A Study of Avastin (Bevacizumab) and Xeloda (Capecitabine) in Patients With Advanced or Metastatic Liver Cancer: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study of Avastin (Bevacizumab) and Xeloda (Capecitabine) in Patients With Advanced or Metastatic Liver Cancer (NCT02013830), a non-randomized phase 2 study. The primary outcome is Percentage of Participants With Objective Response (OR).

2. Clinical Question

This study asks whether objective tumor response was observed during treatment with the two-drug regimen in this specific patient group. It does not directly compare that response with an untreated group or with another treatment. Because treatment assignment was not randomized and there was only one group, the study design cannot separate the regimen's effect from other influences on response, such as the natural course of disease or differences among participants.

The primary outcome focuses on confirmed complete response or confirmed partial response. A response had to persist on repeat imaging at least 4 weeks after the first documentation. The study used the best overall response from first drug administration until progressive disease or the end of the study. The response assessments were scheduled at screening, Weeks 6, 12, 18, 24, and 30, and every 3 months through follow-up.

The registry lists disease control, time to disease progression, and overall survival among secondary outcomes. Their result estimates are not reported in the registry record. The primary question is therefore identifiable from the endpoint definition, but the record does not provide the response estimate needed to describe the observed result.

3. Trial Design

This was a completed phase 2 study sponsored by Hoffmann-La Roche. The registry lists the study start as 2005-05 and primary completion as 2008-03. It enrolled 45 participants with advanced or metastatic liver cancer who had not previously received cytotoxic chemotherapy. The intervention combined bevacizumab, marketed as Avastin, with capecitabine, marketed as Xeloda.

The registry identifies the design as single-group, non-randomized, and open-label. There was no masking and no control arm. Each participant was treated with the combination; the record does not describe a separate group receiving a different regimen or no study treatment. The anticipated time on study treatment was 3-12 months, according to the registry summary. This anticipated duration is not an observed treatment-duration result.

The primary outcome was the percentage of participants with objective response, defined as confirmed complete response or confirmed partial response under RECIST. Imaging was scheduled at screening, Weeks 6, 12, 18, 24, and 30, then every 3 months through follow-up. Confirmation required the response to persist on a repeat imaging study at least 4 weeks after it was first documented. Best overall response was assessed from first administration of study drugs until progressive disease or the end of the study.

The registry also names secondary outcomes related to disease control, progression, and survival. It lists eight outcome measures, but no analysis results are reported for them. The registry record does not report a detailed statistical analysis plan, a sample-size justification, interim analysis procedures, or a randomization procedure. The design should consequently be described using the recorded features rather than inferred protocol details.

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

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Per Protocol (PP) Population included participants who: received greater than or equal to (≥) 1 dose of study medication, ≥6 weeks of treatment (unless excluded for allowed reasons), did not severely violate inclusion/exclusion criteria, had tumor assessment, greater than (\>) 50% of first 6 weeks of treatment, and were not replaced.

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: All enrolled participants who received at least 1 dose of study drug.

6. Primary Endpoint

The registry lists the primary outcome as Percentage of Participants With Objective Response (OR), measured over Screening; Weeks 6, 12, 18, 24, and 30; Every 3 months through follow-up.

The primary endpoint is the percentage of participants with objective response. An objective response meant a confirmed complete response or confirmed partial response according to RECIST. Complete response required disappearance of all target lesions and non-target disease, normalization of tumor marker levels, and no new lesions. Partial response required a reduction of at least 30% from baseline in the sum of the diameters of target lesions, with no new lesions; the description also notes that non-target lesions or tumor marker levels could remain abnormal.

Confirmation was intended to distinguish a response that persisted from one seen only at an initial assessment. The response had to remain present on repeat imaging at least 4 weeks after its initial documentation. The endpoint used the best overall response recorded between first drug administration and progressive disease or the end of the study. Assessments were scheduled repeatedly, rather than at a single post-treatment visit.

In general, an objective response endpoint provides a defined way to summarize whether tumors met prespecified imaging criteria during a specified observation period. It is not the same as disease control, which can include outcomes other than response, and it is not equivalent to longer survival or delayed progression. The registry lists the endpoint's unit as a percentage of participants and its dispersion as a 95% confidence interval, but neither the response percentage nor the interval is reported in the registry record. The complete definition of the endpoint population is also not available because the Per Protocol description ends mid-criterion.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The primary endpoint is a proportion: the number of participants meeting the response definition relative to the relevant analysis population. In general, analysis of a proportion begins with a clearly defined numerator, denominator, and observation rule. Here, the observation rule is described through confirmed complete or partial response under RECIST, while the result itself and the complete endpoint population size are not reported in the registry record. A numerical estimate cannot be reconstructed from the endpoint definition alone.

What does a confidence interval convey?

In general, a confidence interval is a range calculated from data using a statistical procedure. It conveys the uncertainty associated with an estimate under that procedure and its assumptions. It does not mean that the true value is guaranteed to lie in the particular interval, and it does not replace consideration of the study design or endpoint definition. The registry lists a 95% confidence interval as the dispersion measure for the response percentage, but reports neither the interval nor the method used to calculate it. The statistical precision of the primary result therefore cannot be evaluated.

Why does the analysis population matter?

The registry describes the primary analysis as Per Protocol. It states that this population included participants who received at least 1 dose of study medication, had at least 6 weeks of treatment unless excluded for allowed reasons, did not severely violate inclusion or exclusion criteria, and had tumor assessment. The description also begins a further condition concerning more than 50% of the first 6 weeks, but its wording is truncated. The complete eligibility rule and the final number of participants in this population are not reported in the registry record.

In general, a Per Protocol analysis restricts attention to participants meeting specified treatment and assessment conditions. This can make the analysis population differ from everyone enrolled. Such a population can answer a question about outcomes among participants meeting those criteria, but its result should not automatically be interpreted as applying to every enrolled participant. Without the final denominator, exclusions, and handling of missing assessments, the primary analysis cannot be independently reconstructed.

How does repeated assessment affect response measurement?

Response was assessed at screening, Weeks 6, 12, 18, 24, and 30, and every 3 months through follow-up. Confirmation required persistence on repeat imaging at least 4 weeks after initial documentation. This rule makes the endpoint more than a single scan result: an initial response must be followed by another assessment that confirms it. The endpoint also uses the best overall response from first drug administration until progressive disease or study end. The registry record does not report how assessments that were missing, delayed, or not evaluable were handled.

In general, repeated assessments create opportunities for the measured status to change over time. The best-response rule summarizes the most favorable qualifying response during the stated period, while a confirmed-response rule additionally requires persistence. Interpretation depends on both the assessment schedule and the rules for confirmation. The registry provides those broad rules, but it does not report details such as adjudication procedures or handling of assessments after treatment discontinuation.

What can a single-group design estimate?

In general, a single-group study can describe outcomes observed among the treated participants. It lacks a concurrent comparison that would show the outcome distribution under a different treatment or under no study treatment. Since this study was non-randomized and single-group, an observed proportion, if reported, would be descriptive for the studied group. It would not provide a randomized treatment contrast. The open-label designation means masking was absent; the record does not describe whether any other safeguards were used for outcome assessment.

What do the planning illustrations mean?

The registry's planning framework describes illustrations based on the observed effect, a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. It specifies that the effect to detect is a stated fraction of the observed effect. For time-to-event outcomes, the framework describes total events; for other outcomes, it describes participants per arm. These are planning illustrations, not the original trial's design justification and not observed power. The record does not report a sample-size justification for the single-group study. Because the observed primary response estimate is not reported, the planning assumptions should not be mistaken for a posted efficacy result or for evidence about the study's achieved power.

In general, planning calculations describe sample sizes or event totals under stated assumptions about an effect and statistical procedure. They do not show what effect was actually observed, and they do not guarantee that a study will reach a particular conclusion. Their interpretation depends on the assumptions and design to which they apply. A two-arm equal-allocation illustration is not the same design as this single-group trial.

The registry names eight outcome measures and reports no analyses. It does not describe an adjustment method for multiple outcomes, the order in which outcomes were tested, interim analyses, or the detailed statistical methods for progression and survival. These methods are not reported in the registry record. General statistical principles can clarify what the endpoints mean, but cannot substitute for unreported analysis procedures or results.

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("Bevacizumab/Capecitabine")
n      <- c(44)
events <- c(4)
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  Bevacizumab/Capecitabine  44  4  9.1  2.5  21.7  3.6  21.2  3.7  21.2

11. Multiplicity, Missing Data and Analysis Populations

The registry lists a primary response outcome and secondary outcomes concerning disease control, time to disease progression, and overall survival. Several secondary outcomes are expressed as percentages with an event or as the proportion event-free at 12 months. Considering multiple outcomes can raise a multiplicity issue: in general, the more statistical questions examined, the greater the chance that at least one result may appear notable by chance if each is assessed without adjustment. Whether that issue applies to formal testing depends on the analysis plan. The registry record does not report a multiplicity strategy, a hierarchy among outcomes, or whether formal hypothesis tests were planned.

The record does not report interim analysis procedures or a detailed statistical analysis plan. It lists zero analyses in its posted analysis information. Thus, there are no posted outcome estimates or tests from which to assess analysis choices, including whether outcomes were adjusted or which population was used for each secondary analysis.

Missing data handling is also not reported in the registry record. The primary endpoint requires tumor assessment and confirmation on repeat imaging, but the record does not explain how participants without adequate scans, participants who discontinued treatment, or assessments that could not be evaluated were handled. The Per Protocol description is incomplete, and the final number analyzed is not given. These omissions prevent assessment of how missing observations or exclusions may have influenced the response estimate.

In general, missing outcome information can affect both the denominator and the interpretation of a proportion. The direction and size of any effect cannot be inferred without knowing why data were missing and how the analysis treated them. No assumption about those procedures should be made for this study.

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 0% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: abdominal pain (4.4%), pyrexia (4.4%), abdominal pain upper (2.2%), ascites (2.2%), diarrhoea (2.2%).

The registry reports 45 participants at risk for serious adverse events, with 15 affected. It also reports 45 participants at risk for other adverse events, with 44 affected. These are separate registry categories; the record does not explain the operational definitions used for seriousness or how overlapping categories were handled. Deaths are not reported in the registry record.

The listed serious adverse event terms and percentages are abdominal pain, 4.4%; pyrexia, 4.4%; abdominal pain upper, 2.2%; ascites, 2.2%; diarrhoea, 2.2%; and diverticulum intestinal, 2.2%. These are the top serious adverse event terms listed in the record. The listing does not provide event counts for each term, and it does not establish whether participants could contribute more than one term. The percentages should therefore be read as the registry's posted values, not as a basis for deriving additional counts or rates.

In general, adverse-event summaries describe events recorded during a study, while comparisons between treatments require suitable groups and compatible follow-up. This study has no control group, so its safety figures do not provide a comparative estimate. The registry record also does not report event timing, severity grades, attribution to study drugs, or detailed follow-up for these entries. The counts are descriptive and should not be read as proof that an event was caused by either drug.

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

The record describes a phase 2, single-group study of bevacizumab plus capecitabine in participants with advanced or metastatic liver cancer who had not previously received cytotoxic chemotherapy. Its primary endpoint is a confirmed objective response under RECIST, assessed over scheduled imaging visits and follow-up. The study enrolled 45 participants and is listed as completed.

The key limitation for interpreting efficacy is that the registry record does not report the primary response percentage or its 95% confidence interval. It also does not report results for the named disease-control, progression, or survival outcomes. Accordingly, the posted information does not establish the magnitude or precision of response, the timing of progression, or survival outcomes. An outcome definition is not itself an outcome result.

Even if an observed response estimate were available, the single-group, non-randomized, open-label design would limit what it could establish. In general, an estimate from one treated group describes the recorded outcome in that group; without a concurrent comparison arm, it cannot show how outcomes would have differed under an alternative. The design also does not provide a randomized basis for attributing an outcome to the regimen.

The safety record reports 15 participants with serious adverse events among 45 at risk and 44 with other adverse events among 45 at risk. It also lists serious adverse event terms and percentages. These entries provide descriptive safety information, but the record does not give enough detail to infer causation or compare safety with another treatment. Deaths are not reported in the registry record.

In general, statistical interpretation depends on both the observed estimates and the design that generated them. Here, the response endpoint, follow-up schedule, and broad design are described, but the principal efficacy estimate, uncertainty interval, full analysis population, and detailed analysis procedures are absent. The appropriate conclusion is limited to what the registry states about the study design and posted safety counts; efficacy conclusions cannot be drawn from an unreported primary result.

14. Why This Trial Matters Statistically

This trial illustrates why a clearly defined endpoint and a reported estimate are both necessary for statistical interpretation. The registry defines objective response using confirmed complete or partial response under RECIST, specifies a repeated-assessment schedule, and requires confirmation on repeat imaging at least 4 weeks after initial documentation. Those details identify what counted as a response. They do not reveal how many participants met the definition or how uncertain the estimate was.

The difference between enrollment and analysis population is also important. The record lists 45 enrolled participants but describes the primary endpoint for a Per Protocol population. The population criteria are incomplete, and its final size is not reported. Without the analysis denominator and handling of unassessed participants, the response result could not be independently checked even if an estimate were stated elsewhere.

The study also shows how design constrains inference. A non-randomized, single-group study can describe outcomes in the treated group, but it does not create a concurrent treatment comparison. This matters when readers distinguish a descriptive response proportion from a treatment effect. The open-label designation and the absence of a reported masking procedure are additional context for an imaging-based outcome.

Finally, named secondary outcomes and planning assumptions should not be confused with results. The registry lists disease control, progression, and survival outcomes, but their estimates are not reported. Its planning framework describes hypothetical sample-size illustrations, not observed power and not the original study's design justification. Statistical interpretation requires attention to what the record reports, what it leaves unreported, and what the design can support.

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 most consequential limitation in the registry record is the absence of the primary response estimate and its 95% confidence interval. Secondary outcome estimates are also not reported. As a result, the record does not permit quantitative assessment of efficacy, response precision, progression, or survival.

The study had no randomized comparison group. It was single-group, non-randomized, and open-label, so its design does not support a direct treatment contrast. The registry does not report a masking procedure. Since the primary endpoint depends on imaging and response classification, readers should note that the record does not describe whether independent review or another assessment safeguard was used.

The endpoint population is described as Per Protocol, but the criteria are truncated and the number analyzed is absent. The record does not report how missing scans, unconfirmed responses, treatment discontinuation, or participants without evaluable tumor assessments were handled. These details matter because the endpoint is based on repeated imaging and confirmation, and the analysis population may not equal the 45 enrolled participants.

The record lists eight outcome measures and no posted analyses. It does not report a detailed statistical analysis plan, a sample-size justification, multiplicity handling, interim analyses, or methods for calculating the confidence interval. It also does not report detailed safety definitions, event timing, severity grades, attribution, or deaths. These limitations are specific to the information recorded; unreported details should not be inferred.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: A Study of Avastin (Bevacizumab) and Xeloda (Capecitabine) in Patients With Advanced or Metastatic Liver Cancer — NCT02013830.
• Registry record last updated 2014-06-06; status last verified 2014-05.
• Results first posted 2014-06-06.
• 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

NCT02013830 is a completed phase 2 study of bevacizumab plus capecitabine in advanced or metastatic liver cancer among participants with no previous cytotoxic chemotherapy. It was single-group, non-randomized, open-label, and not masked. The registry lists 45 participants enrolled, a study start of 2005-05, and primary completion in 2008-03.

The primary endpoint was the percentage with confirmed objective response under RECIST, using scheduled imaging and repeat confirmation. The response estimate and 95% confidence interval are not reported in the registry record. Secondary outcomes included disease control, progression, and survival, but their results are not reported either.

Safety entries list 15 participants with serious adverse events and 44 with other adverse events, each among 45 at risk. Serious adverse event terms and percentages are recorded, while deaths are not reported in the registry record. The posted information supports a description of the design, endpoint, and safety counts, but not a quantitative interpretation of efficacy.