A Study of Cabozantinib (XL184) vs Everolimus in Subjects With Metastatic Renal Cell Carcinoma: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study of Cabozantinib (XL184) vs Everolimus in Subjects With Metastatic Renal Cell Carcinoma (NCT01865747), a randomized phase 3 study. The primary outcome is Progression-free Survival (PFS).

2. Clinical Question

The study also lists overall survival and objective response rate as secondary outcomes. Numerical results are posted for objective response rate, but a numerical overall-survival result is not reported in the posted outcome data. The question addressed by the primary analysis is therefore about time to progression or death, not overall survival, quality of life, or every possible benefit or harm. The randomized comparison supports interpretation of the groups as assigned, but the registry record does not give enough detail to determine all analysis-population rules or the handling of missing observations.

3. Trial Design

This was a Phase 3, randomized, controlled, parallel-group study with no masking reported. It enrolled 658 participants with renal cell carcinoma. The registry record gives a start date of 2013-06, a primary completion date of 2015-05-22, and a completed status. Exelixis is identified as the sponsor.

The two interventions were oral tablets taken once daily: cabozantinib (XL184) at 60 mg and everolimus (Afinitor) at 10 mg. The comparison addressed disease that had progressed after prior VEGFR tyrosine kinase inhibitor therapy. Random allocation is an important design feature because, in expectation, it helps make the groups comparable at the start of a trial. The record does not describe the allocation sequence, concealment procedures, or any stratification factors, so further details about how randomization was implemented are not reported in the registry record.

The primary endpoint was progression-free survival. The pre-specified primary analysis was based on the first 375 randomized subjects, with 187 in the cabozantinib group and 188 in the everolimus group. The registry record also describes assessment of PFS for up to 17 months in its timeframe text. The primary-analysis description says the endpoint was assessed by investigator according to RECIST 1.1, whereas the timeframe text names an Independent Radiology Committee. The registry record does not reconcile the assessor descriptions.

Objective response rate is a posted secondary outcome, with assessment scheduled at 8 weeks after randomization, every 8 weeks for 12 months, and every 12 weeks thereafter until progression or death, up to May 2015 (approximately 21 months). Overall survival is also listed as a secondary outcome, but its numerical result is not reported in the posted outcome data. The registry record does not report the detailed protocol, the sample-size justification, or the full statistical analysis plan.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Overall Study (ITT)

Reasons for not completing this period:

Primary Intent to Treat (PITT)

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: The pre-specified primary analysis of PFS was based on the first 375 randomized subjects (187 cabozantinib and 188 everolimus).

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

6. Primary Endpoint

The registry lists the primary outcome as Progression-free Survival (PFS), measured over PFS is measured from the date of randomization until the date of first documented disease progression or date of death from any cause as determined by the Independent Radiology Committee (IRC) per RECIST 1.1, assessed for up to 17 months..

Progression-free survival measures the time from randomization until disease progression is documented or death occurs, whichever happens first. The endpoint combines progression and death into a single event definition. This allows an analysis to include the timing of events, rather than only whether a participant had experienced an event by a chosen assessment date. In oncology trials, PFS is used to describe how long participants remain alive without recorded disease progression under the trial's assessment rules.

The registry record describes PFS in two related ways. Its timeframe text names progression assessed by the Independent Radiology Committee under RECIST 1.1, while the primary-analysis description names investigator-determined tumor progression under RECIST 1.1 or death. Both descriptions identify randomization as the starting point and death as an event, but the assessor differs. The record does not explain whether these refer to separate assessments or clarify which description governs the posted primary estimates.

The reported group values are medians estimated with Kaplan-Meier analysis. A median is the estimated time by which half of the modeled event-time distribution has experienced the endpoint; it is not the average time for every participant, nor does it mean that each participant had the same course. The confidence intervals communicate uncertainty around those estimates. The hazard ratio provides a comparison using event timing across follow-up, rather than comparing only the two medians. These measures answer related but different questions, so neither should be treated as a complete account of all participant outcomes.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The registry reports a Kaplan-Meier analysis for the primary PFS endpoint and a log-rank comparison expressed as a hazard ratio. It also reports a Cochran-Mantel-Haenszel analysis for objective response rate. These methods address different outcome types. The registry record does not include enough detail to reconstruct every model specification or analysis decision.

What does Kaplan-Meier estimation do?

In general, Kaplan-Meier estimation describes the probability of remaining free of the endpoint over time while accounting for participants whose observed follow-up ends before an event is recorded. Such observations are commonly called censored: the analysis uses their known time under observation without treating them as if an event had occurred at the end of follow-up. The method relies on assumptions about censoring, including that the reason follow-up ends does not systematically distort the event pattern after accounting for the analysis. The registry record does not report how censoring was handled in detail or provide a description of follow-up completeness.

A Kaplan-Meier curve can be used to estimate a median event time, defined where the estimated event-free probability reaches the midpoint. The posted medians are 7.4 months for cabozantinib and 3.8 months for everolimus. A median condenses a time-to-event distribution to a single point; it does not describe the full shape of the curve. Participants can have varied event times, and uncertainty in the median is reflected by the reported 95% confidence intervals. The record gives intervals of 5.6 to 9.1 months and 3.7 to 5.4 months, respectively.

What does the log-rank comparison represent?

In general, the log-rank test compares event experience between groups across follow-up, using the ordering of event times rather than a comparison limited to a single time point. The reported hazard ratio is 0.58, with a 95% confidence interval of 0.45 to 0.74 and p < 0.001. A hazard ratio summarizes the relative event hazard under a model or test framework; it is not a direct measure of the difference in median months. The registry record does not state whether the proportional-hazards assumption was assessed or report additional model diagnostics.

A confidence interval conveys the range of values compatible with the data under the procedure's assumptions and indicates statistical precision. It is not a guarantee that the true effect lies inside the interval. A p-value evaluates how inconsistent the observed data are with a specified null hypothesis, assuming that hypothesis and the test assumptions. It does not express the probability that the null hypothesis is true, nor does it determine the practical importance of an effect.

Why can over-dispersion matter?

In general, over-dispersion occurs when observed variability in count data is greater than a model's assumed variability. If a model does not account for that extra variation, standard errors can be too small and confidence intervals or tests can appear more certain than warranted. This is a general statistical consideration, not a claim that over-dispersion was present in this trial. The posted primary endpoint is time-to-event PFS, and the registry record does not report an over-dispersion analysis.

How should the response-rate comparison be read?

Objective response rate is a binary outcome: each participant is classified according to whether a response meeting the study's criteria was observed. The posted group percentages are 17.0% for cabozantinib and 3.0% for everolimus. The registry reports a Cochran-Mantel-Haenszel p-value of p < 0.001. In general, this method compares proportions while accounting for strata, if strata are specified. The registry record does not identify the strata or give the full test specification. The recalculated unadjusted comparison is a simple comparison of posted summary counts and is not the same as an adjusted or stratified analysis.

The planning illustrations associated with the recalculated response-rate comparison are hypothetical, not observed trial results and not the trial's own design calculation. They use a two-sided significance level of 0.05, equal allocation, and a normal approximation. For planning for 100% of the observed effect, the illustration gives 72 participants per arm for 80% power and 96 participants per arm for 90% power. For planning for 75% of the observed effect, it gives 149 participants per arm for 80% power and 198 participants per arm for 90% power. For planning for 50% of the observed effect, it gives 377 participants per arm for 80% power and 505 participants per arm for 90% power. These are planning scenarios only; they are not observed power or evidence about the study's original sample-size rationale. The rationale is not reported in the registry record.

What does randomization contribute?

In general, randomization reduces the risk that treatment assignment is systematically linked to measured or unmeasured prognostic factors. It supports a causal interpretation of between-group comparisons when conduct, follow-up, and analysis preserve the intended comparison. Randomization does not eliminate uncertainty, missing data, measurement differences, or limitations in generalizability. The registry describes the study as randomized and unmasked, but does not report allocation concealment or the complete analysis-population rules. Those missing details limit how completely the statistical methods can be assessed from the registry record alone.

10. Analyses Reported by the Sponsor

The sponsor-reported primary comparison is a log-rank analysis with a hazard ratio of 0.58, a 95% confidence interval of 0.45 to 0.74, and p < 0.001. The posted PFS medians are 7.4 months for cabozantinib and 3.8 months for everolimus. The registry marks the primary comparison as a superiority analysis. No recalculated PFS comparison is presented, so the sponsor's time-to-event analysis is the reported inferential result for the primary endpoint.

For objective response rate, the sponsor-reported Cochran-Mantel-Haenszel result is p < 0.001. A separate recalculated, unadjusted comparison gives p < 0.001. These are not identical analyses: the registry names a stratified method for the sponsor-reported test, whereas the recalculation uses posted summary counts without that adjustment. Their p-values therefore should not be treated as interchangeable. The record marks no discrepancy, so there is no stated conflict that would call the sponsor's own analysis into question. If a discrepancy were present, the sponsor's own analysis would take precedence over a simple recalculation.

Recalculations based on summary statistics are useful as checks, but cannot reproduce analyses that depend on unreported individual-level data, strata, censoring, covariates, or other specifications. The registry record does not give enough detail to verify the full sponsor analysis independently.

11. Multiplicity, Missing Data and Analysis Populations

Multiplicity arises when multiple outcomes, comparisons, time points, or analyses are tested. In general, the more opportunities there are to find a small p-value, the greater the chance of at least one false-positive result if every test is interpreted independently. Statistical plans can address this through a hierarchy of testing, adjusted significance thresholds, or other procedures. The registry record does not report whether any such strategy was used for this study, how the primary and secondary outcomes were ordered, or whether interim analyses were planned or conducted. These details are not reported in the registry record, so the posted p-values should not be assumed to account for unreported testing.

The record defines the primary PFS analysis population as the first 375 randomized subjects, comprising 187 assigned to cabozantinib and 188 to everolimus. It does not state the complete rules for inclusion in each analysis, the reason for this primary-analysis population, or whether other PFS populations were analyzed. The response-rate outcome has its own posted group sizes, which differ from those stated for the primary analysis. This should be understood as a difference in the reported analysis populations, not as evidence of a particular reason for that difference.

Missing-data handling is also not described in enough detail to assess its potential influence. For time-to-event outcomes, incomplete follow-up may lead to censoring; for binary response outcomes, missing assessments can affect classification or inclusion. These are general considerations. The registry record does not report the amount or pattern of missing data, the assumptions used, or sensitivity analyses addressing missingness. It also does not report whether analyses followed an intention-to-treat principle or another population definition.

The primary endpoint timeframe names Independent Radiology Committee assessment, while the primary-analysis description names investigator assessment. This difference matters because endpoint classification can depend on who evaluates progression. The registry record does not explain whether the posted result is based on one assessor, both assessments, or distinct analyses. The sponsor-reported and recalculated response-rate p-values also use different methods, and the record does not provide full details of the stratification. These reporting gaps limit independent assessment of multiplicity, missingness, and analysis-population choices.

12. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: pneumonia (4.0%), dyspnoea (4.0%), anaemia (3.7%), renal cell carcinoma (3.4%), abdominal pain (3.0%).

The registry reports serious adverse events among 131 of 331 participants at risk in the cabozantinib group and 139 of 322 at risk in the everolimus group. For other adverse events, it reports 328 of 331 at risk and 270 of 322 at risk, respectively. These are separate summaries: the serious-event counts should not be combined with the other-adverse-event counts. The registry record does not report deaths in the adverse-event data provided.

The listed serious adverse-event terms and percentages include pneumonia (4.0%), dyspnoea (4.0%), anaemia (3.7%), renal cell carcinoma (3.4%), abdominal pain (3.0%), and pleural effusion (3.0%). These entries describe reported terms; they do not establish that the study treatment caused them. The record does not provide enough detail to assess event severity beyond the serious-event classification, timing, recurrence, exposure duration, or the statistical uncertainty around the group figures.

Adverse-event comparisons require care because participants may differ in duration of exposure and observation, and because event definitions and reporting practices affect counts. The registry record does not give those details in this adverse-event summary. The figures therefore describe reported events and participants at risk, but are not a complete comparative safety analysis or a basis for estimating individual risk.

13. What the Result Does, and Does Not, Mean

The posted result describes a randomized comparison of cabozantinib and everolimus on the primary endpoint of PFS in people with advanced renal cell carcinoma after prior VEGFR tyrosine kinase inhibitor therapy. The reported medians, hazard ratio, confidence interval, and log-rank p-value indicate that the registry's stated primary analysis found a difference in PFS between the groups. The figures characterize the trial analysis; they do not predict with certainty what will happen to any individual participant.

A statistical result is not by itself a judgment about clinical importance. The hazard ratio summarizes relative event experience under the analysis, while the medians express estimated time to the endpoint. Neither gives a full account of symptom burden, quality of life, treatment burden, or every possible outcome. The registry record does not report those measures in the information available for this analysis. It also does not provide a numerical overall-survival result, despite listing OS as a secondary outcome.

The result does not show that every participant in one group had a longer PFS than every participant in the other group. Group summaries describe distributions, and individual outcomes vary. Nor does a p-value quantify the probability that the result is due to chance, the probability that a treatment effect is true, or the size of a benefit. Confidence intervals communicate statistical uncertainty under the analysis assumptions, but do not account for every possible limitation in study conduct or reporting.

The interpretation is bounded by the stated primary-analysis population, the mismatch between the two assessor descriptions, and the registry's lack of detail on missing data, multiplicity, and full analysis rules. Secondary results can add context but do not answer the same question as PFS. The safety summaries also require a separate interpretation and do not establish causation. These results should therefore be read as evidence about the posted trial comparisons and outcomes, not as a complete account of all outcomes or a universal conclusion about treatment.

14. Why This Trial Matters Statistically

This trial is statistically notable because it reports a randomized, parallel-group comparison of two treatments for a defined population and uses a time-to-event endpoint as its primary outcome. Randomization supports a fairer comparison than an uncontrolled observation, while Kaplan-Meier estimation and the log-rank approach make use of event timing and incomplete follow-up in ways that simple end-of-study counts do not. The registry reports both group medians and a hazard ratio, allowing readers to see distinct summaries of the PFS result.

The result also illustrates why endpoint definitions and analysis populations matter. The primary analysis is explicitly based on the first 375 randomized subjects, while the assessor is described differently in the timeframe and analysis text. Those are not minor presentation details: the population determines which observations contribute, and assessment rules can determine when progression is counted. The registry record does not clarify these points fully, which limits independent evaluation even though the headline comparison is reported.

Secondary outcomes add another statistical lesson. A response-rate result can align in direction with a PFS result while answering a separate question. Sponsor-reported and unadjusted recalculated tests can also differ in method, so their p-values must be read with the method attached rather than treated as interchangeable. With no reported multiplicity plan and no numerical OS result, the primary PFS finding carries the clearest inferential role among the posted outcome results.

Finally, the record shows why precision and significance are not the same as completeness. Confidence intervals help characterize uncertainty, while missing statistical-plan details, missing-data rules, and safety-context information remain relevant. A careful reading separates what the trial reports from what the record leaves unresolved.

15. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

Several limitations are specific to the registry information for this trial. The primary endpoint descriptions differ on who assessed progression: the timeframe names the Independent Radiology Committee, while the primary-analysis description names the investigator. The record does not clarify the relationship between these assessments or identify which one underlies the posted estimate.

The primary PFS analysis is described as using the first 375 randomized subjects, rather than giving a complete account of the analysis population for every endpoint. The record does not explain the basis for this population, report all population definitions, or describe whether sensitivity analyses were performed. The response-rate outcome has separately reported group sizes, but the registry does not explain the difference in analysis populations.

Statistical-plan details are limited. The registry record does not report a multiplicity strategy, interim-analysis plan, sample-size justification, detailed missing-data rules, or full model specifications. It does not report whether the proportional-hazards assumption was assessed. These omissions prevent a full assessment of how all analyses were planned and how robust the findings are to alternative assumptions.

Overall survival is listed as a secondary outcome but has no numerical result in the posted outcome data. The safety summary gives serious and other adverse-event counts and selected serious terms, but does not provide enough context about exposure time, event timing, or comparative uncertainty for a complete safety assessment. The record also does not provide numerical results for every listed outcome measure. These are limitations of the information reported in the registry record; they do not establish that analyses or information were absent from other study materials.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: A Study of Cabozantinib (XL184) vs Everolimus in Subjects With Metastatic Renal Cell Carcinoma — NCT01865747.
• Registry record last updated 2021-04-27; status last verified 2021-04.
• Results first posted 2017-07-18.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

19. Record Summary

The registry record describes a completed, Phase 3, randomized, unmasked, parallel-group study sponsored by Exelixis. It enrolled 658 participants with renal cell carcinoma after prior VEGFR tyrosine kinase inhibitor therapy. Cabozantinib (XL184) was given as a 60 mg tablet once daily and everolimus (Afinitor) as a 10 mg tablet once daily.

The primary outcome was PFS. The reported medians were 7.4 months for cabozantinib and 3.8 months for everolimus; the log-rank comparison reported a hazard ratio of 0.58, a 95% confidence interval of 0.45 to 0.74, and p < 0.001. The stated primary analysis used the first 375 randomized subjects. The record's descriptions differ on whether progression assessment was by an Independent Radiology Committee or investigator.

Objective response rate is a posted secondary outcome, with a sponsor-reported Cochran-Mantel-Haenszel p-value of p < 0.001. Overall survival is listed as a secondary outcome, but its numerical result is not reported in the posted outcome data. Important details about multiplicity, missing-data handling, and complete analysis populations are also not reported in the registry record.