Trial of Cabozantinib (XL184) in Non-Small Cell Lung Cancer With Brain Metastases: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Trial of Cabozantinib (XL184) in Non-Small Cell Lung Cancer With Brain Metastases (NCT02132598), a na phase 2 study. The primary outcome is Overall Response.

2. Clinical Question

3. Trial Design

This was an open-label, single-arm phase II clinical trial of cabozantinib in patients with previously treated non-small cell lung cancer and brain metastases, including patients with and without C-MET amplification. The registry lists the study as terminated and records enrollment of 5 participants. The start is listed as 2015-12, and primary completion as 2019-08-09. The registry does not report why the trial was terminated.

With only one treatment group, the study was designed to assess outcomes among participants receiving cabozantinib rather than to estimate a randomized treatment contrast. Allocation was listed as not applicable, and masking as none. This means that neither participants nor study personnel were masked to treatment assignment according to the registry entry. The record does not describe any control group, crossover, or randomization procedure.

The intervention was cabozantinib at 60 mg orally once daily, in treatment cycles of 4 weeks. Treatment was intended to continue until disease progression, death, or unacceptable adverse events. Eligibility information in the registry describes previously treated non-squamous disease and brain metastases at any time in the treatment history. Clinically asymptomatic untreated brain metastases could be allowed at investigator discretion; participants with prior brain-directed radiation or surgery were to be clinically stable and recovered from those procedures.

The primary outcome was overall response, defined as the proportion of response-evaluable participants whose best response was complete or partial response by RECIST version 1.1, with assessment up to 12 months. Secondary outcomes included disease control rate, progression-free survival, overall survival, time to progression, and adverse-event summaries. The registry lists 9 outcome measures and 0 analyses. A statistical analysis plan, details of endpoint assessment schedules, and a sample-size justification are not reported in the registry record.

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

Analysis population for the primary outcome, as defined in the registry: Patients evaluable for response: Received at least one dose of study drug, and, underwent at least one radiologic scan follow-up.

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 Overall Response, measured over Up to 12 months.

The primary endpoint was the proportion of response-evaluable participants who achieved complete response or partial response as their best response, assessed by RECIST version 1.1 over a timeframe of up to 12 months. The response-evaluable population was defined as participants who received at least one dose of study drug and underwent at least one radiologic follow-up scan. This population definition matters: it is narrower than all enrolled participants, and the number of participants meeting it is not reported in the registry record.

RECIST provides standardized criteria for classifying changes in measurable disease. In the registry description, complete response requires disappearance of all target lesions and non-target lesions and normalization of tumor-marker level. Partial response requires at least a 30% decrease in the sum of the longest diameters of target lesions, using the baseline sum as the reference. The primary measure reduces these response categories to a proportion, rather than reporting a continuous change in lesion size.

As a general statistical concept, a proportion is directly interpretable as the share of an analysis population meeting a prespecified outcome definition. It is useful in an early-phase single-group study because it can describe observed tumor response without implying a comparison against a control group. Its precision depends on the number of response-evaluable participants and on how many meet the response definition. The registry lists a 95% confidence interval as the measure of uncertainty, but does not report the response estimate, the interval, or the response-evaluable denominator. Therefore, neither the estimated response frequency nor its precision can be assessed from the registry record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

Several statistical ideas help frame the registry information, even though the primary outcome estimate is not reported. The trial uses a single-group design, a binary response endpoint, and a confidence interval as the stated measure of uncertainty. The outcome summaries describe observed participants; they do not create a randomized comparison.

What does a response proportion represent?

A proportion is the number of participants meeting a defined outcome divided by the analysis population for that outcome. In this trial, the intended analysis population comprised participants who received at least one dose and had at least one radiologic follow-up scan. The outcome counted a complete response or partial response as best response under RECIST version 1.1. The registry reports neither the number of response-evaluable participants nor the number with a response, so the proportion cannot be reconstructed.

It is important not to substitute total enrollment for the endpoint denominator without evidence that every enrolled participant met the response-evaluable criteria. The trial enrolled 5 participants, but the endpoint population may be defined differently. The record does not report whether all enrolled participants had a dose and follow-up scan, or how participants without an evaluable scan were handled.

How should a confidence interval be read?

General statistical explanation: a confidence interval expresses uncertainty around an estimated quantity under the method used to construct it. For a response proportion, the interval communicates how much imprecision remains given the observed data and the chosen calculation method. An interval is not a guarantee that the true value lies within its limits, and it does not correct for bias from a non-randomized design, selective follow-up, or incomplete outcome reporting.

The registry names a 95% confidence interval for the primary outcome but does not report its limits or the estimate it would accompany. The interval method is also not reported in the registry record. Therefore, it is not possible to evaluate its width, assess whether the method was appropriate to the observed data, or compare its precision with another estimate.

Why does the single-group design matter?

General statistical explanation: a randomized comparison uses concurrent groups to help distinguish treatment effects from differences that would have occurred even without the intervention. A single-group study lacks that contrast. Participant outcomes may reflect treatment, disease course, previous care, eligibility and selection, or other factors. Descriptive outcomes can be informative about what occurred in the enrolled group, but they do not independently identify what would have happened to the same participants under a different treatment.

Here, allocation is listed as not applicable, masking as none, and the model as single group. There is no control arm against which to estimate a treatment difference. The study’s response endpoint is therefore an observed response proportion, not a comparative effect estimate. The registry’s stated primary result is absent, so even the descriptive response frequency is unavailable.

What can be inferred from the adverse-event percentages?

General statistical explanation: event percentages describe the share of a stated at-risk group with an event, according to the registry’s reporting conventions. They should be read alongside the denominator, the event definition, and the observation period. When the group is small, each participant can have substantial influence on the percentage, and estimates can be imprecise. A percentage alone does not show when an event began, how long it lasted, whether it recurred, or whether it was attributed to treatment.

The registry lists 5 participants at risk in the cabozantinib group for the serious and other adverse-event summaries. It lists serious adverse-event terms with percentages, but the registry entries do not include event-level narratives, timing, or the statistical method used for safety summaries. The listed percentages are descriptive and should not be treated as adjusted comparisons or proof of causation.

What does the planning illustration mean?

General statistical explanation: a sample-size planning calculation estimates how many observations might be needed to detect a specified effect under stated assumptions. It is a design aid, not a measure of evidence generated by the completed trial, and it is not observed power. The planning framework associated with these results assumes a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. Its target is stated as a fraction of the observed effect; time-to-event calculations use a total event count, while other calculations express participants per arm.

Those assumptions describe an illustrative two-arm framework, whereas this trial is single-group. They should not be mistaken for the trial’s actual design or its own sample-size justification. The registry does not report that justification. Since the primary response estimate is not reported, the effect-based planning illustration cannot be interpreted as evidence about the trial’s observed efficacy or precision. Any recalculated estimate would also remain distinct from a sponsor analysis, and where a discrepancy is flagged, the sponsor’s own analysis takes precedence.

What analysis details remain unknown?

The registry lists 9 outcome measures and 0 analyses. It does not report the method for constructing the primary confidence interval, an approach to missing outcomes, adjustment for multiple outcomes, or interim analyses. These omissions limit the ability to assess how planned analyses related to the results. They do not establish that such methods were or were not used outside the registry record.

10. Exact Estimates for Each Arm

No two-arm comparison is shown for this trial, but each arm's result can still be summarized with exact interval estimates computed from the posted counts. The Clopper-Pearson interval is exact and conservative; the Wilson interval is a common closer-to-nominal alternative; the Bayesian credible interval uses a uniform Beta(1, 1) prior. For small groups the three can differ noticeably.

R code and output

arm    <- c("Cabozantinib (XL184)")
n      <- c(4)
events <- c(0)
for (i in seq_along(n)) {
  ci <- binom.test(events[i], n[i])$conf.int          # Clopper-Pearson exact interval
  cr <- qbeta(c(0.025, 0.975), 1 + events[i], 1 + n[i] - events[i])   # Bayesian, uniform Beta(1,1) prior
  wi <- (function(x, n) { p <- x / n; z <- qnorm(0.975); d <- 1 + z^2 / n
    c((p + z^2 / (2 * n) - z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d, (p + z^2 / (2 * n) + z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d) })(events[i], n[i])
  cat(sprintf("ARM|%s|%d|%d|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f\n", arm[i], n[i], events[i], 100 * events[i] / n[i],
              100 * ci[1], 100 * ci[2], 100 * wi[1], 100 * wi[2], 100 * cr[1], 100 * cr[2]))
}

ARM  Cabozantinib (XL184)  4  0  0.0  0.0  60.2  0.0  49.0  0.5  52.2

11. Multiplicity, Missing Data and Analysis Populations

The registry lists overall response as the primary outcome and describes 9 outcome measures in total. Secondary outcomes include disease control rate, progression-free survival, overall survival, time to progression, and several adverse-event summaries. The record lists 0 analyses and does not report a statistical plan for controlling multiplicity, selecting among outcomes, or interpreting several tests together. It is therefore not possible to determine whether or how multiplicity was addressed.

General statistical explanation: when several outcomes or statistical tests are examined, the chance of at least one apparently notable result can increase if each is interpreted in isolation. Prespecified priorities and multiplicity procedures can help organize interpretation. The absence of an adjustment description in the registry is not evidence that no adjustment was used; it means the approach is not reported in the registry record.

Missing outcome information is a separate issue. The primary population required at least one dose and at least one follow-up radiologic scan, but the registry does not report the number meeting that definition or how participants without an evaluable scan were handled. It also does not report missing-data methods, reasons for missing assessments, or whether any participants were excluded from specific analyses. Enrollment of 5 should not automatically be treated as the endpoint denominator.

The record does not report interim analyses or a detailed analysis plan. It also does not provide response results, confidence-interval limits, or results for the listed secondary outcomes. These gaps prevent assessment of the analysis population, the treatment of incomplete follow-up, and the relationship between any planned and reported analyses. They should be understood as limits of what is documented in the registry record, not as proof that particular procedures were omitted from the conduct of the study.

12. Safety Results

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

Deaths

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 0% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: fatigue (40.0%), blood bilirubin increased (40.0%), hypertension (40.0%), infections and infestations - other, specify (20.0%), urinary tract infection (20.0%).

The registry lists one cabozantinib group, with 5 participants at risk and 5 affected in the serious adverse-event summary. It also lists 5 at risk and 5 affected in the other adverse-event summary, and reports 4 deaths. The record does not provide further detail in the registry record about the timing of these events, their duration, or the relationship between the deaths and adverse events.

The serious adverse-event terms with the largest listed percentages are fatigue, increased blood bilirubin, and hypertension, each at 40.0%. “Infections and infestations - Other, specify,” urinary tract infection, and increased alkaline phosphatase are each listed at 20.0%. These are registry-reported percentages, not estimates from a comparison with another group. The record does not clarify whether an individual participant could contribute to more than one listed term, or provide event-level counts alongside each percentage.

General statistical explanation: adverse-event summaries are descriptive. A reported event occurring during a trial does not by itself establish that the study drug caused it. Interpretation also depends on definitions, observation time, baseline health, and attribution procedures, details that are not reported in the registry record for these figures. The small reported group makes the percentages sensitive to the experience of individual participants. The figures should therefore be read as a limited account of reported events in this group, not as a comparative estimate or a complete characterization of safety.

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

The registry record documents a phase II, open-label, single-group study of cabozantinib in previously treated non-small cell lung cancer with brain metastases. It reports that 5 participants enrolled and that the trial is terminated. It also provides a primary endpoint definition and selected safety summaries. Those details establish what the trial set out to study and some information recorded about participants, but they do not supply the primary efficacy result.

The intended primary outcome was the proportion of response-evaluable participants achieving complete or partial response under RECIST version 1.1. The registry does not report that proportion, its 95% confidence interval, or the response-evaluable denominator. Accordingly, the record cannot show the observed response frequency or how precisely it was estimated. The endpoint definition is not a substitute for an observed endpoint value.

Even with a reported response estimate, a single-group trial would not offer a randomized comparison. An observed response could be described among participants in the study, but it could not by itself establish what would have occurred under another treatment or no treatment. Differences in patient characteristics, prior care, disease course, and follow-up can influence outcomes. The lack of a concurrent comparator is therefore central to interpretation, rather than a technical detail that can be resolved by a confidence interval.

The safety figures describe events recorded in a small group, including the listed serious adverse-event terms and 4 deaths. They do not establish the cause of any event or death, and they do not allow a comparative estimate of risk. The record does not report detailed timing or attribution information for those terms. Nor does it provide results for the listed disease control, survival, and progression outcomes in the registry record for this analysis.

These limitations do not mean that the trial yielded no useful information; they define what can be concluded from its registry record. The documented design, enrollment, endpoint criteria, and adverse-event summaries are interpretable as descriptions. Claims about response, comparative effectiveness, subgroup outcomes by C-MET amplification, or treatment causality would go beyond the results reported in the record.

14. Why This Trial Matters Statistically

This trial illustrates why design and reporting details are essential to interpreting clinical results. Its primary endpoint is a response proportion, a measure that is readily understandable in principle. But interpretation requires the actual numerator, the analysis denominator, and an uncertainty estimate. The registry defines the endpoint and the response-evaluable population but does not report the response result or confidence interval. Without those pieces, the main efficacy measure cannot be evaluated.

The single-group design is equally important. It can describe outcomes among people who received cabozantinib, but it does not provide a concurrent control group for separating treatment-associated outcomes from other influences. This distinction is fundamental when moving from a description of what happened to a claim about what the treatment caused. Statistical precision, even if available, would not remove the design’s lack of a randomized comparison.

The record also highlights how small enrollment and incomplete endpoint reporting constrain interpretation. It reports enrollment of 5 participants, while the number who met the response-evaluable criteria is not reported. The listed safety percentages are descriptive for a small group and should not be mistaken for stable population-level estimates or comparisons. The registry also lists secondary outcomes without reporting their results in the registry record for this analysis.

Finally, the distinction between a trial’s own design and a hypothetical planning calculation matters. A planning illustration based on a two-arm framework is not the same as the single-group trial that was conducted, and it is not evidence of observed power. Recognizing that distinction prevents a design scenario from being read as a result. Taken together, these points show why sound statistical interpretation depends not only on endpoint definitions, but also on transparent denominators, estimates, uncertainty, analysis methods, and design limitations.

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.

The primary limitation for interpreting efficacy is that the registry does not report the primary overall-response estimate or its 95% confidence interval. It also does not state how many enrolled participants were response-evaluable under the definition requiring a dose and at least one radiologic follow-up scan. Without these items, the primary endpoint cannot be described quantitatively.

The study was single-group and open-label, with no concurrent comparator. Any reported outcome would therefore be descriptive for this cohort rather than a randomized estimate of treatment effect. The absence of masking can also be relevant to outcome assessment, although the registry description does not report how response assessments were conducted or whether they were independently reviewed.

Enrollment was 5 participants. This limited enrollment constrains the amount of information available for estimating response and characterizing uncommon or variable safety outcomes. The trial is listed as terminated, but the reason for termination is not reported in the registry record. The record also does not provide detailed follow-up information, participant-level outcome status, or results for the listed secondary outcomes in the registry record for this analysis.

The intended population included participants with and without C-MET amplification, but no subgroup enrollment or outcome results are reported. The record therefore cannot support comparisons between those groups. It also does not report a sample-size justification, analysis methods for missing data, multiplicity procedures, interim analyses, or the method used to calculate the primary confidence interval. These limitations describe gaps in the registry documentation and should not be treated as proof about procedures that may have been conducted but are not recorded there.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Trial of Cabozantinib (XL184) in Non-Small Cell Lung Cancer With Brain Metastases — NCT02132598.
• Registry record last updated 2021-03-10; status last verified 2021-03.
• Results first posted 2020-09-16.
• 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 describes a terminated, open-label, single-group phase II trial of cabozantinib in previously treated non-small cell lung cancer with brain metastases, with and without C-MET amplification. Enrollment was 5. Cabozantinib was given at 60 mg orally once daily in 4-week treatment cycles, with treatment intended to continue until disease progression, death, or unacceptable adverse events.

The primary endpoint was the proportion of response-evaluable participants with complete or partial response by RECIST version 1.1, assessed up to 12 months. The response-evaluable population required at least one dose and at least one radiologic follow-up scan. The registry record does not report the response proportion, its 95% confidence interval, or the size of that analysis population.

The record lists selected adverse-event summaries, including 4 deaths and serious adverse-event terms with percentages. There was no randomized comparator. The reason for termination, sample-size justification, detailed statistical analysis plan, and results for the listed secondary outcomes are not reported in the registry record.