1. Introduction
Breast cancer is increasingly diagnosed at an older age. In the United Kingdom (UK), early-stage breast cancer is predominately diagnosed in postmenopausal women (86% aged ≥ 50 years), but 46% are at older ages (age ≥ 70 years) [
1]. The optimisation of treatments for older female patients with breast cancer presents challenges due to the scarcity of robust evidence in treatment effectiveness [
2,
3], constrained clinical capacity for managing complex cases [
4], and the preferences of patients for less invasive treatments [
5].
According to the national clinical guideline in England, irrespective of a patient’s age at diagnosis, surgery and adjuvant systemic therapy are recommended as the first-line treatment for operable breast cancer that is oestrogen receptor positive (ER+) unless significant comorbidity precludes surgery [
6]. Evidence has shown that removing the tumour by surgery significantly extends the survival of postmenopausal women with early-stage operable breast cancer [
7]. Nevertheless, among older female patients deemed unsuitable for surgical intervention, there exists a subset whose life expectancy might be diminished or whose prospects for survival might not be optimally realised through surgery, primarily owing to their compromised physical condition characterised by comorbidities and frailty [
4]. Consequently, this cohort is experiencing a growing shift towards non-surgical therapeutic options [
8].
Primary endocrine therapy (PET) has been considered an alternative to surgery for older female patients diagnosed with ER+ early-stage breast cancer, particularly those for whom surgery is unsuitable [
8]. Nevertheless, its comparative effectiveness has not been comprehensively studied. In the most recent meta-analysis encompassing seven randomised controlled trials (RCTs) and six non-RCTs, a comparative evaluation was conducted between 2829 cases subjected to PET and 11,425 cases subjected to surgical intervention [
9]. The findings elucidated that while PET exhibited non-inferior regarding overall survival compared to surgery, it was non-inferior concerning breast cancer-specific mortality [
9].
However, these results are not seen in clinical practice as physically fit patients are seldom offered PET [
10]. Besides those RCTs, the lone RCT conducted by Reed et al. in 2008, which sought to compare primary endocrine therapy with surgical intervention in older women diagnosed with early-stage breast cancer by various health statuses, was hampered by its inability to recruit an adequate number of participants [
11]. Meanwhile, patients with multi-morbidity or frailty are often excluded from RCTs due to the risk of serious adverse events [
12,
13]; thus, treatment effectiveness in those vulnerable patients is scarce.
Due to the challenge of conducting RCTs in older women with physical functioning impairment, observational studies are considered a feasible approach to evaluate the effectiveness of PET in older women. In line with findings from earlier RCTs, several observational studies investigating older women with breast cancer and impaired physical functioning have reported better overall survival in patients undergoing surgery than those undergoing PET [
14,
15,
16,
17,
18,
19]. However, these findings are susceptible to confounding by indication, presenting a challenging bias to overcome. Additionally, given the presence of frailty or comorbidity in this patient population, there may be an increased risk of non-cancer mortality, leading to higher ’all-cause mortality’ (poor overall survival), a commonly used outcome measure in oncology [
20]. Consequently, the comparative effectiveness of PET versus surgery in older female patients with various degrees of physical functioning impairment remains inconclusive.
To address this knowledge gap, this study aimed to compare all-cause (ACM) and breast cancer-specific mortality (BCSM) associated after PET and surgery in older women (aged ≥ 70 years) with newly diagnosed early-stage oestrogen-receptor-positive breast cancer at different frailty and comorbidity levels. Rigorous examination and sensitivity analysis were conducted to inform appropriate treatment approaches for this patient cohort.
4. Discussion
Surgical intervention continues to serve as the cornerstone therapeutic approach for breast cancer. While surgery is generally considered the preferred method for treating early-stage breast cancer in older women, regardless of comorbidity and frailty levels, PET may still be a viable alternative for those with significant frailty. In this particular patient cohort, no statistically significant difference in breast cancer-specific mortality was observed between surgery and PET arms (
p = 0.251). Similarly, previous studies have reported no significant difference in breast cancer-specific survival when comparing surgery with PET (HR = 0.74, 95%CI: 0.40–1.37,
p = 0.34) [
39]. The similarity in outcomes may be influenced by the fact that mortality in this cohort is often due to non-cancer causes, such as cardiovascular or respiratory diseases, rather than breast cancer itself. Hence, while these results are intriguing, they should be interpreted with caution and further research is needed to validate PET as a comparably effective treatment strategy for surgery for frail older women with early-stage breast cancer.
Although PET is not recommended as the primary strategy for female patients aged over 70 years, 30% of older women with early-stage breast cancer still received PET in routine care. Patients who received PET as the initial treatment strategy had poorer physical functioning than those with surgery. However, after adjusting for relevant covariates, we found the risk gap of ACM between PET and surgery narrowed with increased HFRS and CCI levels. Notably, no statistical difference in BCSM was found between PET and surgery in frail patients. Overall, our study strengthens the national guideline in England [
8] that surgery should be the first-line treatment for operable patients in good physical condition, irrespective of age, since surgery leads to better overall survival and is more cost-effective than PET [
40]. Nevertheless, PET is a potentially effective strategy for older female patients who are physically unfit for surgery (frail or with a higher comorbidity level).
The prevailing preference for PET has its distinctive historical underpinnings. As endocrine therapy emerged and RCTs elucidated its clinical efficacy and effectiveness, PET garnered widespread adoption as a surgical alternative owing to its demonstrated non-inferiority and heightened efficiency relative to surgical interventions. Nowadays, although surgery is still recommended as the first-line treatment for patients who are physically fit for surgery, our study found that a high proportion of older female patients undergoing PET were either non-frail (69.6% of patients who received PET) or had low levels of CCI (70.2% of patients who received PET). These patients could have received surgery instead as their initial treatment. Instead, they received PET due to the challenges in evidence-based clinical decision-making and older female patients’ preference for treatment choices [
41].
As mentioned in the National Audit of Older Women with Breast Cancer UK report [
8], using PET as initial treatment for older female patients who are physically fit for surgery was inconclusive due to a lack of robust evidence to indicate the comparative effectiveness of PET versus surgery by different levels of physical functioning [
42]. The findings from our study address this evidence gap and stratify the characteristics of patients who may benefit from PET to support clinical precision decision-making.
Several previous observational studies have evaluated the clinical effectiveness of surgery and PET [
43,
44,
45,
46]. Still, only one systematic review has summarised the treatment effects in older female patients with breast cancer from such observational studies [
47]. This systematic review notably identified a potential source of bias in the observational studies, primarily stemming from confounding by indication [
47]. Also, all the included studies evaluated the effects of tamoxifen rather than aromatase inhibitors; the latter is more often used in clinical practice for older female patients. Consequently, the findings presented in our study bear greater relevance to the contemporary standard of care for this demographic.
Our study identified the routine clinical practice of different types of PET given to older female patients. Although tamoxifen was still typically prescribed for patients with ER+ breast cancer, aromatase inhibitors (e.g., letrozole) were more prevalently prescribed for female patients aged ≥ 70 years and undergoing PET. A patient-level meta-analysis of four RCTs [
48] evaluated the post-surgical adjuvant treatment effects between aromatase inhibitors and tamoxifen for postmenopausal women with ER+ early-stage breast cancer. The results demonstrated that aromatase inhibitors had superior effects on reducing the likelihood of recurrence (RR: 0.79, 95%CI: 0.69–0.90,
p = 0.0005) and distant recurrence (RR: 0.83, 95% CI: 0.71–0.97;
p = 0.018) [
48]. Thus, the benefit of PET over surgery in very frail patients would be expected to be more marked.
In contrast to the existing body of literature, our study offers a comprehensive investigation into the relative clinical effectiveness of PET versus surgical interventions among elderly female patients, stratified across varying levels of physical functioning (using CCI and HFRS). Our research addresses a notable gap in the current knowledge landscape by leveraging population-based data. These findings constitute a valuable resource for clinicians seeking evidence-based insights to guide treatment optimisation for older women diagnosed with early-stage oestrogen-receptor-positive (ER+) breast cancer, particularly in varying levels of frailty and comorbidity.
This study had some strengths. Firstly, it evaluates the clinical efficacy of PET and surgical interventions utilising a high-quality UK primary care database and enriching its coverage through linkage with cancer registry records, secondary care data, socioeconomic indices, and death certificates. This extensive data integration strategy enhances the study’s sample size, assuring excellent statistical power (>99%). Secondly, the investigation encompasses two pivotal outcome measures to inform clinical decision-making: overall survival (OS) and breast cancer-specific mortality (BCSM). Furthermore, the sensitivity analysis considered competing risks, fortifying the results’ robustness and reliability.
Third, the bias by indication in patients undergoing PET and surgery was considered using the propensity score matching (calculated using IPTW to predict the likelihood of undergoing two treatments) and weighting in regression adjustment [
35]. This doubly robust adjustment makes the survival outcomes comparable to the identical characteristics of two age strata by minimising the confounders (age, frailty, comorbidity, and socioeconomic status) between exposure and outcome [
35]. Fourth, patients were stratified by using CCI and HFRS as proxies for frailty and comorbidities; hence, the results revealed the treatment effectiveness by age-related physical functioning. Finally, this study used aromatase inhibitors as the most common endocrine medication (55.5% of PET medications), which better reflects the current routine practice for early-stage breast cancer. Undeniably, there was still the limitation of the risks of residual bias from IPTW; however, according to the sensitivity analysis, the result may still indicate that PET is an appropriate treatment for older patients who are frail or comorbid.
However, we acknowledge there are some limitations to this study. First, this study only selected three variables to match, and this may not perfectly reflect the influence on the survival consequences. There is some unmeasured confounding (e.g., diet habits, smoking and alcohol status, obesity, and psychological status) that may impact the survival benefits for the cohort [
49]. Nonetheless, regression was used in this matched cohort study to adjust covariate variables (e.g., comorbidity and frailty) that can reflect the impact of such unmeasured confounding as smoking, alcohol status, and obesity on physical conditions.
Second, although frailty and comorbidity commonly represent physical functioning in clinical practice, physical functioning as a complicated physical status may not entirely and accurately be quantified using only two indicators. Physical functioning also includes the body’s movement ability and quality of life [
50]. In addition, although frailty or comorbidity was estimated in this study using the published validated algorithm [
30,
31], the frailty and comorbidity status for the patients in the real world may not be accurately quantified through the limited data source due to other factors that may determine frailty or comorbidity in routine practice. Therefore, this study only analysed the results based on the patients with the represented frailty or comorbidity status to inform the decision-making of treatments in older female patients. Future research on treatment effectiveness in older patients should assess survival outcomes based on their physical conditions, including frailty (HFRS) or comorbidity (CCI) scores.
Finally, this study defined PET and surgery as the initial treatment strategies, irrespective of the subsequent treatment received. Some older female patients who initially received PET would also receive surgery when their breast cancer progressed. However, post-progression management is a different clinical decision problem than early-stage breast cancer management upon diagnosis. To minimise misclassification bias in this study, patients who received PET and then received surgery after one year were excluded from the study cohort.
In forthcoming research endeavours, it is imperative to undertake cost-effectiveness analyses contrasting PET with surgical interventions for frail older female patients. These analyses are pivotal in augmenting the precision of breast cancer management decisions. However, only one investigation conducted by Holmes et al. [
51], derived from the Bridging the Age Gap in Breast Cancer study, has revealed that, except for a select subgroup comprising patients aged 90 years or older with a comorbidity score of 2 or 3, regardless of nodal status, surgical interventions manifest superior cost-effectiveness compared to PET [
51]. Therefore, there is a pressing need for further assessments, stratified by frailty levels, to elucidate nuanced treatment selection paradigms.
Furthermore, it is worth noting that countries beyond the United Kingdom, including the United States, Canada, and various European nations such as the Netherlands and Sweden, confront analogous dilemmas when determining the optimal therapeutic course for older female breast cancer patients [
52]. In light of this, a global perspective should be adopted to facilitate comprehensive investigations aimed at discerning the specific patient characteristics that might derive maximal benefit from PET. Lastly, in the context of geriatric healthcare, where RCTs may not always be feasible or biased by indications, the judicious utilisation of propensity score matching or weighting represents a promising approach for evaluating treatment efficacy and effectiveness in other specific geriatric diseases.