Survival Outcomes in Breast Cancer Patients With Use of Vaginal Estrogen Therapy: A SEER Analysis Authors: Olivia R. Mitchel, Chiu-Hsieh (Paul) Hsu, Leah S. Millheiser, Irene L. Wapnir, Shreya Chandak, Missy Tran, Jennifer Erdrich Published: 2026 ABSTRACT PURPOSEAntihormonal therapy, including tamoxifen or aromatase inhibitors, is central to the treatment of hormone receptor–positive breast cancers. A common adverse effect is genitourinary syndrome of menopause, which may be treated with vaginal estrogen. However, the safety of vaginal estrogen in breast cancer survivors remains controversial, leading to inconsistent clinical guidance and potential underuse.METHODSWe conducted a retrospective cohort study of 25,874 female patients with breast cancer age 65 years and older diagnosed between 2007 and 2017 using the SEER-Medicare Health Outcomes Survey (MHOS) registry. Patients who used vaginal estrogen after diagnosis (n = 1,053) were compared with nonusers (n = 24,821). Multivariable Cox proportional hazard models evaluated associations with overall and cause-specific mortality, adjusting for demographics, comorbidities, and oncologic factors. A secondary analysis modeled vaginal estrogen use as a time-dependent covariate.RESULTSVaginal estrogen use was associated with decreased overall mortality (adjusted hazard ratio [HR], 0.49; P < .01) and cause-specific mortality (adjusted HR, 0.31; P = .01). However, these associations were attenuated and no longer significant in time-dependent analysis. Among patients with hormone receptor–positive breast cancer, vaginal estrogen use was associated with reduced overall mortality (adjusted HR, 0.58; P = .02), although this lost significance in the time-dependent model. A nonsignificant reduction in cause-specific mortality was observed within this subgroup (adjusted HR, 0.45, P = .11).CONCLUSIONVaginal estrogen use was not associated with worse overall survival or breast cancer–specific survival in this cohort SEER-MHOS cohort of patients older than 65 years. Loss of significance in time-dependent models suggests that initial associations may reflect immortal time or selection bias. These findings support growing evidence that local vaginal estrogen is not associated with adverse survival outcomes and may be considered for symptom management in appropriately selected patients. INTRODUCTION Antihormonal therapy with tamoxifen or aromatase inhibitors (AIs) is a cornerstone in the treatment of hormone receptor–positive breast cancers, which account for approximately 80% of all breast cancer diagnoses. Tamoxifen, a selective estrogen receptor (ER) modulator, acts as a competitive antagonist in breast tissue and was approved for use in the United States in 1977 after demonstrating an improved tolerability compared with earlier therapies.1 AIs were later introduced as an alternative strategy, reducing systemic estrogen levels by inhibiting peripheral conversion of androgens to estrogen. AIs are primarily used in postmenopausal patients, in whom estrogen production is largely extragonadal. Despite their efficacy, both tamoxifen and AIs reduce estrogenic stimulation of genitourinary tissues, contributing to genitourinary syndrome of menopause (GSM). GSM is characterized by vaginal dryness, burning, irritation, dyspareunia, and recurrent urinary tract infections.2–6 Vaginal estrogen formulations including creams, inserts, and rings are commonly used to alleviate these symptoms. However, their safety in breast cancer survivors remains controversial.7–9 Patients with breast cancer are cautioned against the use of systemic estrogen therapy due to concern for increased risk of recurrence or mortality, although this has primarily been shown with estrogen plus progestin regimens.10–17 In fact, after the initial Women’s Health Initiative trial published in 2002, a follow-up publication in 2004 showed that estrogen alone was actually associated with a decreased risk of breast cancer; however, this trial excluded women with a history of breast cancer.18 Sometimes, clinical counseling or the patient’s fear extends to a complete ban on any estrogen-based product, including vaginal estrogen.19,20 With low-dose vaginal estrogen formulations, estrogen is absorbed through the vaginal epithelium. Certain vaginal estrogen preparations have been shown to mildly increase serum levels of estrogen, raising concerns about the safety of these products in breast cancer survivors; however, the clinical significance of this small increase is uncertain.21–24 Although the literature from the past decade, including a recent systematic review and meta-analysis, has largely pointed to no evidence of increased risk of breast cancer recurrence or mortality associated with the use of local vaginal estrogen, there is still widespread hesitancy in clinical practice when it comes to counseling patients on options for addressing GSM.4,5,7,25–28 The American College of Obstetricians and Gynecologists guidance has shifted from a cautious 2016 approach, where vaginal estrogen was reserved as a last resort due to theoretical recurrent risk, to a 2021 framework that supports individualized care based on growing safety data.8 Further exploration of the safety of local estrogen therapy in individuals with a history of breast cancer is warranted, as these products have the potential to provide patients symptomatic relief. The objective of this study was to use the SEER Medicare Health Outcomes Survey (MHOS) database to determine whether patients with postmenopausal breast cancer using vaginal estrogen had different outcomes compared with the patients who did not use vaginal estrogen. The primary outcome was overall survival (OS), and the secondary outcome was cause-specific survival. To our knowledge, this is the first SEER-MHOS study to address this research question. We aim to enhance understanding of the safety of vaginal estrogen, recognizing that greater clarity in this area could meaningfully improve patients’ quality of life. METHODS We conducted a retrospective cohort study using SEER-MHOS data from 2007 to 2017, a period selected due to the availability of Medicare Part D prescription data. The SEER-MHOS database includes clinical, demographic, treatment, and patient-reported outcomes for Medicare Advantage enrollees age 65 years and older. Vaginal estrogen search terms and products included in this study were Estrace (cream), Premarin (cream), Vagifem (tablet), Imvexxy (insert), and Estring. Patients who used vaginal estrogen only before breast cancer diagnosis were classified as nonusers. Those with any postdiagnosis use were classified as users. Sensitivity analysis was conducted excluding all patients with prior vaginal estrogen use from statistical analysis to assess the robustness of exposure classification. Hormone receptor positivity was defined as either ER-positive, progesterone receptor–positive, or both. Duration of vaginal estrogen use was calculated based on prescription service dates and categorized as <1 and ≥1 year. Patient characteristics were summarized using medians with interquartile ranges for continuous variables and frequencies for categorical variables. Missing data were excluded by complete case analysis. Group comparisons were performed using Wilcoxon rank-sum and chi-square tests. OS was evaluated using Kaplan-Meier methods and Cox proportional hazard models adjusted for demographics, comorbidities, socioeconomic factors, quality-of-life measures, cancer stage, receptor status, treatments, and year of diagnosis. Cause-specific survival was analyzed using cause-specific Cox models with competing risks handled by censoring at the time of competing events. To address immortal time bias, vaginal estrogen use was also modeled as a time-dependent covariate, allowing exposure status to vary over follow-up. RESULTS Baseline Characteristics A total of 25,874 patients were included, of whom 1,053 (4.06%) used vaginal estrogen after breast cancer diagnosis. Among 13,706 patients with hormone receptor–positive disease, 568 (4.14%) used vaginal estrogen. Use of vaginal estrogen declined over time, from 6.29% in 2007 to 1.86% in 2017. Among vaginal estrogen users, 720 (68.4%) had a duration of use <1 year, and 333 (31.6%) had a duration of use >1 year. Vaginal estrogen users differed from nonusers in several baseline characteristics (Table 1). Users also had lower rates of smoking, diabetes, and hypertension. They were also more likely to have higher socioeconomic status, including greater rates of college education and higher household income, and were more frequently married. Racial distribution differed between groups, with a higher proportion of White patients among users. There were no meaningful differences in quality-of-life measures or prevalence of major cardiovascular comorbidities, including coronary artery disease, heart failure, stroke, or chronic obstructive pulmonary disease. From an oncologic standpoint, vaginal estrogen users were more likely to be diagnosed at earlier stages and to receive definitive local therapy, including surgery and radiation. Tumor receptor profiles were largely similar between groups, although triple-negative disease was less common among users. Chemotherapy use did not differ significantly. Similar patterns were observed in the hormone receptor–positive subgroup, with vaginal estrogen users demonstrating fewer comorbidities, higher socioeconomic status, and greater likelihood of receiving surgery and radiation (Table 2). Of the 4,263 patients with breast cancer who used tamoxifen, 225 (5.28%) also used vaginal estrogen. Among tamoxifen users, a higher percentage of those who used vaginal estrogen also underwent breast surgery compared with those who did not use vaginal estrogen. Of the 12,552 patients with breast cancer who used AIs, 470 (3.74%) also used vaginal estrogen. Among AI users, a higher percentage of those who used vaginal estrogen also underwent breast surgery compared with those who did not use vaginal estrogen. There was no statistically significant difference in the percentage of patients receiving chemotherapy or radiation therapy between vaginal estrogen users and nonusers among those who used either tamoxifen or AIs. Outcomes All Breast Cancer Cohort Overall mortality among all patients with breast cancer was decreased with the use of vaginal estrogen (adjusted hazard ratio [HR], 0.49 [95% CI, 0.32 to 0.75]; P < .01; Fig 1A). There was also a statistically significant decrease in breast cancer–specific mortality with vaginal estrogen use among all patients with breast cancer (adjusted HR, 0.31 [95% CI, 0.13 to 0.76]; P = .01; Fig 1B). Sensitivity analysis excluding all patients with documented vaginal estrogen exposure before diagnosis generated similar results (Data Supplement, Table S1, Fig S1). Analysis by stage of breast cancer at the time of diagnosis showed a decrease in overall mortality with vaginal estrogen use in stage 0, which lost statistical significance after adjustment for confounders (adjusted HR, 0.39 [95% CI, 0.09 to 1.60]; P = .19). The same was true among patients diagnosed at stage 1–2 (adjusted HR, 0.66 [95% CI, 0.4 to 1.09]; P = .1). Among those at stage 3–4 at the time of diagnosis, there was a statistically significant decrease in overall mortality with vaginal estrogen use, which held up after adjustment (adjusted HR, 0.27 [95% CI, 0.09 to 0.88]; P = .03). Duration of use ≥1 year was associated with a reduction in overall mortality compared with duration of use <1 year, although this lost significance after adjustment (adjusted HR, 0.30 [95% CI, 0.06 to 1.65]; P = .17). When vaginal estrogen use was treated as a time-dependent covariate, the association with decreased overall mortality was attenuated and no longer statistically significant (adjusted HR, 0.84 [95% CI, 0.54 to 1.3]; P = .43). Similarly, applying the time-dependent covariate, the decrease in breast cancer–specific mortality associated with vaginal estrogen use was no longer statistically significant (adjusted HR, 0.64 [95% CI, 0.26 to 1.56]; P = .32). Hormone Receptor–Positive Breast Cancer Cohort Among patients with hormone receptor–positive breast cancer, the use of vaginal estrogen was associated with a statistically significant decrease in overall mortality (adjusted HR, 0.58 [95% CI, 0.37 to 0.91]; P = .02; Fig 2A). Similarly, there was a decrease in breast cancer–specific mortality associated with vaginal estrogen use among hormone receptor–positive patients, which lost statistical significance after adjustment (adjusted HR, 0.45 [95% CI, 0.17 to 1.21]; P = .11; Fig 2B). For the subgroup of patients receiving adjuvant tamoxifen therapy, vaginal estrogen use was associated with a decrease in overall mortality that was nonsignificant after adjustment (adjusted HR, 0.86 [95% CI, 0.34 to 2.17]; P = .75), as well as a nonsignificant decrease in breast cancer–specific mortality (adjusted HR, 0.20 [95% CI, 0.02 to 1.87]; P = .16; Fig 3A). For the subgroup of patients receiving adjuvant therapy with AIs, vaginal estrogen use was associated with a decrease in overall mortality, which was attenuated and lost statistical significance after adjustment (adjusted HR, 0.63 [95% CI, 0.33 to 1.18]; P = .15), with a similar trend for breast cancer–specific survival (adjusted HR, 0.66 [95% CI, 0.21 to 2.08]; P = .47; Fig 3B). When vaginal estrogen use was treated as a time-dependent covariate, the association with decreased overall mortality was attenuated and lost statistical significance (adjusted HR, 0.97 [95% CI, 0.62 to 1.51]; P = .89). There was no statistically significant decrease in breast cancer–specific mortality among hormone receptor–positive patients when the time-dependent covariate was applied (adjusted HR, 0.76; 95% CI, 0.28 to 2.07). Additionally, with the application of the time-dependent covariate, there was a nonsignificant increased hazard of overall mortality associated with vaginal estrogen use among those receiving tamoxifen (adjusted HR, 1.32 [95% CI, 0.51 to 3.38]; P = .57), as well as a nonsignificant decrease in breast cancer–specific mortality within this subgroup (adjusted HR, 0.29 [95% CI, 0.03 to 2.74]; P = .28). Among those receiving AIs, vaginal estrogen use in the time-dependent covariate analysis was associated with a nonsignificant increased hazard of both overall (adjusted HR, 1.03 [95% CI, 0.55 to 1.94]; P = .93) and breast cancer–specific mortality (adjusted HR, 1.11 [95% CI, 0.35 to 3.52]; P = .86). Black Patients With Breast Cancer Because Black patients with breast cancer are known to have worse survival outcomes, a subset analysis was performed. There were 3,582 patients with breast cancer who identified as Black. Of those, 76 (2.1%) used vaginal estrogen after diagnosis. Within this subgroup, the use of vaginal estrogen was associated with a decrease in overall mortality that was not statistically significant (adjusted HR, 0.55 [95% CI, 0.08 to 4]; P = .55). Similarly, no statistically significant survival difference could be concluded when the time-dependent covariate was applied (adjusted HR, 0.79 [95% CI, 0.11 to 5.75]; P = .82). DISCUSSION In this retrospective cohort study, vaginal estrogen use was not associated with worse overall survival or breast cancer–specific survival among older breast cancer survivors. Although the time-independent model suggested a survival benefit, this association was attenuated and no longer statistically significant when exposure was modeled as a time-dependent covariate. This finding underscores the importance of addressing immortal time bias in observational studies of treatment effects. In time-fixed analyses, patients classified as exposed must survive long enough to receive treatment, artificially inflating survival estimates. By modeling exposure as time-dependent, we account for the timing of treatment initiation and obtain a more valid estimate of association. Given the older population and the potential for competing risks of mortality, OS may be influenced by deaths unrelated to cancer. Therefore, cause-specific survival analysis was performed to provide a more clinically meaningful estimate of the association with vaginal estrogen use and breast cancer–related deaths. Notably, our findings did not demonstrate an increased risk of breast cancer–specific death associated with vaginal estrogen use, supporting the interpretation that local estrogen therapy does not adversely affect oncologic outcomes in this population. As the concern for potential harmful effects of estrogen therapy may be heightened in patients with a history of hormone receptor–positive breast cancer, we performed subset analysis on these cases. This analysis also demonstrated a statistically significant decrease in overall mortality, as well as a nonsignificant decrease in cause-specific mortality with the use of vaginal estrogen in the time-independent model. Among tamoxifen users who also used vaginal estrogen, there was a nonsignificant decrease in both overall mortality and cause-specific mortality. The same trend was true among those who used AIs with the addition of vaginal estrogen. With the application of the time-dependent covariate, subgroup analyses by endocrine therapy yielded slightly increased hazard, although these were imprecise estimates with wide confidence intervals, likely reflecting limited numbers of events. Vaginal estrogen users were less likely to smoke and had lower rates of diabetes and hypertension. Although adjusted analyses did not attenuate the apparent survival benefit, residual confounding remains possible. Users were also more likely to receive standard-of-care therapy (surgery and radiation), which typically precedes initiation of antihormonal therapy. Although we lack a clear timeline for treatment course due to the nature of the database, it is possible that the observed survival benefit may reflect a healthy user bias, with vaginal estrogen serving as a proxy for better baseline health and fitness for treatment. Vaginal estrogen use may also reflect greater adherence to antihormonal therapy, potentially mediating its association with decreased mortality. A higher proportion of users were diagnosed at earlier stages, suggesting potential selection bias. However, subgroup analysis showed no survival difference among patients diagnosed at Stage 0. In fact, we only observed a statistically significant decrease in OS associated with vaginal estrogen use among patients diagnosed at Stage III-IV, although this lost significance in the time-dependent model. Among all the patients included in this study, there was a greater proportion of Black patients in the nonusers group. As Black women disproportionately experience higher rates of breast cancer mortality, there was concern that the primary results were driven by differences in racial distribution.29 We therefore performed subset analysis including only Black patients and found that vaginal estrogen use in this population was directionally associated with lower hazard; however, this was nonsignificant, with a wide confidence interval. This is likely a reflection of the smaller sample size, fewer events when adjusting for multiple covariates, and perhaps greater heterogeneity within the subgroup. Taken together, these findings add to a rising contemporary paradigm shift in the management of genitourinary symptoms in breast cancer survivors, supporting the safety of vaginal estrogen in postmenopausal women, although further prospective studies are warranted. Our study has several limitations, including a small sample size within an age-restricted patient population. Only a small fraction of our SEER-MHOS breast cancer cohort had documented vaginal estrogen use, and our sample size was further decreased after adjustment and application of the time-dependent covariate, limiting the power of our analysis. Although we performed subset analysis on hormone receptor–positive cases, we had insufficient numbers to further stratify by intensity of hormone receptor status. The average age in our study population was 74 years. It would be important to study this question in younger postmenopausal women and in a premenopausal population, wherein tamoxifen and ovarian suppression may be used.30 We were unable to study recurrence, as this variable is not explicitly captured in the SEER database. This is an important limitation, as recurrence is a critical oncologic end point that may precede differences in mortality. Another limitation of our study was the lack of dosage data. We included several preparations of vaginal estrogen in our study, including creams, rings, and inserts, which vary in their dosing of estrogen. We defined duration of use by the number of years with a documented prescription service date in the MHOS part D database, but we had insufficient information to precisely determine actual use and frequency. Consistent with national trends across the early 2000s, our cohort was notable for a statistically significant decrease in vaginal estrogen use from 2007 to 2017.31,32 This downtrend may be attributed to the response to the WHI trial published in 2002, which suggested that the increased risks associated with postmenopausal estrogen and progesterone hormone replacement therapy—namely cardiovascular disease and breast cancer—outweighed the potential benefits including prevention of osteoporotic fractures.10 Importantly, however, systemic equine estrogen alone was subsequently shown to decrease the risk of breast cancer in women who had a previous hysterectomy. Over the past decade, several studies focusing on local estrogen formulations have demonstrated no increased risk of breast cancer recurrence or mortality.3,5,25,27,33 One group found that fewer ER1 patients received vaginal estrogen compared to ER-patients but showed no significant difference in recurrence rate or recurrence-free survival between these two groups.3 Other groups have demonstrated similar findings suggesting no increased risk in hormone receptor–positive breast cancer survivors.5 Consistent with standard clinical practice, our postmenopausal cohort received primarily AI therapy rather than tamoxifen. Multiple groups have recommended caution in considering local estrogen therapy for patients on AIs, as these drugs do not block the effects of exogenous estrogen.3,34,35 A handful of studies investigating the safety of vaginal estrogen in patients with a history of AI use have reported increased risk of recurrence without any adverse effect on OS, although their applicability today is uncertain.3,5,36,37 The effect of vaginal estrogen on serum estrogen levels is an area of ongoing research. In 2009, one group reported approximately a five-fold increase in serum estradiol in postmenopausal women after 1 week of daily intravaginal treatment with the 25 mg estradiol tablet or 1 g conjugated estrogen cream.38 Of note, the 25 mg dose is no longer in use, calling into question the applicability of these findings. Another group specifically studying patients with breast cancer concurrently taking an AI showed a significant increase in serum estradiol concentrations after 12 weeks for both the 17 beta-estradiol vaginal tablet (25 mg daily for 2 weeks followed by twice weekly insertion) and ring (inserted once every 90 days), which persisted even with long-term use.9 Secondary analysis of a randomized clinical trial published in 2022 found a clinically small, yet statistically significant increase in serum estradiol levels with low-dose (10 mg) vaginal estradiol treatment after 2 weeks of daily use followed by twice weekly insertion for a total of 12 weeks.22 Several additional clinical trials have demonstrated either no change or a transient/negligible increase in serum estrogen levels with concurrent use of AIs and low-dose vaginal estrogen preparations.23,38–42 An important consideration for future research would be to better define the dosage of vaginal estrogen at which systemic levels become high enough to pose risk to patients, both with and without concurrent AI use. A single randomized clinical trial of women with postmenopausal osteoporosis found that those with baseline serum estradiol concentrations higher than 2.7 pg/mL had a significantly greater risk (6.8-fold higher) for breast cancer compared with those with baseline concentrations lower than 2.7 pg/mL, suggesting that concentrations above this threshold may be associated with an increased risk of breast cancer.43 In summary, this SEER-MHOS study of vaginal estrogen use in patients with postmenopausal breast cancer found no increased risk of overall or breast cancer–related mortality. These findings add to growing evidence that locally administered intravaginal estrogen is safe and may improve quality of life in older patients with genitourinary symptoms exacerbated by antihormonal breast cancer therapy. Supplementary Material Data Supplement