A Phase 2 Study of Cabozantinib in Japanese Participants With Advanced Hepatocellular Carcinoma: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Phase 2 Study of Cabozantinib in Japanese Participants With Advanced Hepatocellular Carcinoma (NCT03586973), a non-randomized phase 2 study. The primary outcome is 24-Week Progression-Free Survival Rate (PFSR).

2. Clinical Question

3. Trial Design

This was a Phase 2, open-label, non-randomized study of cabozantinib in Japanese participants with advanced hepatocellular carcinoma who had received prior systemic anticancer therapy. The study was completed. The registry lists enrollment of 34 participants and identifies Takeda as the sponsor. The recorded study start was 2018-08-06, and the primary completion date was 2019-07-17.

The study had two cohorts, both receiving cabozantinib 60 mg as an oral tablet once daily in the fasted state. Cohort A consisted of participants who had received first- or second-line anticancer therapy with sorafenib. Cohort B consisted of participants who had not received first- or second-line anticancer therapy with sorafenib. Treatment was described as continuing for up to approximately 2 years. The registry record does not report the actual treatment duration for participants.

Open-label means that treatment masking was not used. Non-randomized allocation means participants were not assigned to cohorts by chance. Cohort membership reflected prior sorafenib treatment, which is an important difference in participant history. The cohorts therefore do not form a randomized treatment-versus-control comparison, and a difference between their results would not, by itself, establish a treatment effect.

The registry lists progression-free survival, objective response rate, disease control rate, and overall survival as secondary outcomes. It records five outcome measures, but the statistical analysis plan, the details of any additional measure, and the observed primary efficacy result 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.

Before assignment: Participants with a diagnosis of advanced hepatocellular carcinoma (HCC) were enrolled in this study in two Cohorts A and B. Participants who received prior sorafenib were enrolled in Cohort A and participants who received prior systemic anticancer therapy other than sorafenib were enrolled in Cohort B.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Full Analysis Set (FAS) included participants who received at least one dose of study drug.

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.

Population: Full Analysis Set (FAS) included participants who received at least one dose of study drug.

6. Primary Endpoint

The registry lists the primary outcome as 24-Week Progression-Free Survival Rate (PFSR), measured over Week 25 Day 1 plus 7 days.

The primary endpoint was the 24-week progression-free survival rate: the percentage of participants whose progression-free survival lasted at least 24 weeks. The assessment timeframe was Week 25 Day 1 plus 7 days. Progression-free survival was defined as the time from the first day of study drug administration to the earlier of progressive disease or death from any cause. The primary population was the Full Analysis Set, comprising participants who received at least one dose of study drug.

Disease progression was assessed using RECIST 1.1. For target lesions, the registry describes progression as at least a 20% increase in the sum of diameters from the smallest recorded sum, with an absolute increase of at least 5 mm as well. Progression could also be identified through unequivocal progression of nontarget lesions or a qualifying new lesion. The registry notes that modest enlargement of a nontarget lesion alone is not usually sufficient to establish unequivocal progression.

A fixed-time progression-free survival rate expresses the proportion meeting a defined time threshold; it is not the same as a measure of tumor shrinkage. It incorporates both disease progression and death as events. The registry specifies a 90% confidence interval for the endpoint, but the observed rate and interval are not reported in the registry record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The primary outcome is a binary percentage: participants either met the definition of progression-free survival for at least 24 weeks or did not. In general, a percentage summarizes how common an outcome was within a specified analysis population. Its meaning depends on a clear numerator, denominator, follow-up definition, and approach to participants whose outcome status cannot be determined. The registry describes the endpoint and identifies the Full Analysis Set as participants who received at least one dose, but does not report the observed percentage or detailed handling of missing outcome information.

What does a confidence interval convey?

The registry specifies a 90% confidence interval for the primary endpoint. In general, a confidence interval communicates the statistical uncertainty around an estimate under the method used to calculate it. A narrower interval indicates greater precision under that method; a wider interval indicates less precision. It does not show the range of outcomes for individual participants, and it does not remove sources of bias such as non-random cohort assignment or incomplete assessment. The registry record does not report the confidence-interval limits or the calculation method, so its width and the precision of the result cannot be evaluated.

How should the endpoint definition be read?

Progression-free survival is a time-to-event concept: time is measured from the first day of study drug administration until progressive disease or death from any cause, whichever occurs earlier. The primary endpoint then reduces that time-to-event information to whether progression-free survival reached the specified 24-week threshold. This fixed-time percentage is not interchangeable with a median progression-free survival time, a tumor response percentage, or overall survival. The registry lists those other outcomes separately, but does not report their estimates.

Progressive disease was defined with RECIST 1.1 criteria. For target lesions, the stated rule includes a relative increase from the smallest sum of diameters and an absolute increase threshold. Nontarget lesion progression and new lesions are also considered, with the registry noting that modest growth of a nontarget lesion alone is generally not enough for unequivocal progression. These operational rules matter because the classification of progression determines whether a participant meets the primary endpoint. The record does not give further details about image review, assessment schedules beyond the primary endpoint timeframe, or whether assessments were centrally reviewed.

Why does allocation affect interpretation?

This is a non-randomized study, and the cohorts were formed according to prior sorafenib treatment. In general, random allocation helps balance measured and unmeasured participant characteristics across treatment groups, on average. Without randomization, cohort differences may reflect treatment history or other differences, as well as any differences associated with the study treatment. Because both cohorts received cabozantinib and neither is a randomized control group, their comparison cannot isolate the effect of cabozantinib against no cabozantinib or another treatment.

The open-label design means masking was not used. In general, knowledge of treatment can influence some aspects of care, reporting, or outcome assessment, although the extent depends on the outcome and procedures. The registry record does not report additional safeguards or assessment procedures that might reduce such influence. The RECIST definition makes the progression criteria explicit, but a definition alone does not establish how assessments were conducted.

What do the planning calculations represent?

Any sample-size calculations associated with the posted results are planning illustrations based on the observed effect, not the study's original design or observed power. The stated illustration assumptions are a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation; the effect to detect is expressed as a fraction of the observed effect. Such calculations ask how many participants per arm would be needed under those hypothetical assumptions. They do not describe the actual allocation in this non-randomized study, and they do not demonstrate that the study had a particular probability of detecting an effect. The registry record does not report the original sample-size justification or statistical plan.

In general, a planning calculation is sensitive to the assumed effect and the method used. If a planning illustration is based on an observed effect, it should not be read as independent evidence that the effect is real or reproducible. The record does not report a primary endpoint estimate here, so no conclusion about an effect size or the precision of such a planning exercise can be drawn from the registry record.

Why can percentages require careful interpretation?

For a binary endpoint, the observed percentage depends on who is included and how an outcome is classified. Here, the endpoint population is the Full Analysis Set, while the registry record does not describe all details of the analysis denominator or missing assessments. General statistical practice distinguishes an observed percentage from an estimate adjusted for incomplete follow-up or censoring. Those approaches can yield different interpretations, and no particular missing-data method should be assumed without a report of the analysis.

Time-to-event endpoints also involve timing: participants may have different lengths of observation, and some may not experience an event before observation ends. The fixed-time rate is designed to summarize status at a specified threshold, but the method for estimating it and dealing with incomplete follow-up is not reported. The primary endpoint's 90% confidence interval would ordinarily convey uncertainty around the estimated rate, but without the estimate, interval bounds, and method, that uncertainty cannot be described quantitatively.

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("Cohort A: Cabozantinib 60 mg", "Cohort B: Cabozantinib 60 mg")
n      <- c(20, 14)
events <- c(12, 2)
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  Cohort A: Cabozantinib 60 mg  20  12  60.0  36.1  80.9  38.7  78.1  38.4  78.2
ARM  Cohort B: Cabozantinib 60 mg  14  2  14.3  1.8  42.8  4.0  39.9  4.3  40.5

11. Multiplicity, Missing Data and Analysis Populations

The registry record lists progression-free survival, objective response rate, disease control rate, and overall survival as secondary outcomes, in addition to the primary 24-week progression-free survival rate. It records five outcome measures, but does not report a multiplicity strategy, a hierarchy for testing outcomes, or whether any outcomes were treated as exploratory. When several outcomes are examined, general statistical practice considers whether the chance of a false-positive finding is controlled across those tests. No particular adjustment or interpretation strategy can be attributed to this study from the registry record.

The statistical analysis plan is not reported in the registry record. Details not reported include the planned statistical tests, the approach to confidence intervals, analysis populations for the secondary outcomes, and whether any interim analyses were planned or conducted. The record identifies the primary Full Analysis Set as participants who received at least one dose of study drug, but does not specify how missing or unevaluable endpoint assessments were handled. It also does not report how participants who discontinued treatment or had incomplete follow-up contributed to the endpoint analysis.

In general, missing outcome information can affect a percentage if it changes which participants are included or how their status is classified. For time-to-event outcomes, incomplete follow-up can also affect estimation. These are general considerations, not claims about what happened in this study. Without the analysis rules and outcome data, the direction or size of any effect of missing information cannot be determined.

The registry's failure to report these details does not establish that a particular method was or was not used. It means the available record does not permit independent evaluation of multiplicity, missing-data handling, or the full analysis strategy.

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 5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: pyrexia (7.1%), cholangitis (7.1%), hepatic encephalopathy (7.1%), gastrointestinal ulcer (5.0%), ileus (5.0%).

The registry lists serious adverse event counts separately for Cohort A and Cohort B. It reports 7 of 20 participants in Cohort A and 2 of 14 participants in Cohort B as affected by serious adverse events. For other adverse events, the recorded figures are 20 of 20 in Cohort A and 14 of 14 in Cohort B. Deaths are listed as 15 in Cohort A and 8 in Cohort B. These are descriptive registry counts, not a randomized safety comparison.

The registry also identifies selected serious adverse event terms with percentages: pyrexia, cholangitis, and hepatic encephalopathy at 7.1% each; gastrointestinal ulcer and ileus at 5.0% each; and biloma at 5.0%. These term percentages should not be treated as a complete account of all events or as evidence that an event was caused by cabozantinib. The record does not report the attribution criteria or provide a comparison group not receiving study drug.

Serious adverse events, other adverse events, and deaths are distinct categories in the record; the counts should not be added together. A count indicates how many participants were recorded in the relevant category, but does not by itself describe event duration, severity beyond the category definition, or relationship to treatment. Cohort differences are also difficult to interpret because cohort membership was based on prior sorafenib treatment rather than random assignment.

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

The study addresses cabozantinib use in Japanese participants with advanced hepatocellular carcinoma who had received prior systemic anticancer therapy. It provides a defined primary endpoint: the percentage with progression-free survival lasting at least 24 weeks. It also defines progression using RECIST 1.1 and specifies that death from any cause counts as an event for progression-free survival. These details clarify what the intended efficacy measure represents.

The registry record does not report the observed primary endpoint rate or its 90% confidence interval. It therefore does not establish what proportion of participants reached the 24-week threshold or how precise that estimate was. The record also does not provide numerical results for the secondary efficacy outcomes. Without those results, no conclusion about the magnitude of observed efficacy can be drawn from the registry record.

The study does not include a randomized control group. Both cohorts received cabozantinib, and cohort assignment reflected prior sorafenib treatment. As a result, the study cannot provide a randomized comparison of cabozantinib against another treatment or no treatment. Nor can a difference between the cohorts be interpreted as a causal effect of prior sorafenib or of cabozantinib.

The safety entries describe events and deaths recorded within the cohorts. Such counts are useful descriptive information, but they do not by themselves determine whether an event was treatment-related, how risk compares with another treatment, or whether the benefits and harms favor one option. Those judgments would require additional data and study design features not reported in the record.

In short, the record makes the population, treatment, cohort definitions, endpoint, and selected safety observations clear. It does not report the primary numerical efficacy result, detailed statistical analysis, or randomized comparative evidence. Any interpretation should remain within those limits.

14. Why This Trial Matters Statistically

This trial is statistically notable for the distinction between a clearly specified endpoint and a reported endpoint result. The registry describes a fixed-time progression-free survival rate and defines the event criteria in detail, including progressive disease and death from any cause. Yet the observed rate and its 90% confidence interval are not reported in the registry record. Endpoint definition is essential for interpreting an estimate, but it cannot substitute for the estimate itself.

The cohort structure also illustrates why a treatment study is not automatically a treatment comparison. The study was non-randomized, open-label, and organized into cohorts according to prior sorafenib treatment. Although both cohorts received cabozantinib, the cohorts differ by treatment history and do not provide a like-for-like control comparison. This design can describe outcomes in the treated study population, but it cannot isolate a treatment effect through a randomized contrast.

The record also separates a primary endpoint from secondary outcomes, including progression-free survival, objective response rate, disease control rate, and overall survival. When multiple outcomes are considered, the analysis plan and multiplicity approach help readers understand how much weight to place on individual results. Those details are not reported in the registry record, as are the primary efficacy estimate and its interval.

Finally, the safety counts demonstrate the value and limits of descriptive reporting. Cohort-specific counts and selected event percentages convey what the record documents, while not establishing causality or comparative risk. Reading the design, endpoint, analysis information, and safety observations together prevents the common mistake of treating a single-arm study as though it were a randomized efficacy trial.

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 specific to this study constrain interpretation. The study was non-randomized, open-label, and had no randomized control group. Participants entered Cohort A or Cohort B according to prior first- or second-line sorafenib treatment, so the cohorts differ by a clinically relevant element of treatment history. The record itself cautions that the arms are not a like-for-like control comparison. Any between-cohort difference would therefore be descriptive rather than a clean estimate of a treatment effect.

The registry lists enrollment of 34 participants, but the primary endpoint estimate and its 90% confidence interval are not reported. The numerical efficacy result and its precision therefore cannot be evaluated from the record. Although the registry defines the Full Analysis Set as participants who received at least one dose, it does not report the number analyzed for the primary endpoint or explain the handling of missing assessments.

The record does not report the statistical analysis plan, the method used to calculate the primary confidence interval, a multiplicity approach, or whether interim analyses were planned or performed. It also does not report detailed assessment procedures, including whether progression review was independent or centrally reviewed. These omissions limit assessment of how the endpoint definition was implemented and how robust the analysis may be.

The treatment description allows therapy for up to approximately 2 years, but actual treatment exposure and follow-up are not reported. Safety counts are descriptive, with no randomized comparator and no reported attribution criteria. Together, these limitations mean the registry record supports description of the study setup and recorded safety information, but not a full evaluation of comparative efficacy or causal safety effects.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: A Phase 2 Study of Cabozantinib in Japanese Participants With Advanced Hepatocellular Carcinoma — NCT03586973.
• Registry record last updated 2023-04-18; status last verified 2023-04.
• Results first posted 2023-04-18.
• 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

NCT03586973 was a completed Phase 2, open-label, non-randomized study of cabozantinib in Japanese participants with advanced hepatocellular carcinoma after prior systemic anticancer therapy. The registry lists enrollment of 34 participants. Both cohorts received cabozantinib 60 mg orally once daily in the fasted state, with treatment described as lasting up to approximately 2 years. Cohort membership depended on prior first- or second-line sorafenib treatment.

The primary outcome was the percentage with progression-free survival of at least 24 weeks, assessed at Week 25 Day 1 plus 7 days, with a 90% confidence interval specified. The primary endpoint estimate and interval are not reported in the registry record. Secondary outcomes listed are progression-free survival, objective response rate, disease control rate, and overall survival.

The registry reports serious adverse event counts, other adverse event counts, deaths, and selected serious event terms. These entries are descriptive; the non-randomized cohorts do not provide a like-for-like control comparison. The statistical analysis plan, missing-data methods, and multiplicity approach are not reported in the registry record.