SEARCH

SEARCH BY CITATION

Keywords:

  • breast neoplasm;
  • aromatase inhibitor;
  • adverse effects;
  • arthralgia;
  • arthritis postmenopausal;
  • epidemiology

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. Conflict of Interest Disclosures
  8. References

BACKGROUND:

Arthralgia is common in postmenopausal breast cancer survivors (BCS) who are receiving aromatase inhibitors (AIs). The objective of this study was to evaluate the perceived onset, characteristics, and risk factors for AI-related arthralgia (AIA).

METHODS:

In a cross-sectional survey of postmenopausal BCS who were receiving adjuvant AI therapy at a university-based oncology clinic, patient-reported attribution of AIs as a cause of joint pain was used as the primary outcome. Multivariate logistic regression analyses were performed to evaluate risk factors.

RESULTS:

Among 300 survey respondents, 139 (47%) attributed AI as a cause of their current arthralgia. Of those patients, 74% recognized the onset of AIA within 3 months of starting medication, and 67% rated joint pain as moderate or severe in the previous 7 days. In multivariate logistic regression analyses, the time since last menstrual period (LMP) was the only significant predictor of AIA. Controlling for covariates, the women who had their LMP within 5 years had the highest probability of reporting AIA (73%), whereas those who had their LMP ≥10 years previously had the lowest probability of reporting AIA (35%; adjusted odds radio, 3.39; 95% confidence interval, 1.21-9.44; P = .02). Wrists/hands, ankles/feet, elbows, and knees appeared to be associated more strongly with AI-related symptoms than non-AI–related joint symptoms (all P < .01).

CONCLUSIONS:

AIA was common, began within the first 3 months of therapy in most patients, and appeared to be related inversely to the length of time since cessation of menstrual function. These findings suggest that estrogen withdrawal may play a role in the mechanism of this disorder. Cancer 2009. © 2009 American Cancer Society.

Aromatase inhibitors (AIs) have become an indispensable part of standard adjuvant hormone therapy for hundreds of thousands of postmenopausal women with hormone receptor-positive invasive breast cancer. Large, adjuvant, randomized controlled trials (RCTs) have produced improvements in disease-free survival rates as high as 40%, greater than the rates observed with tamoxifen.1-4 The utility of AIs also is being investigated in combination with ovarian suppression in premenopausal women with breast cancer and among healthy postmenopausal women for the prevention of breast cancer.5

With the increase in AI use and potential indications, AI-related arthralgia is emerging as a major source of symptom burden among its users,6, 7 with a 28% relative increase compared with placebo (21.3% for AI vs 16.6%; P < .001).2 In trials comparing AIs with tamoxifen, the incidence of arthralgia was substantially greater in AI groups (range, 5%-36%) than in tamoxifen groups (range, 4%-29%); although the rates varied.8, 9 Arthralgia affected daily function and appeared to decrease adherence, leading to premature discontinuation of AIs.10 One Canadian chart review of 50 patients indicated that 22% of patients discontinued adjuvant AI therapy because of toxicity, including muscle and joint symptoms.11 A study using a health claims dataset indicated that 1 in 5 AI users filled <80% of their prescribed AI medications.12

Because of the significant impact of AI-related arthralgia on quality of life, adherence behavior, and potential survival benefit derived from AIs, research is needed to better define the characteristics of AI-related arthralgia to guide future interventions. Thus, the specific objectives of the current study were 1) to define the rate of AI as a cause for arthralgia in early stage breast cancer survivors (BCS) who currently received AIs, 2) to describe the perceived onset of AI-related arthralgia in relation to initiating AI therapy, 3) to identify the demographic and clinical risk factors associated with AI-related arthralgia, and 4) to explore the joint-specific presentation of AI-related arthralgia.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. Conflict of Interest Disclosures
  8. References

Study Design and Patient Population

We conducted a cross-sectional survey of patients with breast cancer who were receiving care at the Rowan Breast Cancer Center of the Abramson Cancer Center of the University of Pennsylvania (Philadelphia, Pa) between April and October 2007. Potential participants included all postmenopausal women who had a history of histologically confirmed stage I through III, hormone receptor-positive breast cancer and who currently were taking a third-generation AI (anastrozole, letrozole, or exemestane), had completed chemotherapy or radiotherapy at least 1 month before enrollment, and had the ability to understand and provide informed consent in English. Research assistants obtained permission from the treating oncologist, screened medical records, and approached potential study participants for enrollment at their regular follow-up appointments. After informed consent was obtained, each participant was given a self-administered survey. The study was approved by the Institutional Review Board of the University of Pennsylvania and the Scientific Review and Monitoring Committee of the Abramson Cancer Center.

Outcome Measurement

Primary outcome measures included patient self-report of joint pain, particularly self-reported joint pain attributed to AIs; severity of joint pain; and clinical characteristics (ie, onset, location) of joint pain. Because AI-related arthralgia is a relatively new clinical phenomenon, we developed a questionnaire based on limited literature, clinical expert opinion, and patient input. Then, we piloted the questionnaire with 16 BCS who were receiving AIs to determine the clarity of items and patient relevance and to identify additional content. Because arthralgia in this population may be multifactorial,13 we instructed participants to attribute causes for their arthralgia by asking, “What do you believe are the sources of your current joint symptoms?” The response options included “prior osteoarthritis,” “aromatase inhibitors,” “other medical conditions,” “other medications,” and “others.” Respondents were able to choose more than 1 option. To generate the main outcome variable of AI-related arthralgia, any participant who selected “aromatase inhibitors” as a cause of their current joint symptoms were considered to have AI-related arthralgia, and those who did not select that option did not have AI-related arthralgia. This dichotomous variable was used as the main outcome in both bivariate and multivariate analyses.

To assess symptom severity, participants were asked to rate their joint pain over the preceding 7 days on a 5-point Likert scale ranging from “none” to “very severe.” To evaluate perceived onset, we asked the participants, “If you are experiencing joint symptoms which you think are related to aromatase inhibitors, when did you first recognize the symptoms after starting to take the medications?” Response options ranged from “right away” and “within the first week” to “after 6 months.” To evaluate which joints were affected most commonly in those with AI-related arthralgia, we asked participants 2 questions. The participant was asked first to indicate all joints with pain in the preceding 24 hours and then to indicate in which single joint the participant experienced the most pain in the preceding 24 hours.

Information on covariates was collected, including age, race, ethnicity, education level, and employment status. Clinical and treatment characteristics were assessed by either self-report (ie, timing of the last menstrual period [LMP], height, weight before breast cancer and current weight, and comorbidities, including previous arthritis) or medical record abstraction (ie, stage, chemotherapy, tamoxifen use, AI use).

Statistical Analysis

Data analysis was performed using STATA 9.0 for Windows (STATA Corporation, College Station, Tex). We initially performed descriptive statistics and bivariate analyses. Then, we developed multivariate logistic regression to determine the relative impact of each variable on AI-related arthralgia. Variables that were not significant at the .20 level in the bivariate analyses were not included. To explore how AI-related arthralgia may impact specific joints differently, we then performed chi-square analyses to compare individuals with AI-related arthralgia and those without. Statistical tests were 2-sided, and P values <.05 indicated significance except in the exploratory analyses for joint-specific arthralgias. Because multiple comparisons were used in the latter analyses, we lowered the statistical significance to <.01 to indicate significance.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. Conflict of Interest Disclosures
  8. References

Participant Characteristics

Of 484 consecutive patients who were screened, 50 patients (10%) were ineligible because they discontinued AI therapy before screening, 45 patients (9%) had metastatic disease, 64 patients (13%) did not keep their scheduled appointment, and 25 patients (5%) declined enrollment, leaving a total of 300 participants. Characteristics of the study population are listed in Table 1. Among the 300 survey participants, the mean age was 61 years; and, although the majority of patients (84%) was non-Hispanic white, a substantial proportion (13%) was non-Hispanic black. In the analysis, we combined the race categories into white and nonwhite. Among the participants, 51 patients (17.6%) reported having had their LMP within the 5 years before study enrollment, 95 patients (32.8%) had their LMP between 5 years and 10 years before enrollment, and 144 patients (49.7%) had their LMP >10 years before enrollment. One hundred seventy-three patients (57.7%) were taking anastrozole, 69 patients (23%) were taking letrozole, and 58 patients (19%) were taking exemestane (Table 2).

Table 1. Demographic Characteristics of Participants (N=300)
CharacteristicNo. of ParticipantsNo. With AI-Related Pain%*P
  • AI indicates aromatase inhibitor.

  • *

    Values represent the percentage of patients reporting AI-related joint pain in a specific demographic category.

  • P values were based on chi square tests.

  • Nonwhite, mostly black.

Total300139  
Age, y   .001
 <55734562 
 55-651316247 
 >65963233 
Race/ethnicity   .13
 White25312247 
 Nonwhite471736 
Educational level   .4
 High school or less1225243 
 College763445 
 Graduate or  professional school1015252 
Employment   .002
 Full time1146759 
 Part time401948 
 Not currently  employed1425237 
Table 2. Clinical and Treatment Characteristics of Participants (N=300)
CharacteristicNo. of ParticipantsNo. With AI-Related Pain%*P
  • AI indicates aromatase inhibitor; LMP, last menstrual period.

  • *

    Values represent the percentage of patients reporting AI-related joint pain in a specific clinical category.

  • P values were based on chi square tests.

Total300139  
Time since LMP, y   <.001
 <5513773 
 5-10954547 
 >101445236 
Body mass index, kg/m2   .76
 <251125045 
 25-30934245 
 >30954750 
Stage   .45
 I1004242 
 II1427150 
 III321444 
Chemotherapy   .07
 None1114339 
 Chemotherapy but no taxane1014949 
 Chemotherapy included taxane784355 
Previous tamoxifen   .63
 None1517248 
 Yes1366145 
AI   .76
 Anastrozole (Arimidex)1737745 
 Letrozole (Femara)693449 
 Exemestane (Aromasin)582848 
Duration of AI therapy, y   .78
 <11266048 
 1-3793443 
 >3864148 
No. of comorbid conditions   .1
 None874754 
 11034948 
 ≥21104339 
Previous arthritis   .36
 None1979548 
 Yes1034443 
Weight gain since breast cancer, lbs   .1
 No weight gain1335239 
 ≤10834149 
 ≥10814353 
Table 3. Multivariate Logistic Regression Model
Patient CharacteristicBivariate AnalysisMultivariate Analysis
OR (95% CI)PAOR (95% CI)P
  • OR indicates odds ratio; CI, confidence interval; AOR, adjusted odds ratio; LMP, last menstrual period.

  • *

    Nonwhite, mostly black.

Age, y    
 <55 (Reference group)1 1 
 55-650.56 (0.31-1.00).050.99 (0.44-2.21).97
 >650.31 (0.16-0.59)<.001.79 (0.27-2.32).67
Race/ethnicity    
 White (reference group)1 1 
 Nonwhite*0.61 (0.32-1.16).130.59 (0.27-1.27).18
Employment    
 Full time (reference group)1 1 
 Part time0.63 (0.31-1.31).220.63 (0.28-1.41).26
 Not currently0.41 (0.24-0.67)<.0010.62 (0.33-1.13).13
Time since LMP, y    
 >10 (Reference group)1 1 
 5-101.59 (0.94-2.70).081.10 (0.55-2.21).79
 <54.68 (2.32-9.44)<.0013.39 (1.21-9.44).02
Chemotherapy    
 None (reference group)1 1 
 Chemotherapy but no taxane1.49 (0.86-2.57).151.04 (0.55-1.97).89
 Chemotherapy included taxane1.94 (1.08-3.50).0271.06 (0.52-2.16).88
No. of comorbid conditions    
 None (reference group)1 1 
 10.77 (0.44-1.37).380.96 (0.50-1.86).91
 ≥20.55 (0.37-0.97).0370.92 (0.47-1.81).8
Weight gain since breast cancer, lbs    
 None (reference group)1 1 
 ≤101.52 (0.87-2.65).141.12 (0.61-2.07).71
 ≥101.76 (1.01-3.08).0471.25 (0.65-2.38).5
Perceived rate and severity of aromatase inhibitor-related arthralgia

Among the 300 participants, 139 (46.3%) believed that AI was a source of their current joint symptoms, whereas 91 (30.3%) attributed joint symptoms to previous osteoarthritis, 97 (32.3%) attributed joint symptoms to other medical conditions (eg, fibromyalgia, rheumatoid arthritis, spinal stenosis), 13 (4.3%) attributed joint symptoms to other medications (eg, statin, paclitaxel), and 63 (21%) attributed joint symptoms to other causes (eg, mainly “aging,” injury). Compared with participants who did not have AI-related arthralgia, those who reported AI-related arthralgia had significantly more severe pain in the preceding 7 days (P < .001). For example, 26% of those with AI-related arthralgia had pain rated severe or very severe, whereas only 11% among those without AI-related arthralgia had pain on that level. Among those who reported AI-related arthralgia, 67% rated pain of moderate or greater severity.

Perceived onset of aromatase inhibitor-related arthralgia

Among the 139 participants who had AI-related arthralgia, 10 (7%) noticed the onset of arthralgia because of AI right away after starting the medication, 9 (7%) noticed the onset within the first week, 46 (34%) noticed the onset within the remaining first month, 35 (26%) noticed the onset within the first 3 months, 14 (10%) noticed the onset within the first 6 months, 21 (15%) noticed the onset after the first 6 months, and 4 (3%) did not provide a response to this question (see Fig. 1). Thus, about 75% of AI-related arthralgia has been noticed within the first 3 months after the onset of the therapy.

thumbnail image

Figure 1. This chart illustrates the time to onset of aromatase inhibitor-related arthralgia (AI-Arthralgia) in patients with breast cancer.

Download figure to PowerPoint

Factors associated with aromatase inhibitor-related arthralgia

In bivariate analyses, younger age, full-time employment status, and fewer years since LMP were associated with greater AI-related arthralgia (see Tables 1, 2). Nevertheless, in the multivariate regression model (adjusting for variables selected from the bivariate analyses), time since LMP was the only factor associated with reporting AI-related arthralgia (Table 3). Women who had their LMP within the last 5 years were significantly more likely to report AI-related arthralgia than women who had their LMP >10 years previously (adjusted OR, 3.39; 95% CI, 1.21-9.44; P = .02). Adjusting for covariates, the probabilities of reporting AI-related arthralgia were 0.73 (95% CI, 0.59-0.84) for LMP within 5 years, 0.48 (95% CI, 0.37-0.58) for LMP between 5 years and 10 years, and 0.35 (95% CI, 0.28-0.44) for LMP >10 years.

Joint-Specific Aromatase Inhibitor-Related Arthralgia

Because arthralgia in different joints may have different underlying pathophysiologic mechanisms and differential impact on function and quality of life, we examined manifestations of AI-related arthralgia in specific joints. The most common sites of joint pain in individuals who had AI-related arthralgia were wrist/hand (60.4%), knee (59.7%), back (54%), ankle/foot (51.8%), and hip (42.5%), in descending order (see Fig. 2A). The average number of joints affected in patients with joint pain was greater in individuals who had AI-related arthralgia than who did not (3.5 affected joints vs 2.1 affected joints; P < .001). The worst arthralgia was most commonly in the knee (22.3%) and the wrist/hand (21%), followed by the back (16.2%) the ankle/foot (12.3%), and the hip (11.5%) (see Fig. 2B). Compared with the patients who did not have AI-related arthralgia, the patients who reported AI-related arthralgia were significantly more likely to report arthralgia in the wrist/hand (relative risk [RR], 1.97; 95% CI, 1.53-2.54), the elbow (RR, 1.79; 95% CI, 1.42-2.26), the ankle/foot (RR, 1.66; 95% CI, 1.31-2.10), and the knee (RR, 1.54; 95% CI, 1.20-1.99) (see Table 4). It is noteworthy that joints distal to the torso may have greater pain related to AIs (suggested by the increasing RR).

thumbnail image

Figure 2. Common sites of (A) any joint pain and (B) the worst pain are illustrated in patients with breast cancer who had aromatase inhibitor-related arthralgia.

Download figure to PowerPoint

Table 4. Joint Specific Aromatase Inhibitor-Related Arthralgia
Joint Location With PainRR (95% CI)*P
  • RR indicates relative risk; CI, confidence interval.

  • *

    The RRs are based on chi-square comparisons between breast cancer survivors with and without aromatase inhibitor-related arthralgia.

Upper extremity  
 Shoulder1.17 (0.90-1.52).25
 Elbow1.79 (1.42-2.26)<.001
 Wrist/hand1.97 (1.53-2.54)<.001
Core body  
 Neck1.23 (.96-1.57).11
 Back1.49 (1.17-1.90.001
Lower extremity  
 Hip1.45 (1.15-1.84).003
 Knee1.54 (1.20-1.99)<.001
 Ankle/feet1.66 (1.31-2.10)<.001

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. Conflict of Interest Disclosures
  8. References

In this article, we present a comprehensive evaluation of AI-related arthralgia based on patients' perceptions. Despite the multifactorial nature of arthralgia in a postmenopausal BCS population, close to 50% of our sample reported experiencing AI-related arthralgia. Most patients recognized arthralgia within the first 3 months from the initiation of AIs, and >60% reported current arthralgia that was moderate or severe. Controlling for clinical and demographic factors, the time from LMP was related inversely to the report of AI-related arthralgia, and those who experienced menopause in the 5 years before enrollment had a 3-fold greater likelihood of developing AI-related arthralgia compared with those who were ≥10 years postmenopause. Furthermore, despite the finding that multiple joints can be involved, wrists/hands, elbows, ankles/feet, and knees may be particularly vulnerable to AI-related arthralgia.

Recent data suggest that the prevalence of AI-related arthralgia is higher outside of clinical trials.10, 14 In a survey among 200 BCS who were receiving AIs, 94 patients (47%) reported having AI-related arthralgia; and, of those patients, 50% had a new onset of joint pain after the initiation of AI, and 50% reported that their pain worsened since starting AIs.15 Our study, like others that have examined this issue in a practice setting outside a clinical trial, is subject to selection bias. Ten percent of individuals who stopped AIs were not captured by this survey. Although our 5% refusal rate was low, it is possible that those patients who refused to participate were relatively asymptomatic and may have viewed the survey as less relevant. In addition, because patients with prior arthritic conditions may opt not to initiate AIs, our study may underestimate the degree of AI-related arthralgia in patients with pre-existing arthritis. Finally, because our study relied on self-report, some degree of misclassification bias because of patients' perception exists; however, for subjective symptoms like AI-related arthralgia, patient-reported outcome is considered the gold standard. Despite such limitations, our estimated prevalence of ongoing AI-related arthralgia was almost identical to that of Crew et al, suggesting that the prevalence is credible.15 Furthermore, the site(s) at which patients reported joint pain also were very similar to the site(s) reported by Crew et al.15

Our findings with regard to the temporal onset of symptoms confirm a similar observation in a small case series,10 and these data justify and inform prospective studies. Measuring the perceived onset of AI-related arthralgia may be subject to recall bias, because the actual event occurred months if not years before the study. Because most patients reported the onset of AI-related arthralgia within 3 months from the initiation of AIs, more intense follow-up and data collection during this window of opportunity may help evaluate the true timing of the development of AI-related arthralgia as well as elucidate the potential biologic mechanism. These data also may suggest that clinicians should have early follow-up with patients who initiate AIs to discuss the potential onset of arthralgia, so that timely detection and management can occur to prevent premature discontinuation,16 which has been documented as high as 20% in some studies.14, 17

Risk factors for AI-arthralgia are not well characterized. In the only other study to report on this issue in clinical practice, prior tamoxifen use and higher weight appeared to be associated with a lower risk of AI-arthralgia, whereas receiving taxane chemotherapy was associated with increased risk.15 Although these were not significant risk factors in our population, we did observed a trend for the use of taxanes (P = .07). The differences in risk factor profiles between the 2 studies may be the result of sample size or of subtle underlying differences in the study populations. For example, our study population was younger, because most patients were aged <55 years. These findings emphasize the need to replicate such studies in diverse populations to determine whether such associations are robust, reproducible, or population-specific.

In our study, the interval after menopause (measured as the time from LMP) was the only significant factor that was associated with AI-related arthralgia. We hypothesize that those individuals who most recently have transitioned into menopause may have higher residual circulating estrogen; thus, the exposure to AIs may cause a more precipitous absolute drop in estrogen, leading to greater symptom experience. Several recent publications have cited estrogen deprivation caused by AIs as a potential mechanism for this clinical phenomenon.6, 8, 9, 18 The effect of estrogen withdrawal on the clinical syndrome of AI-arthralgia may be multifactorial,19 acting both centrally and peripherally.8 Centrally, acutely reducing estrogen levels may decrease endogenous opioid generation, thereby decreasing pain threshold. Thus, some individuals with subclinical osteoarthritis may experience a rise in AI-associated symptoms because of a decreased pain threshold and an increased awareness of arthralgia, leading some patients to attribute the “cause” of arthralgia to AIs.20 Peripherally, estrogen withdrawal may up-regulate inflammatory cytokines like interleukin-6 and tumor necrosis factor-α, which may accelerate bone loss and bone aging, thereby leading to pain.21 Prospectively measuring appropriate biomarkers (eg, reproductive hormones, drug metabolites, and inflammatory cytokines) in addition to subjective symptom measurement may help elucidate the biology underlying this unexplained clinical phenomenon.

Increasing evidence suggests that those women who may benefit from AI the most also may be the ones who experience the greatest degree of adverse events. Freedman et al observed that younger age was a predictor of benefiting from AIs in a cohort study in women who finished radiation therapy and tamoxifen.22 Cuzick et al recently demonstrated that those with treatment-emergent joint and vasomotor symptoms during the first 3 months of AI therapy were less likely to develop breast cancer recurrence in a retrospective analysis of the Arimidex, Tamoxifen, Alone or in Combination trial.23 Our findings underscore the need to develop mechanistically based treatment options for these women to adequately address AI-related arthralgia so that they can adhere to AIs optimally to maximize the benefit of AIs while maintaining quality of life.

Finally, the pattern of joint pain at specific sites provides important data for both clinical care and future intervention development. Site-specific joint symptoms, as suggested by the rheumatology literature,24, 25 may have a vastly different impact on functions and quality of life; therefore treatment and rehabilitation may differ. Joint-specific symptom and function measures need to be tested in this population to guide further evaluation and treatment development. Some limited self-reported data and clinical review have suggested that pain medications, supplements, and exercise may be helpful for arthralgia,8, 9, 15 although these have not been confirmed by well designed RCTs. Initial research also is underway to evaluate alternative approaches, such as acupuncture, for treating AI-related arthraligia.26, 27

In summary, the current results have shed light on the early onset and pattern of arthralgia related to AIs and have identified the interval from menopause as a novel risk factor for symptom development. Multidisciplinary translational research is critically needed to evaluate the etiology and potential therapeutic options for patients with AI-related arthralgia.

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. Conflict of Interest Disclosures
  8. References

We acknowledge the contributions of the Recruitment, Retention, and Outreach Core Facility of the Abramson Cancer Center for assisting with the development and implementation of the recruitment and retention plan. We also sincerely thank the patients, oncologists, nurse practitioners, and staff for their support of this study.

Conflict of Interest Disclosures

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. Conflict of Interest Disclosures
  8. References

Supported in part by grants from the American Cancer Society (IRG-78-002-30), the Lance Armstrong Foundation, and Pennsylvania Department of Aging. Dr. Mao is also supported by a grant from the American Cancer Society (CCCDA-08-107-01) and by NIH/NCCAM K23 AT004112. The funding agency had no role in the design or and conduct of this study.

References

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. Conflict of Interest Disclosures
  8. References