Neoadjuvant Bevacizumab, Capecitabine and Radiation Therapy With or Without Oxaliplatin Locally Advanced Rectal Cancer: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Neoadjuvant Bevacizumab, Capecitabine and Radiation Therapy With or Without Oxaliplatin Locally Advanced Rectal Cancer (NCT00828672), a randomized phase 2 study. The primary outcome is Pathologic Response at Surgery. Overview of Complete Pathologic Responses, Good and Little Tumour Regression Rates at Surgery..

2. Clinical Question

The question is about whether adding oxaliplatin to the shared preoperative regimen is associated with a different distribution of pathologic response categories in the centrally reviewed surgical specimens. The primary measure uses Dworak tumour regression grades, ranging from no regression to total regression. The primary endpoint timeframe is 4 months.

The registry also lists secondary outcomes concerning resection margins, pathologic complete response, tumour regression, clinical response, adverse events, recurrence, disease-free survival, death and overall survival. These outcomes could address other aspects of treatment and follow-up, but the registry record does not provide numerical results for them. The clinical question is therefore defined more fully than the results can be interpreted from the recorded data.

3. Trial Design

This was a randomized, open-label, parallel-arm phase II study of preoperative treatment for locally advanced rectal cancer. The registry lists Universitaire Ziekenhuizen KU Leuven as sponsor and describes the study as academic and investigator initiated. It enrolled 84 participants. The recorded study start is 2009-06, primary completion is 2014-03, and status is completed.

Participants were assigned to AXE (ARM 1), which combined oxaliplatin, bevacizumab and capecitabine with radiotherapy, or AX (ARM 2), which combined bevacizumab and capecitabine with radiotherapy. Both regimens were given concurrently with radiotherapy, followed by total mesorectal excision as described in the study summary. The registry does not report treatment doses, schedules, eligibility details beyond the condition, or the allocation procedure used to generate the randomized assignments.

Masking was listed as none. In an open-label design, treatment assignment is not concealed from participants and clinical teams as described by the masking entry. The primary outcome nevertheless involved pathologic assessment of operative specimens. Local pathologists assessed response, and investigators’ assessments were reviewed centrally for participants whose surgical materials were available. The record says that the diagnosis of independent central reviewers primed; it does not explain this wording or provide further details about reviewer blinding or adjudication.

The primary outcome population was limited to patients who had surgery and pathology materials available for central review. That differs conceptually from an analysis of all randomized participants: a post-randomization requirement for surgery and available material can result in a smaller outcome population. The number included in this centrally reviewed subset is not reported in the registry record.

The design is parallel rather than crossover: the two arms represent distinct regimens, not successive treatment periods. Although 84 participants were enrolled, the record does not report the allocation counts for the primary endpoint population. The adverse-event entries list 43 participants at risk in AXE and 41 in AX, which sum to the enrolled total, but these entries should not be assumed to be the denominator for every efficacy outcome.

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: Target population was represented by patients with locally advanced rectal cancer (tumour beyond mesorectal fascia (T4) or tumour ≤ 2 mm from mesorectal fascia or T3 tumour \< 5 cm from anal verge by MRI), histologically confirmed. Pts were screened as per incl and excl criteria per protocol. Screening failures were not recorded in the eCRF.

Pre-treatment (Baseline)

Study Treatment

Reasons for not completing this period:

Post-surgery (<30 Days)

Reasons for not completing this period:

Follow-up

Analysis population for the primary outcome, as defined in the registry: Patients who had undertaken surgery and had pathology materials available for central review (centrally reviewed subset)

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 Pathologic Response at Surgery. Overview of Complete Pathologic Responses, Good and Little Tumour Regression Rates at Surgery., measured over 4 months.

The primary endpoint describes pathologic response in tumour tissue assessed at surgery, reported as a percentage of cases. The assessment used Dworak tumour regression grades. TRG0 indicates no regression. TRG1 describes a dominant tumour mass with obvious fibrosis and/or vasculopathy; TRG2 describes dominant fibrotic changes with few tumour cells or groups; and TRG3 describes very few tumour cells, difficult to find microscopically, in fibrotic tissue with or without mucus substance. TRG4 indicates no intact viable tumour cells, with only a fibrotic mass or mucin lakes without associated malignant cells, and represents total tumour regression.

The registry describes three summaries: pathologic complete response, defined as TRG4; good regression, combining TRG3 and TRG4; and little regression, combining TRG0, TRG1 and TRG2. The good and little regression percentages are described as adding to 100%. Complete response is therefore included within the good-regression category rather than being a separate category alongside it.

Pathologic response is used to characterize tumour regression in the resected specimen after preoperative treatment. It gives a tissue-based assessment at surgery, rather than relying only on clinical assessment before surgery. The primary endpoint timeframe is 4 months. The record does not report numerical percentages or confidence intervals for the response categories, nor the size of the centrally reviewed population. As a result, the endpoint’s definition is available, but its observed results and precision cannot be evaluated from the registry record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The main statistical idea is to distinguish the study’s design, its endpoint definition, and the results available for interpretation. The design was randomized and parallel, with two regimens. Randomization is intended, in general, to make treatment groups comparable on average at the start of a trial, including with respect to factors that are measured and those that are not. It does not ensure that every characteristic is exactly balanced in a particular trial. The registry does not report the randomization sequence, allocation concealment procedures, stratification factors or other implementation details.

What does the endpoint summarize?

The primary outcome is categorical: each centrally reviewed specimen is classified by a Dworak tumour regression grade, and the report is intended to summarize selected groupings as percentages. A categorical endpoint differs from a continuous measure such as tumour size because it places observations into named classes. The complete response summary corresponds to TRG4. The broader good-regression summary groups TRG3 with TRG4, while the little-regression summary groups TRG0, TRG1 and TRG2. Since the latter two groupings partition the grades, their percentages are described as adding to 100%. A complete response is contained within good regression, so these summaries overlap rather than forming three mutually exclusive categories.

For a percentage, the denominator is central to interpretation. In this study, the primary population is participants who had surgery and pathology material available for central review. The registry does not state the number of participants in that subset or the number in each response category. Those omissions prevent reconstruction or checking of the percentages and make it impossible to judge how much information supports each estimate.

What does a confidence interval contribute?

The record specifies 95% confidence intervals for the primary response summaries. In general, a confidence interval expresses uncertainty in an estimate under a statistical method and its assumptions. It helps distinguish an estimate that is relatively precise from one that is compatible with a broad range of values. A confidence interval is not a guarantee that the true value lies in a particular realized interval, and it does not by itself establish clinical importance. For this trial, the confidence interval limits are not reported in the registry record, so the precision of the response percentages is unknown.

For a randomized comparison, the question of interest is often a contrast between treatment arms. A clear comparison requires the arm-specific estimates, a defined contrast, and uncertainty around that contrast. The registry description of the primary endpoint lists several categories and does not reduce them to one comparison. It also does not provide a reported primary analysis or arm-level response values. Therefore, no treatment contrast, test statistic or p-value for the primary endpoint can be interpreted from the record.

Why can the analysis population matter?

The primary response summaries are based on a subset defined after randomization: participants had to undergo surgery and have pathology materials available for central review. In general, restricting analysis to a post-randomization subset can affect comparability between arms if inclusion is related to treatment assignment, prognosis or events after assignment. This is a methodological consideration, not evidence that such a problem occurred in this study. The record does not report the number excluded from central review, the reasons for missing materials, or the planned analysis population. It is therefore not possible to assess how these factors were handled.

Central review can support consistent classification by having specimens assessed beyond the local pathology assessment. The registry states that local pathologists read operative specimens and that investigators’ assessments were reviewed centrally for available material. It does not describe reviewer blinding to treatment, methods for resolving disagreement, or inter-reviewer agreement. These details would help readers evaluate how consistently the grades were assigned.

How should the safety counts be read?

The adverse-event record gives, for each arm, the number at risk and the number affected by serious events and by other events. These are descriptive counts. In general, a comparison of event frequencies should consider denominators, follow-up duration, event definitions, severity grading, and uncertainty. The record lists the denominators and affected counts, but does not provide confidence intervals, a formal between-arm test, event-specific arm counts or a detailed definition of the “other” category. The listed counts alone do not establish whether an observed difference is larger than could reasonably arise from sampling variability.

The top serious-event terms are accompanied by percentages, but their denominators and arm attribution are not specified in the registry record. They should not be treated as arm-specific rates without that information. The death entries are also counts, not a time-to-event analysis. The record names overall survival and disease-free survival as secondary outcomes but does not report their estimates, follow-up summaries or analyses.

What can planning calculations tell readers?

A sample-size planning illustration based on an observed effect is not the same as the trial’s original design calculation or an observed-power analysis. In general, planning calculations describe the sample size needed under stated assumptions to detect a chosen effect at a chosen significance level and power. Their results depend on the assumed effect, allocation, statistical model and other inputs. The record describes an illustration using a two-sided significance level of 0.05, equal allocation, a normal approximation and a chosen fraction of the observed effect; it specifies that time-to-event calculations use total events and other calculations use participants per arm. It does not report the trial’s own sample-size justification. Because the primary outcome estimates are not reported in the registry record, a reader cannot evaluate the observed effect underlying such an illustration from this record alone. Any planning result is conditional, not evidence that the completed trial had a particular probability of finding a result.

Overall, the statistical methods that can be identified are limited to random allocation, the categorical pathology grading approach, the stated 95% confidence-interval format, and the descriptive safety summaries. The analysis plan, statistical tests, handling of missing data, multiplicity approach, and detailed sample-size rationale are not reported in the registry record.

10. Multiplicity, Missing Data and Analysis Populations

The primary endpoint contains several response summaries: complete pathologic response, good tumour regression and little tumour regression. These are related rather than wholly independent outcomes because complete response is included within good regression, while good and little regression are complementary groupings. The registry record does not report how multiplicity across these summaries was addressed, whether one summary was designated as the primary inferential comparison, or whether any adjustment to significance thresholds was planned.

Several secondary outcomes are also listed, covering resection margins, pathology response, clinical response, adverse events, recurrence, disease-free survival, death and overall survival. The record does not report a hierarchy among these outcomes, a multiplicity strategy, or which outcomes were formally tested. When many outcomes are considered, general statistical practice recognizes that the chance of at least one apparently notable result can increase unless the analysis plan accounts for multiple testing. This is a general consideration; the record does not establish which approach was used in this trial.

Missing-data procedures are not reported in the registry record. For the primary endpoint, availability of surgical material for central review defines the analyzed subset, but the subset size, missing-material reasons and handling of participants without assessable specimens are not stated. The record also does not report how missing secondary outcome or follow-up information was treated.

The analysis populations are not fully described. The primary endpoint population is specified as operated participants with materials available for central review, while the adverse-event denominators are listed as 43 in AXE and 41 in AX. The record does not state whether an intention-to-treat population, per-protocol population or other prespecified analysis set was used for efficacy outcomes. The distinction matters because results based on different populations can answer different questions.

Details of the statistical analysis plan, interim analyses and sample-size justification are not reported in the registry record. The absence of those details does not show that a particular method was or was not used; it limits what can be assessed from this record.

11. 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: wound infection (11.6%), leak (7.3%), fever (7.0%), diarrhea (7.0%), obstruction (4.7%).

The safety entries list 43 participants at risk in AXE and 41 in AX. Serious adverse events affected 21 participants in AXE and 12 in AX. Other adverse events affected 18 participants in AXE and 15 in AX. The registry also records 0 deaths in AXE and 1 in AX. These figures describe the counts in the listed groups; they are not adjusted comparisons and the record gives no confidence intervals or formal statistical tests for them.

The registry lists wound infection (11.6%), leak (7.3%), fever (7.0%), diarrhea (7.0%), obstruction (4.7%) and urinary tract infection (4.7%) as top serious-event terms. The record does not specify the denominators or attribute these percentages to particular arms. These percentages therefore should not be interpreted as arm-specific rates. Nor does the record explain whether participants could contribute to more than one event term.

In general, adverse-event comparisons depend on definitions, ascertainment, duration of observation and the number of participants at risk. The registry entries do not give follow-up duration for safety, event-level timing, severity grades beyond the serious-event designation, or detailed definitions for “other” adverse events. The counts can be reported as recorded, but they do not by themselves establish a causal relationship between treatment and an event or characterize the full safety profile.

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

The study’s randomized comparison concerned whether adding oxaliplatin to bevacizumab, capecitabine and radiotherapy was associated with a different pathologic response pattern at surgery. The registry record defines how response was graded and identifies the subset used for the primary outcome. It does not report the observed response percentages or their confidence intervals. As a result, the record does not show the magnitude, direction or precision of an efficacy difference between the regimens.

The safety data do report counts of participants affected by serious and other adverse events in each arm, along with death counts. Those entries provide descriptive information, but they do not include a formal comparative analysis or uncertainty estimates. The serious-event terms and percentages also lack arm-specific denominators in the record. These figures can be described without treating them as proof that either regimen is safer or less safe.

The listed secondary outcomes include surgical pathology, clinical response, recurrence and survival measures. Their presence in the registry indicates that they were designated outcomes, but numerical results for them are not reported in the registry record. The completed status of the trial does not substitute for those results.

In general, a trial result should be interpreted in light of the estimated effect, its precision, the analysis population, missing data and the prespecified statistical plan. Here, important elements are absent: the primary outcome estimates, their confidence intervals, the centrally reviewed sample size, details of the analysis plan and missing-data handling. The record therefore supports a description of the question, design and some safety counts, but not a conclusion about comparative efficacy or a definitive treatment effect. It also does not support extrapolation to patients or settings beyond the population and study context recorded.

The result does not mean that the regimens produced identical responses; the registry does not provide the response estimates needed to assess that. It also does not mean that one regimen produced better response or outcomes. Without the outcome data and uncertainty measures, either claim would go beyond what the record reports.

13. Why This Trial Matters Statistically

This trial is statistically informative as an example of the gap between a clearly defined clinical question and the information needed to evaluate its answer. Randomization and a parallel-arm design establish a comparative framework, while the primary endpoint uses a specified pathology grading system and a defined centrally reviewed population. Those elements allow readers to understand what the study intended to compare.

At the same time, the registry record does not report the numerical primary outcome results. A precise endpoint definition cannot replace the observed arm-level percentages, denominators and confidence intervals. Their absence prevents assessment of effect size, uncertainty and the relationship between the complete-response and broader regression summaries. This is especially relevant because multiple related categories are named, yet the statistical approach to multiplicity is not reported.

The endpoint population also illustrates why analysis populations matter. The response summaries apply only to participants who had surgery and available material for central review. Without the subset size and information about participants not included, readers cannot assess the scope of that restriction. In general, post-randomization selection can affect the interpretation of a randomized comparison; identifying and describing the analysis population is therefore important.

The safety entries show how counts can be informative while remaining limited. The arm denominators and affected counts are recorded, but the record does not give comparative uncertainty, event-specific arm details for the listed serious terms or a formal analysis. This distinction helps prevent descriptive figures from being mistaken for a definitive statistical contrast.

Finally, the listed secondary outcomes make clear that efficacy, safety, recurrence and survival are different questions. Naming an outcome is not the same as reporting its result. The registry record supports careful description of the trial’s design and outcome definitions, while also demonstrating why transparent reporting of estimates, uncertainty, analysis populations and methods is necessary for a full statistical interpretation.

14. 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.

A central limitation for interpretation is that numerical results for the primary pathologic response endpoint are not reported in the registry record. Although the endpoint definition and a 95% confidence-interval format are stated, the category percentages, interval limits and arm-specific comparisons are absent. The primary outcome cannot therefore be evaluated for effect size or precision from this record.

The primary endpoint applies to participants who underwent surgery and had pathology material available for central review. The size of this subset, the number without assessable material and reasons for non-inclusion are not reported. The record also does not state how participants without a primary endpoint assessment were handled analytically. These omissions restrict assessment of selection and missing-data concerns.

Pathology assessments were read locally and reviewed centrally, but the record does not describe whether central reviewers were masked to treatment assignment, how disagreements were resolved or how consistent grading was across reviewers. It also does not explain the wording about the independent central reviewers. These details would be relevant to evaluating possible classification variability.

The record does not report the statistical analysis plan, formal tests, multiplicity approach, interim analyses, or the trial’s sample-size justification. It does not state the planned analysis populations for efficacy outcomes. Though adverse-event denominators are listed, the registry does not provide comparative uncertainty, detailed event definitions, safety follow-up duration or arm-specific counts for the listed serious-event terms.

Secondary outcomes are named without numerical results, including recurrence and survival outcomes. Their follow-up summaries and analysis methods are also not reported in the registry record. These limitations concern the available registry information and do not establish what analyses may have been conducted outside the record.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: Neoadjuvant Bevacizumab, Capecitabine and Radiation Therapy With or Without Oxaliplatin Locally Advanced Rectal Cancer — NCT00828672.
• Registry record last updated 2019-07-10; status last verified 2019-07.
• Results first posted 2019-07-10.
• 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.

18. Record Summary

NCT00828672 is a completed randomized, open-label, parallel-arm phase II study sponsored by Universitaire Ziekenhuizen KU Leuven. It enrolled 84 participants with locally advanced rectal cancer. AXE combined oxaliplatin, bevacizumab and capecitabine with radiotherapy; AX combined bevacizumab and capecitabine with radiotherapy. The regimens were used in preoperative treatment followed by total mesorectal excision.

The primary endpoint concerned pathologic response at surgery, assessed using Dworak tumour regression grades in participants with surgical material available for central review. It included complete response, good regression and little regression summaries, with a timeframe of 4 months. Numerical primary outcome results and confidence-interval limits are not reported in the registry record.

Safety entries list serious-event counts of 21 of 43 in AXE and 12 of 41 in AX, other adverse-event counts of 18 of 43 and 15 of 41, and death counts of 0 and 1, respectively. The record names secondary outcomes but does not report their numerical results or the detailed statistical analysis plan.