Effective new treatments for breast cancer, aromatase (estrogen synthetase) inhibitors, powerfully block the conversion of androgen precursors to estrogens, which lowers plasma estradiol levels to close to or below the detection limit of current assays (1–3) and lowers estrogen levels in peripheral tissues (4, 5). Because aromatase inhibitors reduce recurrence rates in women treated for early-stage postmenopausal breast cancer (6), these agents are becoming widely used in breast cancer treatment and are being explored as a treatment to prevent disease in women at high risk. Thus, their use is likely to increase. As noted in the trial data summarized below, women treated with aromatase inhibitors often experience joint pain and musculoskeletal aching, which, in some cases, has necessitated discontinuation of treatment.
We review herein the evidence linking aromatase inhibitors and, more broadly, estrogen deprivation, with arthralgias, alerting physicians and other health professionals to this potential problem. We review the biologic mechanisms by which estrogen deprivation can be tied to joint pain and then examine the evidence associating estrogen deprivation with joint pain in 2 circumstances: pharmacologic deprivation of estrogen and natural menopause.
Although we use the term “arthralgias,” the lack of uniform assessment of musculoskeletal symptoms in studies forces us to label these as such even though the symptoms experienced by women with this syndrome may be broader than arthralgias alone.
Mechanisms of joint pain and potential effects of estrogen
Pain can emanate from a variety of articular structures innervated with nociceptive fibers, including the joint capsule, synovium, periosteal bone, ligaments, and even periarticular structures. In addition, when the joint is inflamed, inflammatory mediators, including prostaglandins and bradykinin, activate receptors on peripheral nociceptors, enhancing their sensitivity to pain such that mechanical stimuli that would not otherwise be painful become so (7). Also, with inflammation in the joint, a nociceptive neuron's receptor fields enlarge, so that pain can be induced by stimuli away from the joint that would not otherwise cause pain.
This expansion of receptor fields, which occurs in the spinal cord (8), provides evidence that neural processing is a critical contributor to joint pain. Most spinal cord neurons that communicate sensory information from articular structures are tonically inhibited by descending neural fibers, and disruption of this inhibition could produce increasing transmission of nociceptive inputs from the spinal cord upward through the spinothalamic tract, where it is ultimately interpreted by the brain as pain.
While estrogen has no specific known effects on articular structures that would cause joint pain, it influences inflammation and neural processing of nociceptive input. Estrogen's effects on inflammation within the joint are not well known. It has tissue-specific effects on inflammatory cytokines (9–12). For example, while estrogen enhances nitric oxide production in endothelial cells, it inhibits nitric oxide and prostaglandin E2 (13) production in microglia. The effects of estrogen on inflammation in the joint could account for enhanced nociception that occurs with estrogen depletion.
Perhaps most relevant to the problem of joint and muscle pain in women with depleted estrogen levels are the antinociceptive effects of estrogen. First, estrogen has direct effects on opioid pain fibers in the central nervous system. Cells containing estrogen receptors express prepro-enkephalin messenger RNA (14), and estrogen given to ovariectomized animals increases enkephalin transcription in the spinal cord. These neurons provide critical inhibition of nociceptive neurons. Further, estrogen receptors are present in opioid-containing neurons (15) in the spinal cord and brain (16), suggesting an even more direct effect of estrogen on opioid generation locally.
In some species, aromatase has been found in dorsal horn cells (17). Androgens there, which are converted by aromatase to estrogens, could provide a source of estrogen in the spinal cord. Two prominent therapies that produce the arthralgias associated with estrogen depletion, aromatase inhibitors and leuprolide, deplete estrogen and testosterone. However, the specific effects of aromatase-containing cells on nociceptive fibers are less well understood than the direct effects of estrogen.
Given the possible effects on inflammation and on nociceptive input, what is the evidence that estrogen affects pain in animals and humans? First, during pregnancy, when estrogen levels are high, women have elevated thresholds for painful stimuli (6). Furthermore, in nonpregnant animals in which pregnancy levels of estradiol and progesterone are simulated, pain thresholds rise. These data support the hypothesis that high levels of gonadal hormones suppress pain sensitivity.
In addition to gestational effects on pain sensitivity, sensitivity to nociception fluctuates in women during the menstrual period but not necessarily according to estrogen levels. Further, sensitivity may be different for different types of nociceptors (18). Ultimately, the data show no clear link in humans between estrogen levels during the menstrual cycle and pain sensitivity (19).
Evidence linking pharmacologically induced low estrogen levels with arthralgias
The strongest evidence linking decreased estrogen production to arthralgias originates from trials of aromatase inhibitors for the treatment or prevention of breast cancer. Compared with women assigned to treatment with placebo or tamoxifen, those treated with aromatase inhibitors have consistently shown higher rates of arthralgia. In trials of women with advanced breast cancer, arthralgias have been more common in women treated with aromatase inhibitors than in the comparator group despite their lower rates of metastases, which might also cause bone pain (20–24).
In secondary prevention trials and trials of women with primary breast cancer, aromatase inhibitor treatment has been linked to higher rates of musculoskeletal complaints than control treatment. For example, in the ATAC (Arimidex, Tamoxifen Alone or in Combination) trial, 27.8% of women treated with anastrozole experienced musculoskeletal disorders, as compared with 21.3% of women treated with tamoxifen (P < 0.0001) (25). (In a recently released report of 5 years of treatment , 35.6% of women receiving anastrozole had arthralgia versus 29.4% of women receiving tamoxifen [P < 0.0001]). In the National Cancer Institute of Canada Clinical Trials Group MA17 (20) trial, 5,187 postmenopausal women who completed a standard course of 5 years of adjuvant tamoxifen therapy were randomized to a further 5 years receiving the aromatase inhibitor letrozole or placebo. A systematic evaluation of joint and muscle pain in women in this trial showed that 21.3% of women who received letrozole reported joint pain versus 16.6% of women who received placebo (P < 0.0001), with muscle aching in 15% versus 12% (P = 0.04), respectively. Ratings of body pain and vitality were worse in women randomized to letrozole than to placebo; this was thought to reflect the musculoskeletal pain noted above. In the Intergroup Exemestane Study of women after 2–3 years of tamoxifen therapy (24), 33% of those who received the aromatase inhibitor experienced generalized aches and pain, as compared with 29.7% who received placebo, with arthralgia also being more common (5.4% versus 3.6%; P = 0.01).
While the arthralgias reported in these trials have generally not been disabling, Donnellan and colleagues (27) reported that 12 of 77 breast cancer patients treated with anastrozole had arthralgias, and 4 of these (5% of the total treated) had to discontinue their treatment because the arthralgias were severe or persistent and unresponsive to treatment. None of these patients had previously had arthritis or joint pain; the areas most commonly affected were the hands, knees, hips, lower back, and shoulders, and early morning stiffness was a common feature. Discontinuation of the aromatase inhibitor led to prompt relief of symptoms.
Leuprolide, a gonadotropin-releasing hormone agonist, is used to treat infertility and a variety of gynecologic disorders in premenopausal women. The majority of adverse effects experienced by treated patients have been attributed to the hypoestrogenemia that leuprolide ultimately produces; with treatment, reproductive hormone production is initially stimulated and then falls substantially. Friedman and colleagues (28) studied 102 premenopausal women treated with leuprolide for leiomyomas, tracking symptoms and corresponding estrogen levels. After 2 weeks of treatment, estradiol levels were in the menopausal range in all women, averaging 23.8 pg/ml (Figure 1). Symptoms of estrogen deprivation, such as vaginal dryness, began at week 2 after beginning leuprolide. Arthralgias and myalgias began between weeks 3 and 7 of treatment. Overall, 25% of the women developed new-onset arthralgias during this period. Joints affected included the knees, ankles, elbows, metacarpals, metatarsals, and costochondrals. Arthralgias resolved in all women 2–12 weeks after cessation of leuprolide therapy. There have been no reports of arthralgias or musculoskeletal symptoms in men treated with leuprolide, although it is not clear that any study investigators have inquired into the occurrence of these symptoms.
Although arthralgias, diffuse pain, and myalgias have all been described, the particular range of syndromes induced by aromatase inhibitors and leuprolide is not clear. It is possible that the syndromes described as “menopausal arthritis” and those produced by estrogen deprivation are different and that some are associated with pain outside the joints, including generalized pain syndromes and prominent myalgias. More study is needed to better define these syndromes. The few data that do exist suggest a process of diffuse transient joint pain that, in the case of estrogen deprivation, is responsive to estrogen repletion. Another explanation for these findings would be that a rapid drop in estrogen levels enhances nociception.
Natural menopause, estrogen replacement therapy, and arthralgia
Whether in the knees, hands, feet, or elsewhere, the prevalence of joint pain and of widespread pain increases progressively with age in women, reaching a maximum in the 50–59-year age group (29, 30). This suggests that hormonal or other changes occurring during menopause may contribute to enhanced pain sensitivity or production of painful stimuli.
If estrogen deprivation sometimes produces arthralgia, then the menopausal transition should also occasionally cause arthralgia, relieved by estrogen therapy. Arthralgias following menopause have been reported in several case series. In 1925, Cecil and Archer (31) described a painful arthritis developing in women 2–5 years after menopause, which they labeled menopausal arthritis. Arthralgias are the predominant somatic symptom of menopause in parts of Asia. For example, among 1,900 women ages 45–55 in Hong Kong, backache or generalized aches and joint stiffness were the most common symptoms reported (more prevalent than hot flashes), and the aches and joint stiffness were more prevalent in perimenopausal and postmenopausal women than in premenopausal women of the same age (32). Although arthralgias have not been reported prominently among menopausal symptoms in non-Asian women, their occurrence has not been well studied.
Recent data show that Asian women in the United States undergoing menopause have lower estradiol levels than Caucasian women (33, 34). Adipose tissue is a major source of estradiol in postmenopausal women, and the higher estradiol levels in Caucasians may be due, in part, to the greater adiposity of Caucasian women (32, 33).
In the Postmenopausal Estrogen/Progestin Intervention trial, women who were randomized to receive conjugated estrogen had a 32–38% decreased chance of developing musculoskeletal symptoms (P = 0.02–0.08) compared with those taking placebo. Symptoms reported by women in this trial included aches and pain, joint pain, muscle stiffness, and skull and neck aching (35). In the estrogen and progesterone arm of the Women's Health Initiative, women assigned to hormone replacement therapy had a significantly decreased occurrence of “bodily pain,” as assessed by the Short Form 36 health survey. In the Heart and Estrogen/Progestin Replacement Study (HERS), an ancillary study evaluating knee pain showed no difference in its prevalence or severity between those randomized to receive estrogen replacement therapy and those taking placebo (36). The mean age of the subjects entering the HERS trial was 66 years, with most women being 10 years beyond menopause.
While estrogen deprivation may produce arthralgias, this does not necessarily mean that estrogen replacement effectively treats musculoskeletal pain. Estrogen replacement therapy delivers a supraphysiologic dose of estrogen, with effects that may extend beyond correcting deficiency. For example, postmenopausal women receiving estrogen therapy may actually have more back pain than those not receiving estrogen therapy, even excluding those with known osteoporotic fractures (37). However, in one observational study, generalized pain was less common in women receiving estrogen therapy than in other postmenopausal women not receiving treatment (38).
Implications of “arthralgia of estrogen deficiency” for common forms of arthritis
The most common forms of arthritis have their highest incidence and prevalence in postmenopausal women. While arthralgias represent only symptoms without objective joint findings, if present in those with objective findings, the arthralgias could encourage a diagnosis of arthritis in those who might not otherwise be diagnosed. The contribution of estrogen deficiency to this high rate of arthritis is unknown. The prevalence of osteoarthritis (OA), one form of which is the menopausal arthritis described earlier, increases around the time of menopause (39). Inconsistent findings have been revealed in studies evaluating the association of estrogen replacement therapy with the prevalence of OA, but many confounding biases, such as the indication for estrogen therapy, make these findings hard to interpret (40). A trial in monkeys (41) suggested that estrogen deprivation produced early structural OA, which was prevented by estrogen treatment. While one study found no association of estrogen levels with hand OA in women (42), this issue has not been studied using sensitive assays of postmenopausal estrogen levels.
Rheumatoid arthritis (RA), a disease predominant in women, has its highest incidence and prevalence in the decade after menopause (43). One explanation for the predilection of these women to develop RA is that enhanced pain sensitivity triggered by estrogen deprivation worsens or precipitates joint pain in women with arthritis and that, in turn, prompts a diagnosis of arthritis. In postmenopausal women, there is no consistent relationship between estrogen therapy and the occurrence of RA. Further, trials have failed to show an effect of estrogen treatment on RA disease activity (44). While it is speculative to link estrogen deprivation in postmenopausal women with the high rates of OA and RA in this population, studies evaluating these associations are warranted.
In summary, the “menopausal arthritis” described in several historical case series and the Asian reports of menopausal arthralgias may be related in their pathogenesis to the symptoms induced by pharmacologic agents, especially aromatase inhibitors and other agents that deplete estrogen. A small but significant proportion of women receiving estrogen-depleting treatments develop this musculoskeletal syndrome, a syndrome that has not been widely appreciated. It is important to recognize that the symptoms are usually transient and resolve with estrogen therapy or when these agents are discontinued.