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
ACT II was a randomized, open-label phase 3 treatment trial in patients with primary anal cancer. The registered title describes radiation therapy plus fluorouracil with or without additional chemotherapy. The registry reports an enrollment of 600 participants.
| Feature | ACT II |
|---|---|
| Phase | Phase 3 |
| Condition | Anal Cancer |
| Trial title | Radiation Therapy Plus Fluorouracil With or Without Additional Chemotherapy in Treating Patients With Primary Anal Cancer |
| Design | Randomized |
| Masking | None |
| Primary purpose | Treatment |
| Enrollment | 600 |
| Status | Completed |
| Start date | 2001-03 |
| Lead sponsor | University College London (UCL) Cancer Institute |
| Sponsor type | Other |
| ClinicalTrials.gov | NCT00025090 |
2. Clinical Question
The registered clinical question concerns treatment of primary anal cancer using radiation therapy plus fluorouracil, with the randomized comparison involving the presence or absence of additional chemotherapy.
Population
Patients with primary anal cancer.
Core treatment
Radiation therapy plus fluorouracil.
Additional treatment
The registered title describes the comparison as radiation therapy plus fluorouracil with or without additional chemotherapy.
Primary statistical question
How does the randomized treatment strategy compare with respect to complete response, acute toxicity, and recurrence-free survival?
3. Trial Design
The registry classifies ACT II as a randomized phase 3 study with no masking and a primary purpose of treatment. The enrollment is 600 participants.
Radiation therapy plus fluorouracil
- Radiation therapy
- Fluorouracil
Additional chemotherapy
- The registered title describes treatment with or without additional chemotherapy.
- The registry lists cisplatin and mitomycin C among the trial interventions.
4. Registered Interventions
The registry lists four interventions associated with ACT II: cisplatin, fluorouracil, mitomycin C, and radiation therapy.
| Intervention | Type |
|---|---|
| Cisplatin | Drug |
| Fluorouracil | Drug |
| Mitomycin C | Drug |
| Radiation therapy | Radiation |
The registered intervention list identifies the treatments evaluated in the study, while the trial title provides the principal comparison as radiation therapy plus fluorouracil with or without additional chemotherapy. The registry information does not provide further numerical treatment-allocation details in the record summarized here.
5. Registered Primary Endpoints
ACT II has three registered primary endpoints. They address different dimensions of treatment effect: tumor response, short-term toxicity, and longer-term disease control.
| Primary endpoint | Time frame / definition | Statistical role |
|---|---|---|
| Complete response rate at 6 months | Complete response rate assessed at 6 months | Binary response endpoint |
| Acute toxicity as measured up to 4 weeks after chemoradiation | Acute toxicity measured through 4 weeks after chemoradiation | Safety endpoint |
| Recurrence-free survival | Recurrence-free survival | Time-to-event endpoint |
Three endpoints, three statistical questions
These endpoints are not interchangeable. Complete response rate asks about the proportion of patients meeting a response criterion at a specified time. Acute toxicity asks about adverse effects during a defined short-term window. Recurrence-free survival incorporates both the timing of recurrence and the timing of censoring or other qualifying events.
Response
What proportion of patients have achieved complete response at 6 months?
Acute safety
What acute toxicity is observed up to 4 weeks after chemoradiation?
Recurrence-free survival
How does the time until recurrence or another protocol-defined event differ between randomized treatment groups?
Different estimands
A response proportion, a toxicity measure, and a time-to-event endpoint require different statistical summaries and should not be collapsed into one measure.
6. Statistical Methodology
Complete response rate
Complete response rate at 6 months is a binary endpoint at a prespecified assessment time. For each randomized treatment group, the basic descriptive quantity is the number of patients achieving complete response divided by the relevant analysis population.
A comparative analysis would normally report the response proportion in each randomized group together with an effect measure such as a risk difference or risk ratio and an appropriate confidence interval. The choice of denominator is important because missing assessments can affect the apparent response rate.
Acute toxicity
Acute toxicity is defined in the registry as being measured up to 4 weeks after chemoradiation. Because this is a safety endpoint rather than a tumor-response endpoint, analysis would ordinarily summarize the frequency and severity of toxicity during the specified window.
For a randomized comparison, a useful presentation would include the number and percentage of patients experiencing the prespecified toxicity outcome in each treatment group. Depending on the exact toxicity definition and its grading structure, categorical methods could include a risk difference, risk ratio, odds ratio, or an appropriate contingency-table test.
Recurrence-free survival
Recurrence-free survival is a time-to-event endpoint. A conventional analysis would estimate the recurrence-free survival distribution with the Kaplan-Meier method and compare randomized treatment groups with a time-to-event comparison such as the log-rank test. A Cox proportional-hazards model can provide a hazard ratio as a relative measure of the event rate over follow-up.
Here, di represents events at time ti, while ni is the number at risk immediately before that time. Patients who remain event-free until their last observation can contribute information through their censoring time.
Hazard ratio interpretation
A hazard ratio is a relative time-to-event measure. It is not the same as a risk ratio, a risk difference, a probability of remaining recurrence-free at a particular time, or the proportion of patients who benefit.
7. Analysis Populations and Randomization
The registry identifies ACT II as randomized. Randomization is central to the statistical interpretation because it creates treatment groups whose differences, in expectation, are attributable to the assigned treatment strategy rather than systematic selection into treatment.
| Design feature | Registry information | Statistical significance |
|---|---|---|
| Allocation | Randomized | Supports an unbiased comparison of randomized treatment assignments under appropriate conduct and analysis. |
| Masking | None | Knowledge of assignment can affect subjective assessment and reporting, particularly for some safety or clinical outcomes. |
| Primary purpose | Treatment | The trial is intended to evaluate treatment strategies rather than prevention, diagnosis, or another primary purpose. |
| Enrollment | 600 | Defines the total registered enrollment available for the trial comparison. |
The registry information summarized here does not report treatment-specific randomized sample sizes, analysis-population definitions, stratification factors, or a detailed statistical analysis plan. Those details therefore cannot be characterized as features of the registered analysis from this record alone.
8. Planned Analysis
Complete response rate at 6 months
The registry says that complete response rate will be assessed at 6 months. A standard analysis for a binary endpoint would compare the proportion of patients with complete response between randomized groups. The analysis would ordinarily report the observed proportion in each group and a confidence interval for the treatment contrast, with the precise method depending on the prespecified statistical analysis plan and handling of patients without an assessable response at 6 months.
Acute toxicity up to 4 weeks
The registry says that acute toxicity will be measured up to 4 weeks after chemoradiation. A standard safety analysis would summarize the frequency of the defined toxicity outcome during this window by randomized treatment group or by treatment received, depending on the prespecified safety population and estimand. For categorical toxicity outcomes, comparative measures such as risk differences or risk ratios are often more directly interpretable than a time-to-event hazard ratio.
Recurrence-free survival
The registry says that recurrence-free survival is a primary endpoint. A standard analysis would use Kaplan-Meier estimation to describe recurrence-free survival over time and a log-rank or related time-to-event comparison to compare randomized groups. A Cox proportional-hazards model is commonly used to estimate a hazard ratio and its confidence interval.
9. Statistical Methods Explained
Why is complete response rate a binary endpoint?
At 6 months, each patient can be classified according to whether the prespecified complete-response criterion has been met. This creates a binary outcome for the primary analysis. The resulting comparison focuses on the probability of complete response by the specified assessment time rather than on the entire trajectory of tumor response.
Why does the 6-month time point matter?
A response rate is meaningful only relative to its assessment time. Complete response at 6 months answers a different question from complete response at an earlier or later assessment. Fixing the time point makes the endpoint definition reproducible and gives both randomized groups the same target assessment window.
Why is acute toxicity given a 4-week window?
The registry specifically defines acute toxicity as being measured up to 4 weeks after chemoradiation. A fixed safety window prevents acute events from being mixed indiscriminately with later toxicity. The statistical analysis should therefore preserve that time boundary when defining who is eligible for the acute-toxicity assessment and which events count.
Why use Kaplan-Meier methods for recurrence-free survival?
Not every participant necessarily experiences recurrence at the same time, and some participants may remain event-free when observation ends. Kaplan-Meier estimation accommodates this right-censoring by using each participant's available follow-up rather than treating patients without an observed event as though they experienced no event throughout an unlimited follow-up period.
What would a hazard ratio mean for recurrence-free survival?
If a Cox model were used, the hazard ratio would compare the estimated instantaneous rate of the recurrence-free-survival event between randomized groups. A value below 1 would indicate a lower estimated event rate for the numerator group under the model. It would not directly state the percentage of patients who remain recurrence-free at 6 months or another specific time.
Why does an open-label design matter statistically?
Because masking is reported as none, participants and investigators may know treatment assignment. That knowledge does not remove the benefit of randomization, but it can be relevant for outcomes that involve clinical judgment or patient reporting. For a response endpoint, standardized assessment procedures can reduce assessment variability; for toxicity, knowledge of treatment can influence reporting or attribution.
Why is the absence of posted results important?
The registry identifies the endpoints but does not provide corresponding statistical estimates, confidence intervals, or p-values. Consequently, the registered record supports understanding of the planned outcome framework but does not provide numerical evidence for the treatment comparison.
10. Interpreting the Three Primary Endpoints Together
The structure of ACT II illustrates why clinical trials often need multiple endpoints. Each endpoint captures a different dimension of treatment experience.
| Endpoint | Question answered | Primary statistical perspective |
|---|---|---|
| Complete response rate at 6 months | How many patients have achieved complete response by 6 months? | Binary proportion and treatment-group contrast |
| Acute toxicity up to 4 weeks after chemoradiation | What acute toxicity occurs during the specified short-term period? | Safety frequency and between-group comparison |
| Recurrence-free survival | How does the time until recurrence-free-survival failure differ between groups? | Kaplan-Meier, log-rank, and potentially Cox modeling |
These outcomes can point in different directions. A treatment strategy might produce a higher response rate while also increasing acute toxicity, or it might affect recurrence-free survival without producing a proportionally similar change in a fixed-time response measure. Statistical interpretation therefore requires examining each endpoint according to its own estimand and time frame.
11. Confidence Intervals, P-values, and Effect Size
Because the registry record summarized here does not post numerical statistical analyses, there are no treatment-effect estimates, confidence intervals, or p-values to interpret for ACT II.
For a binary endpoint such as complete response rate, a confidence interval would quantify uncertainty around the estimated treatment-group contrast. For recurrence-free survival, a confidence interval around a hazard ratio would describe uncertainty around the fitted relative event-rate estimate. A confidence interval is about statistical precision; it is not a prediction interval for what will happen to an individual patient.
A p-value would quantify the compatibility of the observed data with a specified null hypothesis under the statistical model and testing framework. It would not measure the size of the treatment effect, the clinical importance of an effect, or the probability that a treatment hypothesis is true.
A statistically detectable difference can be small, while a clinically meaningful difference can be estimated imprecisely in a smaller dataset. The appropriate interpretation therefore combines the effect estimate, its uncertainty, the endpoint definition, and the clinical context rather than relying on a p-value alone.
12. Recurrence-Free Survival: Statistical Interpretation
Recurrence-free survival is especially different from the 6-month response endpoint because it preserves the timing of events. Two treatment groups could have the same proportion of patients experiencing recurrence by a particular time while having different distributions of when those recurrences occurred.
Censoring
Patients who have not experienced the relevant event when observation ends may be censored rather than treated as event-free indefinitely.
Time matters
An event occurring early contributes differently to a time-to-event analysis from an event occurring much later.
Kaplan-Meier curve
The curve estimates the probability of remaining event-free over follow-up and displays the changing number of participants at risk.
Cox model
A Cox model can summarize the relative event rate with a hazard ratio while accounting for follow-up time.
13. Complete Response Rate: Statistical Interpretation
Complete response rate at 6 months is an absolute proportion. If, for illustration, one randomized group had a complete response rate of 40%, that would mean 40% of patients in the relevant analysis population met the response definition at the 6-month assessment. It would not mean that each patient's tumor burden was reduced by 40% or that treatment reduced recurrence risk by 40%.
The risk difference describes the absolute change in response probability. The risk ratio describes the relative response probability. Neither is equivalent to a survival hazard ratio.
Fixed-time response analysis also requires careful treatment of patients whose response status cannot be established at the specified assessment. Depending on the prespecified analysis plan, missing assessments can be handled in different ways. The registry information summarized here does not report a missing-response or imputation strategy.
14. Acute Toxicity: Statistical Interpretation
Acute toxicity is a safety endpoint defined over a short period: up to 4 weeks after chemoradiation. The time restriction is important because toxicity occurring during this interval represents a different clinical question from toxicity occurring substantially later.
| Statistical issue | Why it matters |
|---|---|
| Definition of toxicity | The exact event or severity threshold determines which observations enter the endpoint. |
| Assessment window | The registry specifies measurement up to 4 weeks after chemoradiation. |
| Denominator | The safety proportion depends on which patients are considered evaluable for the defined window. |
| Severity | If toxicity has multiple grades, a single overall frequency can conceal differences in severity. |
| Between-group comparison | Risk differences and risk ratios can directly quantify absolute and relative differences in toxicity frequency. |
Safety should be interpreted alongside efficacy rather than combined into an informal net-benefit score. Response, recurrence-free survival, and acute toxicity have different definitions, follow-up windows, and clinical meanings.
15. Multiplicity and Three Primary Endpoints
ACT II has three registered primary endpoints. Multiple primary endpoints create an important statistical-design question: how should the overall type I error be controlled when several hypotheses are being evaluated?
The record summarized here does not report an alpha-allocation strategy, hierarchical testing procedure, gatekeeping procedure, or other multiplicity adjustment. Those features therefore cannot be characterized as part of the registered statistical methodology from the available record.
Why multiplicity matters
Testing several hypotheses increases the chance that at least one null hypothesis will appear statistically significant by chance if each is tested independently at the same nominal level.
Possible design approaches
Trials can prespecify methods such as hierarchical testing, alpha allocation, or other familywise-error-control procedures when several confirmatory hypotheses are important.
The existence of three primary endpoints does not by itself reveal how their statistical testing was structured. The appropriate interpretation depends on the prespecified analysis plan.
16. Missing Data and Censoring
Missing data have different implications for the three primary endpoints. A missing 6-month response assessment can directly affect the denominator or classification of the complete-response endpoint. Missing acute-toxicity observations can affect ascertainment during the 4-week window. For recurrence-free survival, incomplete follow-up is commonly represented through censoring when the required conditions are satisfied.
| Endpoint | Potential statistical issue |
|---|---|
| Complete response rate at 6 months | Missing or unevaluable 6-month assessments can affect response classification and the analysis denominator. |
| Acute toxicity up to 4 weeks | Incomplete observation during the specified safety window can affect ascertainment of toxicity. |
| Recurrence-free survival | Loss of follow-up can produce right-censored observations if censoring criteria are met. |
The registry record summarized here does not report a specific imputation method, missing-data sensitivity analysis, or censoring algorithm. These choices can materially affect the interpretation of endpoint estimates and are normally prespecified in the statistical analysis plan.
17. Stratification and Covariate Adjustment
The registry identifies the trial as randomized but does not report stratification factors or a covariate-adjustment strategy in the information summarized here.
In a randomized trial, an unadjusted comparison of the randomized groups is often the principal analysis because treatment assignment provides the foundation for causal comparison. Covariate adjustment can improve precision when important prognostic variables are prespecified, but the exact adjustment variables and model should be defined before examining treatment outcomes.
18. Non-Inferiority, Crossover, and Bayesian Methods
| Design topic | What the registry record reports |
|---|---|
| Non-inferiority margin | Not reported in the ClinicalTrials.gov record summary. |
| Crossover | Not reported in the ClinicalTrials.gov record summary. |
| Bayesian methods | Not reported in the ClinicalTrials.gov record summary. |
| Interim analysis | Not reported in the ClinicalTrials.gov record summary. |
| Multiplicity strategy | Not reported in the ClinicalTrials.gov record summary. |
| Missing-data / imputation strategy | Not reported in the ClinicalTrials.gov record summary. |
| Stratification factors | Not reported in the ClinicalTrials.gov record summary. |
These are important statistical design features, but their absence from the registry information summarized here means they should not be attributed to ACT II without additional documented evidence.
19. What a Complete Statistical Results Table Would Contain
For the three primary endpoints, a complete comparative results presentation would distinguish the estimand from the statistical test and would preserve the endpoint-specific time frame.
| Endpoint | Descriptive result | Comparative effect | Uncertainty | Hypothesis test |
|---|---|---|---|---|
| Complete response rate at 6 months | Response proportion in each randomized group | Risk difference and/or risk ratio | 95% CI for the treatment contrast | Prespecified test, if applicable |
| Acute toxicity up to 4 weeks after chemoradiation | Toxicity frequency by group | Risk difference and/or risk ratio | 95% CI for the treatment contrast | Prespecified safety comparison, if applicable |
| Recurrence-free survival | Kaplan-Meier estimates and event information | Hazard ratio or another prespecified time-to-event contrast | 95% CI | Log-rank or other prespecified test |
This framework demonstrates why a single generic "p-value for the trial" would not adequately describe ACT II. Each primary endpoint has its own definition, estimand, follow-up structure, and appropriate statistical summary.
20. Statistical Interpretation vs Clinical Interpretation
Statistical interpretation
The registered study is a randomized phase 3 comparison with three primary endpoints: a 6-month complete response rate, acute toxicity measured through 4 weeks after chemoradiation, and recurrence-free survival.
Clinical interpretation
The endpoints collectively address tumor response, short-term treatment toxicity, and recurrence over time. Their results would need to be considered separately because they represent different aspects of treatment effect.
21. Important Limitations and Interpretation Issues
- No posted statistical results: the registry record summarized here does not provide numerical results, confidence intervals, or p-values for the three primary endpoints.
- Limited design detail: the available registry information identifies randomization and lack of masking but does not provide treatment-specific allocation counts, stratification factors, or a detailed analysis-population definition.
- Endpoint heterogeneity: complete response, acute toxicity, and recurrence-free survival require different statistical approaches and should not be interpreted as though they measured the same construct.
- Open-label assessment: masking is reported as none, which can be relevant to outcomes involving clinical assessment or reporting.
- Missing-data considerations: the record does not specify how missing 6-month response assessments or incomplete acute-toxicity observations are handled.
- Censoring considerations: recurrence-free survival requires explicit rules for censoring and event classification; those rules are not reported in the registry information summarized here.
- Multiplicity: three registered primary endpoints raise questions about familywise error control, but the record does not state the statistical strategy used to address this issue.
- Unreported statistical specifications: the registry information summarized here does not report a non-inferiority margin, Bayesian methodology, interim-analysis plan, or covariate-adjustment strategy.
22. Why This Trial Matters Statistically
ACT II is a useful statistical teaching case because its registered primary endpoints span three major classes of clinical-trial outcome: a fixed-time binary endpoint, a defined-window safety endpoint, and a time-to-event endpoint.
| Concept | How it appears in ACT II |
|---|---|
| Randomization | The trial is classified as randomized. |
| Open-label design | Masking is reported as none. |
| Binary endpoint | Complete response rate at 6 months. |
| Safety window | Acute toxicity measured up to 4 weeks after chemoradiation. |
| Time-to-event analysis | Recurrence-free survival. |
| Kaplan-Meier estimation | A standard method for describing recurrence-free survival over time. |
| Cox regression | A standard model for estimating a hazard ratio for a time-to-event endpoint. |
| Censoring | Relevant to incomplete follow-up for recurrence-free survival. |
| Multiplicity | Three registered primary endpoints require consideration of the overall testing framework. |
| Endpoint-specific estimands | Response, acute toxicity, and recurrence-free survival answer different statistical questions. |
23. Sources
- ClinicalTrials.gov: NCT00025090 — ACT II.
Continue through the Clinical Biostats knowledge graph
Explore additional Clinical Biostats tutorials, statistical calculators, and clinical trial analyses to deepen the statistical concepts illustrated by randomized clinical trials.
24. Record Summary
ACT II is a randomized, unmasked phase 3 treatment trial with an enrollment of 600 participants and a registered condition of anal cancer. Its three primary endpoints — complete response rate at 6 months, acute toxicity measured up to 4 weeks after chemoradiation, and recurrence-free survival — illustrate three distinct statistical problems. The first is a fixed-time binary outcome, the second is a short-term safety assessment, and the third is a longitudinal time-to-event outcome.
The statistical interpretation of ACT II therefore begins with preserving those distinctions. Complete response is naturally expressed as a proportion and treatment-group contrast; acute toxicity is summarized within its defined 4-week window; and recurrence-free survival is appropriately considered with methods that account for follow-up time and censoring. The registry does not post numerical statistical analyses for these endpoints, so the available record establishes the trial's registered design and endpoint framework without providing treatment-effect estimates.