This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record.
1. Trial at a Glance
PETACC-3 was a randomized, unmasked phase 3 treatment trial in colorectal cancer. The registry title describes a comparison of fluorouracil and leucovorin with or without irinotecan in patients following surgery for stage III colorectal cancer.
| Feature | PETACC-3 |
|---|---|
| Trial acronym | PETACC-3 |
| ClinicalTrials.gov identifier | NCT00026273 |
| Phase | Phase 3 |
| Status | Completed |
| Condition | Colorectal Cancer |
| Allocation | Randomized |
| Masking | None |
| Primary purpose | Treatment |
| Lead sponsor | Pfizer |
| Sponsor type | Industry |
| Start date | 2001-01 |
| Primary completion date | 2004-09 |
2. Clinical Question
The registry title defines the clinical comparison as fluorouracil and leucovorin with or without irinotecan in patients following surgery for stage III colorectal cancer. Statistically, the randomized design is intended to compare outcomes between the treatment strategy containing irinotecan and the corresponding strategy without irinotecan.
Population
Patients following surgery for stage III colorectal cancer.
Intervention
FOLFIRI regimen, including fluorouracil, irinotecan hydrochloride, and leucovorin calcium.
Comparator
Fluorouracil and leucovorin without irinotecan, as specified by the registry title.
Primary question
Among patients following surgery for stage III colorectal cancer, what is the comparative effect of adding irinotecan to fluorouracil and leucovorin?
3. Trial Design
FOLFIRI regimen
- Fluorouracil
- Irinotecan hydrochloride
- Leucovorin calcium
Fluorouracil and leucovorin
- Fluorouracil
- Leucovorin
- Without irinotecan, according to the registry title
The design has two important structural features for statistical interpretation: allocation was randomized and masking was none. Randomization supports a comparison based on treatment assignment, whereas the absence of masking can matter for outcomes that involve clinical assessment or treatment decisions.
4. Registry Timeline
PETACC-3 begins
The registry records the trial start as 2001-01.
Comparative treatment evaluation
The study is classified as a randomized phase 3 treatment trial evaluating fluorouracil and leucovorin with or without irinotecan following surgery for stage III colorectal cancer.
Primary completion
The registry records the primary completion date as 2004-09.
Registry status
The current registry record identifies PETACC-3 as completed.
5. Endpoints
The ClinicalTrials.gov record does not list registered primary endpoints for PETACC-3. Consequently, there is no endpoint definition or registry-specified time frame available here to reproduce as a formal primary endpoint table.
6. Statistical Methodology
Randomized comparison
Randomization is the central statistical feature of PETACC-3. When patients are assigned randomly to treatment strategies, baseline prognostic factors are expected to be distributed between groups in the long run. This permits differences in subsequently measured outcomes to be interpreted as treatment-group differences, subject to adherence, follow-up, missing information, and other features of the trial.
The appropriate numerical form of this contrast depends on the endpoint: it could be a difference in proportions, a difference in means, a ratio, or a time-to-event effect measure. The registry record does not identify the primary endpoint needed to select one of these as the trial's primary estimand.
Intention-to-treat principle
For a randomized treatment comparison, an intention-to-treat analysis generally retains patients in the groups to which they were randomized. This preserves the treatment comparison created by randomization and avoids allowing post-randomization treatment decisions to redefine the primary comparison.
Analysis populations
The registry information available here does not define specific efficacy or safety analysis populations. In a full statistical analysis plan, the analysis population would normally be specified before the outcome analysis, including how treatment discontinuation, protocol deviations, withdrawals, and missing outcome measurements are handled.
Unmasked design
PETACC-3 is recorded as having no masking. Lack of masking does not invalidate randomization, but it can influence the measurement or management of outcomes when knowledge of treatment assignment can affect assessment, reporting, treatment changes, or follow-up procedures.
Phase 3 context
A phase 3 randomized comparison is ordinarily designed to provide a confirmatory assessment of a treatment strategy. The statistical analysis should therefore be anchored to prespecified endpoints, an analysis population, an estimand, and an error-control strategy. Those details are not listed among the registered primary-endpoint information in the registry record available here.
7. Planned Analysis
The registry describes PETACC-3 as a randomized phase 3 treatment trial comparing fluorouracil and leucovorin with or without irinotecan following surgery for stage III colorectal cancer. A complete statistical analysis would normally begin by specifying the endpoint and its measurement time frame before choosing the corresponding model or hypothesis test.
| Statistical question | Typical analysis framework |
|---|---|
| Time-to-event endpoint | Kaplan-Meier estimation, a log-rank comparison, and a Cox proportional-hazards model may be appropriate when the endpoint is defined as time to an event. |
| Binary endpoint | Risk differences, risk ratios, odds ratios, or regression models may be used depending on the endpoint definition and prespecified estimand. |
| Continuous endpoint | A treatment-group difference or regression-based comparison may be appropriate, with the model determined by the scale and distribution of the endpoint. |
| Repeated measurements | A longitudinal model can account for correlations among measurements from the same patient when repeated outcomes are collected. |
| Safety endpoint | Adverse-event incidence is commonly summarized by treatment group, with event definitions, severity categories, exposure, and follow-up specified in advance. |
Because the registry does not list a registered primary endpoint, the exact primary statistical test and effect measure cannot be assigned from the registry record alone. Results have not been posted on ClinicalTrials.gov in the record information available here.
8. Statistical Methods Explained
Why does randomization matter?
Randomization creates the treatment comparison before outcomes are observed. If allocation is properly implemented, prognostic factors—both measured and unmeasured—are expected to be balanced on average between treatment groups. This reduces confounding and makes the randomized treatment assignment the basis for the primary comparison.
Why does the absence of masking matter?
Masking and randomization address different sources of bias. Randomization concerns allocation to treatment, whereas masking concerns knowledge of that allocation after assignment. PETACC-3 is recorded as unmasked, so an analyst should consider whether knowledge of treatment could affect outcome assessment, treatment decisions, reporting, or follow-up.
What would determine the effect measure?
The endpoint determines the appropriate effect measure. For a binary endpoint, an absolute risk difference or relative measure may be useful. For a continuous endpoint, a mean difference may be natural. For a time-to-event endpoint, a hazard ratio or survival probability at a specified time may be appropriate. Without a registered primary endpoint definition, there is no single effect measure that can be identified as the primary PETACC-3 estimate from the registry record.
Why is intention-to-treat important?
Intention-to-treat analysis preserves the original randomized comparison. If patients who discontinue treatment or switch therapy are systematically removed from their randomized groups, the resulting comparison can differ from the treatment strategy that was originally randomized. For this reason, randomized trials generally define the primary analysis population in advance.
When would a Cox model be appropriate?
A Cox proportional-hazards model is appropriate when the primary outcome is a time-to-event endpoint and the analysis is intended to estimate a relative hazard. It does not apply automatically to every clinical-trial endpoint. If used, the proportional-hazards assumption should be considered because a single hazard ratio summarizes a potentially time-varying relationship.
How should missing outcomes be handled?
Missing outcome data can change both the precision and validity of an estimated treatment effect. A statistical analysis plan should define censoring rules, withdrawal handling, and any imputation strategy before examining comparative results. The registry information available here does not specify a missing-data or imputation strategy.
9. Randomization and Causal Interpretation
The randomized allocation is the strongest design feature for causal comparison in this record. Before treatment is assigned, randomization separates the treatment decision from the prognostic factors that could otherwise influence outcomes.
What randomization supports
A comparison of outcomes between the randomized treatment strategies, assuming appropriate conduct and analysis of the trial.
What randomization does not guarantee
It does not guarantee identical baseline characteristics in a particular finite sample, complete follow-up, adherence to assigned treatment, or absence of measurement bias in an unmasked trial.
The distinction is important because a randomized design establishes the framework for causal inference, while the quality of the final estimate still depends on outcome definition, follow-up, missing data, treatment exposure, and the prespecified analysis.
10. Safety Analysis Framework
The registry information available here identifies the treatment interventions but does not provide posted serious adverse-event counts by treatment arm. A numerical safety comparison therefore cannot be reported from the registry record.
| Safety component | Registry information available |
|---|---|
| Intervention regimen | FOLFIRI regimen, including fluorouracil, irinotecan hydrochloride, and leucovorin calcium. |
| Comparator regimen | Fluorouracil and leucovorin without irinotecan, as described in the registry title. |
| Serious adverse events by arm | Not reported in the available registry information. |
| Numerical adverse-event comparison | Not available from the registry information. |
For a randomized safety analysis, event counts should be interpreted alongside the number of treated patients and the amount of exposure or follow-up when relevant. Serious adverse events, treatment-emergent adverse events, treatment discontinuations, and deaths are distinct safety concepts and should not be collapsed into a single measure.
11. Multiplicity and Confirmatory Inference
Multiplicity becomes important when a trial evaluates more than one endpoint, multiple time points, multiple treatment comparisons, or numerous subgroup hypotheses. A prespecified multiplicity strategy controls the probability of making one or more false-positive claims within the relevant family of hypotheses.
For a confirmatory phase 3 trial, the statistical analysis plan would ordinarily define the primary endpoint or endpoints and specify how the type I error is controlled.
The registry information available here does not report a multiplicity strategy. Accordingly, no particular alpha-allocation or hierarchical testing procedure can be attributed to PETACC-3 from the registry record.
12. Interim Analysis
An interim analysis examines accumulating trial information before the planned final analysis. Interim looks can be useful for monitoring efficacy, futility, safety, or other design considerations, but repeated unadjusted efficacy testing can increase the overall type I error.
The registry information available here does not identify an interim-analysis schedule, stopping boundary, alpha-spending procedure, or group-sequential design. Those elements therefore cannot be characterized as features of the PETACC-3 statistical design from this record.
13. Crossover and Treatment Switching
Crossover is a post-randomization change in treatment that can complicate the interpretation of a randomized comparison. If patients assigned to one strategy subsequently receive components of another strategy, the difference between treatment groups can become smaller than it would have been under complete separation of treatment strategies.
The registry information available here does not report a crossover policy or treatment-switching results. No numerical crossover adjustment or treatment-switching analysis is therefore presented.
14. Missing Data and Imputation
Missing data have different statistical consequences depending on why the measurements are missing and how the endpoint is defined. For a longitudinal continuous endpoint, missing observations can require a model or imputation strategy. For a time-to-event endpoint, a patient may instead be censored under prespecified rules. For a binary endpoint, missing observations may require a defined analysis population or sensitivity analysis.
Primary analysis
The analysis plan should define how withdrawals, incomplete assessments, and other missing observations enter the primary estimator.
Sensitivity analysis
Alternative assumptions about missing outcomes can be used to assess whether the principal conclusion is sensitive to departures from the primary missing-data assumptions.
The registry record does not specify an imputation method or missing-data sensitivity analysis.
15. Bayesian Methods
No Bayesian statistical method is identified in the registry information available for PETACC-3. The trial is recorded as a randomized phase 3 study, but the available registry information does not specify a Bayesian prior, posterior probability, Bayesian decision rule, or other Bayesian component.
This distinction matters because Bayesian and frequentist analyses can use the same randomized trial data while answering inferential questions in different ways. A Bayesian analysis would require explicit specification of its prior distribution and decision framework rather than being inferred from the trial phase or randomized design.
16. Non-Inferiority Framework
The registry information available here does not identify a non-inferiority design or non-inferiority margin. The trial title instead describes a randomized comparison of fluorouracil and leucovorin with or without irinotecan.
In a non-inferiority trial, the treatment effect is evaluated against a prespecified margin rather than simply asking whether a conventional two-sided comparison is statistically significant. Without a documented non-inferiority margin, no such framework should be attributed to PETACC-3.
17. Statistical Interpretation of the Treatment Comparison
PETACC-3 is a randomized phase 3 treatment trial comparing fluorouracil and leucovorin with or without irinotecan in patients following surgery for stage III colorectal cancer.
The randomized allocation provides the structural basis for comparing outcomes between the treatment strategies. However, the registry information available here does not provide a registered primary endpoint, effect estimate, confidence interval, or p-value.
The meaning of a numerical treatment estimate depends on the endpoint and estimand. A risk difference describes an absolute difference in event probabilities; a risk ratio describes relative probabilities; a mean difference compares average values; and a hazard ratio compares modeled instantaneous event rates for a time-to-event outcome.
A confidence interval quantifies statistical uncertainty around an estimated effect under the specified analysis framework. It does not describe the range of effects experienced by individual patients and does not substitute for the clinical definition of the endpoint.
A p-value addresses compatibility of the observed data with a specified null hypothesis under the statistical model. It does not measure the magnitude or clinical importance of an effect. Effect estimates and their confidence intervals are needed to describe the size and precision of a treatment comparison.
18. Limitations
- No registered primary endpoint is listed: the registry information available here does not provide a primary endpoint definition or time frame.
- No posted statistical analyses: no formal statistical analysis results are posted in the registry information available here.
- No numerical efficacy results: treatment-effect estimates, confidence intervals, event counts, and p-values are not available in the registry information.
- No enrollment value: the enrollment field is not populated in the available record information.
- Unmasked design: knowledge of treatment assignment can matter for outcomes involving assessment, reporting, or subsequent treatment decisions.
- Unspecified analysis population: the available registry information does not define the efficacy or safety analysis populations.
- Unspecified missing-data strategy: the available registry information does not describe imputation, censoring, or missing-data sensitivity analyses.
- Unspecified multiplicity strategy: the available registry information does not identify an alpha-allocation or multiplicity-control procedure.
- Unspecified interim analysis: no interim schedule or stopping boundary is identified in the available registry information.
19. Why This Trial Matters Statistically
PETACC-3 is a useful statistical teaching case because its registry design illustrates the distinction between a randomized clinical question and the endpoint-specific statistical analysis needed to answer it. The treatment comparison is clearly defined, while the available registry information does not provide the primary endpoint or posted statistical results.
| Concept | How it appears in PETACC-3 |
|---|---|
| Randomization | The registry classifies allocation as randomized. |
| Phase 3 design | The trial is classified as phase 3. |
| Treatment comparison | Fluorouracil and leucovorin are evaluated with or without irinotecan. |
| Unmasked treatment | Masking is recorded as none. |
| Clinical setting | Patients are described as following surgery for stage III colorectal cancer. |
| Primary purpose | The registry classifies the study as a treatment trial. |
| Statistical estimand | Requires an endpoint definition and analysis specification that are not listed in the available registry information. |
| Multiplicity | No registry multiplicity strategy is identified in the available information. |
| Interim monitoring | No registry interim-analysis procedure is identified in the available information. |
| Bayesian analysis | No Bayesian method is identified in the available information. |
20. A Practical Statistical Reading of PETACC-3
A rigorous reading of a randomized trial begins with the estimand rather than the p-value. The analyst first identifies the population, treatment strategies, endpoint, time horizon, and summary measure. Only then can the appropriate statistical model and hypothesis test be selected.
For PETACC-3, the population and randomized treatment comparison are clear from the registry title and design fields. The available record does not, however, identify a registered primary endpoint. That distinction prevents an analyst from treating a generic statistical method as though it were the trial's prespecified primary analysis.
Design question
Were patients randomized between the irinotecan-containing and non-irinotecan treatment strategies? The registry records the allocation as randomized.
Analysis question
What endpoint and time frame were used to quantify the treatment effect? The available registered primary-endpoint information does not specify them.
Inference question
What uncertainty surrounds the treatment estimate? No posted estimate or confidence interval is available in the registry information.
Clinical question
Does adding irinotecan to fluorouracil and leucovorin alter outcomes after surgery for stage III colorectal cancer? The randomized design addresses this question, while the endpoint definition determines how it is quantified.
21. Sources
- ClinicalTrials.gov: PETACC-3, NCT00026273.
Continue through the Clinical Biostats clinical trial library
Clinical trial analyses are most informative when the randomized design, endpoint definition, estimand, statistical model, uncertainty, and limitations are considered together.
22. Record Summary
PETACC-3 is a completed randomized phase 3 treatment trial in colorectal cancer, evaluating fluorouracil and leucovorin with or without irinotecan in patients following surgery for stage III colorectal cancer. The registry records the study as unmasked, with a primary purpose of treatment, a start date of 2001-01, and a primary completion date of 2004-09, with Pfizer as the lead industry sponsor.
From a statistical perspective, the key feature is the randomized treatment comparison. A complete endpoint-level analysis requires the prespecified outcome definition, analysis population, effect measure, and statistical model. The ClinicalTrials.gov record available for this analysis does not list registered primary endpoints or posted statistical analyses, so no treatment-effect estimate, confidence interval, p-value, or outcome result is reported here.