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 contained in the ClinicalTrials.gov record.
1. Trial at a Glance
TROPIC was a randomized, parallel-group phase 3 trial comparing cabazitaxel (XRP6258) plus prednisone with mitoxantrone plus prednisone in hormone-refractory metastatic prostate cancer. The registry identifies overall survival as the single primary endpoint and reports eight statistical analyses across primary and secondary outcomes.
| Feature | TROPIC |
|---|---|
| Trial name | TROPIC |
| Phase | Phase 3 |
| Condition | Neoplasms; Prostatic Neoplasms |
| Brief title | XRP6258 Plus Prednisone Compared to Mitoxantrone Plus Prednisone in Hormone Refractory Metastatic Prostate Cancer |
| Design | Randomized, parallel-group |
| Masking | None |
| Primary purpose | Treatment |
| Enrollment | 755 |
| Arms | 2 |
| Lead sponsor | Sanofi |
| Sponsor type | Industry |
| Status | Completed |
| ClinicalTrials.gov | NCT00417079 |
2. Clinical Question
The central statistical question was whether treatment with cabazitaxel plus prednisone produced a different time-to-death profile from mitoxantrone plus prednisone in the randomized trial population, with overall survival designated as the primary endpoint.
Population
Patients enrolled in the TROPIC trial with the registered condition of hormone-refractory metastatic prostate cancer.
Intervention
Cabazitaxel (XRP6258; RPR116258) plus prednisone.
Comparator
Mitoxantrone plus prednisone.
Primary question
Does cabazitaxel plus prednisone improve overall survival relative to mitoxantrone plus prednisone?
3. Trial Design
Cabazitaxel + Prednisone
- Cabazitaxel (XRP6258; RPR116258)
- Prednisone
Mitoxantrone + Prednisone
- Mitoxantrone
- Prednisone
The registry describes the study as randomized and parallel, with no masking. The analysis population for the efficacy analyses was the intention-to-treat population, meaning patients were analyzed according to the treatment group assigned by randomization.
4. Trial Timeline
Trial start
The registry records the beginning of the TROPIC study in January 2007.
Primary completion
The registry records September 2009 as the primary completion date.
Registered analysis window
The primary overall-survival endpoint was defined from randomization through 104 weeks, using the study cutoff.
5. Primary Endpoint
| Endpoint | Registered definition | Time frame | Analysis |
|---|---|---|---|
| Overall Survival | Overall survival was defined as the time interval from the date of randomization to the date of death due to any cause. In the absence of confirmation of death, the survival time was censored at the last date the patient was known to be alive or at the cut-off date, whichever had come first. | From the date of randomization up to 104 weeks (study cut-off) | Stratified log-rank analysis in the ITT population |
Overall survival is a classic time-to-event endpoint. Unlike a simple binary endpoint, it uses both the occurrence of death and the amount of observed follow-up. Patients who have not experienced death by the relevant censoring time contribute partial follow-up information rather than being treated as if they had experienced the event.
6. Secondary Endpoints
| Secondary endpoint | Time frame | Type | Analysis |
|---|---|---|---|
| Time to Tumor Progression | From randomization up to 104 weeks (study cut-off) | Time-to-event | Stratified log-rank; Cox proportional-hazard effect measure |
| Time to Prostatic Specific Antigen (PSA) Progression | At screening, day 1 of every treatment cycle, up to 104 weeks (study cut-off) | Time-to-event | Stratified log-rank; Cox proportional-hazard effect measure |
| Time to Pain Progression | From baseline up to 104 weeks (study cut-off) | Time-to-event | Stratified log-rank; Cox proportional-hazard effect measure |
| Time to Progression Free Survival (PFS) | From randomization up to 104 weeks (study cut-off) | Time-to-event | Stratified log-rank |
| Overall Tumor Response | From randomization up to 104 weeks (study cut-off) | Binary | Chi-squared test |
| PSA Response | From baseline up to 104 weeks (study cut-off) | Binary | Chi-squared test |
| Pain Response | From baseline up to 104 weeks (study cut-off) | Binary | Chi-squared test |
7. Statistical Methodology
Intention-to-treat analysis
The registry states that the overall-survival analysis was performed in the intention-to-treat (ITT) population. The registry text defines the ITT population as all randomized patients, meaning patients assigned to a treatment group by randomization regardless of subsequent treatment circumstances.
The ITT principle preserves the comparison created by randomization. If patients discontinue treatment, change treatment, or otherwise deviate from the intended treatment course, analyzing them according to their randomized assignment prevents those post-randomization events from redefining the original comparison.
Stratified log-rank testing
The registry states that log-rank comparisons were stratified according to disease measurability and ECOG performance status. Stratification allows the time-to-event comparison to account for these prespecified factors rather than treating every patient as though the distribution of these characteristics were identical across the comparison.
The log-rank test compares the observed pattern of events between groups over follow-up while accounting for the timing of events and censored observations.
Hazard ratios
For secondary time-to-event endpoints, the registry reports a Cox proportional-hazard effect measure, normalized as a hazard ratio. The analysis notes state that an HR less than 1 favors the cabazitaxel group and an HR greater than 1 favors the mitoxantrone group.
A hazard ratio is a relative time-to-event measure. It is not an absolute probability, a median survival time, or the percentage of patients who benefit.
Chi-squared testing
The registry uses chi-squared tests for the binary secondary endpoints of overall tumor response, PSA response, and pain response. For a binary outcome, this approach compares the observed response proportions between treatment groups under the specified analysis population.
Analysis populations can differ by endpoint
Although the registry identifies the ITT population for the time-to-event analyses, some secondary binary endpoints have more specific analytic populations. Tumor response was evaluated only among ITT patients with measurable disease by RECIST. PSA response was evaluated only among ITT patients with a baseline PSA greater than 20 ng/mL. Pain response applied only to ITT patients meeting the stated baseline pain criteria.
8. Statistical Analysis Plan Features Reported by the Registry
| Design feature | Registry-supported information |
|---|---|
| Hypothesis type | Superiority |
| Primary endpoint type | Time-to-event |
| Primary effect measure | Hazard ratio |
| Primary method | Log-rank test |
| Stratification | Disease measurability and ECOG performance status |
| Primary analysis population | Intention-to-treat |
| Planned sample size | 720 patients, 360 per arm |
| Power | 90% |
| Target treatment effect | 25% reduction in the hazard ratio for death in the cabazitaxel group relative to the mitoxantrone group |
| Final analysis trigger | 511 deaths |
The planned sample size was 720 patients, with 360 patients per arm, to detect a 25% reduction in the hazard ratio for death in the cabazitaxel group relative to the mitoxantrone group with 90% power. The final analysis was planned when 511 deaths had occurred.
This illustrates an important principle in survival-trial design: sample size is closely tied to the number of observed events. For time-to-event endpoints, the number and timing of events can provide more direct information about statistical power than the number randomized alone.
9. Primary Result: Overall Survival
The registry reports a formal statistical analysis for the primary endpoint of overall survival. The analysis used the intention-to-treat population and a stratified log-rank test comparing mitoxantrone plus prednisone with cabazitaxel plus prednisone.
Primary hypothesis test
95% confidence level reported for the analysis · Superiority hypothesis
Analysis: stratified log-rank test in the ITT population
| Primary endpoint | Comparison | Method | 95% CI | P-value |
|---|---|---|---|---|
| Overall Survival | Mitoxantrone + Prednisone vs Cabazitaxel + Prednisone | Log-rank | 95% | <0.0001 |
The reported P < 0.0001 indicates that the prespecified log-rank comparison produced a result highly inconsistent with the null hypothesis under the statistical framework used for the primary analysis.
The P-value does not quantify the size of the treatment effect. A very small P-value does not mean that the treatment effect is large, nor does it provide an estimate of how long an individual patient will live.
The registry identifies a hazard ratio as the primary effect measure, but the ClinicalTrials.gov record does not provide the numerical hazard-ratio estimate or its lower and upper confidence limits for overall survival. Those values therefore are not reproduced here.
The absence of the numerical HR in the ClinicalTrials.gov record is important: the P-value establishes the reported hypothesis-test result, but it cannot substitute for the effect estimate or its confidence interval when describing magnitude and precision.
10. Secondary Time-to-Event Results
Time to Tumor Progression
Hazard ratio
95% CI: 0.49–0.76 · P < 0.0001
Analysis: stratified log-rank comparison; Cox proportional-hazard effect measure
An HR of 0.61 means that the estimated instantaneous rate of tumor progression was approximately 61% of the corresponding rate in the comparator group under the fitted time-to-event framework. Equivalently, 0.61 corresponds to a 39% lower estimated hazard, because 1 − 0.61 = 0.39.
The 95% CI of 0.49–0.76 describes statistical uncertainty around the estimated hazard ratio. It does not describe the range of outcomes that individual patients would experience.
The P-value of <0.0001 addresses the evidence against the null hypothesis in the specified analysis; it is not a measure of the magnitude of the hazard ratio.
Because this is a Cox-type hazard-ratio interpretation, proportional-hazards assumptions are relevant. A single HR summarizes the relative event rate over the modeled follow-up and does not imply that the hazard ratio must be identical at every individual time point.
Time to PSA Progression
Hazard ratio
95% CI: 0.63–0.90 · P = 0.0010
Analysis: stratified log-rank comparison; Cox proportional-hazard effect measure
An HR of 0.75 corresponds to an estimated 25% lower instantaneous rate of PSA progression in the cabazitaxel group relative to the mitoxantrone group under the model, because 1 − 0.75 = 0.25.
The 95% CI of 0.63–0.90 provides the uncertainty interval for the hazard-ratio estimate. Because the interval is entirely below 1, the reported analysis is consistent with a lower estimated hazard of PSA progression for cabazitaxel under the specified comparison.
The P-value of 0.0010 is evidence against the null hypothesis for this analysis, but it does not measure clinical magnitude and should not be interpreted as a probability that the null hypothesis is true.
Time to Pain Progression
Hazard ratio
95% CI: 0.69–1.19 · P = 0.5192
Analysis: stratified log-rank comparison; Cox proportional-hazard effect measure
An HR of 0.91 corresponds to an estimated 9% lower instantaneous rate of pain progression in the cabazitaxel group relative to the comparator under the fitted model.
However, the 95% CI of 0.69–1.19 includes 1. This means the confidence interval encompasses values corresponding to a lower hazard, little difference, and a higher hazard for the cabazitaxel group.
The P-value of 0.5192 does not provide evidence against the null hypothesis at conventional significance levels. Importantly, a nonsignificant P-value does not prove that the treatments are identical; the confidence interval is needed to understand what effect sizes remain compatible with the data.
Time to Progression Free Survival (PFS)
Primary reported statistic
Analysis: stratified log-rank comparison in the ITT population
Hypothesis type: superiority
The registry reports a P-value of <0.0001 for the PFS comparison using a stratified log-rank test. This supports a statistically detectable difference in the time-to-event distributions under the specified analysis.
The ClinicalTrials.gov record does not provide a numerical PFS hazard ratio or confidence interval. Therefore, the magnitude of the PFS treatment effect cannot be quantified from the ClinicalTrials.gov record without importing information from another source, which is outside the data rules for this page.
11. Secondary Binary Endpoint Results
The registry also reports three binary efficacy outcomes. These were analyzed with chi-squared tests rather than time-to-event methods.
| Endpoint | Analysis population | Method | 95% CI | P-value |
|---|---|---|---|---|
| Overall Tumor Response | ITT patients with measurable disease by RECIST | Chi-squared | 95% | 0.0005 |
| PSA Response | ITT patients with baseline PSA >20 ng/mL | Chi-squared | 95% | 0.0002 |
| Pain Response | ITT patients with median PPI ≥2 on McGill-Melzack scale and/or mean Analgesic Score ≥10 points at baseline | Chi-squared | 95% | 0.6286 |
Overall Tumor Response
Chi-squared test
Overall tumor response was evaluated among ITT patients with measurable disease by Response Evaluation Criteria in Solid Tumor (RECIST).
The reported P-value indicates a statistically detectable difference in the distribution of the binary tumor-response outcome between the randomized groups under the specified chi-squared analysis. The ClinicalTrials.gov record does not provide the response percentages or the corresponding effect estimate, so neither is reconstructed here.
PSA Response
Chi-squared test
PSA response was evaluated among ITT patients with a baseline PSA >20 ng/mL.
The analysis reports a statistically detectable difference between treatment groups for the binary PSA-response endpoint. The analysis population restriction is important because the result applies to patients meeting the baseline PSA criterion rather than automatically to every randomized patient.
Pain Response
Chi-squared test
Pain response applied to ITT patients with median PPI ≥2 on the McGill-Melzack scale and/or mean Analgesic Score ≥10 points at baseline.
The reported P-value does not provide evidence against the null hypothesis in this specified binary comparison. It does not establish that the two treatment groups have identical pain-response probabilities, because the magnitude and confidence interval of the response difference are not reported in the registry analysis record.
12. Statistical Methods Explained
Why was a log-rank test used for overall survival?
Overall survival records the time from randomization to death and can include patients who have not experienced death by the study cutoff. The log-rank test is designed for this type of censored time-to-event data and compares the survival experience of randomized groups across follow-up rather than reducing every patient to a simple yes/no outcome.
What does an HR of 0.61 mean for time to tumor progression?
An HR of 0.61 means that the fitted model estimates the instantaneous rate of tumor progression in the cabazitaxel group to be approximately 61% of the comparator rate. The corresponding relative reduction in estimated hazard is 39%. It does not mean that 39% of patients avoided progression, nor does it say that each patient experienced exactly the same reduction.
Why is the confidence interval important?
A point estimate such as HR 0.61 is only one estimate from the observed data. The 95% CI of 0.49–0.76 communicates the statistical uncertainty around that estimate. An interval is particularly useful because it shows both the direction and the range of effect sizes that remain compatible with the analysis under its assumptions.
Why does a P-value not measure treatment effect size?
A P-value addresses compatibility with a null hypothesis under a specified statistical model and sampling framework. It depends on both the observed difference and the amount of information available. It therefore cannot be used as a substitute for an effect estimate such as a hazard ratio, risk difference, or response-rate difference.
Why does ITT matter in this trial?
The primary overall-survival analysis was performed using the ITT population. An ITT comparison keeps patients in the group to which randomization assigned them. This preserves the treatment comparison generated by randomization and avoids redefining the groups according to events that occur after assignment.
Why are the response endpoints analyzed differently from survival endpoints?
Overall tumor response, PSA response, and pain response are binary outcomes in the registry analysis. They are therefore analyzed using chi-squared tests. Overall survival, tumor progression, PSA progression, pain progression, and PFS are time-to-event outcomes, for which the registry uses log-rank methods and, for several endpoints, Cox proportional-hazard effect measures.
Why should the pain-progression result not be interpreted simply as "no effect"?
The HR of 0.91 has a 95% CI of 0.69–1.19 and a P-value of 0.5192. These data do not establish a statistically detectable difference in the specified analysis. They also do not prove equivalence or establish that any possible treatment difference is clinically unimportant. The confidence interval describes the range of effects that remains compatible with the analysis.
13. Interpreting the Primary P-value Correctly
The overall-survival analysis produced P < 0.0001 from the prespecified stratified log-rank comparison. Under the reported statistical framework, this is strong evidence against the null hypothesis of no difference in the overall-survival experience between the randomized groups.
The P-value does not state the size of the overall-survival effect. It does not provide a median survival time, an absolute survival difference, or an individual patient's probability of benefit. Those quantities are separate estimands and require corresponding reported estimates.
The registry identifies the hazard ratio as the effect measure for the primary analysis, but the registry-reported primary statistical-analysis record does not contain the numerical HR or its confidence limits. A complete quantitative description of treatment magnitude therefore cannot be reconstructed solely from the posted P-value.
14. Understanding Censoring in TROPIC
The registered definition of overall survival specifies that patients without confirmed death were censored at the last date they were known to be alive or at the study cutoff, whichever occurred first. This is fundamental to the statistical analysis of survival data.
A censored patient is not counted as having survived indefinitely. Instead, the analysis uses the follow-up period during which the patient's event status was known.
This distinction explains why simply calculating the proportion alive at a single date is not equivalent to performing a Kaplan-Meier or log-rank analysis. Time-to-event methods incorporate the timing of events and censoring throughout follow-up.
15. Stratification and Randomization
The registry analysis notes that log-rank comparisons were stratified according to disease measurability and ECOG performance status. These factors are relevant to the statistical comparison because they can be associated with prognosis and may affect the distribution of time-to-event outcomes.
Disease measurability
The stratification factor distinguishes patients according to the registry-defined disease-measurability categories used for the analysis.
ECOG performance status
The analysis accounts for ECOG performance status through stratified time-to-event comparisons.
Stratification does not turn a randomized trial into an observational comparison. Rather, it incorporates prespecified strata into the statistical comparison while maintaining the randomized treatment assignment as the basis of inference.
16. Primary Analysis and Statistical Power
The registry-reported analysis notes provide unusually useful design information. The study required an estimated sample size of 720 patients, 360 per arm to detect a 25% reduction in the hazard ratio for death in the cabazitaxel group relative to the mitoxantrone group with 90% power.
| Planning quantity | Registered value |
|---|---|
| Estimated sample size | 720 patients |
| Planned patients per arm | 360 |
| Target hazard reduction | 25% |
| Planned power | 90% |
| Final-analysis event target | 511 deaths |
| Actual enrollment | 755 patients |
The relationship between these quantities illustrates why event-driven planning is central to survival studies. The analysis is designed around the amount of information generated by deaths rather than merely the number of people entering the trial.
17. Secondary Endpoint Population Definitions
| Endpoint | Population restriction | Statistical implication |
|---|---|---|
| Overall Tumor Response | ITT patients with measurable disease by RECIST | The comparison concerns the subset meeting the measurable-disease criterion. |
| PSA Response | ITT patients with baseline PSA >20 ng/mL | The response analysis is conditional on the baseline PSA criterion. |
| Pain Response | ITT patients with median PPI ≥2 on McGill-Melzack scale and/or mean Analgesic Score ≥10 points at baseline | The analysis applies to patients meeting the prespecified baseline pain criteria. |
These distinctions are statistically important because the denominator for an endpoint determines the population to which its result applies. A P-value from a restricted endpoint population should not automatically be described as though it came from all randomized patients.
18. Safety Results
The ClinicalTrials.gov record reports serious adverse events by randomized arm using affected patients over the number at risk.
| Safety measure | Affected / at risk | Observed proportion from registry-reported count |
|---|---|---|
| Mitoxantrone + Prednisone | 77 / 371 | 77 / 371 |
| Cabazitaxel + Prednisone | 145 / 371 | 145 / 371 |
The registry data identify 77 of 371 patients in the mitoxantrone plus prednisone group and 145 of 371 patients in the cabazitaxel plus prednisone group as affected by serious adverse events.
Because the ClinicalTrials.gov record provides the affected and at-risk counts rather than a formal comparative statistical analysis, the safety figures should be read descriptively. No inferential P-value or confidence interval for the between-arm serious-adverse-event comparison is reported here.
The bars above are scaled to the two registry-reported affected counts for visual comparison; they are not percentages. The underlying reported values remain 77/371 and 145/371.
19. Multiplicity and Multiple Endpoints
TROPIC has one registered primary endpoint and seven secondary statistical analyses in the ClinicalTrials.gov record. This distinction matters when interpreting the collection of reported P-values.
| Endpoint class | Number in the ClinicalTrials.gov record | Interpretive role |
|---|---|---|
| Primary | 1 | Overall survival |
| Secondary | 7 | Additional efficacy outcomes |
| Total statistical analyses posted | 8 | Primary plus secondary analyses |
Because multiple secondary endpoints are analyzed, their P-values should not automatically be interpreted as though each were the sole confirmatory test in the study. Whether a particular multiplicity adjustment was prespecified would require the relevant statistical-analysis plan or protocol information. The ClinicalTrials.gov record does not provide such an adjustment scheme.
20. What the Secondary Hazard Ratios Mean
| Endpoint | HR | 95% CI | P-value | Statistical reading |
|---|---|---|---|---|
| Time to Tumor Progression | 0.61 | 0.49–0.76 | <0.0001 | Estimated hazard lower in cabazitaxel group; CI excludes 1 |
| Time to PSA Progression | 0.75 | 0.63–0.90 | 0.0010 | Estimated hazard lower in cabazitaxel group; CI excludes 1 |
| Time to Pain Progression | 0.91 | 0.69–1.19 | 0.5192 | CI includes 1; no statistically detectable difference in the reported test |
The pattern demonstrates why effect estimates and uncertainty intervals are more informative than P-values alone. The three secondary time-to-event analyses have different point estimates and different degrees of statistical precision.
The tumor-progression estimate of 0.61 is farther from the null value of 1 than the PSA-progression estimate of 0.75, while the pain-progression estimate of 0.91 is close to 1 and has an interval extending on both sides of 1. These are descriptive statistical differences among the reported estimates, not evidence that the endpoints should be considered interchangeable.
21. Time-to-Event Endpoints Versus Binary Endpoints
TROPIC provides a useful contrast between two major classes of clinical-trial outcomes.
Time-to-event outcomes
Overall survival, time to tumor progression, time to PSA progression, time to pain progression, and PFS incorporate both whether an event occurred and when it occurred.
Binary outcomes
Overall tumor response, PSA response, and pain response classify participants according to whether the prespecified response occurred during the defined analysis period.
The choice of statistical method follows the endpoint structure. Log-rank testing is used for the time-to-event outcomes, while chi-squared testing is used for the binary response outcomes in the registry.
22. Why the Primary Endpoint Is More Than a P-value
The primary overall-survival result is statistically compelling in the ClinicalTrials.gov record because its P-value is <0.0001. But a statistically complete interpretation requires several distinct pieces of information:
- Estimand: what outcome is being compared — here, time from randomization to death from any cause.
- Analysis population: the ITT population.
- Comparison: mitoxantrone plus prednisone versus cabazitaxel plus prednisone.
- Statistical test: stratified log-rank test.
- Effect measure: hazard ratio, identified by the registry but not numerically reported in the primary-analysis record provided here.
- Uncertainty: a 95% confidence interval would normally accompany the HR, but the numerical limits are not present in the registry-reported primary analysis.
- Evidence against the null: P < 0.0001.
This separation is important. The P-value answers one inferential question; the hazard ratio addresses magnitude; the confidence interval addresses precision; and the endpoint definition determines what clinical event the analysis actually represents.
23. Limitations
- Incomplete numerical reporting in the registry-reported primary analysis: the registry identifies hazard ratio as the primary effect measure but the ClinicalTrials.gov record does not contain the numerical overall-survival HR or its confidence interval.
- No median survival values in the ClinicalTrials.gov record: median overall survival and other median time-to-event estimates are not included in the provided record and therefore are not reported here.
- Secondary endpoint multiplicity: seven secondary statistical analyses are reported, but the ClinicalTrials.gov record does not specify a complete multiplicity-adjustment procedure.
- Endpoint-specific populations: several binary response outcomes apply only to ITT subsets meeting prespecified baseline criteria.
- Censoring: survival analyses depend on assumptions about the censoring process and the quality of follow-up information.
- Proportional-hazards interpretation: hazard ratios from Cox-type analyses require care if relative hazards vary materially over time.
- Safety comparison: the registry-reported serious-adverse-event information is descriptive; a formal comparative test or confidence interval is not provided.
- Registry-level detail: the ClinicalTrials.gov record does not provide the complete protocol, statistical analysis plan, detailed baseline table, event counts for every endpoint, or individual-patient data.
24. Why This Trial Matters Statistically
TROPIC is a useful teaching example because it combines randomized treatment allocation, an event-driven survival endpoint, stratified log-rank testing, hazard-ratio interpretation, ITT analysis, binary response outcomes, and endpoint-specific analysis populations.
| Statistical concept | How it appears in TROPIC |
|---|---|
| Randomization | 755 patients assigned to two parallel treatment arms |
| Intention-to-treat analysis | Primary overall-survival analysis performed in the ITT population |
| Time-to-event analysis | Overall survival and several secondary endpoints |
| Log-rank test | Used for the primary and several secondary time-to-event comparisons |
| Stratified analysis | Time-to-event comparisons stratified by disease measurability and ECOG performance status |
| Hazard ratio | Effect measure for overall survival and several secondary time-to-event outcomes |
| Confidence interval | 95% intervals reported for the secondary hazard ratios |
| Chi-squared test | Used for tumor, PSA, and pain response endpoints |
| Event-driven design | Final analysis planned after 511 deaths |
| Power calculation | 90% power based on the planned sample size and target hazard reduction |
25. Statistical Interpretation in Context
Evidence for overall survival
The primary analysis reports P < 0.0001 from a stratified log-rank comparison in the ITT population.
Magnitude information
The registry identifies the hazard ratio as the primary effect measure, but the registry-reported primary analysis does not provide its numerical estimate or confidence limits.
Secondary progression endpoints
Time to tumor progression and time to PSA progression have HRs below 1 with 95% confidence intervals entirely below 1.
Pain progression
The HR is 0.91 with a 95% CI of 0.69–1.19 and P = 0.5192, illustrating why a point estimate should always be interpreted together with its uncertainty.
The overall statistical story therefore cannot be reduced to a single number. The primary endpoint establishes the formal trial question, the secondary endpoints show how different disease-related outcomes were analyzed, and the endpoint-specific populations determine precisely to whom each result applies.
26. Sources
- ClinicalTrials.gov: TROPIC — NCT00417079. Official registry record and source for the trial data used on this page.
- Linked publication: PubMed record — PMID 25538172.
- Linked publication: PubMed record — PMID 23723295.
- Linked publication: PubMed record — PMID 22743148.
- Linked publication: PubMed record — PMID 20888992.
The linked publications are included as references associated with the registry record. The numerical trial content and interpretations on this page remain restricted to the ClinicalTrials.gov record.
27. Related Tutorials
Learn more about the methods used in this trial:
28. Related Calculators
29. Record Summary
TROPIC is a randomized phase 3 comparison of cabazitaxel plus prednisone versus mitoxantrone plus prednisone with 755 enrolled patients and a primary endpoint of overall survival. The primary analysis used the ITT population and a stratified log-rank test, with disease measurability and ECOG performance status as stratification factors. The registry reports P < 0.0001 for the primary overall-survival comparison.
The secondary analyses demonstrate the complementary roles of survival and categorical methods. Time to tumor progression, time to PSA progression, time to pain progression, and PFS were analyzed as time-to-event outcomes, while overall tumor response, PSA response, and pain response were analyzed using chi-squared tests. The registry provides numerical hazard ratios and 95% confidence intervals for three secondary time-to-event endpoints, while the numerical primary overall-survival HR is not present in the ClinicalTrials.gov record.
The most important statistical lesson is that a clinical-trial result is a combination of endpoint definition, randomized analysis population, statistical test, effect measure, uncertainty, and study design. TROPIC illustrates why the P-value should be read together with the hazard ratio and confidence interval when those quantities are available, and why the absence of an effect estimate should be acknowledged rather than filled in from an outside source.
Continue with the statistical methods
Explore the survival-analysis, hypothesis-testing, and clinical-trial methods that underpin the TROPIC analysis.