CAPOX, Bevacizumab and Trastuzumab for Patients With HER2-Positive Metastatic Esophagogastric Cancer: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for CAPOX, Bevacizumab and Trastuzumab for Patients With HER2-Positive Metastatic Esophagogastric Cancer (NCT01191697), a na phase 2 study. The primary outcome is Objective Response Rate (ORR).

2. Clinical Question

This study asks whether objective tumor responses were observed among patients treated with the combination. It is a question about response in a single treated group, not a direct test of whether the combination improves outcomes relative to another treatment. The registry also names median overall survival, median duration of response and median progression-free survival as secondary outcomes, but their results are not reported in the registry record. Neither the primary nor secondary outcome descriptions establish a comparative treatment effect.

3. Trial Design

The registry classifies this as a Phase II, single-group study with no masking. Allocation is listed as NA, consistent with the absence of a randomized allocation between treatment arms. The study is marked completed. Dana-Farber Cancer Institute is listed as sponsor. The registry gives a start date of 2011-02 and a primary completion date of 2021-12.

The treatment arm combines capecitabine and oxaliplatin (CAPOX) with bevacizumab and trastuzumab for patients with HER2-positive metastatic esophagogastric cancer. The arm description says each cycle is 21 days. It describes an initial trastuzumab loading dose, followed by later-cycle bevacizumab and trastuzumab doses, but the description ends partway through the treatment details. The complete regimen, including any additional instructions, is not reported in the registry record.

The registry reports enrollment of 37. That figure describes enrollment, not necessarily the number included in every endpoint analysis. The adverse-event record lists 36 participants at risk. The analysis population for the primary endpoint, and the reason the safety count differs from enrollment, are not reported in the registry record.

A single-group design can describe outcomes among people who received the regimen. Without a concurrent control group, it cannot separate treatment-associated outcomes from the course of disease, other care, or differences in the patients treated. The registry also lists no statistical analyses. Details such as the planned analysis population, handling of incomplete outcome assessments, and any model-based adjustment are not reported in the registry record. The record therefore supports a descriptive account of the design and endpoint, but not a reconstruction of the full statistical protocol.

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:

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 Objective Response Rate (ORR), measured over Patients received a median of 19 cycles of therapy (Interquartile range (IQR): 8 - 34.5 cycles). Median duration of follow-up of 23.2 months (IQR: 11.0 - 46.9 months )..

The primary endpoint is objective response rate (ORR), defined as the percentage of participants who achieved complete response or partial response on treatment according to RECIST 1.1. The registry defines complete response for target lesions as disappearance of all target lesions. Partial response requires at least a 30% decrease in the sum of the longest diameters of target lesions relative to baseline. The description also specifies that an overall response of partial response or better requires, at minimum, incomplete response or stable disease for non-target lesions and no new lesions.

Response rate is used to summarize how often a prespecified tumor-response category is achieved during treatment. It is a binary outcome at the participant level: each participant is categorized according to whether the response criterion was met. That summary can be clinically interpretable as a measure of observed tumor shrinkage under the stated criteria, but it is not itself a comparison with another treatment and does not establish how long a response lasts.

The registry describes treatment exposure as a median of 19 cycles, with an interquartile range (IQR) of 8 - 34.5 cycles, and median follow-up as 23.2 months, with an IQR of 11.0 - 46.9 months. These descriptions provide context for observation and treatment exposure; they do not replace the endpoint result. The ORR estimate, its denominator, and a confidence interval are not reported in the registry record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The statistical approach can be understood only in part from the registry record. It identifies a single-group Phase II study, a binary primary endpoint, and response criteria based on RECIST 1.1. It does not report an analysis plan, an endpoint estimate, a confidence interval, or a statistical test. The explanations below describe general statistical concepts and do not imply that a particular procedure was used in this study.

What does a single-group response estimate describe?

For a binary endpoint such as objective response, a response proportion summarizes the share of assessed participants who meet the response definition. The interpretation depends on who was assessed and how participants without an evaluable response were classified. A denominator is therefore essential: the enrollment count is not automatically the analysis denominator. In this record, the denominator for ORR and the rules for classifying unevaluable participants are not reported in the registry record.

A confidence interval is a range of values compatible with the observed data under a specified statistical method. It communicates uncertainty around an estimate; it does not guarantee that the true value lies within the interval. The width depends on the amount of information and the method used. No confidence interval for ORR is reported in the registry record, so its precision cannot be assessed. A point estimate alone, if one were available, would not show how much statistical uncertainty surrounds it.

Why does the design matter for inference?

A single-group study can estimate or describe outcomes in the group treated, subject to complete and appropriate outcome ascertainment. A comparative treatment effect requires a comparison that can distinguish outcomes under one treatment strategy from outcomes under another. Random allocation, when used, helps balance measured and unmeasured prognostic factors on average; this study is not reported as randomized and has no listed concurrent comparator. Thus, even a reported response proportion would not by itself identify what response would have occurred under an alternative treatment.

Observed outcomes also reflect more than the assigned treatment. Patient selection, disease characteristics, accompanying care, and the timing or completeness of assessments can influence a descriptive rate. The registry does not provide enough information about these features to adjust or account for them. This is not evidence that any particular bias occurred; it marks what cannot be evaluated from the record.

What does the response definition contribute?

RECIST 1.1 supplies explicit criteria for categorizing tumor response. The registry describes complete response, partial response, and requirements involving non-target lesions and new lesions. Such rules make the endpoint more reproducible than an informal judgment, but they do not eliminate uncertainty about measurement or assessment timing. The registry does not report the assessment schedule, independent review procedures, or whether response assessments were centrally adjudicated. Those details are not reported in the registry record.

ORR is distinct from duration of response, progression-free survival, and overall survival. A response-rate summary classifies whether a response occurred under the endpoint rules; it does not describe how long that response persisted. The record lists duration of response and survival outcomes as secondary measures, but their estimates and methods are not reported. A median is a summary of the middle of an outcome distribution, while time-to-event methods may also account for participants whose event status is not observed through the end of follow-up. The registry record does not describe which methods were used for these outcomes.

What can be learned from the planning illustration?

Any sample-size figures accompanying the results are planning illustrations based on an observed effect, not the study's own design justification. The stated assumptions describe a hypothetical two-arm calculation with equal allocation, a two-sided significance level of 0.05, and a normal approximation; the target effect is expressed as a fraction of the observed effect. Those assumptions do not match the registry's single-group design, so such calculations should not be read as the trial's actual allocation or protocol. They are not observed power. The protocol's sample-size justification and planned power are not reported in the registry record.

In general, power is a design property tied to a specified effect, variability, sample size, and analysis; it is not a measure of whether an observed result is true. Recalculating a planning requirement from an observed effect uses the effect estimate as an input and should not be mistaken for a prospective design analysis. No planning-illustration values are stated in the registry record, so no particular required sample size can be interpreted.

What analysis details are absent?

The registry lists no analyses. It does not report a statistical model, testing procedure, confidence-interval method, treatment of missing data, analysis population, or interim-analysis plan. It also does not report whether any adjustment for prognostic factors or multiplicity was planned. These omissions prevent a detailed audit of the inferential method. They do not establish that a method was or was not used outside the registry entry; they mean that it is not reported in 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("Trastuzumab, Bevacizumab, Oxaliplatin and Capecita")
n      <- c(36)
events <- c(29)
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  Trastuzumab, Bevacizumab, Oxaliplatin and Capecita  36  29  80.6  64.0  91.8  65.0  90.2  64.8  90.2

11. Multiplicity, Missing Data and Analysis Populations

The registry identifies objective response rate as the primary endpoint and lists median overall survival, median duration of response and median progression-free survival as secondary outcomes. It does not report how these outcomes were prioritized analytically, whether formal hypothesis tests were planned, or whether any adjustment for multiple testing was intended. When several outcomes are examined, the chance of at least one apparently notable result can increase if each is tested separately. Methods to address multiplicity depend on the aims and prespecified analysis plan. No such plan is reported in the registry record.

The registry also does not report whether analyses were interim or final, how the analysis population was defined, or how incomplete assessments were handled. For ORR, missing response assessments could affect the denominator and the classification of participants, depending on the prespecified rules. Those rules are not reported in the registry record. The recorded enrollment is 37, whereas the adverse-event record lists 36 at risk; the reason for the difference and the endpoint analysis denominator are not reported.

It is not possible to infer whether all enrolled participants were included in each outcome analysis, whether exclusions occurred, or whether sensitivity analyses were performed. The registry lists no analyses. Consequently, the data entry does not establish how missing outcome information, deaths, withdrawals, or differing follow-up might have affected efficacy estimates. These are reporting gaps, not evidence that any specific handling choice was made.

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: dehydration (13.9%), pneumothorax (8.3%), abdominal pain (5.6%), pulmonary embolism (2.8%), pneumonia (2.8%).

The adverse-event record lists 36 participants at risk for serious adverse events and 13 participants affected. For other adverse events, it lists 36 at risk and 36 affected. These counts are descriptive entries for the single treatment group; they do not compare risk with a control group. The registry lists dehydration (13.9%), pneumothorax (8.3%), abdominal pain (5.6%), pulmonary embolism (2.8%), pneumonia (2.8%) and hyponatremia (2.8%) among the top serious-event terms.

The percentages should be read as registry-reported figures for the listed terms, not as estimates of the causal effect of treatment. The record does not describe the event-attribution rules, observation windows, or whether a participant could contribute more than one event term. It also does not report event-level counts for each listed term in the information available for this analysis. The terms therefore should not be added together or interpreted as mutually exclusive categories.

The registry records 31 deaths. Cause, timing, and relationship to treatment are not reported in the registry record, so the death count cannot be interpreted as treatment-related mortality or as a survival estimate. The difference between enrollment and the safety population is also unexplained. Safety interpretation is limited to these recorded counts and percentages.

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

This record establishes that a completed Phase II study evaluated a combination of capecitabine, oxaliplatin, bevacizumab and trastuzumab in patients with HER2-positive metastatic esophagogastric cancer. It specifies objective response rate as the primary endpoint and describes the RECIST 1.1 criteria used to define response. It also records treatment exposure and follow-up summaries, names secondary outcomes, and includes adverse-event counts and selected serious-event percentages.

Those details do not establish the magnitude or precision of the primary efficacy result. The observed ORR is not reported in the registry record, and no confidence interval or analysis denominator is given. The secondary outcome results are also not reported. Without these results, readers cannot assess the response frequency, survival outcomes, or duration of response from the registry entry.

Even if a response estimate were reported, this design would support a description of outcomes in the treated group rather than a direct comparison against an alternative. The study is single-group, has no listed comparator, and is not reported as randomized. It therefore cannot determine from its design alone whether any observed outcomes differ from what would have occurred under another treatment or without the study regimen.

The safety counts and event percentages describe events recorded for the treatment group. They do not establish causation, and the registry does not report enough detail about attribution, timing, or event handling to resolve those questions. The recorded death count likewise has no reported cause or treatment relationship.

The most defensible reading is therefore limited: the registry records a study design, treatment combination, response definition, exposure and follow-up summaries, and selected safety information, but omits the numerical primary and secondary efficacy results and key analysis details. Conclusions about comparative effectiveness, treatment causation, or the precision of response estimates are not supported by the information reported.

14. Why This Trial Matters Statistically

This study is statistically instructive because a clearly defined clinical endpoint and a completed study record do not necessarily provide enough information to evaluate an effect. The registry specifies ORR and the RECIST 1.1 response rules, but does not report the actual ORR estimate, its denominator, or its uncertainty. The distinction between endpoint definition and endpoint result is essential: knowing what was intended to be measured does not reveal what was observed.

The single-group design also illustrates the difference between describing outcomes and estimating a comparative effect. A response proportion can be a useful descriptive summary, but a control group is needed to make a direct comparison with another strategy. Randomization, when present, can support causal comparisons by balancing prognostic factors on average. Neither a randomized allocation nor a concurrent control group is listed for this study.

The registry's safety record offers another important lesson. Counts and percentages provide a compact account of recorded events, but interpretation depends on the population at risk, event definitions, observation period, attribution, and handling of participants with incomplete follow-up. Some of those details are absent. The difference between enrollment and the safety count is not explained, reinforcing why denominators and analysis populations should be reported clearly.

Finally, the record lists multiple outcomes but does not describe a statistical analysis plan or multiplicity strategy. These omissions limit the ability to interpret formal inference even if outcome estimates were available. The study is therefore a useful example of why design, endpoint definitions, denominators, uncertainty, missing-data handling, and complete results must be considered together when judging the statistical evidence.

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 main trial-specific limitation is the lack of reported efficacy results. Although the primary endpoint is defined, the registry does not report the ORR estimate, its denominator, or an uncertainty interval. The secondary outcomes are named but their estimates are not reported. This prevents assessment of the study's efficacy findings from the registry record.

The single-group design has no concurrent control and is not reported as randomized. It can describe outcomes among treated participants but cannot directly estimate a comparative treatment effect. The registry does not report an analysis plan, statistical model, confidence-interval method, or any adjustment strategy. It also does not report whether interim analyses were conducted.

There is incomplete information about who contributed to each analysis. Enrollment is 37, while the safety record lists 36 at risk. The reason for this difference, the primary endpoint population, and the handling of missing or unevaluable response assessments are not reported in the registry record. The treatment description is truncated, so the full regimen cannot be reconstructed from the recorded arm description.

Follow-up and exposure are summarized, but the timing and methods of response assessments, event attribution procedures, and detailed survival analysis methods are not reported. The registry lists 31 deaths without cause or treatment relationship. These gaps restrict interpretation of both efficacy and safety. They are limitations of what can be established from the registry entry and do not, on their own, show that the trial lacked procedures that were not recorded.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: CAPOX, Bevacizumab and Trastuzumab for Patients With HER2-Positive Metastatic Esophagogastric Cancer — NCT01191697.
• Registry record last updated 2025-11-03; status last verified 2025-09.
• Results first posted 2023-09-05.
• 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 completed Phase II, single-group study sponsored by Dana-Farber Cancer Institute of CAPOX, bevacizumab and trastuzumab in HER2-positive metastatic esophagogastric cancer. It lists enrollment of 37, a primary endpoint of objective response rate under RECIST 1.1, and secondary outcomes of overall survival, duration of response and progression-free survival.

The record reports median treatment exposure of 19 cycles and median follow-up of 23.2 months. It includes adverse-event counts, selected serious-event percentages, and 31 deaths. However, the primary and secondary efficacy results, endpoint analysis denominator, uncertainty estimates, statistical plan, missing-data rules, and explanation for the safety population count are not reported in the registry record. The single-group design does not provide a concurrent comparative treatment estimate.