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
AMAROS was a randomized, parallel-group, open-label phase 3 trial comparing complete axillary lymph node dissection with axillary radiation therapy in women with invasive breast cancer. The registry lists axillary recurrence rate from randomization as the primary endpoint.
| Feature | AMAROS |
|---|---|
| Acronym | AMAROS |
| ClinicalTrials.gov identifier | NCT00014612 |
| Phase | Phase 3 |
| Status | Completed |
| Condition | Breast Cancer |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | None |
| Primary purpose | Treatment |
| Enrollment | 4,813 |
| Number of arms | 2 |
| Lead sponsor | European Organisation for Research and Treatment of Cancer - EORTC |
| Sponsor type | Network |
2. Clinical Question
The registered study title describes a direct comparison of complete axillary lymph node dissection with axillary radiation therapy in women with invasive breast cancer. Statistically, the central question is whether the randomized treatment strategies differ with respect to the rate of axillary recurrence during follow-up beginning at randomization.
Population
Women with invasive breast cancer enrolled in the phase 3 AMAROS trial.
Intervention
Axillary lymph node dissection, with the registry also identifying lymphoscintigraphy and therapeutic conventional surgery among the study interventions.
Comparator
Radiation therapy directed to the axillary treatment strategy.
Primary question
How does randomized assignment to complete axillary lymph node dissection compare with axillary radiation therapy with respect to axillary recurrence from randomization?
3. Trial Design
The registry identifies AMAROS as a randomized, parallel-group, unmasked phase 3 treatment trial. There were two study arms and 4,813 enrolled participants.
Axillary lymph node dissection
- Axillary lymph node dissection
- Lymphoscintigraphy is identified among the registered procedures
- Therapeutic conventional surgery is identified among the registered interventions
Axillary radiation therapy
- Radiation therapy
- The study is registered as a randomized comparison of axillary treatment strategies
- The design is parallel-group and unmasked
4. Trial Timeline
Trial start
The registry records February 2001 as the study start date.
Randomized phase 3 study
The trial evaluated two axillary treatment strategies in a parallel randomized design.
Primary completion
The registry records November 2010 as the primary completion date.
Registry status
The ClinicalTrials.gov record identifies the study as completed.
5. Endpoints
The registry lists one primary endpoint for AMAROS. The endpoint is explicitly tied to time from randomization.
| Endpoint | Registry definition / time frame |
|---|---|
| Axillary recurrence rate | From randomization |
Why the endpoint wording matters
The phrase “from randomization” establishes the time origin for the primary endpoint. This is statistically important because an event rate is not fully defined without specifying the population at risk and the period over which events are observed.
“Axillary recurrence rate” is also different from a generic tumor-response percentage. It is a recurrence outcome tied to the axilla and therefore requires follow-up over time. A statistical analysis should preserve that distinction rather than treating the endpoint as if it were simply a cross-sectional binary response measured at one fixed visit.
6. Statistical Methodology
No formal statistical analyses are posted in the ClinicalTrials.gov record. The registry therefore establishes the primary endpoint and trial design, but it does not provide numerical estimates, confidence intervals, p-values, event counts, or a posted statistical-analysis result for axillary recurrence.
For an endpoint defined as axillary recurrence from randomization, the natural statistical framework is a time-to-event analysis when the registry's follow-up structure records the timing of recurrence and censoring. The analysis would compare the distribution of time to axillary recurrence between randomized treatment groups.
Participants who have not experienced axillary recurrence by the end of their available follow-up contribute information until their censoring time.
Kaplan-Meier estimation
A Kaplan-Meier estimator can describe the probability of remaining free of the event over time. This is useful when participants have different lengths of follow-up, because it does not require every participant to be observed for the same duration.
For an axillary-recurrence endpoint, S(t) can be interpreted as the probability of remaining free from axillary recurrence beyond time t, subject to the assumptions of the estimator and censoring process.
Comparison between randomized groups
A log-rank test is a conventional method for comparing time-to-event distributions between randomized groups. A Cox proportional-hazards model is another common approach for estimating a hazard ratio, provided its assumptions are appropriate and the necessary event-time data are available.
Rate versus time-to-event probability
The wording “recurrence rate” can be interpreted in more than one statistical way. If recurrence is reported at a fixed follow-up time, a cumulative incidence proportion may be appropriate. If recurrence times and censoring are central to the analysis, a time-to-event method provides a more complete description. The registry's statement “from randomization” makes the time origin explicit but does not itself specify a particular statistical estimator.
7. Planned Analysis
The registered primary endpoint is axillary recurrence rate from randomization. For this type of endpoint, a typical analysis would compare the occurrence of axillary recurrence between the two randomized treatment groups over follow-up.
Primary estimand
The treatment contrast concerns axillary recurrence following randomized assignment to axillary lymph node dissection versus axillary radiation therapy.
Time origin
Randomization is the registered starting point for measuring the primary endpoint.
Typical analysis
If event timing and censoring are available, Kaplan-Meier methods and a time-to-event comparison such as the log-rank test would ordinarily be considered.
Effect estimate
A time-to-event analysis could summarize the relative difference with a hazard ratio and confidence interval, while fixed-time analyses could report an absolute recurrence probability.
8. Statistical Methods Explained
What does “from randomization” mean?
It defines the time origin for the primary endpoint. Each participant enters the risk period at randomization, so the comparison is anchored to the point at which randomized treatment assignment occurs. This is important for causal interpretation because randomization establishes the treatment groups before the subsequent outcome experience.
Why would a time-to-event method be useful for axillary recurrence?
Participants in a clinical trial may have different amounts of observed follow-up. Some may experience recurrence, while others remain recurrence-free when their observation ends. A time-to-event analysis can use the available follow-up without treating every participant as though they had identical observation time.
What would a Kaplan-Meier curve show?
A Kaplan-Meier curve would estimate the probability of remaining free of axillary recurrence as follow-up progresses. Two curves could then be compared to describe how recurrence-free experience differs between the randomized treatment groups.
What would a hazard ratio mean in this setting?
If a Cox model were used, a hazard ratio below 1 would indicate a lower estimated instantaneous rate of axillary recurrence in the numerator treatment group relative to the reference group, under the model. It would not mean that the same percentage of patients had their recurrence risk reduced by that amount.
Why is censoring important?
Censoring occurs when a participant's recurrence status is no longer observed after a particular point in time. A time-to-event analysis uses information up to that point. Its validity depends on assumptions about the relationship between censoring and the event process.
Would a p-value measure the size of the treatment effect?
No. A p-value addresses evidence against a specified null hypothesis under a statistical model. It does not describe the magnitude of the treatment effect. Effect estimates and their confidence intervals are needed to describe magnitude and precision.
9. Interpreting the Primary Endpoint
The primary endpoint is a clinically specific recurrence outcome rather than a generic measure of survival or tumor shrinkage. Its interpretation depends on both whether recurrence occurs and when it occurs after randomization.
A relative effect, such as a hazard ratio, answers a different question from an absolute recurrence probability at a particular time. A complete statistical presentation would ideally make both perspectives available when the underlying results support them.
A confidence interval around an effect estimate describes statistical precision under the specified analysis framework. It does not describe the range of outcomes that individual patients will experience.
A p-value is not a measure of clinical importance, treatment benefit, or the probability that one treatment is superior. It is evidence about compatibility with a null hypothesis under the assumptions of the statistical test.
10. Statistical Design Features
| Design feature | Registered description | Statistical implication |
|---|---|---|
| Phase | Phase 3 | The study is a confirmatory-stage randomized clinical trial rather than an exploratory early-phase study. |
| Allocation | Randomized | Randomization supports comparison of outcomes according to assigned treatment. |
| Design model | Parallel | Participants are evaluated within one of two parallel treatment strategies rather than being repeatedly assigned to different interventions. |
| Masking | None | The trial was unmasked, so treatment assignment was not concealed through a blinded design. |
| Primary purpose | Treatment | The registered purpose was evaluation of treatment strategies. |
| Arms | 2 | The primary comparison is between two randomized treatment strategies. |
Why the unmasked design matters
An unmasked trial does not invalidate randomization. Randomization and masking answer different methodological questions. Randomization concerns how treatment assignment is generated; masking concerns whether participants, clinicians, assessors, or analysts know the assignment.
For an outcome involving recurrence, masking can matter particularly when assessment requires clinical judgment. The appropriate interpretation therefore depends on how axillary recurrence was defined and assessed, information that is not specified in the registry material summarized here.
11. Population and Interventions
The registry identifies Breast Cancer as the study condition and describes the trial as a comparison of complete axillary lymph node dissection with axillary radiation therapy in women with invasive breast cancer.
| Category | Registered information |
|---|---|
| Condition | Breast Cancer |
| Study population described in title | Women with invasive breast cancer |
| Intervention | Axillary lymph node dissection |
| Additional registered procedure | Lymphoscintigraphy |
| Additional registered procedure | Therapeutic conventional surgery |
| Comparator treatment | Radiation therapy |
The registry therefore establishes the broad clinical comparison but does not provide, in the record summarized here, a baseline demographic table or numerical description of disease characteristics. Those characteristics should not be inferred from the trial title alone.
12. Analysis Population Considerations
Because AMAROS is randomized, the principal efficacy comparison would ordinarily be anchored to randomized treatment assignment. This preserves the causal contrast created at randomization and avoids redefining the treatment groups according to events that occur after assignment.
Randomized population
The registry reports 4,813 enrolled participants in a randomized two-arm phase 3 trial.
Primary endpoint
The primary outcome is axillary recurrence rate measured from randomization.
Post-randomization events
Subsequent clinical events occur after treatment assignment and should not be used to redefine the original randomized comparison without an explicit estimand and analysis plan.
Missing follow-up
For time-to-event analysis, loss of follow-up is handled through censoring rather than automatically treating the participant as having experienced recurrence.
13. Censoring and Follow-Up
Time-to-event analyses require a rule for what happens when axillary recurrence has not been observed by the end of available follow-up. Such a participant is typically censored at the last time their recurrence status is known.
Censoring does not mean that the participant is assumed to have remained recurrence-free forever. It means that the statistical analysis uses the information available up to the censoring time.
A valid survival analysis distinguishes an observed recurrence-free period from the unobserved period after a participant's last available assessment.
The validity of the resulting estimate depends on assumptions about censoring. If the probability of being censored is related to the underlying recurrence risk in ways not adequately addressed by the analysis, the estimated recurrence distribution can be biased.
14. Confidence Intervals and Precision
A treatment comparison should be interpreted through both its point estimate and its uncertainty. If the primary analysis were expressed as a hazard ratio, the point estimate would summarize the relative treatment effect while the confidence interval would describe its statistical precision.
| Quantity | What it communicates |
|---|---|
| Effect estimate | The estimated magnitude and direction of the treatment contrast. |
| Confidence interval | The statistical uncertainty surrounding the effect estimate under the specified model. |
| P-value | The degree of incompatibility between the observed data and a specified null hypothesis under the test assumptions. |
| Absolute recurrence probability | The estimated probability of experiencing the recurrence event by a specified time, when such a time-specific estimate is available. |
These quantities should not be treated as interchangeable. In particular, a small p-value does not establish that a treatment effect is large, and a large p-value does not prove that the treatments are equivalent.
15. Proportional-Hazards Considerations
If a Cox proportional-hazards model were used for the axillary recurrence endpoint, its hazard ratio would conventionally be interpreted under a proportional-hazards framework. This means the relative event rate is assumed to be reasonably represented by a common hazard ratio over the relevant follow-up.
If hazards vary substantially over time in a nonproportional manner, a single hazard ratio may compress important temporal information. In that situation, Kaplan-Meier estimates at clinically relevant times, restricted mean survival time, time-varying effects, or other prespecified approaches may provide complementary information.
16. What Would Constitute a Complete Statistical Result?
For the registered primary endpoint, a complete numerical statistical presentation would ideally include the number of axillary recurrence events in each randomized group, the follow-up framework, an effect estimate, a measure of uncertainty, and the statistical comparison used.
| Result component | Purpose |
|---|---|
| Event count | Shows how much direct event information contributes to the comparison. |
| Time-to-event estimate | Describes the recurrence experience over follow-up. |
| Effect estimate | Quantifies the relative or absolute treatment contrast. |
| Confidence interval | Shows statistical precision around the estimate. |
| P-value | Provides the formal test result for the specified null hypothesis, when one is used. |
| Analysis population | Defines which randomized participants contribute to the efficacy comparison. |
The ClinicalTrials.gov record does not post these numerical statistical results for the primary endpoint.
17. Limitations and Interpretation Issues
- No posted statistical analysis: the registry lists the primary endpoint but does not post a formal statistical analysis or numerical result for axillary recurrence.
- Limited endpoint detail in the registry record: the primary endpoint is identified as axillary recurrence rate from randomization, but the record summarized here does not specify a detailed estimator, test, or model.
- Unmasked design: the trial was not masked, which can matter for outcomes involving clinical assessment or treatment decisions.
- Time-to-event assumptions: if recurrence is analyzed with survival methods, interpretation depends on assumptions concerning censoring and, for Cox models, proportional hazards.
- Event definition: recurrence endpoints require a consistent operational definition and ascertainment process. The registry summary does not provide those additional details here.
- Absence of numerical results: without event counts and estimates, the magnitude and precision of the randomized treatment contrast cannot be characterized from the registry record alone.
- Generalizability: the registry identifies women with invasive breast cancer as the study population, but the summarized record does not provide enough baseline detail to characterize applicability across specific clinical subgroups.
18. Why This Trial Matters Statistically
AMAROS is a useful statistical teaching case because the design creates a clear randomized comparison while the primary endpoint introduces the central issues of recurrence analysis and time-to-event methodology.
| Concept | How it appears in AMAROS |
|---|---|
| Randomization | Participants were randomized to two treatment strategies. |
| Parallel-group design | The registry identifies a parallel design with two arms. |
| Phase 3 confirmation | The study is registered as phase 3. |
| Time origin | The primary endpoint is measured from randomization. |
| Recurrence analysis | The primary endpoint is axillary recurrence rate. |
| Censoring | Relevant if recurrence is analyzed as a time-to-event outcome. |
| Kaplan-Meier estimation | A conventional approach for describing recurrence-free time when event timing and censoring are available. |
| Log-rank testing | A conventional method for comparing time-to-event distributions between randomized groups. |
| Cox modeling | A conventional framework for estimating a hazard ratio for a time-to-event endpoint. |
| Unmasked design | The registry identifies no masking. |
19. Statistical Methods in Context
Randomization answers the causal question
The randomized allocation creates the primary comparison between treatment strategies. The statistical analysis then quantifies how the observed recurrence experience differs between those groups.
Time matters
Axillary recurrence is not simply a count of events. The timing of recurrence and the amount of follow-up are central to interpreting the endpoint.
Censoring preserves usable information
Participants without a recurrence observed during available follow-up can still contribute information before their censoring time in an appropriate survival analysis.
Effect size needs context
A relative effect estimate is most informative when accompanied by uncertainty and, where appropriate, absolute recurrence estimates at clinically meaningful times.
20. Registry Record Summary
AMAROS is registered as a completed phase 3 randomized parallel-group treatment trial involving 4,813 participants and two arms. The condition is Breast Cancer, the registered population is women with invasive breast cancer, and the study compares complete axillary lymph node dissection with axillary radiation therapy.
The primary endpoint is axillary recurrence rate, from randomization. That wording naturally leads to a statistical framework centered on recurrence over time, with randomization defining the treatment comparison and follow-up defining the opportunity for recurrence to occur.
The registry does not post a formal statistical analysis or numerical result for the primary endpoint. Consequently, the most important statistical lesson available from the record is methodological: recurrence outcomes in randomized trials require careful definition of the time origin, event, censoring rules, analysis population, effect measure, and uncertainty.
21. Sources
- ClinicalTrials.gov: AMAROS — NCT00014612.
- PubMed record: PubMed.
- PubMed record: PubMed.
- PubMed record: PubMed.
- PubMed record: PubMed.
- PubMed record: PubMed.
Continue through the Clinical Biostats clinical-trial library
Explore additional clinical-trial statistical analyses and educational resources covering study design, endpoints, survival analysis, and interpretation.