Introduction

The incidence of breast cancer has been steadily rising and represents the most common cancer in women. Over two million women are diagnosed with breast cancer worldwide every year. This accounts for over 30% of female neoplasms. Lifetime risk of developing breast cancer in women has been shown to be nearly one in eight with the 5-year overall survival estimated to be 80% or higher in high-income countries [1].

Early diagnosis and the employment of efficacious personalised treatments such as endocrine therapies have conferred significant mortality benefit [2]. Considering 80% of breast cancers are oestrogen receptor positive in women aged over fifty years, this makes drugs such as aromatase inhibitors (AIs) a highly attractive option. Both steroidal (exemestane) and non-steroidal (anastrozole and letrozole) AIs achieve over 98% aromatase inhibition in post-menopausal women with 80–90% resultant drop in peripheral oestradiol level. This is crucial in delivering improved survival for these patients [3].

Whilst conveying major benefits, the oestrogen deficiency induced by AIs leads to a significant increase in bone resorption and accelerated bone loss, especially at the trabecular bones [4]. The effects of adjuvant therapy on bone mineral density (BMD) have been quantified by several studies that have reported an annual bone loss in healthy postmenopausal women of 1–2% per year, while AIs therapy alone causes 2–3% BMD loss per year, greater during the first year, and progressively lower in the following years [5]. This translates into higher fracture risk with up to one in five women prescribed AIs reported to sustain a clinical fracture and nearly a third with incident vertebral fracture on morphometric analysis [6].

To address the AIs associated bone health concerns, a consensus statement of seven international bone and cancer societies was published in 2017 proposing an algorithm based on clinical risk factors and different BMD threshold for bone active therapeutic intervention [7]. Our study aims to determine the real-world impact of the 2017 consensus guidelines on AIs induced bone loss and whether bone sparing therapy utilising proposed risk stratification model is effective in fracture prevention compared to the 2008 UK recommendations [8].

Methods

We undertook a retrospective study of patients prescribed AI for breast cancer over a seven-year period at our university teaching hospital. All the data was recorded electronically with full access to demographics, disease parameters, investigations, and drug management. Dual energy X-ray absorptiometry (DXA) scans performed prior to initiation of AI were compared with subsequent imaging over a mean follow up of 3 years. Outcome data for cancer and all fractures was collected. Descriptive statistics were employed to investigate significant relationships amongst the variables of interest. The project was approved on Jan 7, 2022 (approval number 13/2021-22/Medicine/Rheumatology).

Over a 7-year study period, 1001 women were prescribed AI at our university teaching hospital by rheumatologists working in the rheumatology department. The new guidelines were adopted in July 2017. We split the participants in two groups: 361 (36%) women had commenced their AI prior to the adoption of guidelines (Group One) and 640 (64%) were in the post implementation group (Group Two).

First group were offered bone active treatment based on National Osteoporosis Society (NOS) 2008 guidelines whereas the second group followed the 2017 consensus guidelines. The inhouse protocols for treatment were based on these guidelines. Women with osteoporosis were all offered treatment, however the difference in guideline is pertinent to osteopenia and we compared the results of that group. All patients with osteopenia or osteoporosis were advised to take calcium and vitamin D.

Patients were followed for bone health by having a repeat DXA scan. Fractures whilst on AIs were either diagnosed by imaging or a verbal report was received from a patient that they had sustained a fracture.

Results

1001 women were included. Mean age was 64 years (range 29–93). 929 (93%) were Caucasian, 57 (6%) were Asian and 15 (1%) were Afro-Caribbean. 723 women (72%) had invasive ductal carcinoma and 863 women (86%) were postmenopausal. At diagnosis, 428 women (43%) had node positive disease and 35 women (4%) had metastases. 91 women (9%) had sustained fractures prior to their cancer diagnosis [Table 1].

Table 1 Demographics of the two cohorts.
Full size table

276 women (83%) were offered oral bisphosphonates based on DEXA result, with 58 (17%) offered parenteral therapy.

First group

361 women had a baseline DEXA with a mean femoral neck left (FN left). BMD of 0.888 g/cm2 (range 0.552–1.222) [Table 2].

Table 2 Baseline DXA measurements of the two groups at diagnosis.
Full size table

143 (40%) women had a normal DXA, 174 (48%) had osteopenia and 44 (12%) had osteoporosis.

Of the women with osteopenia (n = 174), 22 (12.6%) suffered fractures [Fig. 1]. 44 (25%) women were offered treatment based on 2009 NOS guidelines and 33 women had a repeat DXA after a mean of 4 years. In the treatment group, FN left mean BMD remained relatively unchanged from 0.814 g/cm2 to 0.812 g/cm2 at the repeat DEXA (p = 0.94) [Table 3]. 7/44 (16%) had fractures with no prior history of fractures with a median time to fracture of 1.5 years (range of 1–7 years). One lady was already prescribed bisphosphonates.

Fig. 1
figure 1

Group one DXA results spread.

Full size image
Table 3 Baseline DXA scores of individuals found to have osteopenia, and DXA scores at follow up for those who were offered bone active treatment.
Full size table

130/174 (75%) did not qualify for bone active treatment. 15/130 (11.5%) suffered fractures during follow up [Fig. 2].

Fig. 2
figure 2

Group one fracture spread.

Full size image

Second group

640 women had a baseline DEXA with a mean FN left BMD of 0.888 g/cm2 (range 0.512–1.390) [Table 2].

216 (33%) women had normal DXA, 322 (50%) had osteopenia and 107 (17%) had osteoporosis.

Of the 322 women with osteopenia, 8 (2.5%) women had a fracture [Fig. 3].

Fig. 3
figure 3

Group two DXA results spread.

Full size image

127/322 (39%) women were offered treatment, and 56 women had a repeat DXA after a mean of 3 years. In the treatment group, FN left mean BMD remained relatively unchanged from 0.822 g/cm2 to 0.829 g/cm2 at the repeat DEXA (p = 0.6169) [Table 3]. 3/127 (2.3%) suffered fractures with a median time to fracture of 2 years (range 1–9 years).

195/322 (61%) did not qualify for bone active treatment. 5/195 (2.5%) had a fracture during the follow up [Fig. 4].

Fig. 4
figure 4

Group two fracture spread.

Full size image

Discussion

Our study provides real world evidence of the success of 2017 consensus statement in lowering fracture risk. Though there has been data for positive impact on BMD decline with this approach, evidence for fracture prevention has been limited. This study showcases the success of lowering bone active therapy threshold employing alternative risk modelling strategy for women with breast cancer commenced on AI. A significant reduction in fractures pre (12.5%) and post guidelines change (2.5%) was demonstrated (absolute risk reduction of 10%) which has implications for healthcare systems worldwide as we have demonstrated this approach can reduce morbidity.

Poor bone health risk is well established in women with breast cancer owing to higher bone remodelling secondary to advanced age, hypovitaminosis D and employment of corticosteroids in chemotherapeutic regimens [9]. AIs are reported to nearly double this risk when annual bone loss rates are compared to healthy postmenopausal women [6]. The concern is even higher in premenopausal women prescribed endocrine therapy particularly those with premature ovarian failure due to chemotherapy [10].

Irrespective of the degree of BMD loss, AIs have been shown to confer higher fracture risk compared to placebo or Tamoxifen. The risk seems to be similar for all the approved agents [11]. Furthermore, the longer the exposure to AIs, the higher the risk of fractures. This is important as most guidelines recommend 5–10 years of AI treatment to improve cancer outcomes [12]. As it is mainly trabecular bone loss seen with AIs, vertebral fractures are the predominant manifestation with up to a third of women affected in a recent study [13].

Therefore, it’s imperative that fracture risk assessment tools consider the significant risk conferred by AIs especially when the probability of fracture is partly independent of BMD. This is pivotal in women with osteopenia as those with osteoporosis will be offered fracture prevention therapy anyway. Fracture risk assessment tool (FRAX) has been shown to work reasonably well in a large registry-based cohort study however the threshold to commence bone active therapy is subject to discussion captured in the variability of the guidelines [14].

In the UK, most centres follow 2008 Consensus position statement from a UK Expert Group which recommended bisphosphonates for postmenopausal women younger than 75 years with T-scores <−2.0 or with bone loss ⩾ 4% per year in pre-existing osteopenia (T scores between −1.1 to −2.4). At our centre, this translated to only a quarter of women with osteopenia qualifying for bisphosphonates leading to over one in ten women fracturing during follow up [7].

In 2017, The Joint position statement of the International Osteoporosis Foundation, Cancer and Bone Society, European Calcified Tissue Society, International Expert Group for AIBL, European Society for Clinical and Economics Aspects of Osteoporosis, Osteoarthritis, and Musculoskeletal Diseases, International Menopause Society and International Society for Geriatric Oncology recommended starting anti-osteoporotic therapy when T-score is <−2.0 or with ⩾ 2 risk factors (including T-score <−1.5). When we switched to these guidelines, an extra 15% women qualified for bone active therapy thereby achieving absolute fracture reduction of 10% compared to the earlier cohort [8]. Our study confirms the superiority of the 2017 joint position statement which were reinforced by European Society of Medical Oncology (ESMO) [15].

All guidelines concur on the need to offer bone sparing therapy for women with osteoporosis however the differences exist in recommendations for osteopenia. Our study reaffirms that a significant number of fractures occur in women with osteopenia. The differential improvement in fractures observed is largely down to the extra 15% of women qualifying for antiresorptive therapy based on 2017 guidelines. This endorses the need for holistic risk assessment and lower threshold for intervention in women with low bone mass considering all risk factors.

There are several caveats to consider including monocentric, retrospective nature of the study with two unmatched cohorts in a diverse ethnic setting thus making generalisability of the findings difficult. Additionally, data was not collected on markers of bone turnover, vitamin D levels or administration of concomitant glucocorticosteroids or chemotherapy which may have been confounding factors. However, the strengths include a large cohort with a long follow up period, availability of comprehensive clinical data and ability to provide hard outcome of fracture rather than BMD data.