This page separates reported trial results from statistical interpretation. 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
TROPiCS-02 was a randomized, parallel-group, open-label phase 3 trial evaluating sacituzumab govitecan-hziy versus treatment of physician's choice (TPC) in participants with HR+/HER2- metastatic breast cancer.
| Feature | TROPiCS-02 |
|---|---|
| Trial name | TROPiCS-02 |
| ClinicalTrials.gov identifier | NCT03901339 |
| Phase | Phase 3 |
| Status | Completed |
| Condition | Metastatic breast cancer |
| Population described in the brief title | Participants with HR+/HER2- metastatic breast cancer |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | None |
| Primary purpose | Treatment |
| Enrollment | 543 |
| Primary endpoint type | Time-to-event |
| Registered primary endpoints | 1 |
| Outcome measures posted | 13 |
| Statistical analyses posted | 10 |
| Lead sponsor | Gilead Sciences |
| Sponsor type | Industry |
2. Clinical Question
The central statistical question was whether sacituzumab govitecan-hziy differed from treatment of physician's choice with respect to progression-free survival by blinded independent central review in participants with HR+/HER2- metastatic breast cancer.
Population
Participants with HR+/HER2- metastatic breast cancer, as specified by the trial's brief title and condition.
Intervention
Sacituzumab govitecan-hziy.
Comparator
Treatment of physician's choice (TPC), with the listed drugs including eribulin, capecitabine, gemcitabine, and vinorelbine.
Primary question
How does sacituzumab govitecan compare with TPC for progression-free survival by blinded independent central review?
3. Trial Design
The registry describes TROPiCS-02 as a randomized, parallel-group, unmasked phase 3 treatment trial. The study enrolled 543 participants and compared two treatment strategies.
Sacituzumab Govitecan-hziy
- Sacituzumab govitecan-hziy
- Compared with treatment of physician's choice
Treatment of Physician's Choice
- Eribulin
- Capecitabine
- Gemcitabine
- Vinorelbine
4. Trial Timeline
Trial start
The registry lists May 8, 2019 as the study start date.
Primary completion
The registry lists October 20, 2023 as the primary completion date.
Results posted
ClinicalTrials.gov has posted 13 outcome measures and 10 statistical analyses for the trial.
5. Primary Endpoint
| Endpoint | Registry definition / time frame | Analysis |
|---|---|---|
| Progression-Free Survival (PFS) by Blinded Independent Central Review (BICR) Assessment | Time from the date of randomization to the date of the first documentation of disease progression or death, whichever occurred first, according to BICR using RECIST 1.1. The registry definition specifies disease progression as an increase of greater than 20% in the sum of the longest diameter of target lesions together with a 5 mm absolute increase component. | Log-rank test; hazard ratio; ITT population |
The registered primary endpoint is a classic time-to-event outcome. Its clock begins at randomization, and the first qualifying event is either documented disease progression or death, whichever occurs first. Because participants can remain alive and progression-free at the end of their observed follow-up, censoring is an essential part of the statistical analysis.
6. Results: Primary Progression-Free Survival
The posted primary analysis compares sacituzumab govitecan with TPC in the ITT population. The analysis used a log-rank test and reported the treatment effect as a hazard ratio.
Progression-free survival by BICR
95% CI: 0.526–0.812 · P = 0.0001
Time frame: up to 42.8 months
| Primary endpoint | Sacituzumab Govitecan | Treatment of Physician's Choice | Effect estimate | P-value |
|---|---|---|---|---|
| PFS by BICR | ITT population | ITT population | HR 0.653 (95% CI 0.526–0.812) | 0.0001 |
The hazard ratio of 0.653 is a relative time-to-event measure comparing the estimated hazard of progression or death between the two randomized groups. Numerically, an HR of 0.653 corresponds to approximately a 34.7% lower estimated hazard for the sacituzumab govitecan group relative to TPC, because 1 − 0.653 = 0.347.
The HR does not mean that 34.7% of participants avoided progression, that progression was delayed by exactly 34.7% of a given number of months, or that every participant experienced the same relative reduction. It is a model-based relative comparison of event rates over the analyzed follow-up.
The two-sided 95% confidence interval of 0.526–0.812 describes the statistical uncertainty around the estimated hazard ratio under the analysis framework. It is not a range containing the individual treatment effects experienced by participants.
The P-value of 0.0001 addresses the evidence against the null hypothesis associated with the reported statistical comparison. It does not measure the size or clinical importance of the treatment effect. Effect magnitude is described by the hazard ratio and its confidence interval.
Because this is a time-to-event analysis, interpretation also depends on censoring and on the appropriateness of summarizing the treatment difference with a hazard ratio. A single HR is most naturally interpreted under the proportional-hazards framework; it should not automatically be translated into a constant individual-level risk reduction.
7. Understanding the Primary Analysis
Why use a log-rank test?
The primary endpoint is time-to-event PFS, so simply comparing the proportion of participants who had progressed at one fixed time would discard much of the available follow-up information. The log-rank test instead compares the observed and expected pattern of events across the follow-up period while accounting for the fact that some participants are censored before an event occurs.
The log-rank framework repeatedly compares the treatment groups at observed event times and aggregates those differences across follow-up.
Why is PFS a time-to-event endpoint?
PFS incorporates both the timing and occurrence of disease progression or death. A participant who remains progression-free for a longer period contributes information for a longer period than a participant who experiences an event early. A participant who has not experienced an event by the end of observation can contribute censored follow-up rather than being treated as if an event had occurred.
Why does BICR matter?
The primary endpoint was assessed by blinded independent central review. This is statistically relevant because the study itself was unmasked, while the central review process provides an independent mechanism for assessing radiographic progression. The registry definition specifically ties the endpoint to BICR and RECIST 1.1.
8. Secondary Endpoint Results
ClinicalTrials.gov reports nine secondary statistical analyses in addition to the posted primary analysis. These cover overall survival, response, clinical benefit, an additional PFS assessment, and quality-of-life deterioration endpoints.
| Secondary endpoint | Method | Effect measure | Estimate | 95% CI | P-value |
|---|---|---|---|---|---|
| Overall Survival (OS) | Log-rank | HR | 0.788 | 0.652–0.952 | 0.0133 |
| ORR by BICR Assessment | Cochran-Mantel-Haenszel | OR | 1.662 | 1.058–2.609 | 0.0268 |
| ORR by LIR Assessment | Cochran-Mantel-Haenszel | OR | 1.989 | 1.174–3.369 | 0.0098 |
| CBR by BICR Assessment | Cochran-Mantel-Haenszel | OR | 1.796 | 1.227–2.628 | 0.0025 |
| CBR by LIR Assessment | Cochran-Mantel-Haenszel | OR | 1.834 | 1.237–2.717 | 0.0024 |
| PFS by LIR Assessment | Log-rank | HR | 0.728 | 0.602–0.881 | 0.0010 |
| TTD of Global Health Status/QoL Scale by EORTC QLQ-C30 | Log-rank | HR | 0.751 | 0.612–0.922 | 0.0059 |
| TTD of Pain Score by EORTC QLQ-C30 | Log-rank | HR | 0.918 | 0.748–1.126 | 0.4151 |
| TTD of Fatigue Score by EORTC QLQ-C30 | Log-rank | HR | 0.732 | 0.598–0.894 | 0.0021 |
All confidence intervals in the registry's posted statistical analyses are two-sided 95% confidence intervals. The registry lists the hypothesis type as "Other / not stated" for these analyses. Accordingly, the numerical P-values should be interpreted as reported statistical results rather than assigned an unreported hypothesis hierarchy.
Overall Survival
Overall survival
95% CI: 0.652–0.952 · P = 0.0133
Time frame: up to 42.8 months
The OS analysis used the ITT population and a log-rank test. The HR below 1 indicates a lower estimated hazard of death in the sacituzumab govitecan group relative to TPC under the reported analysis.
Objective Response Rate
| Assessment | Odds ratio | 95% CI | P-value |
|---|---|---|---|
| BICR | 1.662 | 1.058–2.609 | 0.0268 |
| LIR | 1.989 | 1.174–3.369 | 0.0098 |
The ORR analyses used the ITT population with available data and a Cochran-Mantel-Haenszel test. The effect measure was an odds ratio. An OR above 1 indicates higher estimated odds of the response outcome in the sacituzumab govitecan group relative to TPC under the reported comparison.
Clinical Benefit Rate
| Assessment | Odds ratio | 95% CI | P-value |
|---|---|---|---|
| BICR | 1.796 | 1.227–2.628 | 0.0025 |
| LIR | 1.834 | 1.237–2.717 | 0.0024 |
As with ORR, CBR was analyzed as a binary outcome using the Cochran-Mantel-Haenszel test. The BICR and LIR analyses give two related but distinct assessments rather than a single pooled estimate.
Progression-Free Survival by Local Investigator Review
PFS by LIR assessment
95% CI: 0.602–0.881 · P = 0.0010
Time frame: up to 42.8 months
The LIR analysis produced an HR of 0.728, compared with 0.653 for the BICR primary analysis. Both estimates are below 1, but they are based on different assessment approaches. The distinction is important: agreement between assessment methods can provide useful context, while the estimates should not be treated as if they were independent randomized treatment comparisons.
Quality-of-Life Time-to-Deterioration Endpoints
| Endpoint | Analysis population | HR | 95% CI | P-value |
|---|---|---|---|---|
| TTD of Global Health Status/QoL | HRQOL-Evaluable Population | 0.751 | 0.612–0.922 | 0.0059 |
| TTD of Pain Score | HRQOL-Evaluable Population with baseline pain score ≤ 90 | 0.918 | 0.748–1.126 | 0.4151 |
| TTD of Fatigue Score | HRQOL-Evaluable Population with baseline fatigue score ≤ 90 | 0.732 | 0.598–0.894 | 0.0021 |
These endpoints shift the statistical question from tumor progression or death to deterioration in patient-reported health-status measures. The registry defines the HRQOL-Evaluable Population as participants with an evaluable baseline assessment and at least one evaluable postbaseline assessment. That population is therefore different from the unrestricted ITT population used for the primary PFS analysis.
9. How to Read the Secondary Hazard Ratios
An HR of 0.788 for OS corresponds to approximately a 21.2% lower estimated hazard of death, because 1 − 0.788 = 0.212. An HR of 0.732 for fatigue deterioration corresponds to approximately a 26.8% lower estimated hazard of fatigue deterioration. These are relative hazard interpretations, not absolute reductions in the probability of an event.
The confidence interval for OS, 0.652–0.952, is narrower than a hypothetical interval extending far from the point estimate, but it still reflects uncertainty in the treatment-effect estimate. The confidence interval for pain deterioration, 0.748–1.126, spans 1.0, illustrating that the reported estimate is compatible with both a lower and a higher hazard under the statistical uncertainty represented by the interval.
The reported P-values quantify the statistical evidence under the corresponding test. They do not rank the endpoints by clinical importance and do not indicate the probability that the treatment is effective. For example, the P-value of 0.4151 for pain-score deterioration should not be translated into a 41.51% probability that the treatment has no effect.
10. Statistical Methodology
Intention-to-treat analysis
The primary PFS analysis used the ITT population, which the registry defines as including all participants who were randomized, regardless of whether they received study treatment. This preserves treatment assignment as the basis for the efficacy comparison.
Log-rank testing
The log-rank test was used for the primary BICR PFS analysis and for the posted OS, LIR PFS, global-health-status/QoL, pain, and fatigue time-to-event analyses. The method compares survival experience between groups across the observed event times rather than comparing only a single fixed time point.
Hazard ratio
The hazard ratio is the principal effect measure for the time-to-event analyses in the registry. An HR below 1 indicates a lower estimated event hazard in the sacituzumab govitecan group relative to TPC; an HR above 1 indicates a higher estimated hazard.
This ratio is a relative measure of event rates. It is not an absolute risk difference and is not equivalent to a ratio of median survival times.
Cochran-Mantel-Haenszel test
The registry reports the Cochran-Mantel-Haenszel test for ORR and CBR. This is a categorical-data method that can compare treatment groups while accounting for stratification variables when such strata are specified for the analysis. The ClinicalTrials.gov record does not identify the specific stratification variables used for these analyses, so no additional stratification factors are assumed here.
Odds ratio
ORR and CBR use odds ratios rather than hazard ratios. An odds ratio compares the odds of the binary outcome between treatment groups. An OR of 1 indicates equal odds; an OR above 1 indicates higher odds in the numerator group, while an OR below 1 indicates lower odds.
Confidence intervals
The posted effect estimates use two-sided 95% confidence intervals. The interval gives a measure of precision around the estimated effect under the statistical model and sampling framework. It is not a prediction interval for future participants and does not describe the range of individual treatment responses.
11. Statistical Methods Explained
Why was a log-rank test used for the primary endpoint?
PFS is defined by the timing of progression or death, so the analysis needs to account for both event timing and censoring. The log-rank test is designed for this type of survival comparison and uses information across the follow-up period rather than reducing the endpoint to a single binary status at one arbitrary date.
Why is the primary analysis based on the ITT population?
The ITT principle keeps participants in the group to which they were randomized. This maintains the randomized comparison and avoids changing the treatment groups simply because participants did or did not receive treatment as planned.
What does an HR of 0.653 mean?
An HR of 0.653 means that the estimated hazard of progression or death was approximately 34.7% lower in the sacituzumab govitecan group relative to TPC under the reported analysis. It does not mean that 34.7% of participants avoided progression or that every participant experienced a 34.7% reduction.
Why is the confidence interval important?
The 95% CI of 0.526–0.812 shows that the point estimate of 0.653 is not known with perfect precision. The width of the interval provides information about uncertainty that a single HR cannot convey. An effect estimate should therefore be read together with its confidence interval.
Why are ORR and CBR analyzed differently from PFS?
ORR and CBR are binary outcomes: each participant is classified according to whether the specified outcome occurred. PFS records the timing of an event and can include censored observations. The statistical methods therefore address different data structures: Cochran-Mantel-Haenszel for the posted binary analyses and log-rank testing for the posted time-to-event analyses.
Why does BICR differ from LIR?
BICR and LIR represent different assessment processes. The primary endpoint used blinded independent central review, while a secondary analysis used local investigator review. Differences between their HR estimates do not automatically indicate a contradiction; they reflect the fact that the underlying assessments are not identical.
What does a P-value of 0.0001 tell us?
It reports the statistical evidence associated with the specified primary comparison under the test used. It does not quantify treatment benefit, does not give the probability that the null hypothesis is true, and does not indicate how large the treatment effect is. The HR and its confidence interval are needed to describe effect magnitude and precision.
12. Time Frames and Analysis Populations
The registry uses different time frames and analysis populations for different endpoints. This is important because results from one population or follow-up window should not be silently substituted for another.
| Endpoint group | Time frame | Population |
|---|---|---|
| Primary BICR PFS | Up to 42.8 months | ITT population |
| Overall Survival | Up to 42.8 months | ITT population |
| ORR | Up to 42.8 months | ITT population with available data |
| CBR | Up to 42.8 months | ITT population with available data |
| PFS by LIR | Up to 42.8 months | ITT population |
| Global Health Status/QoL TTD | Up to 37.8 months | HRQOL-Evaluable Population |
| Pain TTD | Up to 37.8 months | HRQOL-Evaluable Population with baseline pain score ≤ 90 |
| Fatigue TTD | Up to 37.8 months | HRQOL-Evaluable Population with baseline fatigue score ≤ 90 |
13. Safety Results
The ClinicalTrials.gov record reports serious adverse events by randomized treatment arm using affected participants over participants at risk. These counts should be distinguished from the efficacy analyses because safety populations and denominators can differ from the ITT population.
| Safety measure | Sacituzumab Govitecan | Treatment of Physician's Choice |
|---|---|---|
| Serious adverse events, affected / at risk | 74 / 268 | 48 / 249 |
The reported serious-adverse-event counts therefore identify 74 affected participants among 268 at risk in the sacituzumab govitecan arm and 48 among 249 at risk in the TPC arm. The ClinicalTrials.gov record does not provide a formal statistical comparison for this safety measure, so no hazard ratio, odds ratio, confidence interval, or P-value is assigned to these counts here.
14. What the Binary Effect Measures Mean
Odds ratio for ORR
The BICR ORR analysis reported an OR of 1.662. Conceptually, this compares the odds of objective response between the two groups. It is not the same as a risk ratio or percentage-point difference.
An OR of 1 corresponds to equal odds. An OR above 1 indicates higher odds of the outcome in the numerator group.
Why the ORR BICR and LIR estimates differ
The BICR ORR analysis produced an OR of 1.662, while the LIR analysis produced an OR of 1.989. These are based on different assessment approaches. The difference illustrates why endpoint definitions and assessment mechanisms matter when comparing clinical-trial results.
CBR uses the same general statistical framework
The two CBR analyses also used Cochran-Mantel-Haenszel testing and odds ratios. Their estimates were 1.796 for BICR and 1.834 for LIR. The similarity in direction does not make the two assessments interchangeable; each remains a separate reported analysis.
15. The Primary Endpoint in Statistical Context
Event definition
PFS counts the first documented progression or death, whichever occurs first, according to the registry definition.
Assessment
The primary endpoint was assessed by blinded independent central review using RECIST 1.1.
Population
The primary analysis used the ITT population, including all randomized participants regardless of whether study treatment was received.
Comparison
The reported treatment effect was estimated with a hazard ratio and tested with a log-rank test.
This combination of design elements is statistically coherent: randomization establishes the comparison, ITT preserves that randomized assignment, BICR provides an independent endpoint-assessment mechanism, and survival methods account for the timing of progression or death and incomplete follow-up.
16. Interpreting the Relationship Between PFS and OS
The registry reports both PFS and OS as time-to-event outcomes, but they represent different clinical events. PFS ends at the first documented progression or death, whereas OS concerns death. Consequently, an HR for PFS and an HR for OS should not be interpreted as measuring the same event.
| Feature | PFS | OS |
|---|---|---|
| Event | First disease progression or death | Death |
| Primary assessment | BICR using RECIST 1.1 | ITT time-to-event analysis |
| Reported HR | 0.653 | 0.788 |
| 95% CI | 0.526–0.812 | 0.652–0.952 |
| P-value | 0.0001 | 0.0133 |
| Time frame | Up to 42.8 months | Up to 42.8 months |
The two estimates should therefore be read as complementary pieces of evidence rather than as duplicate measures. PFS captures disease-control time before progression or death, while OS captures survival until death.
17. Censoring and Time-to-Event Interpretation
A central feature of the TROPiCS-02 analyses is that participants can contribute information even when they have not experienced the endpoint by the time their observation ends. This is why PFS, OS, and time to deterioration are analyzed as time-to-event outcomes rather than ordinary binary outcomes.
The Kaplan-Meier framework estimates the event-free survival function while accounting for participants whose event status is censored during follow-up.
The ClinicalTrials.gov record does not report the underlying event and censoring counts needed to reconstruct a Kaplan-Meier curve. A statistically valid curve cannot therefore be generated from the summary hazard ratios and P-values alone.
18. Why BICR and LIR Are Both Informative
The primary PFS endpoint was assessed by BICR, while the registry also reports PFS by local investigator review. The two analyses address the same broad clinical outcome but rely on different assessment processes.
| Assessment approach | Endpoint | HR | 95% CI | P-value |
|---|---|---|---|---|
| BICR | Primary PFS | 0.653 | 0.526–0.812 | 0.0001 |
| LIR | Secondary PFS | 0.728 | 0.602–0.881 | 0.0010 |
The estimates differ numerically, but both are below 1. The important statistical point is that the analysis method and assessment source are part of the endpoint definition. The estimates should therefore be interpreted within their respective assessment frameworks rather than treating the difference between 0.653 and 0.728 as a separate test of disagreement.
19. Quality-of-Life Time to Deterioration
The HRQOL analyses extend the statistical story beyond tumor control. They evaluate time to deterioration in global health status/quality of life, pain, and fatigue using the EORTC QLQ-C30.
Global Health Status/QoL
HR 0.751; 95% CI 0.612–0.922; P = 0.0059. Time frame: up to 37.8 months.
Pain
HR 0.918; 95% CI 0.748–1.126; P = 0.4151. Time frame: up to 37.8 months.
Fatigue
HR 0.732; 95% CI 0.598–0.894; P = 0.0021. Time frame: up to 37.8 months.
Population restriction
Pain and fatigue analyses required baseline scores ≤ 90, in addition to the HRQOL-evaluable criteria.
These analyses illustrate why endpoint-specific populations matter. The HRQOL-evaluable population requires baseline and postbaseline evaluable assessments, so it is not identical to the ITT population. The resulting estimates answer a narrower question about participants with evaluable quality-of-life data.
20. Statistical Evidence vs Effect Size
One of the most important lessons from the TROPiCS-02 results is that statistical significance and effect size are different quantities.
| Endpoint | Effect estimate | 95% CI | P-value | What each component contributes |
|---|---|---|---|---|
| Primary PFS | HR 0.653 | 0.526–0.812 | 0.0001 | HR describes relative effect; CI describes precision; P-value describes evidence under the test. |
| OS | HR 0.788 | 0.652–0.952 | 0.0133 | Same distinction for the death endpoint. |
| Pain TTD | HR 0.918 | 0.748–1.126 | 0.4151 | The point estimate, uncertainty, and test result provide different information. |
A small P-value does not necessarily imply a large treatment effect, while a treatment effect that is clinically interesting may be estimated imprecisely. The correct statistical reading therefore considers the effect estimate, confidence interval, analysis population, endpoint definition, and test together.
21. Important Limitations and Interpretation Issues
- Open-label treatment assignment: the registry lists masking as none. This makes the independent central review of the primary PFS assessment particularly relevant to endpoint assessment.
- Hazard-ratio interpretation: a single HR is a relative time-to-event summary and should not be translated into an absolute risk reduction or individual-level treatment effect.
- Proportional-hazards considerations: interpretation of a single hazard ratio is most straightforward when the relative hazards are reasonably represented by the model over time. The ClinicalTrials.gov record does not report a formal assessment of the proportional-hazards assumption.
- Different analysis populations: the primary PFS analysis used ITT, while ORR and CBR used ITT participants with available data, and HRQOL analyses used an HRQOL-evaluable population.
- BICR versus LIR: the two assessment methods produced different numerical estimates for PFS and response outcomes. They should be understood as distinct assessment frameworks.
- Safety denominators: serious adverse events are reported using 268 and 249 participants at risk, which are not presented in the ClinicalTrials.gov record as the overall enrollment denominator of 543.
- Unreported design details: the ClinicalTrials.gov record does not specify stratification factors, a non-inferiority margin, crossover rules, factorial structure, Bayesian methods, interim-analysis boundaries, or missing-data imputation methods. Those features should not be inferred from the posted results.
- Multiplicity: the ClinicalTrials.gov record identifies multiple statistical analyses but do not provide a multiplicity-control strategy. The posted P-values should therefore be reported as registry results without inventing an unreported endpoint hierarchy or alpha-allocation procedure.
- No subgroup estimates reported: the provided statistical-analysis data do not include subgroup-specific effect estimates, so subgroup conclusions cannot be drawn from this record.
22. What the Registry Does Not Establish From the Supplied Data
The posted data support a detailed analysis of the reported effect measures, but they do not provide every quantity commonly found in a full clinical-trial publication.
| Topic | What can be stated from the ClinicalTrials.gov record |
|---|---|
| Median PFS | Not reported in the ClinicalTrials.gov record. |
| Median OS | Not reported in the ClinicalTrials.gov record. |
| Kaplan-Meier curves | Not reconstructable from the registry-reported summary estimates alone. |
| Subgroup effects | Not included in the statistical analyses posted on ClinicalTrials.gov. |
| Baseline characteristics | Not reported in the ClinicalTrials.gov record. |
| Exact randomized arm sizes | Not registry-reported separately from total enrollment. |
| Interim-analysis design | Not reported in the ClinicalTrials.gov record. |
| Multiplicity strategy | Not reported in the ClinicalTrials.gov record. |
| Missing-data/imputation strategy | Not reported in the ClinicalTrials.gov record. |
| Non-inferiority margin | Not applicable to the reported hypothesis information; no non-inferiority margin is reported. |
| Bayesian analysis | No Bayesian method is listed among the registry-reported normalized methods. |
This distinction is important for reproducibility. A statistical analysis should not fill missing registry information with assumptions simply because a particular design feature is common in similar trials.
23. Why This Trial Matters Statistically
TROPiCS-02 is a useful teaching case because it places several core clinical-trial methods in one randomized comparison: time-to-event analysis, central endpoint review, ITT analysis, hazard ratios, confidence intervals, log-rank testing, categorical-outcome analysis, odds ratios, and patient-reported time-to-deterioration endpoints.
| Concept | How it appears in TROPiCS-02 |
|---|---|
| Randomization | The trial is randomized with two parallel treatment arms. |
| Intention-to-treat analysis | The primary PFS analysis includes all randomized participants regardless of whether they received study treatment. |
| Time-to-event endpoint | Primary PFS measures time from randomization to progression or death. |
| Log-rank test | Used for the primary PFS comparison and multiple secondary time-to-event analyses. |
| Hazard ratio | Used for PFS, OS, LIR PFS, and quality-of-life deterioration endpoints. |
| Confidence interval | All registry-reported effect estimates have two-sided 95% confidence intervals. |
| Cochran-Mantel-Haenszel test | Used for ORR and CBR analyses. |
| Odds ratio | Used to quantify the reported ORR and CBR comparisons. |
| BICR | The primary PFS endpoint was assessed by blinded independent central review. |
| LIR | Secondary PFS, ORR, and CBR analyses also used local investigator review. |
| Quality-of-life analysis | Time to deterioration was evaluated for global health status/QoL, pain, and fatigue. |
24. Clinical Interpretation vs Statistical Interpretation
Statistical interpretation
The primary BICR PFS comparison reported an HR of 0.653 with a two-sided 95% CI of 0.526–0.812 and P = 0.0001. The analysis used the ITT population and a log-rank test.
Effect-size interpretation
The HR below 1 indicates a lower estimated hazard of progression or death in the sacituzumab govitecan group. The point estimate alone does not describe absolute PFS probabilities or individual treatment benefit.
Assessment interpretation
The primary PFS result was based on BICR, while a separate LIR analysis reported an HR of 0.728. Assessment method is therefore part of the statistical context.
Patient-reported outcomes
Global health status/QoL and fatigue deterioration analyses reported HRs below 1, while pain deterioration had an HR of 0.918 with a 95% CI spanning 1.
25. A Compact Statistical Summary
| Domain | Reported result | Statistical reading |
|---|---|---|
| Primary PFS by BICR | HR 0.653; 95% CI 0.526–0.812; P = 0.0001 | Lower estimated hazard of progression or death with sacituzumab govitecan under the reported analysis. |
| Overall Survival | HR 0.788; 95% CI 0.652–0.952; P = 0.0133 | Lower estimated hazard of death under the reported ITT log-rank analysis. |
| ORR by BICR | OR 1.662; 95% CI 1.058–2.609; P = 0.0268 | Higher estimated odds of objective response under the reported binary comparison. |
| ORR by LIR | OR 1.989; 95% CI 1.174–3.369; P = 0.0098 | Higher estimated odds of objective response under the local-assessment analysis. |
| CBR by BICR | OR 1.796; 95% CI 1.227–2.628; P = 0.0025 | Higher estimated odds of clinical benefit under the BICR analysis. |
| CBR by LIR | OR 1.834; 95% CI 1.237–2.717; P = 0.0024 | Higher estimated odds of clinical benefit under the LIR analysis. |
| PFS by LIR | HR 0.728; 95% CI 0.602–0.881; P = 0.0010 | Lower estimated hazard of progression or death under local investigator assessment. |
| Global Health Status/QoL TTD | HR 0.751; 95% CI 0.612–0.922; P = 0.0059 | Lower estimated hazard of deterioration in global health status/QoL. |
| Pain TTD | HR 0.918; 95% CI 0.748–1.126; P = 0.4151 | Point estimate below 1, with the reported confidence interval spanning 1. |
| Fatigue TTD | HR 0.732; 95% CI 0.598–0.894; P = 0.0021 | Lower estimated hazard of fatigue deterioration under the reported analysis. |
26. Related Tutorials
Learn more about the methods used in this trial:
27. Related Statistical Calculators
28. Sources
- ClinicalTrials.gov: TROPiCS-02, NCT03901339.
- PubMed: PMID 40657894.
- PubMed: PMID 33093337.
- PubMed: PMID 32946924.
Continue exploring clinical-trial statistics
Connect trial endpoints with deeper statistical tutorials and practical analysis tools.
29. Record Summary
TROPiCS-02 provides a useful example of a randomized phase 3 clinical-trial analysis built around a time-to-event primary endpoint. The primary PFS endpoint was defined from randomization to first documented progression or death, assessed by blinded independent central review using RECIST 1.1, and analyzed in the ITT population with a log-rank test. The reported HR was 0.653 with a two-sided 95% CI of 0.526–0.812 and P = 0.0001.
The statistical evidence extends beyond the primary endpoint. Overall survival was analyzed with the same broad time-to-event framework and produced an HR of 0.788. ORR and CBR were analyzed as binary outcomes using Cochran-Mantel-Haenszel methods and odds ratios, while local investigator assessments provided additional PFS, ORR, and CBR analyses. Time to deterioration in global health status/QoL, pain, and fatigue added patient-reported outcome dimensions to the analysis.
The most important statistical lesson is that these results should be interpreted according to their endpoint definitions, analysis populations, assessment methods, effect measures, confidence intervals, and P-values. A hazard ratio is not an absolute risk difference; an odds ratio is not a risk ratio; and a P-value is not an effect-size measure. Keeping those distinctions explicit is essential for an accurate reading of the trial.