | up to 1496 days | PFS was defined as the time from the date of the first dose of study drug after transition until the earliest date of progressive disease or death due to any cause. |
| Transition Period: Kaplan-Meier Estimates of PFS of the Indicated Lengths of Time | up to 1496 days | PFS was defined as the time from the date of the first dose of study drug after transition until the earliest date of progressive disease or death due to any cause. Kaplan-Meier estimates indicate the percent probability of a participant being alive at the indicated time after treatment start. |
6. Statistical Methodology
Stratified log-rank test
The primary randomized-treatment PFS comparison used a stratified log-rank test. The registry states that the P-value was calculated using a stratified log-rank test and that the strata information was based on the Interactive Response Technology (IRT) randomization data.
Stratified Cox proportional-hazards model
The hazard ratio for PFS was estimated using a stratified Cox proportional-hazards model. The registry specifies Efron's method for accounting for ties in event times, with strata information based on the IRT randomization data.
The hazard ratio is a relative time-to-event measure. It summarizes the estimated relationship between randomized treatment assignment and the instantaneous event rate under the fitted Cox model. It is not an absolute difference in the percentage of participants who experience progression or death.
Kaplan-Meier estimation
The registry identifies Kaplan-Meier estimation as the method used to estimate median survival time for overall survival and duration of response. The primary PFS endpoint also has a separately registered Kaplan-Meier-estimate endpoint. Kaplan-Meier methods are appropriate for time-to-event outcomes because participants can be followed for different lengths of time and can be censored before an event occurs.
Here, di is the number of events at time ti, while ni is the number of participants at risk immediately before that time.
Cochran-Mantel-Haenszel test
The secondary binary endpoints ORR and DCR used the Cochran-Mantel-Haenszel (CMH) test. The registry states that the treatment comparison and P-value used the CMH test, with strata information based on IRT randomization data. The corresponding odds ratios were also calculated using the CMH approach.
Exact confidence intervals for binary outcomes
For ORR and DCR, the registry states that confidence intervals were calculated using the exact method for the binomial distribution. This is distinct from the CMH test used for the treatment comparison: the confidence interval for an individual response proportion is an interval for that proportion, while the CMH analysis supplies the stratified comparison between treatment groups.
7. Primary PFS Result
The registry provides a formal statistical analysis for the randomized-treatment-period PFS endpoint. The analysis population was the ITT population, and treatment groups were defined by randomized assignment.
Progression-free survival
95% CI: 0.39–0.87 · P = 0.0078
Stratified log-rank test; hazard ratio estimated with a stratified Cox proportional-hazards model.
| Element | Reported result |
|---|---|
| Endpoint | Randomized Treatment Period: Progression-free Survival (PFS) |
| Time frame | up to 1422 days |
| Analysis population | Intent-to-Treat Population: all randomized participants |
| Comparison | Pemigatinib 13.5 mg QD vs gemcitabine 1000 mg/m2 plus cisplatin 25 mg/m2 to pemigatinib 13.5 mg QD |
| Method | Stratified log-rank test |
| Effect measure | Hazard ratio |
| Estimate | 0.58 |
| 95% CI | 0.39–0.87 |
| P-value | 0.0078 |
| Hypothesis type | Superiority |
The estimated hazard ratio of 0.58 means that, under the stratified Cox model, the estimated instantaneous rate of progression or death was about 42% lower with pemigatinib than with the comparator treatment. That is a relative model-based comparison, not a statement that 42% of participants avoided progression or death.
The 95% confidence interval of 0.39–0.87 describes the statistical uncertainty around the estimated hazard ratio under the analysis framework. It does not describe the range of individual patient outcomes. The interval is entirely below 1, which is consistent with the reported superiority hypothesis test.
The P-value of 0.0078 measures the strength of evidence against the null hypothesis under the specified stratified testing framework. It does not measure the size of the treatment effect. Effect size is described by the hazard ratio and its confidence interval.
Because the effect estimate comes from a Cox proportional-hazards model, interpretation also depends on the model's proportional-hazards framework. A single hazard ratio should not automatically be interpreted as a constant reduction in event risk at every point in time.
The analysis was conducted according to randomized assignment in the ITT population, which preserves the treatment comparison established by randomization. The registry also states that the P-value used a stratified log-rank test and that the hazard ratio used a stratified Cox model, with strata derived from IRT randomization data.
Registered Kaplan-Meier PFS endpoint
The registry also lists Kaplan-Meier estimates of PFS of the indicated lengths of time as a primary endpoint for the randomized treatment period, with a time frame of up to 1422 days. Results are posted for this endpoint, but the ClinicalTrials.gov record does not provide the individual time-point estimates. The formal inferential analysis available in the ClinicalTrials.gov record is therefore the PFS hazard-ratio analysis reported above.
8. Secondary Efficacy Results
The registry contains four posted secondary statistical analyses: objective response rate, overall survival, duration of response, and disease control rate. Each uses the ITT population, although the specific endpoint definitions and methods differ.
Objective Response Rate
Odds ratio for objective response
95% CI: 2.54–12.24 · P < 0.0001
Cochran-Mantel-Haenszel test with strata based on IRT randomization data.
An odds ratio of 5.57 indicates that the estimated odds of objective response were 5.57 times as high with pemigatinib as with the comparator in the stratified analysis.
Odds are not the same as probabilities. An odds ratio of 5.57 therefore should not be read as saying that the response probability was 5.57 times higher. Converting an odds ratio into a risk or probability ratio requires the underlying response proportions.
The 95% CI of 2.54–12.24 indicates substantial uncertainty around the estimated odds ratio, while remaining above 1. The P-value <0.0001 indicates strong evidence against the null hypothesis under the specified CMH testing framework; it does not quantify how large the response difference is.
The registry specifies an exact binomial method for confidence intervals for the response measure and a CMH test for the treatment comparison. These are complementary pieces of the analysis rather than interchangeable calculations.
Overall Survival
Hazard ratio for overall survival
95% CI: 0.73–1.64 · P = 0.6581
Stratified log-rank test; hazard ratio estimated with a stratified Cox proportional-hazards model.
The reported overall-survival hazard ratio of 1.10 means that the estimated hazard of death was higher in the pemigatinib group relative to the comparator under the fitted model, but the estimate is close to 1 and has substantial uncertainty.
The 95% CI of 0.73–1.64 includes 1. Consequently, the ClinicalTrials.gov record does not establish a statistically detectable difference in overall survival under the reported superiority test.
The P-value of 0.6581 does not mean that the two treatments are proven equivalent or identical. It means that the observed result is not statistically persuasive against the null hypothesis under this particular testing framework. A nonsignificant result is not the same as evidence of equivalence or non-inferiority.
The registry states that the analysis used a stratified log-rank test for the P-value and a stratified Cox proportional-hazards model for the hazard ratio, with strata based on IRT randomization data. As with the PFS analysis, the hazard ratio is a model-based relative measure rather than an absolute survival difference.
Duration of Response
Hazard ratio for duration of response
95% CI: 0.16–1.01 · P = 0.0526
Stratified log-rank test; hazard ratio estimated with a stratified Cox proportional-hazards model.
The estimated hazard ratio of 0.40 corresponds to an estimated 60% lower instantaneous hazard of the duration-of-response event with pemigatinib under the fitted model. However, the confidence interval is relatively wide, extending from 0.16 to 1.01.
Because the confidence interval includes 1 and the reported P-value is 0.0526, the registry-reported analysis does not meet a conventional 0.05 threshold for statistical significance. More importantly, the point estimate and confidence interval should be considered together rather than reducing the result to the P-value alone.
Duration-of-response analysis is also intrinsically conditional on response: the registry states that only participants meeting the response-related criteria were included in this analysis. It therefore answers a different question from PFS in the full randomized population.
Disease Control Rate
Odds ratio for disease control
95% CI: 1.76–9.54 · P = 0.0007
Cochran-Mantel-Haenszel test with strata based on IRT randomization data.
An odds ratio of 4.10 means that the estimated odds of disease control were 4.10 times as high with pemigatinib as with the comparator in the stratified analysis.
As with ORR, this is an odds ratio rather than a risk ratio. Without the underlying disease-control proportions, it should not be translated directly into a percentage-point difference or a multiplicative change in probability.
The 95% CI of 1.76–9.54 indicates uncertainty around the treatment effect while remaining above 1. The P-value of 0.0007 provides the evidence against the null hypothesis under the reported CMH framework; it is not a measure of clinical magnitude.
| Secondary endpoint | Effect measure | Estimate | 95% CI | P-value | Method |
|---|---|---|---|---|---|
| Objective Response Rate | Odds ratio | 5.57 | 2.54–12.24 | <0.0001 | Cochran-Mantel-Haenszel |
| Overall Survival | Hazard ratio | 1.10 | 0.73–1.64 | 0.6581 | Stratified log-rank / stratified Cox model |
| Duration of Response | Hazard ratio | 0.40 | 0.16–1.01 | 0.0526 | Stratified log-rank / stratified Cox model |
| Disease Control Rate | Odds ratio | 4.10 | 1.76–9.54 | 0.0007 | Cochran-Mantel-Haenszel |
9. Why the Different Effect Measures Matter
FIGHT-302 illustrates an important statistical distinction between hazard ratios and odds ratios. They are not alternative ways of expressing the same quantity.
Hazard ratio
Used for PFS, overall survival, and duration of response. It compares instantaneous event rates within a time-to-event framework and accounts for censoring through survival-analysis methods.
Odds ratio
Used for ORR and DCR. It compares the odds of a binary outcome between randomized groups using the stratified CMH framework.
Kaplan-Meier
Estimates the distribution of a time-to-event endpoint over follow-up and can provide time-specific survival probabilities and median event times.
Confidence interval
Describes statistical uncertainty around an estimated treatment effect. Its interpretation depends on the effect measure and the statistical model used to obtain it.
The PFS HR and ORR OR cannot be compared numerically to decide which outcome is "larger." They quantify different aspects of treatment effect using different probability models.
10. Statistical Methods Explained
Why was a stratified log-rank test used for PFS?
The primary PFS endpoint is a time-to-event outcome. Some participants can experience progression or death during follow-up while others remain event-free at their last assessment and are therefore censored. The log-rank test is designed to compare survival distributions between randomized groups while accounting for these different follow-up times. FIGHT-302 used a stratified log-rank test, with strata information derived from the IRT randomization data.
What does a PFS hazard ratio of 0.58 mean?
A hazard ratio of 0.58 indicates an estimated instantaneous event rate that is 58% of the comparator rate under the fitted Cox model, corresponding to approximately a 42% lower estimated hazard. It does not mean that 42% of participants avoided progression, nor does it imply that each participant experienced the same reduction in risk.
Why is the confidence interval important?
The point estimate alone does not reveal how precisely the treatment effect was estimated. The PFS confidence interval of 0.39–0.87 shows the uncertainty around the reported HR of 0.58. Confidence intervals should be interpreted as statements about statistical uncertainty under the specified analysis framework, not as a prediction interval for individual patients.
Why was the Cochran-Mantel-Haenszel test used for ORR and DCR?
ORR and DCR are binary endpoints. The CMH framework provides a way to compare treatment groups while accounting for stratification. In FIGHT-302, the registry states that the strata were based on IRT randomization data. The same framework was used to calculate the reported odds ratios.
Why is an odds ratio not the same as a response-rate ratio?
Odds are defined as probability divided by one minus probability. The odds ratio therefore compares odds rather than probabilities. When outcomes are common, an odds ratio can differ substantially from a risk ratio. The reported ORR odds ratio of 5.57 should consequently not be described as a 5.57-fold increase in response probability.
Why does the overall-survival P-value not prove equivalence?
The overall-survival HR was 1.10, with a 95% CI of 0.73–1.64 and P = 0.6581. The nonsignificant result means the analysis did not provide sufficient evidence against the null hypothesis under the specified superiority framework. It does not establish that the treatments have identical survival effects, because an equivalence or non-inferiority conclusion requires a different prespecified framework and margin.
Why is duration of response analyzed differently from ORR?
ORR is binary: it asks whether a participant achieved the defined response outcome. Duration of response is a time-to-event endpoint among participants meeting the response criteria. Its analysis therefore uses survival-analysis methods such as Kaplan-Meier estimation, the log-rank test, and Cox modeling rather than a binary-outcome comparison alone.
11. Safety Results
The ClinicalTrials.gov record reports serious adverse events by treatment period and arm. These counts should be interpreted as the number affected among the corresponding number at risk, rather than as efficacy outcomes.
| Treatment period | Arm | Serious adverse events affected | At risk |
|---|---|---|---|
| Randomized Period | Pemigatinib 13.5 mg QD | 23 | 83 |
| Randomized Period | Gemcitabine 1000 mg/m2 | 17 | 73 |
| Transition Period | Pemigatinib 13.5 mg QD | 9 | 42 |
The registry does not supply a formal comparative effect estimate or P-value for these serious-adverse-event counts in the ClinicalTrials.gov record. The appropriate interpretation is therefore descriptive rather than an inferential comparison.
12. Study Status and Termination
Study start
The registry lists June 3, 2019 as the study start date.
Randomized treatment comparison
The study was a randomized, parallel-group, open-label phase 3 trial with two arms and a treatment primary purpose.
Primary completion
The registry lists July 7, 2025 as the primary completion date.
Enrollment-related termination
The registry states that the study was terminated because enrollment was lacking after a change in the standard of care for first-line treatment of patients with cholangiocarcinoma. It explicitly states that there were no safety concerns contributing to the decision.
13. What the PFS Result Does — and Does Not — Mean
The PFS hazard ratio of 0.58 represents a relative comparison of the instantaneous rate of progression or death under the stratified Cox model. Expressed descriptively, 0.58 corresponds to an estimated 42% lower hazard relative to the comparator.
It does not mean that 42% of participants were protected from progression, that every participant had exactly 42% less risk, or that median PFS was reduced or increased by 42%. Those would require different quantities.
The 95% CI of 0.39–0.87 communicates uncertainty around the HR estimate. Its width reflects the precision of the available information under the statistical model. It should not be interpreted as the range of individual treatment effects.
The P-value of 0.0078 is evidence against the null hypothesis in the reported stratified log-rank analysis. It is not a measure of effect magnitude, clinical importance, or probability that the treatment hypothesis is true.
14. Comparing the Primary and Secondary Findings
| Outcome | Effect | 95% CI | P-value | Statistical interpretation |
|---|---|---|---|---|
| PFS | HR 0.58 | 0.39–0.87 | 0.0078 | Reported superiority analysis favors pemigatinib under the specified time-to-event framework. |
| ORR | OR 5.57 | 2.54–12.24 | <0.0001 | Reported stratified binary analysis shows higher estimated odds of response with pemigatinib. |
| Overall survival | HR 1.10 | 0.73–1.64 | 0.6581 | Does not establish a statistically detectable difference under the reported superiority framework. |
| DOR | HR 0.40 | 0.16–1.01 | 0.0526 | Point estimate favors pemigatinib, but the CI includes 1 and the P-value is above 0.05. |
| DCR | OR 4.10 | 1.76–9.54 | 0.0007 | Reported stratified binary analysis shows higher estimated odds of disease control with pemigatinib. |
The results answer different clinical questions. PFS measures time until progression or death, ORR measures a binary response outcome, DCR measures a binary disease-control outcome, DOR measures the duration of response among qualifying responders, and OS measures time to death. A treatment effect should therefore be interpreted in the context of the endpoint rather than by comparing the numerical size of 0.58, 5.57, 1.10, 0.40, and 4.10 as though they were the same statistic.
15. Stratification and the IRT Randomization Data
The registry analyses repeatedly state that strata information was based on the Interactive Response Technology (IRT) randomization data. This is important because the inferential analyses were not simply unstratified comparisons.
Stratified PFS analysis
The PFS P-value came from a stratified log-rank test, while the HR came from a stratified Cox proportional-hazards model.
Stratified binary analysis
ORR and DCR used the CMH test and CMH-derived odds ratios with the same IRT-based strata information.
Why preserve the strata?
Stratification allows the treatment comparison to account for the factors represented in the randomization strata rather than ignoring the trial's randomized structure.
Interpretation
The reported estimates are therefore tied to the specified stratified analysis rather than being generic unadjusted treatment contrasts.
16. Transition-Period Endpoints
FIGHT-302 includes a second set of registered primary endpoints for a transition period. These endpoints use a different time origin from the randomized treatment period.
| Feature | Randomized Treatment Period | Transition Period |
|---|---|---|
| Time origin | Date of randomization | First dose of study drug after transition |
| Event definition | Disease progression according to RECIST v1.1 and assessed by ICR, or death, whichever occurs first | Earliest date of progressive disease or death due to any cause |
| Registered time frame | up to 1422 days | up to 1496 days |
| Registered KM endpoint | Yes | Yes |
The transition-period definition is not simply a continuation of the randomized-period clock. Because its time origin is the first dose after transition, transition-period PFS should be analyzed and interpreted separately from randomized-period PFS.
17. Limitations
- Study termination: the registry states that the study was terminated because enrollment was lacking after a change in the standard of care for first-line cholangiocarcinoma. This limits the amount of information available from the randomized comparison.
- Open-label design: masking was none. Knowledge of treatment assignment can potentially affect treatment management and some assessments, although the primary PFS definition incorporated Independent Central Review.
- Limited registry-reported sample information: the ClinicalTrials.gov record reports enrollment of 167 but do not provide a complete baseline-characteristics table or a full accounting of randomized participants by arm for all analyses.
- No median PFS reported: although PFS is a time-to-event endpoint and Kaplan-Meier methods are part of the registered endpoint structure, the registry-reported formal statistical analysis reports the HR, CI, and P-value rather than a median PFS estimate.
- Hazard-ratio interpretation: Cox proportional-hazards estimates rely on the model framework. A single HR should not automatically be interpreted as a constant relative difference throughout follow-up.
- Binary-effect interpretation: odds ratios for ORR and DCR are not probability ratios. Without the underlying response proportions, they cannot be translated directly into absolute response differences.
- Overall survival: the reported OS analysis does not establish a statistically detectable difference under the stated superiority framework, and the confidence interval remains compatible with effects in either direction.
- Duration of response: DOR is conditional on response and therefore represents a different analysis population and clinical question from the randomized PFS analysis.
- Safety denominators: serious-adverse-event counts are reported for different treatment periods and denominators, so they should not be pooled or treated as directly interchangeable groups.
- Unreported design details: the ClinicalTrials.gov record does not support conclusions about a non-inferiority margin, interim analysis, multiplicity strategy, missing-data imputation, factorial structure, or Bayesian methods, so these features are not assumed here.
18. Why This Trial Matters Statistically
FIGHT-302 is a useful teaching example because several core clinical-trial methods appear within one randomized phase 3 study, while the registry also illustrates why endpoint definitions and analysis populations matter.
| Concept | How it appears in FIGHT-302 |
|---|---|
| Randomization | The study was randomized with two parallel treatment arms. |
| Intention-to-treat analysis | The primary PFS analysis used all randomized participants and preserved treatment assignment from randomization. |
| Time-to-event analysis | PFS, overall survival, and duration of response were analyzed as time-to-event outcomes. |
| Kaplan-Meier estimation | Kaplan-Meier methods were specified for time-to-event estimation, including overall survival and duration of response. |
| Log-rank testing | The primary PFS P-value used a stratified log-rank test. |
| Cox modeling | The PFS, OS, and DOR hazard ratios were estimated using stratified Cox proportional-hazards models with Efron's method for ties. |
| Stratified analysis | IRT randomization data registry-reported the strata information used in the reported analyses. |
| Cochran-Mantel-Haenszel test | ORR and DCR used the CMH test for stratified binary treatment comparisons. |
| Odds ratio | ORR and DCR were reported using CMH-derived odds ratios. |
| Confidence intervals | Time-to-event and binary endpoints were reported with 95% confidence intervals using their respective methods. |
| Endpoint time origin | Randomized-period PFS begins at randomization, while transition-period PFS begins at the first dose after transition. |
| Trial termination | Enrollment-related termination provides an important context for interpreting the amount of randomized evidence available. |
19. Clinical Interpretation vs Statistical Interpretation
Statistical interpretation
The primary randomized-period PFS analysis reported HR 0.58 with a 95% CI of 0.39–0.87 and P = 0.0078 using a stratified log-rank test and stratified Cox model. Secondary endpoints produced different effect estimates and levels of statistical evidence.
Endpoint interpretation
The PFS result concerns progression or death, whereas ORR and DCR concern binary tumor-control outcomes, DOR concerns duration among responders, and OS concerns death from any cause. These endpoints should not be collapsed into a single statistic.
The most important statistical lesson is that a clinical-trial result is defined by more than its P-value. The estimand, endpoint definition, analysis population, effect measure, confidence interval, censoring framework, stratification, and study design all contribute to the meaning of the result.
20. Sources
- ClinicalTrials.gov: FIGHT-302, NCT03656536. Official registry record used for the trial information and statistical results presented on this page.
- Linked publication: PubMed record for PMID 42223137.
- Linked publication: PubMed record for PMID 32677452.
Continue through the Clinical Biostats statistical pathway
Use the methods illustrated by FIGHT-302 to explore survival analysis, stratified treatment comparisons, binary endpoints, confidence intervals, and clinical-trial interpretation.
21. Record Summary
FIGHT-302 provides a compact example of randomized clinical-trial methodology spanning time-to-event analysis, stratified log-rank testing, Cox proportional-hazards modeling, Kaplan-Meier estimation, and Cochran-Mantel-Haenszel analysis. The primary randomized-period PFS analysis reported a hazard ratio of 0.58 with a 95% CI of 0.39–0.87 and P = 0.0078. Secondary analyses reported an ORR odds ratio of 5.57, an OS hazard ratio of 1.10, a DOR hazard ratio of 0.40, and a DCR odds ratio of 4.10, each with its corresponding confidence interval and P-value.
The statistical story is therefore best understood endpoint by endpoint. PFS, ORR, DCR, DOR, and OS use different estimands and answer different questions. The interpretation also requires attention to the ITT population, IRT-based stratification, open-label design, different time origins for randomized and transition periods, safety denominators, and the registry's stated enrollment-related termination.