Randomized Phase III Study Of Exemestane (Aromasin) For 5 Years Versus Tamoxifen for 2.5 to 3 Years Followed By Exemestane: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Randomized Phase III Study Of Exemestane (Aromasin) For 5 Years Versus Tamoxifen for 2.5 to 3 Years Followed By Exemestane (NCT00036270), a randomized phase 3 study. The primary outcome is Disease Free Survival (DFS): Number of Events (Disease Relapse or Death) From Baseline up to 2.75 Years.

2. Clinical Question

The trial asked whether disease-free survival differed between adjuvant exemestane used for 5 years and a sequential regimen of tamoxifen followed by exemestane, with treatment extending over a total of 5 years. Its target population was postmenopausal patients with receptor-positive breast cancer, with either node-negative or node-positive disease. The brief summary describes the setting as early breast cancer.

The primary endpoint combines disease relapse and death from any cause. Relapse includes primary tumor recurrence, whether locoregional or distant, and ipsilateral or contralateral breast cancer. The registry also lists outcomes concerning overall survival, new primary breast cancers, time to relapse, and new primary non-breast cancers. These outcome descriptions provide context, but the primary question addressed by the reported event counts is disease-free survival during the specified observation period.

3. Trial Design

This was a randomized, phase III, parallel-group study with no masking. The registry reports an enrollment of 9779 and a completed status. The recorded start was 2001-08 and primary completion was 2008-10. The record identifies Pfizer as the sponsor and breast neoplasms as the condition.

The comparison was between exemestane and tamoxifen followed by exemestane. The trial title describes exemestane for 5 years in one arm, versus tamoxifen for 2.5 to 3 years followed by exemestane, for a total of 5 years, in the other. The brief summary describes the study as comparing these approaches in adjuvant treatment of postmenopausal women with early breast cancer. Detailed arm descriptions are not reported in the registry record.

The primary endpoint is reported as the number of disease-free survival events from baseline up to 2.75 years. The record describes participants as censored at 2.75 years for this endpoint. The arm-level entries list 352 events among 4898 for exemestane and 388 among 4868 for tamoxifen followed by exemestane. These are event-count entries, not estimates of the time until an event for each participant.

The brief summary places this study within an international collaboration of investigators conducting 7 similar but independent studies in 9 countries. It says the study was designed to be part of the larger TEAM trial, with data from the 7 studies to be combined and a pre-specified analysis of pooled data conducted. The registry record does not give the pooled analysis results in the trial entry.

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.

Randomized to Study Treatment

Reasons for not completing this period:

Study Treatment to 5 Year Report

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Intent-to-Treat (ITT) population: participants randomized to one of the two study arms (all randomized participants); (n) = number of participants at observation for exemestane and tamoxifen, respectively. All participants were censored at 2.75 years.

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 Disease Free Survival (DFS): Number of Events (Disease Relapse or Death) From Baseline up to 2.75 Years, measured over Baseline (Month 0) up to 2.75 years.

The primary endpoint is disease-free survival, recorded as the number of events consisting of disease relapse or death from any cause. The registry defines disease-free survival as the time from randomization to the earliest documentation of relapse or death. Relapse includes primary tumor recurrence at a locoregional or distant site and ipsilateral or contralateral breast cancer. A death without a previously recorded relapse is also an event.

The stated observation window runs from baseline, identified as Month 0, up to 2.75 years. Participants were censored at 2.75 years. The registry describes the analysis population as intent-to-treat: participants randomized to either study arm. The event-count entries are 352 for exemestane and 388 for tamoxifen followed by exemestane, with arm-level participant counts of 4898 and 4868, respectively.

Combining relapse and death into a time-to-event endpoint captures more than recurrence alone: either event marks the endpoint. In general, this kind of composite can summarize the time until the first qualifying event, while the event counts alone do not show when events occurred or how follow-up was distributed. The registry's reported log-rank hazard ratio provides a separate time-to-event comparison. The record does not report a breakdown of the primary endpoint into relapse events and deaths.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The registry reports a log-rank analysis with a hazard ratio as its measure of comparison. In general, the log-rank test compares the pattern of time-to-event outcomes between groups over follow-up. It uses information about when events occur and how participants contribute time before an event or censoring, rather than relying only on the final event counts. The registry's reported estimate is a hazard ratio of 0.89, with a 95% confidence interval from 0.77 to 1.03 and p = 0.118.

What does a hazard ratio describe?

In general, a hazard is the instantaneous rate of an event among participants who have not yet experienced it, at a particular point in follow-up. A hazard ratio compares these rates between groups. It is not a ratio of the event counts, a difference in the probability of an event, or a statement about how long each participant lived without an event. The ratio also does not by itself show whether an observed difference is clinically important.

The order in which groups are listed in a record does not always make the numerator and reference-group coding explicit. The registry gives the groups as exemestane and tamoxifen followed by exemestane, but does not separately describe the hazard-ratio coding in the result text. The estimate should therefore be interpreted as the reported comparison, with care not to assert more about direction than the record establishes.

How should the confidence interval be read?

In general, a confidence interval communicates statistical uncertainty around an estimate under the method used. The reported 95% interval, 0.77 to 1.03, includes 1, the reference value for no relative difference in hazard. Its lower and upper limits span values on either side of that reference. Thus, the data do not pinpoint a single relative effect, and the interval allows for uncertainty that includes no difference. The interval is not a range in which 95% of individual outcomes fall.

What does the p-value contribute?

In general, a p-value assesses how unusual data at least as incompatible with a specified null hypothesis would be if that null hypothesis and the test assumptions held. The registry reports p = 0.118 for the log-rank analysis. It is not the probability that the null hypothesis is true, nor is it a measure of the probability that the treatment comparison will be replicated. The record does not report the full statistical analysis plan, including the planned significance threshold for this result.

The registry separately supplies event counts: 352 for exemestane and 388 for tamoxifen followed by exemestane. For a time-to-event endpoint, counts omit the ordering and timing of events and the time participants remained under observation. This explains why the count entries and the log-rank hazard-ratio analysis answer related but different statistical questions. The record flags the count-other endpoint comparison as having values missing or not estimable for a two-arm comparison in recalculated outputs. The posted log-rank result is therefore the available comparative analysis; the arm counts should not be used to invent another hazard ratio or significance test.

In general, censoring means that a participant's event time is not observed beyond a specified point or loss of observation, so the participant contributes information up to that point under the analysis method. Here, the primary endpoint description says all participants were censored at 2.75 years. It does not give additional details about censoring before that point, the handling of incomplete outcome information, or assumptions used for the time-to-event analysis. Those details are not reported in the registry record.

Planning calculations based on an observed effect are distinct from trial results. The stated planning assumptions describe a two-sided significance level of 0.05, equal allocation, and a normal approximation, with the target effect set as a fraction of the observed effect. For time-to-event outcomes, the assumptions describe total events; other outcomes are expressed as participants per arm. Such calculations are planning illustrations, not observed power and not evidence that the trial was designed with those assumptions. The registry record does not report the trial's own sample-size justification or design calculations.

10. Analyses Reported by the Sponsor

The registry reports a log-rank analysis with a hazard ratio of 0.89, a 95% confidence interval from 0.77 to 1.03, and p = 0.118. This is the available comparative time-to-event result. The registry also lists the primary endpoint event counts, but it flags count-other endpoint comparisons as having values missing or not estimable for a two-arm comparison in recalculated outputs. As a result, there is no separate unadjusted re-calculation to compare with the reported hazard ratio.

The posted event counts of 352 and 388 do not reproduce the log-rank analysis: they do not encode event timing or censoring information. The record marks the discrepancy status as false, so there is no stated conflict between a recalculation and the registry-reported result. If a discrepancy were present, the sponsor's own analysis would take precedence; that condition is not indicated in this record.

The comparative interpretation should therefore rest on the registry's reported analysis and its uncertainty interval. The count entries remain descriptive endpoint information, not an independent estimate of the time-to-event effect. Further details on the sponsor's full statistical analysis, including model specifications beyond the listed log-rank method, are not reported in the registry record.

11. Multiplicity, Missing Data and Analysis Populations

The registry lists secondary outcomes concerning overall survival, time to new primary breast cancers, time to relapse, and new primary non-breast cancers. It also reports 4 analyses. The record does not specify how the primary and secondary outcomes were arranged into a testing hierarchy, whether statistical testing was adjusted for multiple comparisons, or how multiplicity was handled across outcomes or analyses. No conclusion about multiplicity control can be drawn from the listed outcome names alone.

The brief summary says that a pre-specified analysis was planned for pooled data from 7 similar but independent studies in the larger TEAM collaboration. The pooled analysis results and details of its analysis plan are not reported in this registry record. The existence of a stated pooled-analysis plan does not establish how the separate trial result relates to any later combined analysis.

The primary endpoint description identifies an intent-to-treat population consisting of participants randomized to either arm. That describes the stated analysis population, but does not explain how every randomized participant's endpoint information was handled. The record says all participants were censored at 2.75 years; it does not give details about earlier loss to follow-up, missing endpoint information, withdrawals, protocol deviations, or any sensitivity analyses for missing data. These matters are not reported in the registry record.

The registry includes a reported event count and arm-level participant count for each group, but does not provide a participant-flow account or explain whether the counts used in the log-rank analysis differed from the entries in the primary outcome. The analysis population and missing-data procedures beyond the stated intent-to-treat description and censoring at 2.75 years are therefore incompletely documented in this record. The sample-size planning assumptions described for planning illustrations should not be mistaken for the trial's own design or its multiplicity plan.

12. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 2% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: osteoarthritis (0.7%), dyspnoea (0.6%), fracture (0.6%), pneumonia (0.5%), cerebrovascular accident (0.5%).

The registry lists serious adverse event figures for both treatment groups. For tamoxifen followed by exemestane, it records 784 affected among 4814 at risk; for exemestane, it records 831 affected among 4852 at risk. These are the posted serious-adverse-event counts and at-risk counts, not a separately calculated comparison. The record does not provide a statistical test or adjusted estimate comparing these figures.

The listed serious adverse event terms and percentages are osteoarthritis, 0.7%; dyspnoea, 0.6%; fracture, 0.6%; pneumonia, 0.5%; cerebrovascular accident, 0.5%; and pulmonary embolism, 0.5%. These are the top serious terms listed in the record, not a complete description of all safety outcomes. The record does not indicate whether the terms are mutually exclusive or whether a participant could contribute to more than one term.

In general, adverse-event counts describe recorded events or affected participants according to the registry's reporting definitions; they do not by themselves establish that the study treatment caused an event. The record has no reported death counts for the serious-adverse-event entry. Details about event timing, severity beyond the serious classification, attribution, and statistical comparison are not reported in the registry record.

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

The primary result is a time-to-event comparison of disease-free survival, defined as time to the earliest documentation of relapse or death from any cause. The registry reports a hazard ratio of 0.89, a 95% confidence interval from 0.77 to 1.03, and p = 0.118. The interval includes 1, so the estimate is not precise enough to rule out no relative difference in hazard. The point estimate should not be treated as proof of a difference, and the p-value is not a measure of treatment benefit or harm.

The event counts—352 for exemestane and 388 for tamoxifen followed by exemestane—refer to recorded events during follow-up through 2.75 years. They do not show the timing of those events and should not be interpreted as a direct risk comparison. The endpoint description also says participants were censored at 2.75 years. The finding therefore concerns the recorded disease-free survival comparison over that specified period, not outcomes after it.

The registry record does not report the full statistical analysis plan, detailed missing-data procedures, or the trial's own sample-size justification. It also does not provide a breakdown of the primary endpoint into relapse and death components. These omissions limit how fully the analysis can be assessed from the registry entry alone. The planned pooled analysis described in the brief summary is not a result reported in this record.

This result does not establish that either regimen is clinically preferable, nor does it establish equivalence. It also does not address outcomes outside those reported, or support conclusions for populations that differ from the stated postmenopausal, receptor-positive breast cancer population. The statistical evidence should be read as the reported estimate with its uncertainty, within the endpoint definition, population, and observation period recorded.

14. Why This Trial Matters Statistically

This trial illustrates the distinction between event counts and time-to-event analysis. The registry gives counts of disease-free survival events in each arm, but its comparative result is a log-rank hazard ratio. The count entries alone cannot convey when events occurred or how censoring contributed to follow-up. For readers assessing clinical research, keeping these quantities distinct helps prevent raw counts from being mistaken for a time-to-event effect estimate.

The endpoint is a composite of disease relapse and death from any cause. Its interpretation depends on the registry's definition of relapse, which includes locoregional or distant recurrence and ipsilateral or contralateral breast cancer. A composite endpoint can be informative about the first qualifying event, but an overall estimate does not reveal the separate contributions of its components. The record does not report those component-specific counts for the primary result.

The result also demonstrates why an estimate should be considered together with its uncertainty. The reported hazard ratio is 0.89, while the 95% confidence interval ranges from 0.77 to 1.03 and includes 1. The p-value is 0.118. These statistics do not provide a precise distinction between no relative difference and effects on either side of the reference value. Reporting the estimate without its interval, or treating a p-value alone as a verdict, would omit important information.

Finally, the registry describes this study as part of a planned collaboration involving 7 similar but independent studies. The brief summary's mention of a pre-specified pooled analysis makes clear that a single study entry and a pooled analysis are different evidence units. The pooled result is not reported in this record, so the registry's single-study result should not be conflated with the larger collaboration's planned analysis.

15. 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.

The registry provides a defined primary endpoint and a log-rank hazard-ratio result, but it does not report the full statistical analysis plan. Details not reported include the planned significance threshold for this comparison, model specifications beyond the named log-rank method, the sample-size justification, and any interim analyses. The planning assumptions associated with sample-size illustrations are not evidence of the trial's actual design assumptions.

The primary endpoint is reported as a count of relapse or death events up to 2.75 years, with participants censored at 2.75 years. The record does not give separate counts for relapse and death, nor does it describe the timing of each event. It also does not provide detailed information on early censoring, missing outcome data, or sensitivity analyses. These gaps limit assessment of how the endpoint information was distributed and handled.

The registry reports the result for the study entry and notes that pooled data from 7 independent studies were planned for analysis in the broader TEAM collaboration. Results of that pooled analysis are not reported in this record. The single-study result should therefore not be treated as the pooled estimate. The record also lists secondary outcomes but does not explain multiplicity adjustment or a testing hierarchy.

For safety, the registry lists serious adverse event counts and selected terms, but does not provide a statistical comparison, event attribution details, or reported death counts for the serious-adverse-event entry. The arm descriptions are blank, although the trial title and brief summary describe the regimens. These limitations concern the detail available in the registry record; they do not establish that the information is absent from all trial documents.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Randomized Phase III Study Of Exemestane (Aromasin) For 5 Years Versus Tamoxifen for 2.5 to 3 Years Followed By Exemestane — NCT00036270.
• Registry record last updated 2015-10-28; status last verified 2015-10.
• Results first posted 2009-12-04.
• 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.

19. Record Summary

This completed randomized phase III parallel-group trial compared adjuvant exemestane for 5 years with tamoxifen followed by exemestane over a total of 5 years in postmenopausal patients with receptor-positive, node-negative or node-positive breast cancer. The registry reports enrollment of 9779, no masking, and an endpoint window from baseline up to 2.75 years.

The primary endpoint was disease-free survival, defined by relapse or death from any cause. The posted event counts were 352 for exemestane and 388 for tamoxifen followed by exemestane. The reported log-rank analysis gave a hazard ratio of 0.89, a 95% confidence interval from 0.77 to 1.03, and p = 0.118. The interval includes 1, leaving uncertainty about the relative hazard.

The record also lists secondary outcomes and serious adverse event figures. It does not report full analysis-plan details, detailed missing-data procedures, or results of the planned pooled analysis described for the TEAM collaboration.