Sulindac and Breast Density in Women at Risk of Developing Breast Cancer: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Sulindac and Breast Density in Women at Risk of Developing Breast Cancer (NCT04542135), a randomized phase 2 study. The primary outcome is Change in Percent Breast Density by MRI Between Treatment Arms.

2. Clinical Question

3. Trial Design

This was a randomized, parallel-group phase 2 trial comparing sulindac with placebo in postmenopausal women at elevated risk of breast cancer. The registry record describes the study as single-masked but does not specify who was masked. It reports enrollment of 61 participants. The study status is completed, with a start date of 2020-11-20 and a primary completion date of 2025-04-22.

The planned treatment period in the primary objective was 12 months. Sulindac was described as 150 mg twice a day, and the comparator was a placebo pill. The registry record identifies the primary outcome as change in percent breast density by MRI between treatment arms. It also lists secondary outcomes concerning MRI-measured density change within and between treatment arms, including analyses stratified by aromatase inhibitor use. Results for those secondary outcomes are not included in the posted primary comparison.

The primary outcome analysis used a linear mixed effect model for longitudinal data. The analysis population was limited to participants who completed 6 months and had percent breast density data available. The registry record does not report the number screened, the number randomized to each arm, reasons for withdrawal, or the extent of missing outcome data. The posted arm estimates are based on 28 participants in the sulindac arm and 26 in the placebo arm.

The registry record does not report the randomization procedure, allocation concealment, the exact masking arrangements, or details of the treatment schedule beyond the stated dose and duration. It also does not report a sample-size justification or a detailed statistical analysis plan. These omissions limit how fully the design and conduct can be assessed from the registry record alone.

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

Analysis population for the primary outcome, as defined in the registry: Linear mixed effect model for longitudinal data is used for testing the relative change in percent breast density. Participants who completed 6 months in the study and for whom percent breast density data were available were included in analysis of primary outcome.

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 Change in Percent Breast Density by MRI Between Treatment Arms, measured over 12 months.

The primary endpoint was relative change in percent breast density measured by MRI, compared between the sulindac and placebo arms at 12 months. Percent breast density is described as the estimated percentage of breast volume that appears dense rather than fatty on MRI. Relative change is defined as the difference between follow-up and baseline values divided by the baseline value. The reported unit is relative change (%) of percent density.

This endpoint turns MRI measurements at different times into a change relative to each participant’s baseline. In general, measuring change can help describe whether an outcome moved over the treatment period, while comparing arms helps distinguish changes associated with treatment assignment from changes also seen in the control group. The endpoint is a continuous measure, so the analysis summarizes average changes and their uncertainty rather than counting participants with or without an event.

The registry record identifies breast density as the primary measure of interest, not breast cancer occurrence. The results therefore address the MRI density outcome during the stated follow-up period. They do not directly report whether participants developed breast cancer, and the density measure should not be treated as a substitute for a clinical outcome not reported in the primary results.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The primary outcome is continuous: it records relative change in MRI-measured percent breast density rather than a yes-or-no event. The registry record reports least-squares mean estimates for the arms and a mixed-model comparison. Such models can use measurements collected over time and account for the relationship between repeated observations from the same participant. The record identifies a linear mixed effect model for longitudinal data, but does not give the model formula, covariance structure, adjustment variables, or details of how the reported final-value comparison was obtained.

What does the arm estimate describe?

The sulindac estimate is -4.85, with a 95% confidence interval from -11.29 to 1.59. The placebo estimate is -6.29, with a 95% confidence interval from -13.36 to 0.78. These are mean estimates for the continuous relative-change outcome. They are not counts of participants who improved, and they do not show the distribution of individual responses. A mean can conceal variation among participants, and the registry record does not report participant-level results or a measure of that variation.

Why focus on the between-arm contrast?

The primary objective asks whether change differs between sulindac and placebo. The registry reports a mean difference in final values of 1.44, with a 95% confidence interval from -6.23 to 9.11. The contrast is the direct comparison relevant to the randomized treatment question. Looking separately at whether each arm changed from baseline would not answer whether the arms differed from one another. In general, a randomized comparison helps control for influences shared across groups, subject to the trial’s conduct, follow-up, and analysis.

How should the confidence interval be read?

A confidence interval expresses statistical uncertainty around an estimate under the assumptions of the analysis. A narrower interval indicates greater precision; a wider interval indicates less precision. The interval for the between-arm estimate spans zero, so it is compatible with no difference and with differences in either direction. Its endpoints also show that the estimate is not precise enough to restrict the plausible contrast to a narrow range. The interval is not a guarantee about the true effect, and it does not account for unreported sources of bias such as selective missingness.

What does the p-value add?

The reported p-value is 0.706. In general, a p-value measures how unusual data at least as incompatible with a null hypothesis would be if that hypothesis and the statistical model were correct. It is not the probability that the null hypothesis is true, nor does it measure the size or clinical importance of an effect. This value does not provide evidence against the no-difference hypothesis in the reported model. A large p-value does not prove equivalence or rule out effects that the confidence interval still permits.

Why can over-dispersion matter?

Over-dispersion is a general concern for count outcomes when observed variation exceeds what a basic count model assumes. It can make standard errors too small if ignored. This trial’s primary endpoint is continuous, not a count, and the registry identifies a linear mixed effect model. Over-dispersion is therefore not the central issue for interpreting this primary result. More relevant general considerations include model fit, the dependence among repeated measurements, and the treatment of missing observations. The record does not report diagnostic checks or enough model detail to assess those features.

What can be said about recalculation?

The posted arm summaries and the sponsor’s mixed-model contrast are different kinds of estimates. The contrast comes from the reported model, while an unadjusted recalculation from summary statistics would not necessarily reproduce a longitudinal mixed-model result. No recalculated contrast or discrepancy is reported. The sponsor’s own analysis therefore remains the relevant posted inferential result; the arm estimates provide context but should not be substituted for that analysis.

The registry record states that participants who completed 6 months and had percent breast density data available were included in the primary analysis. It does not report how the model handled incomplete follow-up, the number of repeated observations contributing per participant, or whether any covariates were included. Those details matter because the validity and precision of a mixed-model estimate depend on the data available and on assumptions about the outcome process and missingness. The numerical results should be interpreted within the limits of the analysis details that are reported.

10. Analyses Reported by the Sponsor

The registry reports the sponsor’s own primary comparison as a mixed-model analysis. It gives a mean difference in final values of 1.44, a 95% confidence interval from -6.23 to 9.11, and p = 0.706. The arm-level estimates are -4.85 for sulindac and -6.29 for placebo, but those summaries do not by themselves reproduce the model-based comparison. A simple unadjusted recalculation from arm summaries, if performed, could differ because it would not necessarily use the repeated measurements or modeling choices used in the sponsor’s analysis.

No recalculated contrast or discrepancy is reported alongside these results. The sponsor’s own analysis therefore takes precedence as the posted inferential result. Its interval spans zero and allows effects in either direction, while the p-value does not indicate evidence against no between-arm difference under the reported model. This is not proof that the groups are equivalent, and the result should be interpreted with the analysis population and unreported model details in mind.

11. Multiplicity, Missing Data and Analysis Populations

The registry record lists the primary MRI density comparison and secondary outcomes concerning changes within and between arms, including results stratified by aromatase inhibitor use. It records 3 outcome measures and 2 analyses, but does not report a multiplicity-adjustment procedure or explain how the primary and secondary questions were prioritized for statistical testing. When several outcomes or subgroup comparisons are examined, the chance of at least one apparently notable result can increase. In general, prespecified testing strategies or multiplicity adjustments can address that concern, but their use cannot be assumed.

The record identifies the primary analysis population as participants who completed 6 months and for whom percent breast density data were available. This is an availability-based description, not a full account of all randomized participants. The posted arm estimates include 28 participants assigned to sulindac and 26 assigned to placebo. Enrollment was 61, but the record does not explain the difference between enrollment and these analysis counts, report the number randomized to each arm, or give reasons for missing data or non-completion.

Missing outcomes can affect both precision and bias. In general, a model may use incomplete repeated measurements under particular assumptions, but the registry record does not say whether that occurred or state the assumptions used. It does not report imputation, sensitivity analyses for missing data, or comparisons between participants included and excluded from the primary analysis. As a result, readers cannot determine from the posted record whether missingness might have shifted the arm estimates.

The record also does not report whether interim analyses were planned or conducted, whether the analysis was based on intention to treat, or whether the analysis population was defined before results were examined. These details are not reported in the registry record. Their absence limits assessment of selective analysis and the robustness of the primary result, but does not by itself establish that any such problem occurred.

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 0% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: colitis (3.3%), appendicitis perforated (3.2%).

The posted safety figures list serious adverse events for 31 participants at risk in the sulindac group, with 1 affected, and 30 participants at risk in the placebo group, with 1 affected. Deaths are reported as 0 in each group. These are descriptive counts in the registry record; they do not establish whether an event was caused by treatment. The record does not provide a causal assessment or enough detail to evaluate the timing and clinical context of the events.

For other adverse events, 27 of the 31 participants at risk in the sulindac group were affected, and 22 of the 30 participants at risk in the placebo group were affected. The registry record does not give a full account of event severity, duration, recurrence, or how events were collected. These figures should therefore be read as the posted counts, not as a complete characterization of comparative safety.

The listed top serious-event terms are colitis, reported as 3.3%, and appendicitis perforated, reported as 3.2%. The registry record does not specify in this listing which treatment group each term occurred in or provide enough context to interpret these percentages. Small event counts and limited reporting make comparative safety conclusions especially uncertain. The counts describe what was recorded; they do not establish that either treatment caused or prevented an event.

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

The central result is a model-based between-arm estimate of 1.44 for relative change in MRI-measured percent breast density, with a 95% confidence interval from -6.23 to 9.11 and p = 0.706. The estimate is uncertain, and the interval spans zero. The posted analysis therefore does not show a clear difference between sulindac and placebo on the primary endpoint. At the same time, the interval includes effects in either direction, so the result does not establish that there is no effect or that the treatments are equivalent.

The endpoint is a surrogate-like imaging measure described as the estimated share of breast volume appearing dense rather than fatty. The registry’s primary question concerns that measure at 12 months. The result does not directly answer whether sulindac changes the chance of developing breast cancer, and the record does not report breast cancer incidence as the primary outcome. It also does not establish effects on outcomes beyond the MRI density measure.

The point estimate should not be interpreted in isolation. Its precision, the analysis population, and the model used all matter. The arm-level estimates suggest that the mean relative-change estimates were negative in both groups, but the randomized question is whether the groups differed. The reported contrast and its uncertainty are more directly relevant to that question than separate within-arm descriptions.

These results also cannot settle the broader question of whether the observed data would be reproduced in another population or under different study conditions. The registry record does not report a sample-size justification, detailed eligibility criteria, full missing-data handling, or the complete statistical plan. Those omissions prevent a complete assessment of precision planning, generalizability, and the robustness of the analysis. The findings should be understood as the posted result for this trial and this endpoint, not as a general conclusion about breast cancer prevention.

14. Why This Trial Matters Statistically

This trial raises a useful statistical distinction between a change within each arm and a treatment difference between arms. The sulindac and placebo estimates are both negative, but those values alone do not answer whether the assigned treatments produced different changes. The relevant contrast is the reported mean difference of 1.44. Its confidence interval spans zero and extends in both directions, making the uncertainty in the comparison central to interpretation.

The endpoint also illustrates how continuous imaging measures are analyzed differently from clinical event counts. MRI-measured percent breast density is summarized as relative change, and the registry reports least-squares means and a mixed-model comparison. That approach is intended to analyze longitudinal measurements, but its interpretation depends on the repeated-measure structure, model specification, and handling of incomplete observations. The registry record gives the model family and a brief description of the analysis population, while leaving many implementation details unreported.

Randomization and a placebo comparator provide a framework for comparing assigned groups, but they do not eliminate uncertainty from limited analysis counts, missing outcomes, or incomplete reporting. The primary contrast is imprecise, and the available record does not explain the gap between enrollment and the analysis counts. Readers therefore need to distinguish the numerical estimate from the strength and completeness of the evidence supporting it.

Finally, the study question concerns breast density, not breast cancer incidence. Keeping the endpoint distinct from the broader clinical motivation avoids overstating what the trial’s primary result can establish. This is a statistical lesson in matching conclusions to the outcome measured, the comparison analyzed, and the uncertainty reported.

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 record provides a limited account of the trial methods. It identifies randomization, parallel groups, single masking, a 12-month treatment objective, and a longitudinal mixed model, but does not describe the randomization sequence, allocation concealment, who was masked, or the model specification. These details affect assessment of potential bias and the reproducibility of the analysis.

The primary analysis population is described as participants who completed 6 months and had percent breast density data available. The record does not state how many participants were randomized to each arm, why some enrolled participants are not reflected in the posted analysis counts, or how missing measurements were handled. It also does not report sensitivity analyses for missingness. Consequently, it is difficult to assess whether the analysis population differs systematically from all randomized participants.

The reported endpoint is MRI-measured relative change in percent breast density, and the primary comparison is at 12 months. The registry record does not report the full distribution of individual outcomes, model diagnostics, or additional covariate details. The confidence interval is broad enough to leave substantial uncertainty about the direction and size of the between-arm contrast. The posted result should not be interpreted as evidence of equivalence.

The record lists secondary outcomes, including a stratified analysis by aromatase inhibitor use, but does not report multiplicity handling or a detailed plan for interpreting multiple comparisons. It also does not report a sample-size justification, interim analyses, or the complete analysis plan. These items are not reported in the registry record, so their absence should be treated as a limit on evaluation rather than evidence that a particular procedure was or was not used.

Safety reporting is also brief. Serious-event counts, other adverse-event counts, deaths, and two event terms are listed, but causal attribution and detailed event context are not reported. These limits prevent a comprehensive assessment of comparative safety from the posted figures alone.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Sulindac and Breast Density in Women at Risk of Developing Breast Cancer — NCT04542135.
• Registry record last updated 2026-07-17; status last verified 2025-04.
• Results first posted 2026-07-17.
• 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

The registry record describes a completed randomized, single-masked, parallel-group phase 2 trial of sulindac versus placebo in postmenopausal women at elevated risk of breast cancer. Enrollment was 61. The primary endpoint was relative change in MRI-measured percent breast density at 12 months, analyzed with a linear mixed effect model among participants who completed 6 months and had density data available.

The posted comparison reports a mean difference in final values of 1.44, with a 95% confidence interval from -6.23 to 9.11 and p = 0.706. The interval spans zero and permits effects in either direction, so the result does not establish a clear between-arm difference or equivalence. The registry record does not report full details of missing-data handling, the sample-size justification, or the complete statistical plan. The primary result concerns breast density, not breast cancer incidence.