Aromatase inhibitor-associated bone and musculoskeletal effects: new evidence defining etiology and strategies for management

Aromatase inhibitors are widely used as adjuvant therapy in postmenopausal women with hormone receptor-positive breast cancer. While the agents are associated with slightly improved survival outcomes when compared to tamoxifen alone, bone and musculoskeletal side effects are substantial and often lead to discontinuation of therapy. Ideally, the symptoms should be prevented or adequately treated. This review will focus on bone and musculoskeletal side effects of aromatase inhibitors, including osteoporosis, fractures, and arthralgias. Recent advances have been made in identifying potential mechanisms underlying these effects. Adequate management of symptoms may enhance patient adherence to therapy, thereby improving breast cancer-related outcomes.


Introduction
In 2010, it is estimated that more than 200,000 women will be newly diagnosed with invasive breast cancer in the United States [1], making it the most commonly diagnosed cancer in women. Th e majority of women are post-menopausal at the time of diagnosis. Adjuvant endocrine manipulations reduce the risk of breast cancer-related recurrence and death in women with hormone receptor-positive disease. Th e introduction of aromatase inhibitors (AIs) to the adjuvant treatment of postmenopausal women with hormone receptor-positive breast cancer has signifi cantly changed the management of the disease. Th ese agents are commonly used instead of or in sequence with tamoxifen because of the demonstrated improvement in disease-free survival compared to tamoxifen alone [2]. Since long-term survival rates are high in patients with early-stage breast cancer who receive AIs and treatment may continue for many years, the complications arising from therapy in this patient population can have long-term eff ects and may greatly impact patient quality of life.
Th e three third-generation AIs in routine clinical useanastrozole (Arimidex), letrozole (Femara), and exemestane (Aromasin) -have similar effi cacy and toxicity profi les when evaluated in cross-study comparisons. Th e primary adverse eff ects include menopausal symptoms, vaginal dryness, sexual dysfunction, and musculoskeletal symptoms, including bone demineralization with risk of osteoporosis and fracture, arthralgias, and myalgias. Th is review will focus on AI-associated bone and musculoskeletal toxicities, including prevalence, typical symptoms, potential etiologies, and strategies for management of these side eff ects.

Aromatase inhibitor effi cacy and safety
Estrogen is primarily produced in the ovary prior to menopause. After menopause, estrogen production occurs in peripheral tissues (skin, muscle, fat, and benign and malignant breast tissue) through the conversion of androgens to estrogens by the P450 cytochrome enzyme aromatase (CYP19) [3][4][5][6]. Th ere are two primary approaches to the hormonal treatment of estrogen receptor (ER)-positive breast cancers: selective ER modulators (for example, tamoxifen) that directly interact with the ER and inhibit its activity in breast tissue; and AIs that reduce post-menopausal production of estrogen [2]. Th e nonsteroidal AIs anastrozole and letrozole competitively inhibit aromatase, while the steroidal AI exemestane irreversibly inhibits the enzyme; however, both types of inhibitors suppress plasma and tissue estrone concentrations, the dominant estrogen in post-menopausal women, by >93% [7][8][9]. AIs are ineff ec tive in women with functional ovaries because of their inability to block ovarian production of estrogen [10].
Numerous large randomized controlled trials have evaluated AIs in the treatment of early-stage hormone receptor-positive breast cancer. Th e studies have consistently demonstrated improved disease-free survival when used in multiple settings: upfront in place of tamoxifen, following 2 to 3 years of tamoxifen (sequential strategy), or after completion of 5 years of tamoxifen therapy (extended strategy) [11][12][13][14][15][16][17][18][19]. However, there has been no overall survival advantage when compared to tamoxifen.
Results of these clinical trials have also demonstrated a favorable safety profi le for the AIs compared to tamoxifen. In the long-term safety analysis of the Anastrozole, Tamoxifen Alone or in Combination (ATAC) trial, signifi cantly fewer treatment-related adverse events were observed resulting in fewer withdrawals due to drug-related adverse events in the anastrozole group compared to tamoxifen alone. In comparison to tamoxifen, anastrozole was associated with fewer thromboembolic events, cerebrovascular events, and diagnoses of endometrial cancer [11]. However, reports of osteopenia, osteoporosis, and fracture rates were increased in the anastrozole group as were rates of dyspareunia and decreased libido secondary to vaginal dryness, increased lipidemia, and worsening joint symptoms. Similar results were seen in the major trials of each of the third generation AIs [11][12][13][14][15][16][17][18]20].

Bone demineralization and aromatase inhibitors
Numerous reports have demonstrated that aromatase sup pression leads to clinically signifi cant bone de minerali zation resulting in increased rates of osteopenia, osteoporosis, and fractures (Table 1). In the ATAC study, higher fracture rates were reported in the anastrozole arm when compared to tamoxifen (2.93% versus 1.9%, respectively, P < 0.0001, after a median follow-up of 100 months) [21]. However, after treatment was completed, fracture rates were equivalent. Th e fracture rate in anastrozole-treated women appeared to plateau after 24 months, with no progressive increase in fracture risk, although the fracture risk remained signifi cant [22]. In the Breast International Group (BIG) 1-98 trial, which directly compared 5 years of adjuvant tamoxifen with 5 years of letrozole, the fracture rate was signifi cantly higher in the letrozole group (8.6% versus 5.8%, P < 0.001) at 60 months follow-up [13]. Th e Intergroup Exemestane Study is a sequential dosing study designed to compare 5 years of tamoxifen with 2 to 3 years of tamoxifen followed by 2 to 3 years of exemestane [23]. After a median follow-up of 55.7 months, fracture rates and new diagnoses of osteoporosis were increased in patients receiving exemestane versus tamoxifen alone (4.3% versus 3.1%, respectively, for fractures, P = 0.03; and 7.3% versus 5.5%, respectively, for osteoporosis, P = 0.01) [14]. In each of these studies, the AI was compared to tamoxifen, which is thought to have a weak estrogenic eff ect on bone tissue, reducing bone resorption and maintaining bone mineral density [24,25]. Th e diff erence in fracture rates becomes less apparent when compared to placebo. In the National Cancer Institute of Canada Clinical Trials Group MA.17 study in which 5 years of letrozole were compared to placebo in women who completed 5 years of tamoxifen, there was no diff erence in the incidence of clinical fractures in the letrozole group compared with the placebo group (5.3% versus 4.6%, P = 0.25); however, more women receiving letrozole reported new diagnosis of osteoporosis in the 2 years following initiation of therapy (8.1% versus 6%, P = 0.003) [15]. Overall, fracture and osteoporosis rates were increased regardless of which AI or dosing strategy was used.

Bone mineral density as a marker of AI-induced bone fragility
In a substudy of the ATAC trial, there was a signifi cant reduction in lumbar spine and hip bone mineral density (BMD; 2.2% and 1.3%, respectively) in patients receiving anastrozole over the fi rst year of treatment; while BMD signifi cantly increased in women treated with tamoxifen over the same time period (1.0% and 0.5% increase in lumbar spine and hip, respectively) [26]. Over the 5 years of the study, the median decrease in lumbar spine BMD was -6.08% in the anastrozole-treated group compared with an increase of +2.77% in the tamoxifen-treated group [27]. Similar results were seen in the total hip measurements (-7.24% and +0.74% in the anastrozole and tamoxifen groups, respectively). After 2 years, patients in the MA.17 bone subprotocol receiving letrozole had a signifi cant decrease in total hip (-3.6% versus -0.71%, P = 0.044) and lumbar spine BMD (-5.35 versus -0.7%, P = 0.008) compared with placebo [28]. Whether BMD can be used as a surrogate for fragility fracture risk is controversial [29].

Proposed mechanism of bone loss
Bone metabolism is a balance between osteoblastic and osteoclastic activity. Estrogen defi ciency has been identifi ed as the key factor in mediating age-related bone loss [30]. Th ere is a clear association between post-menopausal estrogen defi ciency and the development of osteoporosis. ERs and aromatase are both expressed in bone, and estrogen has been shown to regulate bone remodeling by stimulating the expression of antiresorptive factors such as osteoprotegerin. Th is results in the attenuation of receptor activator of NF-kappa-B (RANK) and RANK ligand (RANKL) signaling, leading to inhibition of osteoclastogenesis and attenuated bone turnover [31,32]. Indeed, estrogen defi ciency is associated with increased expression of measurable markers of bone resorption and bone formation [33].

Molecular markers of bone turnover in AI-treated patients
Markers of bone remodeling were evaluated in several studies and the results support AI-associated increase in bone remodeling. In the ATAC bone substudy, at one year patients receiving anastrozole had a signifi cant increase in markers of bone resorption, including Ctelopeptide (CTX; +26%) and N-telopeptide (NTX; +15%) along with an increase in markers of bone formation, including bone alkaline phosphatase (ALP; +20%) and procollagen type-I N-propeptide (PINP; +18%) [26]. In contrast, patients receiving tamoxifen had a decrease in both resorption and formation markers (CTX -56%, NTX -52%, ALP -16%, PINP -72%). In the MA.17 bone substudy, an increase in the bone resorption markers NTX and CTX were observed in patients treated with letrozole at 24 months (+57% and +17%, respectively, compared to +16% and -12%, respectively, in patients treated with tamoxifen) [28]. In a double-blind placebo-controlled study comparing bone turnover markers following 2 years of exemestane to placebo in women with early breast cancer, exemestane was associated with a signifi cant increase in the markers compared to placebo (ALP +52% and +25%, respectively, and CTX +35% and -5%, respectively) [34].
Together, the data demonstrate that all AIs have potentially deleterious eff ects on measures of bone health with a decrease in BMD and increase in bone-remodeling. However, the overall incidence of fractures during 5 years of AI therapy is quite low. In the Anastrozole versus Letrozole, an Investigation of Quality Of Life and Tolerability (ALIQUOT) study, both anastrozole and letrozole were associated with similar eff ects on bone metabolism and turnover in postmenopausal women with ER-positive breast cancer [35]. In this study, discontinuing tamoxifen therapy and initiating an AI was associated with an increased rate of turnover compared to starting an AI in a patient who had never received tamoxifen. At the same time, the administration of tamoxifen after AI therapy is associated with a decrease in markers of bone resorption.

Risk factors
In the bone substudy of the BIG1-98 trial, several risk factors for the development of fractures were identifi ed, including increased age, prior fractures, diagnosis of osteoporosis at baseline, and previous hormone therapy [36]. Similarly, another study identifi ed eight risk factors among women with breast cancer: AI therapy, T-score <-1.5, age >65 years, low body mass index (<20 kg/m 2 ), family history of hip fracture, personal history of fragility fracture after age 50 years, oral corticosteroid use >6 months, and smoking [37]. Bone mineral loss was also increased in women who received letrozole in the 4 years since menopause compared to women who were more than 4 years since their menopause (median percent change -11.32 in women <4 years from last menstrual period and -5.41 in women >4 years since last menstrual period) [27]. Each of these factors may be an important consideration in assessing the most appropriate adjuvant therapy with the least toxicity for an individual woman.

Guidelines for management of AI-associated bone loss
Given the risk of developing skeletal-related events in otherwise healthy women with early-stage breast cancer treated with AIs, there has been a signifi cant interest in determining the best preventative measures and treatment strategies. Recent guidelines have been published with recommendations for the management of AIinduced bone loss [2,37,38]. As in all postmenopausal women, adequate dietary vitamin D and calcium intake are important for maintaining BMD [39]. Resistance and aerobic exercise also slows BMD loss in women with early breast cancer receiving cytotoxic chemotherapy [40]; however, the eff ects on AI-associated BMD are unknown. Reduction of other risk factors, such as cessation of smoking and minimization of other drugs associated with decreasing BMD (for example, corticosteroids), are also likely to have a positive impact on bone health. Dual energy X-ray absorptiometry (DEXA) scan to assess BMD is recommended at the initiation of therapy and at least every 2 years while receiving an AI [37].
Th e American Society of Clinical Oncology (ASCO) guidelines on the management of bone health issues in women with breast cancer recommend initiation of bisphosphonate therapy if osteoporosis is present on DEXA scan (T-score <2.5) [41]. Th e UK guidelines recom mend more aggressive treatment of bone mineral loss. Bisphosphonate therapy is recommended in all elderly (>75 years of age) women with one or more risk factors for osteoporotic fracture irrespective of BMD [38]. Bisphosphonate therapy should be considered for any post-menopausal woman whose T-score falls below -2 or if the rate of bone loss in a woman with pre-existing osteopenia exceeds 4% per year. In premenopausal women receiving ovarian suppression and an AI, the threshold for intervention is a T-score <-1 (because of very rapid bone loss averaging 17% over 3 years) [42].
Which bisphosphonate to use in the treatment of AIassociated bone loss has not been determined. Studies have demonstrated that intravenous zoledronic acid, oral ibandronate, and oral risedronate increase bone mineral density in AI-treated patients [42][43][44][45][46][47]. Another unanswered question is whether bisphosphonates should be initiated at the start of AI therapy rather than delaying until osteoporosis develops. Th e Zometa-Femara Adjuvant Synergy (Z-FAST and ZO-FAST) trials were designed to evaluate an immediate versus delayed strategy of bone protection with zoledronic acid [43,44]. Immediate therapy was more eff ective in preserving BMD at 12 months than delaying bisphosphonate therapy until the lumbar spine or total hip T-score was below -2.0 or when a non-traumatic fracture occurred. Neither study was powered to show a diff erence in the number of fractures. In the ARIBON trial (Arimidex-Bondronate), all patients received anastrozole but osteopenic patients were randomized at the start of therapy to receive either oral ibandronate or placebo. Patients receiving iban dronate gained rather than lost BMD (lumbar spine: +2.98% compared to -3.22% in patients receiving placebo) [45]. Similar results were found in the Study of Anastrozole with the Bisphosphonate Risedronate (SABRE) [46] and Arimidex Bone Mass Index and Oral Bisphosphonates (ARBI) [47] trials, showing the BMD loss can eff ectively be reduced or even completely mitigated by the addition of a bisphosphonate. However, whether the increase in bone mineral density and decrease in bone turnover translates into reduced fracture risk is under debate. Recent metaanalyses of the bisphosphonates in AI patients have called their use into question, particularly for prevention of BMD loss. While bisphosphonates were associated with improved BMD, there was no eff ect on fracture risk [48,49].
In addition to bisphosphonates, other treatment options are emerging. Denosumab, a RANKL targeted antibody that prevents bone resorption, was shown to increase BMD in AI-treated patients [50]. Combination therapy of AIs with inhibitors of Src, a non-receptor tyrosine kinase with roles in growth, metastasis, and bone metabolism, have shown promise in restoring sensitivity to endocrine-resistant cells [51]. Th e eff ects of this combination, and in particular evaluation of the eff ect on markers of bone resorption, are under evaluation in phase II clinical trials.

Aromatase inhibitor-induced arthralgias
Musculoskeletal symptoms have arisen as important adverse eff ects of AIs. In the major phase III clinical trials that compared AI to tamoxifen, the reported incidence of musculoskeletal symptoms ranged from 5 to 36% [11][12][13][14][15][16][17][18]20] (Table 2). However, case series have reported an even higher incidence of emergence of new or worsening joint symptoms in up to 61% of AI-treated women [52][53][54][55]. By contrast, tamoxifen has not been associated with increased joint symptoms [56,57]. While AI-induced arthralgias were reported as mild to moderate in severity and did not result in signifi cant discontinuation of medication in the large trials [58,59], in more recent analyses, severe AI-induced arthralgias resulted in therapy interruption in up to 20% of patients [52,55,60]. Th erefore, AI-associated arthralgia may account for reduced medication compliance, leading to decreased effi cacy and an increase in recurrence rates. Despite the frequent reporting of AI-induced arthralgias, the etiology of this adverse eff ect remains unknown.
Th e most commonly reported symptoms include morning stiff ness and pain of the hands, knees, hips, lower back, and shoulders [54,60], impairing ability to perform activities of daily living as well as work-related tasks [53,61]. In a cross-sectional analysis of postmenopausal women treated with adjuvant AI therapy at a university-based oncology clinic, the most common sites of joint pain were wrist/hand (60.4%), knee (59.7%), back (54%), ankle/foot (51.8%), and hip (42.5%) [62]. Digital stiff ness, trigger fi nger, and carpal tunnel syndrome have been frequently reported clinical symptoms [11,55,61,63]. Surgery for carpal tunnel syndrome was found to be up to seven times more frequent in patients receiving an AI than those receiving tamoxifen [11,64]. In initial studies exemestane was associated with dramatically increased risk of carpal tunnel syndrome compared to tamoxifen (2.8% versus 0.3%, respectively). In a 100-month followup of the ATAC trial, symptoms were typically reported within the fi rst few months of therapy, to be of mild to moderate intensity, and of short duration [65]. Th ere was increased reporting of carpal tunnel symptoms, although the incidence remained low (2.6%) and generally did not require surgical intervention. In a prospective evaluation of 92 postmenopausal patients with early stage breast cancer taking adjuvant AIs, 32% of patients reported new or worsening arthralgia most commonly aff ecting the knees (70%), wrists (70%), and small joints of the hands (63%) [66]. Most patients reported mild to moderate symptoms that were easily managed with analgesics and very few patients discontinued therapy due to emergence of symptoms. In contrast, in a small prospective study of 25 patients, 15 patients developed AI-induced arthralgia within the fi rst 12 months of treatment and 13 patients discontinued therapy as a result of the musculoskeletal symptoms [67].

Estrogen deprivation
Post-menopausal status and estrogen defi ciency are associated with the development of joint pain and joint symptoms and frequently improve with hormone supplemen tation [68]. Estrogen deprivation has been hypothesized as the major cause of AI-induced arthralgias. Indeed, development of arthralgia has also been seen in patients treated with the gonadotropin-releasing agonist leuprolide, which results in menopausal range estrogen concentrations [69]. Approximately 25% of women developed arthralgia within 3 weeks of initiation of leuprolide. Alternatively, estrogen-based therapy is associated with reduced incidence of radiologic knee osteoarthritis and decreased incidence of joint pain/swelling [70][71][72]. However, this eff ect has not been seen in all studies of estrogen therapy [73]. Whether the eff ect is secondary to systemic or localized estrogen defi ciency is unclear. ERs have been identifi ed in cartilage and estrogen defi ciency in ovariectomized rats accelerated cartilage turnover and increased cartilage surface erosion while administration of estrogen suppressed cartilage degradation signifi cantly [74][75][76][77][78]. Surgically ovariectomized primates similarly develop osteoarthritic changes that can be prevented by estrogen therapy [79]. Estrogen is associated with chondroprotective eff ects by decreasing collagen degradation [80,81]. In addition, aromatase is expressed in synovial cells and chrondrocytes of articular cartilage with evidence of local conversion of androstenedione to estrone and estradiol [82,83]. Th erefore, both systemic and local AI-induced estrogenic defi ciency may impair cartilage maintenance.

Anti-nociceptive eff ects
Estrogen has also been associated with anti-nociceptive eff ects and it has been postulated that estrogen defi ciency results in increased sensation of pain [84]. Th is hypothesis mainly stems from the observation that pain thresholds are aff ected by various hormonal states (increased pain thresholds during pregnancy) [85]. Th is estrogen-dependent eff ect is mediated through the spinal cord kappa-opioid analgesic system. Th e absence of estrogens thus would be expected to result in a reduction in analgesic eff ect [86]. ERs and aromatase are expressed in the central nervous system and local estrogen production may modulate pain and sensory perception [87]. In contrast, several studies have reported that pain thresholds are actually decreased when estrogen levels are high [88,89]. Given the radiologic fi ndings associated with AI-induced arthralgia (see below), this eff ect is not likely solely related to pain perception.

Tenosynovial changes and joint eff usions
Several studies have recently identifi ed characteristic radiologic changes associated with AI-induced arthralgia.
In a small study that evaluated 12 patients with severe AI-associated arthralgia, ultrasound evaluations revealed fl uid in the tendon sheath surrounding the digital fl exor tendons and MRI showed increased intra-articular fl uid as well as enhancement and thickening of the tendon sheath in all 12 patients [61]. In a larger prospective trial, patients who developed AI-related arthralgia were evaluated with musculoskeletal sonography and electromyo graphy [66]. Patients with AI-induced arthralgia had higher rates of joint eff usions and more electromyography fi ndings consistent with carpal tunnel syndrome. Interestingly, a retrospective analysis of women treated with adjuvant AI therapy showed that women who were on chronic diuretic treatment for heart disease or hypertension were less likely to have symptoms of arthralgia, muscular or skeletal stiff ness (6.97% versus 15.85%, P = 0.01), suggesting that fl uid retention within joints may play a role in AI-induced arthralgia [90].

Autoimmunity
Another possible etiology involved a potential link between AI therapy and autoimmunity. In one study, 24 women who developed disabling joint pain were referred for rheumatological consultation, radiological evaluation, and immunologic investigations [91]. Nineteen of the 24 patients were found to have infl ammatory pain of multiple joints. Nine of the 19 had elevated antinuclear antibodies, four had increased rheumatoid factor serum concentrations, and two had laboratory abnormalities consistent with a systemic infl ammatory syndrome. Ten patients had symptoms consistent with sicca syndrome, and one met diagnostic criteria for Sjogren's syndrome.
In support of a possible autoimmune mechanism, there appears to be an association between estrogen defi ciency and increased secretion of proinfl ammatory cytokines [92]. Estrogens have also been shown to have signifi cant anti-infl ammatory properties by repressing the trans cription of proinfl ammatory genes through the ER [93]. In a prospective randomized study designated Exemestane and Letrozole Pharmacogenetics (ELPh), patients who developed worsening joint symptoms were referred for rheumatologic evaluation [55]. Only a small fraction of the participants had elevated concentrations of infl ammatory or rheumatologic markers (5 to 18%). Th ey were most likely to be diagnosed with a moderate intensity, non-infl ammatory regional musculoskeletal disorder, including tendonitis/tenosynovitis (37%), osteoarthritis (29%), and carpal tunnel syndrome (21%). In a small cohort of the ELPh study, evaluation of concentrations of circulating infl ammatory markers in patients with AI-induced arthralgia showed no signifi cant change in the tested markers relative to pre-treatment concentrations or compared to women who did not report symptoms [94]. Although a small preliminary study, it supports other reports that AI-induced arthralgia is probably not associated with a systemic infl ammatory response.

Predictive factors mediating risk of developing AI-induced arthralgia
Several studies have evaluated the risk factors associated with AI-induced arthralgia. Overweight patients and those who had previously been treated with tamoxifen were at lower risk for AI-induced arthralgia, while patients who had previously been treated with taxanes were four times more likely to develop the symptoms [54]. In a separate study, interval since menopause was the only signifi cant risk factor (possibly linked to cytokine activity or to a more precipitous drop in estrogen levels), with women who had their last menstrual period within 5 years of starting therapy more likely to develop joint symptoms compared to those whose last menstrual period was 10 years prior to starting therapy (73% versus 35%, adjusted odds radio, 3.39; 95% confi dence interval, 1.21 to 9.44; P = 0.02) [62]. Th e majority of patients (75%) developed symptoms within 3 months of starting therapy.
In a prospective evaluation of musculoskeletal symptoms that develop in women treated with AI, the median time to onset of symptoms was 1.6 months and 13% of patients discontinued AI therapy after a median of 6.1 months secondary to musculoskeletal toxicity [55]. Type of surgery, radiation therapy, chemotherapy, or tamoxifen use did not predict the development of symptoms, although the report focused on the fi rst 100 participants only. A retrospective analysis of the ATAC trial identifi ed several risk factors for the development of arthralgia: previous hormone therapy, hormone receptor positivity, previous chemotherapy, obesity, and treatment with anastrozole. Only women without baseline joint symp toms at the outset of the trail were included in the analysis; thus, the study does not evaluate risk factors associated with worsening joint symptoms in patients with baseline arthralgia [58]. Th is study reports that women with joint symptoms at the outset reported fewer symptoms during treatment, which is in contrast to other studies.
In a retrospective analysis from the ATAC trial, treatment-induced vasomotor or joint symptoms were associated with improved effi cacy of the treatment, suggesting that adequate management of the symptoms is particularly important in maintaining medication compliance [95]. Women who experienced joint symptoms (with or without vasomotor symptoms) after 3 months of endocrine therapy (anastrozole or tamoxifen) had a signifi cantly reduced risk of developing recurrent disease than those without joint symptoms (adjusted hazard ratio 0.60 (0.50 to 0.72) P < 0.0001). While other preliminary investigation failed to show an association between AIrelated symptoms and outcomes [96], until prospective data are available, it is important to develop better sympto matic management of these symptoms and to improve adherence in women receiving endocrine treatment.

Management
No large study has focused on the optimal management of AI-induced arthralgia (Table 3). Th e majority of patients in the ATAC retrospective analysis had received some kind of treatment for their joint symptoms that consisted typically of non-steroidal anti-infl ammatory drugs and/or other analgesics [58]. Other reports have also described successful treatment of a subset of patients with analgesics, including non-steroidal anti-infl ammatory drugs, acetaminophen, and opioids [54,55,91,97]. Low dose corticosteroids were reported to be eff ective in one study, but the toxicity profi le and long-term side eff ects of corticosteroids make them an unappealing choice for treatment of AI-induced arthralgia [91].
Dietary supplementation with vitamins, glucosamine and chondroitin, omega fi sh oils, and Chinese herbal remedies have shown variable effi cacy [54,63]. A small study evaluated the eff ect of vitamin D and calcium supplementation started at the outset of AI therapy [98]. Although it was not a randomized placebo-controlled study, the authors showed that maintaining vitamin D levels >66 ng/ml resulted in lower rates of joint disability. Similarly, in a prospective cohort study [99], vitamin D levels ≥40 ng/ml were associated with lower risk for development of joint pain, although the authors found that despite supplementation many women on the study did not achieve adequate vitamin D levels. A larger prospective randomized placebo-controlled trial to evalu ate vitamin D supplementation in AI-induced arthralgia is ongoing (NCT00263185). Bisphosphonates were identifi ed in a retrospective study as an inverse risk factor for developing AI musculoskeletal symptoms [100]; however, this has not been evaluated in a prospective randomized placebo-controlled trial. Diuretics were recently reported to reduce arthralgia symptoms in a retrospective study consistent with the fi nding of joint eff usion and fl uid in the tendon sheaths [90]. Duloxetine, a serotonin and norepinephrine reuptake inhibitor, was shown to signifi cantly reduce AI-associated pain in a single-arm, open-label phase II study and improved functional status [101]. Other antidepressants and anticonvulsants are often used in the treatment of chronic pain disorders; however, their use has not been evaluated in AI-induced arthralgia. Acupuncture has been shown to be a feasible and eff ective treatment modality for AI-associated arthralgia [102]. In a randomized, single-blinded sham-controlled acupuncture trial, women treated with true acupuncture reported a two-point improvement in pain score compared to women treated with sham acupuncture (80% versus 22%) [103]. Both these studies suggest nonpharma ceutical approaches may be benefi cial for women with AI-induced arthralgia and enhance adherence. Other non-pharmaceutical approaches, such as exercise, yoga, and massage, may also be benefi cial but have not been evaluated.
Th e eff ect of switching aromatase inhibitors on musculoskeletal symptoms was recently evaluated in the Articular Tolerance of Letrozole (ATOLL) study, a 6-month, prospective, non-randomized, multicenter trial [104]. Patients who discontinued anastrozole because of musculoskeletal symptoms were started on letrozole and assessed for recurrence of symptoms, severity, and discontinuation of therapy. At the end of 6 months after switching from anastrozole to letrozole, 71.5% of patients continued therapy with letrozole while 28.5% discontinued therapy secondary to severe joint pain. Although the joint symptoms were more tolerable and did not result in as many discontinuations, the majority of patients continued to have joint symptoms despite switching therapy. However, this study suggests that patients who are intolerant to one AI may benefi t from switching to another AI to continue to receive the benefi ts of the hormonal adjuvant therapy. Whether switching classes of AI (from steroidal to non-steroidal or vice versa) will be associated with improvement of symptoms has not been reported. Other reasonable alternatives include a drug holiday and/or switching to tamoxifen if clinically appropriate.

Conclusion
AIs are widely used in the treatment of early-stage breast cancer. While results from the defi nitive phase III randomized clinical trials comparing AI use to tamoxifen initially suggested that AI may result in a reduced toxicity profi le compared to tamoxifen, patient-reported outcomes in prospective studies demonstrate that the musculo skeletal side eff ects of these agents are substantial, increasing treatment-related morbidity and result ing in treatment discontinuation. Given the significantly increased risk of osteoporosis and fractures associated with AIs, a thorough assessment of risk factors prior to the start of therapy is indicated with consideration to initiating preventative measures (calcium, vitamin D, bisphosphonates as indicated) at the outset of treatment.
In our practice, we encourage all women on AIs to participate in weight bearing exercise and take calcium and vitamin D supplements, and generally follow the United States Preventative Task Force guidelines for initiation of bisphosphonate therapy if osteoporosis is present on DEXA scan [105]. No guidelines are available for the treatment of AI-associated arthralgia. We approach treating these patients on a case-by-case basis, reserving discontinuation of therapy or switching to tamoxifen for refractory cases [2]. In our experience, there is a wide variability in response to non-steroidal anti-infl ammatory drugs, but newer approaches, including non-pharmacologic treatments, hold promise for improving the tolerability of AIs.
Musculoskeletal symptoms in women treated with AIs represent a signifi cant burden whose etiology is still unexplained. Th ere is a need to identify the mechanisms underlying the development of toxicity with a focus on determining predictive factors and prospective assessment of interventional approaches. Eff ective management and symptomatic treatment of these symptoms is imperative to enhance adherence to therapy, improve outcomes, and decrease breast cancer recurrences.