S0307 Phase III Trial of Bisphosphonates as Adjuvant Therapy for Primary Breast Cancer.: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for S0307 Phase III Trial of Bisphosphonates as Adjuvant Therapy for Primary Breast Cancer. (NCT00127205), a randomized phase 3 study. The primary outcome is Disease-free Survival.

2. Clinical Question

The trial record frames the clinical uncertainty as whether zoledronate is more effective than clodronate or ibandronate in this postoperative setting. The endpoint combines recurrence of disease with death from any cause, so it addresses time until either of those events rather than recurrence alone. Overall survival and sites of first recurrence were listed among secondary outcomes. The registry does not report results for those outcomes in the posted result data. The question is therefore about comparative disease-free survival across the assigned regimens, not about a single drug compared with no treatment or placebo.

3. Trial Design

NCT00127205, titled S0307 Phase III Trial of Bisphosphonates as Adjuvant Therapy for Primary Breast Cancer, is recorded as a completed phase III study sponsored by SWOG Cancer Research Network. It began in 2005-07, with primary completion in 2017-12. Enrollment was 6097. The design was randomized, parallel-group, and unmasked; masking is listed as none. These features identify a three-arm treatment comparison, but the registry record does not describe the randomization procedure or any stratification factors.

Arm I received zoledronate intravenously over 15 minutes once a month for 6 months and then once every 3 months for 2.5 years. Arm II received oral clodronate once daily for 35 months. Arm III received oral ibandronate once daily for 35 months. The registry reports disease-free survival estimates for all three arms, and Cox-model results for zoledronate compared with each oral regimen. It also reports an overall log-rank comparison across the three groups.

For the primary outcome, the listed analysis population includes only eligible patients. The record defines disease-free survival from registration until recurrence or death from any cause. Participants last known to be alive without recurrence were censored at their last contact date. Disease assessments were scheduled every 6 months for 5 years and then annually for 5 years, or until death or recurrence. The result is presented as a 5 year survival rate, with a 95% confidence interval for each arm.

The registry record does not report the allocation ratio, specific eligibility criteria, adherence, treatment discontinuation, or details of follow-up completeness. It also does not report a statistical analysis plan or sample-size justification. Those omissions limit how fully the design and analysis decisions can be reconstructed from the posted information.

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.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Only eligible patients will be included in the analysis.

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.

Population: Only eligible patients are included in the analysis.

6. Primary Endpoint

The registry lists the primary outcome as Disease-free Survival, measured over Disease assessments are completed every 6 months for 5 years then annually for 5 years or until death or recurrence.

Disease-free survival is the time from registration to the first recorded recurrence or death from any cause. A participant who was last known to be alive and free of recurrence was censored at the date of last contact. Censoring means that the event time was not observed during the available follow-up; the participant contributes information up to the last known event-free contact rather than being treated as having had an event then.

The registry presents the primary outcome as the 5 year survival rate, expressed as a percentage of analyzed participants, with a 95% confidence interval. The listed assessment schedule was every 6 months for 5 years and then annually for 5 years, or until death or recurrence. Disease-free survival is useful in this setting because it counts recurrence and death, rather than relying only on recurrence or only on survival. It therefore summarizes the time participants remain alive without a documented recurrence under the endpoint definition.

The endpoint does not distinguish the relative contribution of recurrence and death to the combined result. The registry lists sites of first recurrence and overall survival among secondary outcomes, but their results are not reported in the posted data. The primary result should consequently be read as an estimate of the combined disease-free survival outcome, not as a separate estimate of recurrence prevention or overall survival.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

What does the survival estimate represent?

A survival estimate at a specified time point describes the estimated proportion of participants who have not yet had the event by that time, under the endpoint definition. For this trial, the event was the first recurrence or death from any cause, and the reported time point was 5 years. Because participants could be censored at their last known event-free contact, the estimate uses follow-up information that may end before the event is observed. A confidence interval communicates uncertainty in the estimated rate. It is not a guarantee that the true value lies within the particular interval, nor does it describe the range of possible outcomes for an individual.

The registry reports a 95% confidence interval for each arm: 87.0% to 90.0% for zoledronate, 86.0% to 89.0% for clodronate, and 86.0% to 89.0% for ibandronate. The interval widths indicate that the estimates are not exact. Comparing whether arm-specific intervals overlap is not a substitute for a formal test of the between-arm contrast. The trial also reports comparative analyses, which address the groups jointly or in pairs.

What does the log-rank test assess?

The log-rank test compares time-to-event experience between groups over follow-up. It uses information about when events occur and accounts for censoring, rather than comparing only the percentage with an event at one selected time. The reported three-arm test had p = 0.49. The p-value is the probability, under the test's null model and its assumptions, of obtaining a test statistic at least as incompatible with that null as the observed statistic. It is not the probability that the null hypothesis is true, and it does not give the probability that the treatments are equal.

A single overall test across three arms asks whether the observed time-to-event patterns show evidence of any group difference. It does not identify a particular pair as different. Pairwise comparisons answer more specific questions, but testing multiple contrasts can increase the chance of at least one false-positive result if no adjustment or prespecified testing strategy is used. The registry record does not report a multiplicity adjustment or the planned testing hierarchy.

How should the Cox-model results be understood?

A Cox regression model compares event hazards over time. The hazard is the instantaneous event rate among participants who have not yet had the event, conditional on the model's structure. The reported hazard ratios were 1.09 for the zoledronate-clodronate comparison and 1.06 for the zoledronate-ibandronate comparison. Their respective 95% confidence intervals were 0.94 to 1.26 and 0.90 to 1.24. The registry lists p = 0.24 and p = 0.50 for these analyses.

A hazard ratio is a relative measure, not a difference in 5 year survival percentages. Its interpretation depends on which group is the reference and how the event contrast is coded. Although the record identifies the compared groups, it does not explicitly describe the numerator and denominator direction. The confidence intervals include the no-difference value, so the posted results do not provide strong evidence of a hazard difference in either comparison. A confidence interval also indicates a range of effect sizes compatible with the data and model, rather than proving there is no difference.

The usual Cox-model interpretation relies on assumptions, including proportional hazards: the relative hazards are assumed to be reasonably stable over time. The registry record does not report whether this assumption was assessed, what covariates were included, or whether the displayed hazard ratios were adjusted. The model specification and diagnostics are therefore not available for independent evaluation from this record.

Why does censoring matter?

Survival methods can use incomplete follow-up by including each participant's observed time until an event or last known event-free contact. This approach relies on censoring being adequately handled and on the censoring process not making the remaining observed participants systematically unrepresentative in a way that biases the estimate. The registry defines the censoring date but does not report the amount or pattern of censoring, or analyses assessing its impact. The event counts and detailed follow-up distribution are not available in the posted primary result data.

What can planning illustrations say?

Sample-size calculations are planning exercises that depend on assumptions about the effect to detect, significance level, allocation, and statistical method. They are not observed power, and they do not show that the trial had a particular probability of success after results were observed. The registry record does not report its own sample-size justification or its original planning assumptions. No numerical planning illustration is reported in the result data. Any planning values should therefore be treated as hypothetical design illustrations rather than as a reconstruction of the trial's design rationale.

10. Analyses Reported by the Sponsor

The registry reports the trial's own primary outcome estimates, overall log-rank test, and Cox-model comparisons. The posted analyses give the three 5 year disease-free survival estimates and the corresponding confidence intervals, alongside a three-arm log-rank p-value of 0.49. The Cox-model results are reported for zoledronate compared with clodronate and zoledronate compared with ibandronate.

No recalculated comparative estimate is reported in the registry record. The registry does not provide enough information about the Cox-model coding, covariate adjustment, event data, or other model details to reproduce those analyses from the arm-level percentages. A simple unadjusted recalculation from summary statistics would not necessarily match a time-to-event analysis, which uses event timing and censoring information. It would also answer a different statistical question from the sponsor's fitted model.

No discrepancy is flagged between the sponsor's posted result and a recalculated result. Accordingly, the posted sponsor analysis remains the relevant reported comparison; there is no conflicting recalculation to resolve. This does not make the model details transparent: the registry record does not report the full statistical plan, diagnostics, or hazard-ratio coding direction. The results should be described as reported, with those details kept in view rather than inferred from the point estimates alone.

11. Multiplicity, Missing Data and Analysis Populations

The registry reports one primary endpoint, disease-free survival, and lists three secondary outcomes: overall survival, distributions of sites of first recurrence across the arms, and the number of patients with grade 3 through 5 adverse events related to study drugs. The record also reports an overall log-rank test and two pairwise Cox-model comparisons. It does not report how the primary and secondary outcomes were prioritized in the analysis, whether the pairwise comparisons were prespecified, or whether any multiplicity adjustment was used.

Multiplicity arises when several hypotheses or outcomes are tested. Without an adjustment or a clearly defined testing sequence, the chance of at least one false-positive finding across a collection of tests can be higher than for a single test. This issue does not change the numerical p-values reported, but it affects how a set of tests should be interpreted together. The registry record does not report an interim-analysis plan, stopping boundaries, or whether any outcome analyses were performed at interim reviews. These matters are not reported in the registry record.

The primary analysis population is described as eligible patients only. The registry does not report the detailed eligibility rules, the number excluded from that analysis, or whether the analysis followed an intention-to-treat principle. It also does not report how missing outcome assessments were handled beyond the stated censoring rule for participants last known to be alive and recurrence-free. The extent and reasons for loss to follow-up, missing assessments, and treatment discontinuation are not reported.

For time-to-event analysis, censoring at last contact preserves observed follow-up, but interpretation depends on the censoring mechanism and adequate outcome ascertainment. The record gives the censoring definition but does not describe sensitivity analyses or tests of assumptions. The Cox-model adjustment variables, proportional-hazards assessment, and handling of any missing covariates are also not reported. These gaps limit assessment of robustness, but they do not justify assuming that an unreported method was or was not used.

12. Safety Results

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

Deaths

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: inf w/normal anc or gr 1-2 neutrophils - skin (0.8%), fracture (0.7%), thrombosis/thrombus/embolism (0.7%), osteonecrosis (avascular necrosis) (0.6%), dyspnea (shortness of breath) (0.6%).

The registry lists serious adverse events for each arm as numbers affected among participants at risk: 21 of 2125 for zoledronic acid, 190 of 2186 for clodronate, and 141 of 1530 for ibandronate. These are counts, not adjusted comparisons. The at-risk group sizes differ, and the record does not provide exposure-adjusted rates, event timing, or a statistical comparison for these figures. The counts should not be converted into a claim about comparative safety without those additional details.

The registry also lists deaths of 238, 263, and 182 for the zoledronic acid, clodronate, and ibandronate groups, respectively. These counts are reported separately from the serious adverse event figures. The record does not specify in this result summary how the deaths relate to the adverse-event attribution process or the timing of safety follow-up. They should not be treated as a comparative mortality analysis.

Serious event terms listed include infection with normal ANC or grade 1-2 neutrophils at 0.8%, fracture at 0.7%, thrombosis/thrombus/embolism at 0.7%, osteonecrosis (avascular necrosis) at 0.6%, dyspnea at 0.6%, and cholecystitis at 0.5%. The listed percentages are not assigned to specific arms in the term list. The registry separately identifies grade 3 through 5 adverse events related to study drugs as a secondary outcome, but the complete arm-specific results for that outcome are not reported in the posted data. These safety summaries are descriptive and do not establish causation or a between-group difference.

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

The posted results describe estimated disease-free survival after surgery among eligible participants assigned to one of three bisphosphonate regimens. The 5 year estimates were 88.0% for zoledronate, 88.0% for clodronate, and 87.0% for ibandronate, with the reported 95% confidence intervals. The overall log-rank test and two Cox-model comparisons did not provide strong statistical evidence of a difference among the groups. This is a description of the evidence in these analyses, not proof that the regimens have identical effects.

A failure to detect a difference is not the same as establishing equivalence. Equivalence requires a margin defined in advance and an analysis designed to assess whether the plausible difference is sufficiently small. The registry record does not report such a margin or an equivalence analysis. The confidence intervals leave room for a range of comparative effects, and their interpretation depends on the analysis model and its assumptions.

The endpoint combines recurrence and death, so it cannot by itself distinguish which event type accounts for a particular result. The record lists overall survival and sites of first recurrence as secondary outcomes, but their results are not fully reported in the posted data. Nor do the primary results establish effects in patient subgroups, because subgroup findings are not reported.

These results apply to the study population and treatment regimens described in the registry. The available record does not give detailed eligibility criteria, adherence, or the full analysis plan, which constrains assessment of applicability and robustness. The statistics should therefore be read as estimates with uncertainty, under the reported endpoint definition and analyses, rather than as a complete answer about every outcome or every clinical context.

14. Why This Trial Matters Statistically

This trial is statistically informative because it compares three active regimens in a randomized parallel-group design and reports both an overall time-to-event comparison and pairwise Cox-model comparisons. The primary outcome is defined in terms of time to recurrence or death, with censoring at the last known event-free contact. That makes the result more than a simple comparison of event proportions: follow-up timing and incomplete observation are central to the analysis.

The reported 5 year estimates are similar in magnitude, while the confidence intervals communicate uncertainty around each arm-specific estimate. The overall log-rank test and pairwise Cox-model results provide distinct views of the evidence: one evaluates the time-to-event patterns across all three arms, and the others report comparative hazard ratios for two pairs. Their interpretation depends on model assumptions and contrast coding, details that the registry does not fully report.

The record also illustrates an important distinction between no statistically persuasive evidence of a difference and proof of equivalence. The reported tests do not establish that the regimens have the same effect. A formal equivalence or noninferiority claim would require a specified margin and suitable analysis, neither of which is reported.

Finally, the trial's posted statistics are only part of the evidence needed to understand its results. Analysis-population details, multiplicity decisions, censoring patterns, model diagnostics, and complete secondary-outcome results are not reported in the registry record. Identifying those gaps is part of sound statistical interpretation: conclusions should track what the estimates and tests address, while leaving unreported questions unresolved.

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.

Several trial-specific details needed to assess the results are not reported in the registry record. The eligibility criteria are not fully described, although the primary analysis is limited to eligible patients. The record does not state whether the primary analysis followed an intention-to-treat approach, how many enrolled participants were excluded from the eligible-patient analysis, or the reasons for any exclusions.

The registry gives the censoring rule but not the extent, timing, or reasons for censoring and loss to follow-up. It does not report event counts for the primary outcome in the posted result data, nor detailed follow-up distributions. These omissions limit assessment of how much information supports the 5 year estimates and whether censoring could affect comparisons.

The Cox-model results lack reported details on reference-group coding, covariate adjustment, model diagnostics, and assessment of proportional hazards. The overall log-rank result and pairwise Cox results are reported, but the testing hierarchy and any multiplicity adjustment are not. The registry also does not report an interim-analysis plan or a sample-size justification. These are not grounds to assume a particular method; they are limits on what can be checked from the record.

Only the primary outcome and selected comparative analyses are described in the posted results. Although secondary outcomes include overall survival, sites of first recurrence, and drug-related grade 3 through 5 adverse events, complete results for these outcomes are not reported in the data. Safety counts are not accompanied by exposure-adjusted comparisons or timing information. Collectively, these limitations mean that the registry record supports a bounded reading of the reported disease-free survival analyses, not a full reconstruction of trial conduct or all outcome evidence.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: S0307 Phase III Trial of Bisphosphonates as Adjuvant Therapy for Primary Breast Cancer. — NCT00127205.
• Registry record last updated 2021-07-02; status last verified 2021-06.
• Results first posted 2019-08-28.
• 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

NCT00127205 is a completed, randomized, unmasked, parallel-group phase III trial sponsored by SWOG Cancer Research Network. It enrolled 6097 participants and compared postoperative zoledronate, clodronate, and ibandronate regimens for stage I, stage II, or stage III breast cancer. The primary outcome was disease-free survival, defined as time from registration to recurrence or death from any cause and reported as the 5 year survival rate.

The reported estimates were 88.0% for zoledronate, 88.0% for clodronate, and 87.0% for ibandronate. The overall log-rank p-value was 0.49. Cox-model comparisons reported hazard ratios of 1.09 for zoledronate versus clodronate and 1.06 for zoledronate versus ibandronate, each with a 95% confidence interval that includes the no-difference value. The record does not report a formal equivalence analysis, full model details, or complete results for all listed secondary outcomes.