This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record. Numerical trial facts on this page are restricted to the ClinicalTrials.gov record. The registry reports that results have been posted, but the ClinicalTrials.gov record contains no formal statistical-analysis estimates, confidence intervals, or p-values for the registered primary endpoint.
1. Trial at a Glance
NSABP C-08 was a randomized, double-masked, parallel phase 3 treatment trial involving 2710 patients with colon cancer following surgery. The registered primary endpoint was 3-year disease-free survival, defined using recurrence, second primary cancer, or death from any cause.
| Feature | NSABP C-08 |
|---|---|
| Trial name | NSABP C-08 |
| NCT identifier | NCT00096278 |
| Phase | Phase 3 |
| Status | Completed |
| Therapeutic area | Oncology |
| Enrollment | 2710 |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | Double |
| Primary purpose | Treatment |
| Primary endpoint type | Time-to-event |
| Primary endpoints registered | 1 |
| Outcome measures posted | 9 |
| Lead sponsor | National Cancer Institute (NCI) |
| Sponsor type | NIH |
Registry result status: ClinicalTrials.gov reports that results have been posted for this trial. The registry-reported statistical-analysis section, however, contains no formal analysis records. This distinction is important: a posted outcome measure is not the same as a registry-posted treatment-effect estimate with a confidence interval and p-value.
2. Clinical Question
The central question was whether adding bevacizumab to fluorouracil, leucovorin, and oxaliplatin affected disease-free survival in patients who had undergone surgery for stage II or stage III colon cancer.
Population
Patients who had undergone surgery for stage II or stage III colon cancer, including the registered stage IIA, IIB, IIC, IIIA, IIIB, and IIIC categories.
Intervention
Fluorouracil, leucovorin, and oxaliplatin with bevacizumab, as described by the trial's brief title and intervention list.
Comparator
Fluorouracil, leucovorin, and oxaliplatin without bevacizumab.
Primary question
How does treatment assignment affect disease-free survival over the registered 3-year time frame?
3. Trial Design
With bevacizumab
- Bevacizumab
- Fluorouracil
- Leucovorin calcium
- Oxaliplatin
Without bevacizumab
- Fluorouracil
- Leucovorin calcium
- Oxaliplatin
The ClinicalTrials.gov record does not provide dosing schedules, treatment durations, randomization ratios, stratification factors, crossover rules, or detailed treatment sequencing.
4. Trial Timing and Registry Status
Trial start
The registry start date was 2004-09-15.
Primary completion
The registry lists 2009-03-12 as the primary completion date.
Completed
The ClinicalTrials.gov record identifies the trial status as completed and reports that results have been posted.
5. Endpoints
| Endpoint | Registry definition | Time frame | Type |
|---|---|---|---|
| Disease-free Survival | Where events are defined as recurrence, second primary cancer, or death from any cause | 3 years | Time-to-event |
This is the only registered primary endpoint reported in the ClinicalTrials.gov record. The registry reports one primary endpoint and nine posted outcome measures, but the ClinicalTrials.gov record does not provide definitions and results for the other eight outcome measures.
6. Statistical Methodology
Because the registered primary endpoint is a time-to-event outcome, the natural statistical framework focuses on the distribution of time until the first qualifying disease-free-survival event. Randomization supplies the principal design basis for comparing the two treatment groups.
Time-to-event analysis
A disease-free-survival analysis generally begins with estimation of the event-time distribution in each randomized group. Patients who have not experienced a qualifying event by the end of available follow-up are typically right-censored at their last evaluable time.
Here, di is the number of events at event time ti, while ni is the number of patients at risk immediately before that time.
Comparing randomized treatment groups
For a two-arm randomized time-to-event trial, a log-rank test is commonly used to compare the survival curves, while a Cox proportional-hazards model can provide a hazard-ratio estimate and confidence interval. These are typical methods for a disease-free-survival endpoint, not statistical-analysis results reported in the registry-reported NSABP C-08 data.
An HR below 1 would indicate a lower estimated instantaneous event rate in the numerator group relative to the comparator under the fitted model. It would not by itself identify the absolute probability of remaining disease-free at 3 years.
Why censoring is important
Not every patient in a time-to-event study necessarily experiences an event during the observation period. Censoring allows patients to contribute information up to the point at which their event status is no longer observed. The validity of standard survival methods depends on assumptions concerning the censoring mechanism and the information available before censoring.
Why the randomized design matters
Randomization creates the framework for comparing outcomes according to assigned treatment rather than according to post-randomization treatment choices. In an appropriately conducted randomized trial, this helps balance measured and unmeasured prognostic factors in expectation and supports causal interpretation of the treatment comparison.
7. Statistical Methods Explained
Why is disease-free survival a time-to-event endpoint?
The endpoint is defined not simply by whether a patient eventually experiences recurrence or another event, but by the time from the relevant trial starting point until the first qualifying event. This preserves information about when events occur and allows patients with different lengths of follow-up to contribute appropriately.
Why would Kaplan-Meier estimation be useful?
Kaplan-Meier estimation is designed for right-censored time-to-event data. Instead of treating every patient who has not yet had an event as if they had the same known outcome, it uses each patient's observed follow-up period to estimate the disease-free-survival function over time.
Why would a log-rank test be used?
A log-rank test provides a formal comparison of the event-time distributions between randomized treatment groups. Its calculation incorporates the numbers at risk and the observed events at successive event times rather than reducing the entire trial to a single binary outcome.
What would a Cox hazard ratio tell us?
A Cox model can summarize the relative event rate between treatment groups as a hazard ratio. For example, an HR below 1 would describe a lower modeled instantaneous event rate in the treatment group. It would not mean that the same percentage of patients avoided events, nor would it directly equal a difference in 3-year disease-free-survival probabilities.
Why is the 3-year time frame important?
The registry specifically defines the primary endpoint over 3 years. A 3-year disease-free-survival probability is an absolute time-specific quantity, whereas a hazard ratio summarizes relative event rates over the analyzed follow-up. These measures are related but answer different statistical questions.
Why does the composite definition matter?
The registry defines disease-free-survival events as recurrence, second primary cancer, or death from any cause. A treatment effect on the composite endpoint can therefore reflect differences in any of these event types. The endpoint should not be reinterpreted as recurrence alone.
8. Results
ClinicalTrials.gov reports that results have been posted for NSABP C-08. However, no formal statistical analyses were posted. As a result, the available trial data do not provide a treatment-effect estimate, confidence interval, or p-value for the registered primary endpoint.
| Primary result field | registry registry information |
|---|---|
| Primary endpoint | Disease-free Survival |
| Time frame | 3 years |
| Event definition | Recurrence, second primary cancer, or death from any cause |
| Results posted | Yes |
| Formal statistical analyses reported | None |
| Hazard ratio reported | Not reported |
| Confidence interval reported | Not reported |
| P-value reported | Not reported |
What analysis would normally be used?
For this registered time-to-event endpoint, a typical analysis would estimate disease-free survival using Kaplan-Meier methods, compare randomized groups using a log-rank framework, and potentially estimate a hazard ratio with a Cox proportional-hazards model. Those are methodological explanations rather than reported NSABP C-08 results.
A formal statistical conclusion about the magnitude or uncertainty of the treatment effect requires the actual event counts or the registry's statistical-analysis results.
The registry's "results posted" status is therefore a statement about data availability in the registry, not itself evidence of statistical significance, clinical benefit, or lack of benefit.
9. Safety Results
The ClinicalTrials.gov record includes serious adverse events by arm as affected patients over patients at risk. The registry labels posted on ClinicalTrials.gov for the two entries are identical in the source data, so the entries are reproduced without assigning an additional treatment label that is not present in the registry-reported field.
| Registry arm label | Serious adverse events affected / at risk |
|---|---|
| Oxaliplatin + Leucovorin + 5-Fluorouraci | 81/1326 |
| Oxaliplatin + Leucovorin + 5-Fluorouraci | 270/1334 |
The two registry-reported entries therefore indicate 81 affected patients among 1326 at risk and 270 affected patients among 1334 at risk. Because the arm labels are identical in the ClinicalTrials.gov record, this page does not infer which numerical entry corresponds to bevacizumab versus no bevacizumab.
10. Statistical Interpretation of the Primary Endpoint
The primary endpoint combines recurrence, second primary cancer, and death from any cause. This makes the endpoint broader than recurrence-free survival defined only by recurrence and means that the first qualifying event determines the endpoint event time.
If a formal Cox analysis were available, its HR would summarize relative instantaneous event rates between randomized groups under the model. It would not directly state the percentage of patients who remain disease-free at 3 years.
A confidence interval around a treatment-effect estimate would describe statistical uncertainty associated with the estimate under the specified analysis framework. A narrow interval would indicate greater statistical precision than a wide interval, but neither interval width nor the point estimate alone describes the clinical importance of every individual patient's outcome.
If a formal hypothesis test were reported, its p-value would quantify the compatibility of the observed data with a specified null hypothesis under the statistical model. It would not tell us how large or clinically important the treatment effect is. Effect estimates and confidence intervals are needed to describe magnitude and precision.
11. Limitations
- No formal statistical-analysis records reported: although the registry reports that results have been posted, the ClinicalTrials.gov record contains no treatment-effect estimates, confidence intervals, or p-values.
- Limited endpoint detail: the ClinicalTrials.gov record provides the registered primary endpoint but do not provide the other eight posted outcome-measure definitions.
- No event counts for disease-free survival: the ClinicalTrials.gov record does not state the number of primary endpoint events in either randomized group.
- No randomization ratio reported: the trial is identified as randomized with two arms, but the allocation proportion is not provided in the ClinicalTrials.gov record.
- No stratification factors reported: the ClinicalTrials.gov record does not identify any stratification variables used at randomization or analysis.
- No censoring rules reported: the endpoint is time-to-event, but the ClinicalTrials.gov record does not specify detailed censoring conventions.
- No missing-data or imputation strategy reported: the available data do not identify a prespecified imputation approach.
- No multiplicity procedure reported: the registry extract does not specify alpha allocation, multiplicity adjustment, or an endpoint hierarchy beyond identifying the single primary endpoint.
- No interim-analysis strategy reported: the available data do not describe interim looks, stopping boundaries, or alpha spending.
- No subgroup estimates reported: subgroup-specific treatment effects cannot be evaluated from the provided statistical-analysis field.
12. Why This Trial Matters Statistically
NSABP C-08 is a useful teaching example because its registry description combines randomized allocation, double masking, a parallel phase 3 design, a two-arm treatment comparison, and a time-to-event primary endpoint. The available data also illustrate an important principle in statistical reporting:
| Concept | How it appears in NSABP C-08 |
|---|---|
| Randomization | The study is registered as randomized. |
| Parallel design | The registry identifies a parallel design model with two arms. |
| Double masking | The registry identifies the masking as double. |
| Phase 3 | The trial is registered as phase 3. |
| Time-to-event analysis | The primary endpoint is disease-free survival. |
| Composite endpoint | Events are recurrence, second primary cancer, or death from any cause. |
| Fixed time frame | The registered primary endpoint is assessed over 3 years. |
| Safety denominators | Serious adverse events are reported as affected patients over patients at risk. |
| Statistical transparency | The registry reports results posted, while the registry-reported statistical-analysis field contains no formal analyses. |
13. What Can Be Learned From the Endpoint Definition
The disease-free-survival definition is statistically important because it determines what counts as an event. A patient who experiences recurrence, develops a second primary cancer, or dies from any cause reaches the endpoint. The statistical analysis therefore needs to preserve the ordering and timing of these events rather than collapsing them into a simple treatment-group percentage.
Event occurrence
A qualifying recurrence, second primary cancer, or death constitutes an event under the registry-reported definition.
Event timing
Because the endpoint is time-to-event, the timing of the first qualifying event contributes information to the analysis.
Censoring
Patients without an observed qualifying event during their available follow-up would ordinarily contribute censored follow-up information under a standard survival analysis.
Three-year assessment
The registered time frame is 3 years, so a reported 3-year disease-free-survival probability would be an absolute time-specific measure.
14. Randomization and Blinding as Statistical Design Features
The registry identifies NSABP C-08 as both randomized and double masked. These are design features rather than statistical test results, but they affect the quality and interpretability of the comparison.
Randomization
Randomization assigns treatment independently of investigators' and patients' prognostic preferences according to the trial's allocation mechanism. Its principal statistical role is to establish comparable treatment groups in expectation, allowing subsequent differences in outcomes to be attributed to treatment assignment under the assumptions of the randomized design.
Double masking
Double masking can reduce the opportunity for knowledge of treatment assignment to influence aspects of trial conduct or outcome assessment. The exact individuals covered by the registry's "double" designation are not specified in the ClinicalTrials.gov record, so the term should not be expanded into a more detailed masking description than the record supports.
Why these features matter for disease-free survival
Because disease-free survival can involve events that occur during follow-up, preserving treatment concealment and maintaining randomized comparisons can help reduce systematic differences in follow-up and assessment between treatment groups. The ClinicalTrials.gov record does not, however, provide enough information to quantify any such effects.
15. Planned Analysis
Because formal statistical-analysis records are absent from the ClinicalTrials.gov record, the following describes the analysis framework that would ordinarily be appropriate for the registered endpoint rather than reporting an NSABP C-08 treatment effect.
| Analytic step | Typical approach for this endpoint | What it would provide |
|---|---|---|
| Estimate disease-free survival | Kaplan-Meier method | Estimated disease-free-survival function over time |
| Compare randomized groups | Log-rank test | Formal comparison of event-time distributions |
| Estimate relative effect | Cox proportional-hazards model | Hazard ratio and confidence interval |
| Assess the 3-year endpoint | Time-specific survival estimate | Absolute disease-free-survival probability at 3 years |
| Handle incomplete follow-up | Right censoring under prespecified rules | Retention of available follow-up information |
16. Serious Adverse Events and Denominator-Based Interpretation
The safety data posted on ClinicalTrials.gov for NSABP C-08 use the format affected patients divided by patients at risk. That denominator is important because a raw event count alone cannot describe the frequency of serious adverse events.
The registry-reported values are 81/1326 and 270/1334. The registry extract does not provide a unique arm label for these two entries, so this page does not convert them into treatment-specific comparative rates.
A formal safety comparison would ordinarily need clearly identified treatment groups, consistent safety definitions, denominators, exposure information, and potentially confidence intervals or hypothesis tests. None of those additional comparative details are reported here.
17. What Is Not Appropriate to Infer
Do not infer an HR
The enrollment number and endpoint definition do not determine a hazard ratio. A numerical HR requires event and follow-up information or a reported analysis.
Do not infer a p-value
The fact that results were posted does not establish statistical significance. A p-value must come from a defined statistical comparison.
Do not relabel safety counts
The two serious-adverse-event entries have identical registry-reported labels, so their treatment assignment cannot be safely reconstructed from the field.
Do not treat the endpoint as recurrence alone
The registered definition explicitly includes recurrence, second primary cancer, and death from any cause.
18. Statistical Concepts in This Trial
Learn more about the methods used in this trial:
19. Related Statistical Calculators
20. Sources
- ClinicalTrials.gov: NCT00096278 — NSABP C-08.
- Linked publication: PubMed record: PMID 38315963.
- Linked publication: PubMed record: PMID 33356421.
- Linked publication: PubMed record: PMID 31811950.
- Linked publication: PubMed record: PMID 31268130.
- Linked publication: PubMed record: PMID 28006055.
Source policy for this page: the ClinicalTrials.gov record determine the trial facts and numerical content. The linked PubMed records are provided as source navigation only; no additional numerical trial results have been imported from them.
Continue through the Clinical Biostats statistical library
Explore tutorials and calculators covering the statistical concepts that arise in randomized clinical trials, including randomization, blinding, survival analysis, and time-to-event methods.
21. Record Summary
NSABP C-08 is a completed randomized, double-masked, parallel phase 3 trial with 2710 enrolled patients evaluating fluorouracil, leucovorin, and oxaliplatin with or without bevacizumab in patients who had undergone surgery for stage II or stage III colon cancer. Its registered primary endpoint is 3-year disease-free survival, with events defined as recurrence, second primary cancer, or death from any cause.
The principal statistical lesson from the ClinicalTrials.gov record is equally important: a well-defined randomized time-to-event design identifies an appropriate analytical framework, but it does not supply the treatment effect by itself. ClinicalTrials.gov reports that results are posted, yet the ClinicalTrials.gov record contains no formal estimates, confidence intervals, or p-values for the primary endpoint. A rigorous analysis therefore distinguishes the registered design and endpoint from the numerical statistical results actually available in the ClinicalTrials.gov record.