Adjuvant Radiation vs Endocrine Therapy After Lumpectomy for Early-Stage Breast Cancer in Older Women: Analysis of Real-World Survival Outcomes.
Jun 6, 2025DOI: 10.1245/s10434-025-17532-1Journal ArticleASO
Written by: Tori C NierenbergKerri-Anne CrowellTon WangLaura H RosenbergerGayle A DiLallaSusan G R McDuffGretchen KimmickE Shelley HwangJennifer K Plichta
The average age of a breast cancer diagnosis is 62 years old, with breast cancer becoming increasingly prevalent in women over 65 years of age [1]. Older patients are more likely to be diagnosed with favorable tumor phenotypes, with 80% of patients in this cohort being diagnosed with hormone receptor positive disease [2]. While their cancers tend to have favorable outcomes, older women still tend to have worse overall survival (OS) outcomes due to other risk factors and comorbidities [3]. This leads these women to be at risk for over treatment of their cancers. Several studies from our group have sought to investigate potentially de-escalating surgical treatment for older patients with breast cancer, particularly in terms of surgical considerations [4–6] and the use of genomic assays to help tailor treatment recommendations [7]. However, other prospective studies have also investigated de-escalation strategies of non-surgical therapies.Previous prospective clinical trials have shown that it is reasonable to omit certain therapies for older women with favorable early-stage breast cancer. For example, the CALGB 9343 trial investigated women over 70 with T1N0M0 breast cancer who underwent lumpectomy followed by tamoxifen plus radiation therapy (RT) verses tamoxifen alone [8]. The study found that while there were no significant differences between the groups in terms of OS, patients in the tamoxifen plus irradiation group had significantly longer times to local or regional recurrences with 99% of patients in this group not having any recurrence [8]. Despite these differences, given the shortened residual life span of this population, it was concluded that it was safe to omit radiation for these patients. Similarly, PRIME II, a randomized controlled trial, once again found that while adding RT to endocrine therapy (ET) reduced rates of local recurrence, it did not actually impact OS in older (>65 years) women [9]. Based on these trials, it was concluded that de-escalation via omission of RT is a valid and safe option for these patients.However, this suggestion assumes that patients will adhere to at least 5 years of ET. In clinical practice, medication tolerance and compliance are difficult to predict, with studies showing that approximately 30% of older women do not adhere to their 5-year ET regimens [10]. Further, ET may be difficult for some older patients with comorbidities to tolerate and may have a significant impact on quality of life (typically worse) for the duration of the therapy. Therefore, it is possible that older patients with favorable disease may prefer a shorter course of RT over the longer ET course, if it has similar survival benefits. However, currently there is minimal data investigating this treatment approach. A recently published preplanned interim analysis of 207 patients comparing RT-only and ET-only in an older population (EUROPA Study) found that patient’s receiving RT-only reported greater quality of life and less adverse effects compared to those receiving ET-only [11]. However, survival events were not reported in the published preliminary report. As such, the aim of this study is to determine whether adjuvant treatment solely with ET compared to RT alone is associated with comparable OS in women aged ≥65 diagnosed with early-stage, ER+/HER2− breast cancer after breast conserving surgery.
METHODS
Older women (ages ≥65 years) diagnosed between 2004–2020 with prognostic clinical stage 1, ER+/HER2− invasive breast cancer who underwent lumpectomy were selected from the National Cancer Database (NCDB, Participant User File 2022). The NCDB is a clinical oncology database based on data from >1500 Commission on Cancer (COC)-accredited facilities, which captures >70% of all cancer diagnoses and >80% of all breast cancer diagnoses each year [12, 13]. Patients with node positive disease (clinically N+ and/or pathologically N+) or larger tumors (clinically T3–4 and/or pathologically T3–4) were excluded, as were those with missing TNM categories, estrogen receptor (ER) status, progesterone receptor (PR) status, human-epidermal-growth-factor-receptor-2 (HER2) status, or survival. Patients who had chemotherapy and/or immunotherapy at any time were excluded, as well as those who received any neoadjuvant therapy. Patients with stage IV disease were also excluded. ‘No adjuvant therapy’ was defined as receiving neither ET nor RT. Per NCDB definitions, ER and PR status were defined as positive if ≥1% of cells stained positive. For select subgroup analyses, ER status was further stratified as low vs high if ER staining was 1–10% (low) or >10% (high).
Patient, disease, and treatment characteristics were summarized by N (%) for categorical variables and median (interquartile range, IQR) for continuous variables. Differences across groups were tested using the chi-square test for categorical variables and analysis of variance (ANOVA), with appropriate Tukey or Benjamini & Hochberg correction for multiple comparisons, for continuous variables. The reverse Kaplan-Meier method was used to estimate follow-up time. The Kaplan-Meier method was used to estimate OS, and log-rank tests were used to test for differences in survival between groups. A Cox Proportional Hazards model was used to estimate the association of response with OS after adjustment for select covariates, including demographic and disease variables, available in the NCDB with clinical relevance (specified within each analysis). This model included a sandwich covariance estimator to account for the correlation of patients treated at the same facility. Due to NCDB administrative censoring, patients diagnosed in 2021 were excluded from all survival analyses. Only patients with complete data were included in each analysis, and effective sample sizes are included for each table and figure. Of note, HER2 status was not routinely recorded in the NCDB until 2010, and thus, 17,340 patients were excluded based on missing HER2 status. As such, an additional multivariable subgroup analysis was conducted including only patients diagnosed 2010–2020. All statistical analyses were conducted using SAS version 9.4 (SAS Institute, Cary NC) and R studio (Version 4.2.2). Given the use of retrospective, de-identified data from a national tumor registry, this study was deemed exempt by our local institutional review board and did not require informed consent.
RESULTS
The final cohort included 91,505 patients (Figure 1), with a median follow-up time of 67.6 months (95% CI 67.15 – 68.01). Within the cohort, 26,981 patients received ET-only (29.5%) and 10,805 patients received RT-only (11.8%, Table S1). Comparing the RT-only and ET-only groups, patients were slightly older in the ET-only group [median 75yo (IQR 71–80)]; compared to the RT-only group [median 73yo (IQR 69–79); p<0.001]. Patients in the ET-only group were less likely to have a comorbidity score of 0 (comorbidity score of 0: ET-only 75.4% vs RT-only 80.8%; p<0.001; Table 1). Patients in the ET-only group had negligibly larger tumors [median tumor size (IQR): ET-only 1 (0.7–1.5) vs RT-only 0.9 (0.6–1.3); p<0.001], were less likely to have grade 3 tumors (ET-only 7.01% vs RT-only 8.38%; p<0.001), and were more likely to have PR+ disease (ET-only 91.4% vs RT-only 89.6%; p<0.001; Table 1).
In the unadjusted Kaplan-Meier analysis, when comparing ET-only vs RT-only directly, RT-only had a better 5-year OS rate compared to the ET-only group [RT-only 88.9% (88.2–89.6%) vs ET-only 85.8% (85.3–86.3%); log rank p<0.001; Figure 2]. When stratifying the monotherapy treatment groups by ER status (low 1–10% vs high >10%), patients with high ER status who received RT-only still had a better 5-year OS [RT-only 93.0% (88.7–97.5%) vs ET-only 86.1% (83.0–89.3%); p=0.03; Figure 3A]. Furthermore, when stratified by PR status (positive vs negative), the RT-only treatment group continued to show better 5-year OS regardless of PR status [PR-, RT-only 86.8% (84.6–89.0%) vs ET-only 85.1% (83.4–86.9%); PR+, RT-only 89.2% (88.5–89.9%) vs ET-only 85.9% (85.4–86.4%); p<0.001; Figure 3B]. Stratifying by T status (T1 vs T2) continued to show greater 5-year OS in the RT-only group for patients with both T1 and T2 tumors [T1, RT-only 89.5% (88.9–90.2%) vs ET-only 86.8% (86.4–87.3%); T2, RT-only 80.4% (77.3–83.7%) vs ET-only 73.0% (70.7–75.4%); p<0.001; Figure 3C]. Finally, a similar trend was observed when stratifying ET-only vs RT-only by age group, although notably less pronounced (all log rank p<0.05; Figure 3D). Patients receiving both ET and RT had the greatest 5-year OS [ET & RT 93.1% (92.8–93.3%)] with patients receiving no adjuvant therapy having the worst 5-year OS [No adjuvant therapy 79.9% (79.0–80.7%); Figure S1].
In the adjusted analysis, comparing patients who received either ET-only or RT-only (n=30,177), and accounting for demographics, comorbidities, tumor characteristics, insurance status, facility type and location, RT-only remained associated with a better OS than ET alone [ET-only ref, RT-only HR 0.91 (95% CI 0.85–0.97); p=0.002; Table 2]. In an adjusted subgroup analysis of only patients diagnosed 2010–2020 (when HER2 status was reliably recorded in the NCDB), we similarly found that RT-only was associated with better OS compared to ET-only [ET-only ref, RT-only HR 0.91 (95% CI 0.85–0.97); p=0.003; Table S2]. When including all 4 treatment groups in the adjusted analysis, patients who received only one adjuvant treatment modality or no adjuvant therapy had worse OS compared to patients receiving dual therapy [ET & RT ref, ET-only HR 1.45 (95% CI 1.38–1.52), RT-only HR 1.34 (95% CI 1.27–1.42), No adjuvant therapy HR 1.89 (95% CI 1.78–2); all p<0.001); Table S3].
DISCUSSION
Overall, our study showed that older women with ER positive disease are more likely to receive ET monotherapy compared to RT monotherapy, following lumpectomy. Patients in both treatment groups were of similar age, suggesting that age did not influence the treatment decision in this population of older women. The RT-only group had slightly better OS compared to the ET-only treatment group, even when stratified by age. In the adjusted multivariable analysis which included age, RT-only remained associated with an improved OS.
While ET and RT are commonly used therapies for breast cancer patients and often used together, there is less data comparing them as monotherapies directly. The NSABP B21 trial found that receiving RT-only following lumpectomy significantly reduced frequency of local recurrence compared to patients who only received tamoxifen [14]. The BASO II study built upon these findings, demonstrating that while both ET and RT independently reduced the likelihood of local recurrence, the combination of both treatments provided the greatest reduction in recurrence risk [15]. While studies have shown that RT decreases the chances of local recurrence, there appears to be less evidence that the addition of RT to ET provides an OS benefit. The CALGB 9343 trial found that the addition of RT to ET did not improve OS in women aged ≥70 years old with early-stage, ER-positive breast cancer who are taking adjuvant ET (median follow-up: 12.6 years) [16]. Their results provided evidence for the potential de-escalation of RT in this specific population given their tumor biology, comorbidities, and potentially limited life expectancy. Similarly, the PRIME II study investigated de-escalation of RT in older women taking ET, finding that RT decreases local recurrence rates in women over 65, but again, did not improve OS (median follow-up: 9.1 years) [9]. While both studies investigated the de-escalation of RT, there is a gap in the literature regarding the potential de-escalation of ET [17].
However, the EUROPA trial prospectively compared ET-only versus RT-only in older patients with low-risk early-stage breast cancer [11]. At the 2-year preplanned interim analysis, investigators concluded that receiving ET-only was associated with a greater decrease in quality of life scores and increased treatment related adverse effects compared to RT-only [11]. However, as an ongoing study, the sample size was limited and only included the first 207 patients enrolled, who were notably 99% Caucasian [11]. Our study, using a much larger and more diverse cohort, similarly suggests that RT-only may be a viable treatment option, and taken together with the findings from the EUROPA trial, possibly even a preferable approach in this unique population.
While ET is relatively easy and accessible (an oral pill for 5–10 years), it often comes with side effects, sometimes making it difficult for patients to complete the treatment course and receive its full potential benefits. Some studies estimate that over 30% of women do not adhere to their ET treatment course [10]. In addition, ET works systemically to prevent distant recurrences, which are most likely to occur 5–20 years after diagnosis [18]. Unfortunately, these benefits may be beyond the life expectancy of select patients. Newer studies have investigated the use of a lower dose of tamoxifen to aid in the prevention of breast cancer and to minimize side effects [19]. This is also being studied in a prospective, international, open-label, randomized, non-inferior trial for patients with low-risk DCIS, comparing low dose tamoxifen to conventional fraction RT and hypofractionated RT (TBCC-ARO DCIS Trial, ClinicalTrials.gov, ID: NCT04046159) [20]. However, low dose tamoxifen (or other types of endocrine therapy) has not yet been investigated specifically in an older patient population. Although a lower dose of ET could potentially reduce side effects and lead to greater treatment adherence, its comparative efficacy remains understudied.
On the other hand, RT offers a shorter treatment course, and while not without side effects, newer data for hypofractionation, ultra-hypofractionation, and partial breast approaches have made modern RT courses more tolerable with improved side effect profiles compared to older treatment courses. Patients receiving RT often experience short-term skin irritation and fatigue. In the longer term, RT can result in changes in breast cosmesis, with rare pulmonary, cardiac, and secondary malignancy risks. However, RT is a much shorter treatment course compared to ET and reduces the chance of local recurrence significantly [21]. While it is shorter, patients do have to go into the clinic to receive the treatment, which could be a potential barrier for patients struggling with transportation. Newer research has continued to demonstrate efficacy with shorter and lower dose radiation courses, including more targeted approaches, such as partial breast or hypo-fractionated regimens [22]. The FAST-Forward trial found that patients can safely and effectively achieve the same local tumor control and results with a 1-week course of RT (26 Gy) compared to the traditional 3-week course (40 Gy) [23]. This personalization and condensing of RT appear to provide women with comparable efficacy and may make RT a more appealing treatment option for select patients. Based on these data, patient reported outcomes may also serve as an important endpoint for future studies as well.
Given our study results suggesting that radiation monotherapy results in similar, or potentially improved, OS as endocrine monotherapy, and the recent findings from the EUROPA trial demonstrating a better quality of life with RT, shared decision making will be critical. Different patients may have unique goals and concerns involving their treatment that should be considered when deciding the best adjuvant therapy for each individual patient. If future studies continue to demonstrate equivalent OS outcomes with RT-only vs ET-only approaches, an educational tool could be developed to help facilitate informed decision making. While our study explored the potential association of ER and PR status with outcomes, genomic assays could also potentially be performed to help determine if patients should receive ET-only vs RT-only. For example, the Ultralow MammaPrint assay can help to identify patients whose long-term risk of death from breast cancer is extremely low, allowing for the omission of adjuvant treatments [24]. Ultimately, additional prospective studies are needed to further clarify the risks and benefits of each approach, as well as barriers to each approach.
As with all retrospective database studies, our study has limitations. For example, the NCDB is dependent on data entry, and although quite standardized, it is possible that there were errors in the data entry or that there was incomplete or missing data that could potentially have impacted our results. Missing data from databases has been shown to falsely inflate OS [25], and some would suggest that cancer-specific-survival may be a better end-point, which is not routinely available in the NCDB, although data entry is improving for this variable [26], and it may become available in the future. Furthermore, our RT-only group was a relatively small sample size (N=10,805) compared to the entire study cohort (N=91,505). While this was not ideal, it is reflective of current treatment guidelines where older patients are much more likely to receive ET-only compared to RT-only. Additionally, our study was not able to account for actual treatment compliance for patients receiving ET. The NCDB only records whether a patient initiates ET or not, it does not provide any further details on follow-up. Therefore, it is possible that some patients in the ET-only group did not actually finish their treatment course, potentially leading to worse OS. Finally, it is likely there was some selection bias and treatment bias within our dataset. This bias may have led us to find that the combined ET+RT treatment group had the greatest OS benefits, which contrasts with the findings of the PRIME II and CALGB9343 studies. While we attempted to mitigate for selection and treatment bias by adjusting for age and comorbidities via a multivariable Cox regression model, it is possible that we were not able to eliminate biases completely. Further analyses may benefit from propensity score matching and/or inverse probability of treatment weighting, although generalizability may be more limited with these statistical approaches. Using real-world datasets, it is difficult to fully account for other life factors that may guide the treatment decisions that were made. A frailty score or ECOG score may have helped to determine overall patient health status and further reduce possible selection bias present. Furthermore, it is possible that the NCDB did not track treatments accurately, and patients may have received additional therapies at other facilities that were not accounted for in our study. This is particularly pertinent for RT therapies, where 1/8th of patients in the NCDB have been found to have anomalous radiation data [27].
In conclusion, survival outcomes were overall similar for older patients with ER+/HER2− disease in this retrospective cohort study of >90,000 patients in the NCDB, regardless of whether they received ET or RT monotherapy after lumpectomy, although treatment selection bias could have influenced these findings. Additional prospective studies are needed to further investigate this finding, and survival outcomes from the EUROPA study will be critical. In the interim, providers could consider these findings when having treatment discussions with older women to help personalize treatment plans.
JP
Dr. Jennifer Plichta
Surgery · Breast Surgical OncologyDuke Medical Centerjenniferp