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 ClinicalTrials.gov record.
1. Trial at a Glance
KATE2 was a randomized, double-blind, phase 2, parallel-group trial evaluating trastuzumab emtansine with either atezolizumab or atezolizumab-placebo in participants with HER2-positive locally advanced or metastatic breast cancer who had received prior trastuzumab and taxane-based therapy. The registry reports progression-free survival as a primary endpoint and adverse events as a second primary endpoint.
| Feature | KATE2 |
|---|---|
| Phase | Phase 2 |
| Condition | Metastatic Breast Cancer |
| Population | Participants with HER2-positive locally advanced or metastatic breast cancer who received prior trastuzumab and taxane-based therapy |
| Design | Randomized, double-blind, parallel-group |
| Allocation | Randomized |
| Primary purpose | Treatment |
| Enrollment | 202 |
| Primary endpoints | Progression-Free Survival (PFS) and Percentage of Participants With Adverse Events |
| Results posted | Yes |
| Statistical analyses posted | 3 |
| ClinicalTrials.gov | NCT02924883 |
| Lead sponsor | Hoffmann-La Roche |
2. Clinical Question
The central statistical question was whether adding atezolizumab to trastuzumab emtansine was associated with a different time-to-progression-or-death profile than trastuzumab emtansine with placebo in participants with HER2-positive locally advanced or metastatic breast cancer who had received prior trastuzumab and taxane-based therapy.
Population
Participants with human epidermal growth factor-2 (HER2) positive locally advanced or metastatic breast cancer who received prior trastuzumab and taxane-based therapy.
Intervention
Trastuzumab emtansine in combination with atezolizumab.
Comparator
Trastuzumab emtansine in combination with atezolizumab-placebo.
Primary question
Does the trastuzumab emtansine plus atezolizumab combination produce a different PFS experience from trastuzumab emtansine plus placebo under the prespecified superiority framework?
3. Trial Design
Trastuzumab Emtansine + Atezolizumab
- Trastuzumab emtansine
- Atezolizumab
- Double-blind randomized treatment assignment
Trastuzumab Emtansine + Placebo
- Trastuzumab emtansine
- Atezolizumab-placebo
- Double-blind randomized treatment assignment
The registry describes the study as randomized, double-blind, and parallel. These design features are important statistically. Randomization establishes the treatment comparison before outcomes occur; blinding can reduce differential expectations or assessment behavior; and the parallel design means participants remain associated with their randomized treatment group rather than switching between randomized treatment strategies as a defining feature of the study design.
4. Endpoints
The registry lists two primary endpoints: a time-to-event efficacy endpoint and a binary safety endpoint. The ClinicalTrials.gov record also contain formal statistical analyses for overall survival and duration of objective response as secondary endpoints.
| Endpoint | Registry definition / time frame | Type |
|---|---|---|
| Progression-Free Survival (PFS) as Determined by Investigator's Tumor Assessment Using RECIST Version 1.1 (v1.1) | PFS was defined as the time from randomization to the first occurrence of disease progression or death from any cause, whichever occurred first, on the basis of investigator assessments. Progression was defined as at least a 20% increase in the sum of diameters of target lesions with an absolute increase of at least 5 millimeter (mm) or the appearance of one or more new lesions. Time frame: Baseline up to approximately 15 months. | Time-to-event |
| Percentage of Participants With Adverse Events | An adverse event is any untoward medical occurrence in a participant administered a pharmaceutical product and which does not necessarily have to have a causal relationship with the treatment. An adverse event can therefore be any unfavorable and unintended sign, including an abnormal laboratory finding, symptom, or disease temporally associated with the use of a pharmaceutical product. Time frame: Baseline up to study completion, approximately 40 months. | Binary |
Secondary endpoints with formal analyses
| Endpoint | Time frame | Analysis | Effect measure |
|---|---|---|---|
| Overall Survival (OS) | Baseline up to study completion or death, whichever occurs first, approximately 40 months | Log-rank test; stratified analysis | Hazard ratio |
| Duration of OR as Determined by Investigator's Tumor Assessment Using RECIST v1.1 | Baseline up to approximately 15 months | Log-rank test; stratified Cox proportional hazards model | Hazard ratio |
5. Analysis Populations and Treatment Comparison
The primary PFS analysis was conducted in the intent-to-treat (ITT) population, defined in the registry as including all randomized participants grouped according to the treatment assigned at randomization. The same ITT definition was posted on ClinicalTrials.gov for the OS and duration-of-objective-response analyses.
| Analysis | Population | Comparison |
|---|---|---|
| Primary PFS | ITT: all randomized participants, grouped according to treatment assigned at randomization | Trastuzumab emtansine + placebo vs trastuzumab emtansine + atezolizumab |
| Secondary OS | ITT: all randomized participants, grouped according to treatment assigned at randomization | Trastuzumab emtansine + placebo vs trastuzumab emtansine + atezolizumab |
| Secondary duration of OR | ITT; participants with OR were considered for duration of OR | Trastuzumab emtansine + placebo vs trastuzumab emtansine + atezolizumab |
The ITT principle is particularly important for a randomized comparison. Once participants have been randomized, analyzing them according to their assigned group preserves the treatment contrast created by randomization. It also means that the resulting estimate answers a question about the effect of assignment to the randomized strategy, rather than simply the effect among participants who remained continuously exposed to treatment.
6. Statistical Methodology
Stratified log-rank test
The registry reports a 2-sided log-rank test for the primary PFS analysis. The test was stratified by world region (Western Europe vs U.S. vs Rest of World) and PD-L1 status (IC 0 vs IC 1/2/3).
A log-rank test compares the observed pattern of event occurrence between treatment groups over follow-up. Rather than comparing only a single time point, it incorporates the ordering of events across the study period. Stratification allows the comparison to account for prespecified groups that may have different underlying event patterns.
The log-rank framework tests whether the survival experience differs between the randomized groups, with the registry-reported analysis additionally stratified by world region and PD-L1 status.
Stratified Cox proportional-hazards analysis
The duration-of-objective-response analysis used a stratified Cox proportional hazards model stratified by world region (Western Europe, U.S., Rest of World) and PD-L1 status (IC 0, IC 1/2/3). The registry reports the hazard ratio as the effect measure for this analysis.
A Cox model describes the relationship between treatment assignment and the instantaneous event rate. A hazard ratio below 1 indicates a lower estimated instantaneous event rate in the first group relative to the reference comparison, whereas a hazard ratio above 1 indicates a higher estimated instantaneous event rate.
For this page, the reported comparisons are presented in the registry's stated order: trastuzumab emtansine + placebo versus trastuzumab emtansine + atezolizumab.
Intention-to-treat analysis
The ITT population included all randomized participants and grouped them according to treatment assigned at randomization. This is the principal efficacy population reported for the three registry-reported statistical analyses.
Stratification
For the registry-reported time-to-event analyses, the registry identifies two stratification dimensions: world region and PD-L1 status. Stratification does not mean that the treatment effect is estimated separately and independently within every stratum. Instead, the statistical comparison accounts for the specified strata while estimating the overall randomized treatment contrast.
Binary safety endpoint
The percentage of participants with adverse events is a binary endpoint because each participant can be classified according to whether the specified adverse-event outcome occurred. The registry identifies this as a primary endpoint, but the ClinicalTrials.gov record does not provide a formal statistical analysis for this endpoint beyond the arm-specific serious-adverse-event counts reported below.
7. Primary Result: Progression-Free Survival
The primary formal efficacy analysis compared progression-free survival between trastuzumab emtansine plus placebo and trastuzumab emtansine plus atezolizumab. The analysis used the ITT population and a 2-sided stratified log-rank test, with stratification by world region and PD-L1 status.
Hazard ratio for progression or death
95% CI: 0.55–1.23 · P = 0.3332
Analysis: 2-sided stratified log-rank test
| Primary endpoint | Estimate | 95% CI | P-value | Analysis |
|---|---|---|---|---|
| Progression-Free Survival | HR 0.82 | 0.55–1.23 | 0.3332 | 2-sided log-rank; stratified by world region and PD-L1 status |
The reported HR of 0.82 is below 1. In the registry's comparison order, this corresponds to an estimated instantaneous rate of progression or death that is approximately 82% of the corresponding rate in the comparator group under the time-to-event analysis. Equivalently, 0.82 corresponds to a 18% lower estimated hazard relative to the comparator when expressed as 1 − HR.
The HR does not mean that 18% of participants avoided progression, that progression was prevented in exactly 18% of patients, or that each participant experienced an 18% reduction in individual risk. It is a relative time-to-event measure, not an absolute difference in the probability of an event.
The 95% CI of 0.55–1.23 shows the uncertainty around the reported HR. Because the interval includes 1, the ClinicalTrials.gov record is compatible with hazard ratios both below and above the null value. The width of the interval also indicates that the estimate is not highly precise.
The P-value of 0.3332 is a measure of the evidence against the specified null hypothesis under the stated statistical test. It is not a measure of how large the treatment effect is, nor is it the probability that either treatment is effective or ineffective. Effect size and uncertainty are better represented by the HR and its confidence interval.
Because PFS is a time-to-event endpoint, interpretation also depends on censoring and on the assumptions underlying the hazard-based analysis. The ClinicalTrials.gov record does not provide enough information to independently assess the proportional-hazards assumption or the detailed censoring rules.
What the PFS result does and does not establish
What it establishes
The registry reports a prespecified superiority analysis with an HR of 0.82, 95% CI 0.55–1.23, and P = 0.3332 using a 2-sided stratified log-rank test.
What it does not establish
The HR is not an absolute risk difference, a median PFS difference, a probability of individual benefit, or a measure of treatment effect at one particular time point.
8. Secondary Result: Overall Survival
Overall survival was analyzed as a secondary time-to-event endpoint. The registry analysis used the ITT population and compared the same two randomized groups using a log-rank test with stratification by world region and PD-L1 status.
Hazard ratio for death
95% CI: 0.42–1.30 · P = 0.2934
Analysis: 2-sided stratified log-rank test
| Secondary endpoint | Estimate | 95% CI | P-value | Analysis |
|---|---|---|---|---|
| Overall Survival | HR 0.74 | 0.42–1.30 | 0.2934 | Log-rank; stratified by world region and PD-L1 status |
The reported HR of 0.74 corresponds to an estimated instantaneous rate of death approximately 74% of the comparator rate in the model-based time-to-event comparison. Expressed as 1 − HR, this corresponds to a 26% lower estimated hazard in the direction of the comparison.
The HR does not mean that 26% of participants survived, that mortality was reduced by exactly 26% for every individual, or that the absolute probability of death differed by 26 percentage points.
The 95% CI of 0.42–1.30 spans the null value of 1. This interval is relatively broad and includes values corresponding to both lower and higher estimated hazards. Thus, the point estimate alone should not be treated as a precise description of the underlying treatment effect.
The P-value of 0.2934 does not measure effect magnitude. It quantifies the result of the specified statistical test under its null hypothesis. A P-value should be interpreted together with the HR, confidence interval, endpoint definition, analysis population, and trial design.
As with PFS, the analysis is subject to the considerations that accompany time-to-event methods, including censoring and the proportional-hazards framework. The ClinicalTrials.gov record does not report enough detail to independently evaluate those assumptions.
9. Secondary Result: Duration of Objective Response
Duration of objective response was another secondary time-to-event endpoint. Participants with objective response were considered for the duration-of-response analysis. The registry-reported analysis used the ITT framework with responders considered for this endpoint and reported a stratified Cox proportional-hazards model.
Hazard ratio for duration of objective response
95% CI: 0.52–3.03 · P = 0.6099
Analysis: stratified Cox proportional-hazards model
| Secondary endpoint | Estimate | 95% CI | P-value | Analysis |
|---|---|---|---|---|
| Duration of OR as determined by investigator's tumor assessment using RECIST v1.1 | HR 1.26 | 0.52–3.03 | 0.6099 | Stratified Cox proportional-hazards model |
The reported HR of 1.26 is above 1 in the stated comparison order. Under the model, that corresponds to a higher estimated instantaneous event rate for the first-listed group relative to the second-listed group, conditional on the response population used for this analysis. It should not be translated into a 26% shorter duration for individual responders.
The 95% CI of 0.52–3.03 is wide and crosses 1. This indicates substantial uncertainty around the point estimate. The interval includes values compatible with a lower event hazard as well as values compatible with a higher event hazard.
The P-value of 0.6099 does not measure the magnitude of the observed HR. It describes the result of the specified statistical test relative to its null hypothesis and should not be interpreted as the probability that the treatment has no effect.
Duration-of-response analysis also has an important conditioning feature: only participants with an objective response were considered for duration of OR. Consequently, it answers a different question from the primary PFS analysis, which begins with the randomized population and includes progression or death as the event definition.
10. Comparing the Three Reported Time-to-Event Analyses
| Endpoint | Role | HR | 95% CI | P-value | Method |
|---|---|---|---|---|---|
| Progression-Free Survival | Primary | 0.82 | 0.55–1.23 | 0.3332 | 2-sided stratified log-rank |
| Overall Survival | Secondary | 0.74 | 0.42–1.30 | 0.2934 | Stratified log-rank |
| Duration of Objective Response | Secondary | 1.26 | 0.52–3.03 | 0.6099 | Stratified Cox model |
The three estimates should not be interpreted as interchangeable measures. PFS counts progression or death from the time of randomization; OS measures death; and duration of objective response begins with the subset of participants who achieved an objective response and examines the subsequent duration of that response. Different endpoints therefore condition on different clinical events and populations.
11. Safety Results
The registry identifies Percentage of Participants With Adverse Events as a primary binary endpoint. The ClinicalTrials.gov record does not provide arm-specific percentages for that endpoint or a formal statistical analysis of the overall adverse-event endpoint. They do, however, provide serious-adverse-event counts by randomized arm.
| Safety measure | Trastuzumab Emtansine + Atezolizumab | Trastuzumab Emtansine + Placebo |
|---|---|---|
| Serious adverse events, affected / at risk | 52 / 133 | 16 / 67 |
The reported serious-adverse-event counts are descriptive. The ClinicalTrials.gov record does not provide a formal between-group hypothesis test, confidence interval, or adjusted effect measure for serious adverse events. Accordingly, the counts should not be converted into an inferred statistical comparison.
12. How to Read the PFS Hazard Ratio
The PFS HR of 0.82 is a relative time-to-event measure. In the reported comparison order, it corresponds to an estimated instantaneous progression-or-death rate that is 0.82 times the comparator rate under the model. Expressed as a relative reduction, 1 − 0.82 = 0.18, or 18%.
An HR of 0.82 does not mean that the probability of progression or death was reduced by exactly 18 percentage points. Absolute risks depend on follow-up time and the underlying event distribution. The ClinicalTrials.gov record does not report median PFS or time-specific PFS percentages, so those quantities are not added here.
The 95% CI of 0.55–1.23 describes uncertainty around the estimated relative hazard. Because it crosses 1, it includes both a possible lower hazard and a possible higher hazard relative to the comparator.
The P-value of 0.3332 is evidence from the specified 2-sided stratified log-rank test. It is not a measure of clinical importance, effect size, or the probability that the null hypothesis is true.
13. Statistical Methods Explained
Why was a log-rank test used for PFS?
PFS is a time-to-event endpoint because participants can experience progression or death at different times, while some participants may not experience either event during the relevant observation period. A log-rank test is designed to compare survival-type event patterns across randomized groups while using information throughout follow-up rather than reducing the endpoint to a single binary status at one arbitrary date.
What does an HR of 0.82 mean?
An HR of 0.82 means that, under the time-to-event model and in the stated comparison order, the estimated instantaneous event rate is 0.82 times the comparator rate. It does not mean that 82% of patients avoided progression or that every patient's individual risk was reduced by 18%.
Why does the confidence interval matter more than the point estimate alone?
The point estimate is only one estimate of the underlying treatment contrast. The 95% CI of 0.55–1.23 shows that the observed data are compatible with a substantially lower hazard as well as a hazard above the null value of 1. A reader therefore needs both the HR and its interval to understand the precision of the estimate.
Why was the analysis stratified?
The registry-reported analysis was stratified by world region and PD-L1 status. Stratification allows the time-to-event comparison to account for these specified categories rather than treating all participants as if the underlying event experience were necessarily identical across those strata.
Why is the ITT population important?
The ITT population includes all randomized participants and keeps them in the group to which they were assigned. This maintains the treatment contrast generated by randomization and avoids redefining the primary efficacy population based on subsequent treatment behavior.
Why is duration of response different from PFS?
PFS begins at randomization and ends at progression or death. Duration of objective response, by contrast, considers participants with an objective response and measures the time that the response persists. A duration-of-response HR therefore cannot be interpreted as another version of the PFS HR; the endpoints start from different clinical states and answer different questions.
Why should the safety counts not be treated as a formal statistical test?
The ClinicalTrials.gov record reports serious adverse events as 52/133 versus 16/67, but do not provide a formal comparison, confidence interval, or P-value for that safety measure.
14. Randomization, Blinding, and Stratification
These features work together. Randomization determines treatment assignment; blinding helps control information-related sources of bias; the parallel design defines the treatment comparison; and stratified analysis accounts for the registry-specified strata in the time-to-event comparison.
15. Time-to-Event Analysis and Censoring
PFS and OS are not ordinary binary outcomes because the timing of the event matters. A participant who progresses early and a participant who progresses much later are both classified as having an event, but they contribute different information to a time-to-event analysis. Similarly, a participant who remains event-free at the end of available observation may contribute censored follow-up rather than being treated as if an event never could occur.
The survival function represents the probability of remaining free of the specified event beyond time t. For PFS, the event is the first occurrence of progression or death.
The registry's PFS definition is especially important because it specifies that progression is based on investigator assessments using RECIST v1.1 and that the event is the first occurrence of disease progression or death from any cause. Those details determine exactly what the time-to-event analysis is attempting to compare.
16. Why the Analysis Is Stratified by PD-L1 Status and World Region
The primary PFS analysis was stratified by world region (Western Europe vs U.S. vs Rest of World) and PD-L1 status (IC 0 vs IC 1/2/3). The same factors appear in the registry-reported OS analysis, while the duration-of-response Cox model uses the same categories.
| Stratification factor | Categories reported in the registry analysis | Role |
|---|---|---|
| World region | Western Europe vs U.S. vs Rest of World | Stratification of time-to-event comparison |
| PD-L1 status | IC 0 vs IC 1/2/3 | Stratification of time-to-event comparison |
Stratification is different from presenting separate subgroup treatment effects. The ClinicalTrials.gov record specifies stratification factors for the formal analyses, but they do not provide separate HRs or interaction tests for those categories. Therefore, this page does not infer treatment-effect heterogeneity from the stratification scheme.
17. Primary Endpoint Structure and Multiplicity
KATE2 has two registered primary endpoints: PFS and the percentage of participants with adverse events. The statistical analyses posted on ClinicalTrials.gov include one primary analysis for PFS, while the other two posted formal analyses concern secondary time-to-event endpoints.
| Endpoint | Registry role | Formal statistical analysis reported? |
|---|---|---|
| Progression-Free Survival | Primary | Yes — 2-sided stratified log-rank test |
| Percentage of Participants With Adverse Events | Primary | No formal statistical analysis reported in the ClinicalTrials.gov record |
| Overall Survival | Secondary | Yes — log-rank test |
| Duration of OR | Secondary | Yes — stratified Cox proportional-hazards model |
Because there are two registered primary endpoints, interpretation of the complete confirmatory framework would normally require the prespecified protocol or statistical analysis plan to determine how type I error was allocated across those endpoints. The ClinicalTrials.gov record does not state an alpha-allocation or multiplicity-adjustment procedure.
18. Superiority Hypothesis
The analyses posted on ClinicalTrials.gov identify the hypothesis type as superiority. This means the statistical framework was directed toward detecting a difference between the randomized treatment groups rather than demonstrating that two treatments were sufficiently similar within a prespecified non-inferiority margin.
Superiority framework
The question is whether the randomized groups differ with respect to the specified endpoint under the prespecified hypothesis-testing framework.
No non-inferiority margin reported
The ClinicalTrials.gov record contain no non-inferiority margin, so no margin-based conclusion should be constructed for this trial.
The distinction matters because a confidence interval that includes the null value does not automatically become evidence of non-inferiority. Non-inferiority requires a prespecified clinically meaningful margin and an analysis specifically designed to test that hypothesis.
19. What the P-Values Mean
| Endpoint | P-value | What it represents |
|---|---|---|
| Primary PFS | 0.3332 | Result of the registry-reported 2-sided stratified log-rank test |
| Secondary OS | 0.2934 | Result of the registry-reported log-rank analysis |
| Secondary duration of OR | 0.6099 | Result of the registry-reported stratified Cox analysis |
A P-value is calculated under a specified statistical model and null hypothesis. It is useful for quantifying the compatibility of the observed data with that null hypothesis, but it does not tell the reader how clinically large an effect is.
For example, the PFS analysis combines a P-value of 0.3332 with an HR of 0.82 and a 95% CI of 0.55–1.23. The HR communicates the estimated relative effect; the confidence interval communicates uncertainty around that estimate; and the P-value communicates the result of the specified hypothesis test. These are complementary quantities, not interchangeable ones.
20. Limitations
- Limited registry detail: the ClinicalTrials.gov record does not include arm-specific randomized sample sizes, median PFS, median OS, event counts for efficacy endpoints, or time-specific survival estimates.
- Primary safety analysis detail: the registry identifies adverse events as a primary endpoint, but the ClinicalTrials.gov record does not provide a formal statistical comparison for that endpoint.
- Safety denominators: serious-adverse-event counts are reported using denominators of 133 and 67, rather than the overall enrollment of 202.0.
- Confidence-interval width: the registry-reported intervals for PFS, OS, and duration of OR are broad enough that substantial uncertainty remains around the corresponding point estimates.
- Proportional-hazards assumption: hazard-ratio interpretation relies on a time-to-event modeling framework. The ClinicalTrials.gov record does not provide diagnostics for assessing proportional hazards.
- Multiplicity: two primary endpoints are registered, but the ClinicalTrials.gov record does not state how type I error was allocated across them or among secondary analyses.
- Missing-data methods: the ClinicalTrials.gov record does not specify imputation procedures or detailed rules for handling missing assessments.
- Interim analysis: no interim-analysis or alpha-spending procedure is provided in the ClinicalTrials.gov record.
- Crossover: no crossover strategy is reported in the ClinicalTrials.gov record, so none is inferred.
- Subgroup interpretation: although PD-L1 status and world region were used for stratification, separate subgroup effects and interaction tests are not reported.
- Duration-of-response conditioning: duration of OR considers participants with an objective response, so its interpretation differs fundamentally from the randomized-population PFS analysis.
21. Why This Trial Matters Statistically
KATE2 is a useful teaching case because its registry data bring together randomized treatment assignment, double blinding, stratified time-to-event analysis, an ITT efficacy population, a primary PFS endpoint, secondary OS and duration-of-response analyses, and a separately defined safety endpoint.
| Concept | How it appears in KATE2 |
|---|---|
| Randomization | The allocation is randomized across 2 parallel treatment arms. |
| Blinding | The study is double-blind. |
| ITT analysis | The PFS, OS, and duration-of-OR analyses use the ITT population as defined in the registry. |
| Time-to-event endpoint | PFS is defined as time from randomization to progression or death. |
| RECIST v1.1 | Investigator tumor assessment using RECIST v1.1 determines progression and objective response-related endpoints. |
| Log-rank test | The primary PFS analysis uses a 2-sided stratified log-rank test. |
| Hazard ratio | PFS, OS, and duration of OR are summarized using hazard ratios. |
| Stratified analysis | World region and PD-L1 status are used as stratification factors. |
| Confidence interval | Each registry-reported efficacy HR is accompanied by a 95% confidence interval. |
| P-value | Formal P-values are posted on ClinicalTrials.gov for the three posted statistical analyses. |
| Safety endpoint | Percentage of participants with adverse events is a registered primary binary endpoint. |
The most important statistical lesson is that a clinical-trial result is not a single number. The endpoint definition, randomized population, analysis method, effect measure, confidence interval, and hypothesis-testing framework all determine what that number actually means.
22. Trial Timeline
Study start
The registry lists September 26, 2016 as the study start date.
Primary completion
The registry lists December 11, 2017 as the primary completion date.
Final registry status
The trial is listed as completed, with results posted on ClinicalTrials.gov.
23. Clinical Interpretation vs Statistical Interpretation
Statistical interpretation
The primary PFS analysis reports HR 0.82 (95% CI 0.55–1.23; P = 0.3332) using a 2-sided stratified log-rank test. The secondary OS analysis reports HR 0.74 (95% CI 0.42–1.30; P = 0.2934), while duration of OR reports HR 1.26 (95% CI 0.52–3.03; P = 0.6099).
Clinical interpretation
The ClinicalTrials.gov record supports describing the estimated relative treatment effects and their uncertainty, but they do not provide enough additional outcome detail to construct median survival comparisons, absolute survival differences, subgroup conclusions, or long-term efficacy claims.
This distinction is essential. Statistical interpretation concerns what the reported estimates and tests say under their stated analytical framework. Clinical interpretation requires considering endpoint definitions, the magnitude and precision of effects, safety, patient population, and other evidence. The ClinicalTrials.gov record supports only the specific conclusions presented on this page.
24. A Practical Reading of the Confidence Intervals
| Endpoint | Point estimate | 95% CI | Null value inside interval? |
|---|---|---|---|
| PFS | HR 0.82 | 0.55–1.23 | Yes |
| OS | HR 0.74 | 0.42–1.30 | Yes |
| Duration of OR | HR 1.26 | 0.52–3.03 | Yes |
For hazard ratios, the null value is 1. All three registry-reported 95% confidence intervals include 1. This is an important descriptive property of the reported intervals. It does not mean that all possible treatment effects are equally plausible, nor does it mean that the point estimate should be ignored. Rather, it indicates that the interval includes the no-difference value as well as a range of alternative values.
The intervals also differ in width. The duration-of-response interval of 0.52–3.03 is particularly broad relative to its point estimate, illustrating why precision matters when interpreting a hazard ratio. A point estimate without its confidence interval can give a misleading impression of certainty.
25. What Is Not Reported in the ClinicalTrials.gov record
| Topic | Availability in the ClinicalTrials.gov record |
|---|---|
| Median PFS | Not reported |
| Median OS | Not reported |
| Arm-specific PFS event counts | Not reported |
| Arm-specific OS event counts | Not reported |
| Time-specific PFS or OS percentages | Not reported |
| Detailed baseline characteristics | Not reported |
| Separate PD-L1 subgroup HRs | Not reported |
| Formal interaction tests | Not reported |
| Interim-analysis procedure | Not reported |
| Multiplicity procedure | Not reported |
| Missing-data/imputation strategy | Not reported |
| Crossover strategy | Not reported |
Keeping these omissions explicit is part of a statistically responsible trial summary. A complete clinical-trial analysis should distinguish between what the registry reports and what a statistician might ordinarily want to know from a full protocol, statistical analysis plan, clinical study report, or publication.
26. Statistical Concepts in This Trial
Learn more about the methods used in this trial:
27. Related Statistical Calculators
28. Sources
- ClinicalTrials.gov: NCT02924883 — KATE2.
- Linked publication: PubMed record for PMID 33002436.
Continue through Clinical Biostats
Explore statistical tutorials, calculators, and additional clinical-trial analyses covering randomized designs, survival analysis, confidence intervals, and clinical-trial methodology.
29. Record Summary
KATE2 provides a useful example of randomized clinical-trial time-to-event analysis. The trial was a completed, randomized, double-blind, parallel phase 2 study with 202.0 enrolled participants and two treatment arms. The primary PFS endpoint was defined from randomization to the first occurrence of disease progression or death and was analyzed in the ITT population using a 2-sided log-rank test stratified by world region and PD-L1 status.
The reported PFS analysis produced an HR of 0.82 with a 95% CI of 0.55–1.23 and P = 0.3332. The secondary OS analysis produced an HR of 0.74 with a 95% CI of 0.42–1.30 and P = 0.2934. The secondary duration-of-objective-response analysis produced an HR of 1.26 with a 95% CI of 0.52–3.03 and P = 0.6099.
The key statistical lesson is to read these estimates together with their endpoint definitions, analysis populations, stratification factors, confidence intervals, and hypothesis-testing methods. The ClinicalTrials.gov record does not provide enough information to reconstruct median survival, absolute survival probabilities, detailed subgroup effects, or the complete multiplicity and interim-monitoring framework, so those conclusions are intentionally not added.