Fax: (734) 763-7672
A prospective study of aromatase inhibitor-associated musculoskeletal symptoms and abnormalities on serial high-resolution wrist ultrasonography
Article first published online: 19 AUG 2010
Copyright © 2010 American Cancer Society
Volume 116, Issue 18, pages 4360–4367, 15 September 2010
How to Cite
Henry, N. L., Jacobson, J. A., Banerjee, M., Hayden, J., Smerage, J. B., Van Poznak, C., Storniolo, A. M., Stearns, V. and Hayes, D. F. (2010), A prospective study of aromatase inhibitor-associated musculoskeletal symptoms and abnormalities on serial high-resolution wrist ultrasonography. Cancer, 116: 4360–4367. doi: 10.1002/cncr.25385
- Issue published online: 3 SEP 2010
- Article first published online: 19 AUG 2010
- Manuscript Accepted: 23 MAR 2010
- Manuscript Revised: 22 MAR 2010
- Manuscript Received: 20 JAN 2010
- breast cancer;
- aromatase inhibitor;
Nearly half of women treated with aromatase inhibitors (AI) develop AI-associated musculoskeletal symptoms (AIMSS) such as arthralgias, but to the authors' knowledge the etiology is unclear. The upper extremities are frequently affected, especially the wrists, hands, and fingers. AI use may also increase the risk of developing carpal tunnel syndrome. Tendon sheath fluid and tenosynovial changes have been demonstrated by imaging symptomatic patients who were treated with AIs. The authors hypothesized that these abnormalities are correlated with AIMSS.
Thirty consecutive patients in whom adjuvant therapy with letrozole or exemestane was initiated on a prospective clinical trial enrolled in a pilot study evaluating tendon and joint abnormalities at baseline and after 3 months of AI therapy. Patients underwent high-resolution ultrasonography of the wrists bilaterally and completed the Health Assessment Questionnaire (HAQ) and pain Visual Analog Scale (VAS). AIMSS were defined as an increase in the HAQ or VAS score during AI therapy that exceeded a predefined cutoff.
Twenty-five patients completed both the baseline and 3-month assessments. During the first 12 months of AI therapy, 15 patients developed AIMSS, and 13 discontinued therapy because of musculoskeletal symptoms. There was a trend toward an association between the presence of tendon sheath abnormalities on wrist ultrasound at baseline and the development of AIMSS (P = .06).
Clinically relevant musculoskeletal symptoms develop in women treated with AIs, leading to treatment discontinuation in a substantial percentage of these patients. However, in the current study, patient-reported symptoms were not found to be associated with changes visible on wrist ultrasonography. Cancer 2010. © 2010 American Cancer Society.
Aromatase inhibitors (AI) substantially decrease circulating estrogen concentrations in postmenopausal women via blockade of aromatase, the enzyme responsible for the conversion of androgens to estrogens. This class of drugs is increasingly used for the adjuvant treatment of postmenopausal women with hormone receptor-positive invasive breast cancer, in part because of data to support their superiority compared with tamoxifen, a selective estrogen receptor modulator.1-9 Although AIs were reported to be relatively well tolerated in the large prospective randomized clinical trials,4, 8, 10, 11 more recent studies have demonstrated that up to half of treated patients develop AI-associated musculoskeletal symptoms (AIMSS) such as arthralgias, joint stiffness, and carpal tunnel syndrome (CTS), which can lead to treatment discontinuation.12-15 A previous study found that symptoms of AIMSS appear, on average, within 3 months of the initiation of AI therapy.14
To the best of our knowledge, the pathophysiology of AIMSS remains unclear. Previously reported studies using magnetic resonance imaging (MRI) have demonstrated tenosynovitis in a substantial proportion of patients with severe wrist symptoms.16 A prospective study of 12 women who were treated with AIs demonstrated a high incidence of MRI-detected wrist tenosynovitis after 6 months of AI therapy, and new abnormalities on imaging studies were associated with a decrease in grip strength.17 However, in these studies, associations between the objective findings and the development of patient-reported musculoskeletal symptoms were not evaluated.
High-resolution ultrasound is an effective and readily accessible tool that identifies abnormalities of the tendon sheath, can quantify changes in the median nerve in CTS,18 and may be applicable for the evaluation of musculoskeletal changes that occur with AI therapy. The objective of the current pilot study was to prospectively evaluate the anatomy of the wrist using serial high-resolution ultrasonography in women initiating treatment with AIs as part of their adjuvant breast cancer treatment plan. We hypothesized that the presence of ultrasound-detected abnormalities of the wrist before treatment initiation, or their appearance during AI therapy, may be correlated with the development of AIMSS.
MATERIALS AND METHODS
Women who enrolled in the multicenter, prospective, randomized Exemestane and Letrozole Pharmacogenomics (ELPh) clinical trial (ClinicalTrials.gov #NCT00228956) between October 2007 and September 2008 at the University of Michigan Comprehensive Cancer Center were offered concomitant enrollment in this ultrasound substudy. As per the parent study, patients were randomized to receive exemestane (Aromasin; Pfizer, New York, NY) at a dose of 25 mg orally daily or letrozole (Femara; Novartis, East Hanover, NJ) at a dose of 2.5 mg orally daily. Key eligibility criteria included female gender, postmenopausal status, age ≥18 years, early stage hormone receptor-positive breast cancer, and intent to initiate adjuvant therapy with an AI. No prior AI therapy was permitted. All participants indicated that surgery, radiotherapy, and chemotherapy for breast cancer were completed before study enrollment. This substudy excluded patients with a history of prior significant wrist injury, wrist surgery, or CTS. The protocol was approved by the Institutional Review Board, and all enrolled patients provided written informed consent.
Study Design and Interventions
This prospective study of high-resolution wrist ultrasonography was conducted as a substudy of the ELPh trial. The study design of the ELPh trial has been reported previously.14 Briefly, patients were evaluated at baseline and after 1, 3, 6, and 12 months of AI therapy to assess changes in medical history and concomitant medications, and to complete symptom questionnaires and the modified Health Assessment Questionnaire (HAQ),19 which includes a pain Visual Analog Scale (VAS).
AIMSS were previously reported to develop in the majority of patients treated with AIs within 3 months of the initiation of therapy.14 Therefore, all patients in this substudy also completed the Disabilities of the Arm, Shoulder, and Hand (DASH) questionnaire20 and underwent evaluation of the bilateral wrists with high-resolution ultrasonography using 12 to 17 megahertz linear transducers (Model iU22; Philips, Bothell, WA) at baseline and after 3 months of AI therapy.
Definition of AIMSS
In the ELPh trial, AIMSS were defined as meeting ≥1 of the following criteria at any point during study participation: 1) HAQ score increased by >0.4 over the baseline score; 2) pain VAS score ≥5 cm (of 10-cm VAS) for patients with no pain (VAS of 0) at baseline; and 3) the pain VAS score increased and the pain was rated as much worse or very much worse on a self-rated clinical global impression scale (graded as 6 or 7)21 for patients reporting pain at baseline (VAS of >0).14
Patients underwent baseline and 3-month bilateral, high-resolution ultrasound imaging of the wrist. The ultrasound examination included static images and video clips, as well as a written report of imaging findings. Imaging data included: 1) the presence or absence of fluid (compressible anechoic or hypoechoic distention), hyperemia on color or power Doppler imaging, or synovitis (noncompressible, hypoechoic distention) in the volar or dorsal joint spaces (distal radioulnar, radiocarpal, and midcarpal); 2) the presence or absence of fluid, hyperemia, or synovitis in the flexor or extensor tendon sheaths; 3) the presence or absence of tendinosis (hypoechoic and enlarged), full-thickness tear (complete discontinuity), or partial thickness tear (partial anechoic cleft) in the flexor or extensor tendons; 4) median nerve area (in mm2) measured at the wrist crease (using circumferential trace) at the level of the pisiform; and 5) the presence of subcutaneous edema (hypoechoic, linear, and branching distended lymphatics). Images were analyzed in a blinded manner by an academic musculoskeletal radiologist after all patients had undergone both evaluations.
The primary objective of the current study was to estimate the 1) incidence of tendon sheath, joint space, and tendon abnormalities; and 2) increase in median nerve cross-sectional area before and 3 months after the initiation of AI therapy, and to quantify those changes based on high-resolution ultrasonography of the wrist. The secondary objective was to evaluate associations between objective, ultrasound-detected changes in the median nerve and tendon sheaths and treatment-emergent musculoskeletal symptoms.
For analysis, the ultrasound abnormalities were grouped into those affecting the tendon, the tendon sheath, and the joint space. Change in the average median nerve area before and 3 months after the initiation of AI therapy was tested using the Student t test for paired data. Univariate associations between baseline findings and the development of AIMSS or early treatment discontinuation were tested using chi-square tests. Logistic regression was used to model the probability of developing AIMSS or early treatment discontinuation (2 separate models) as a function of baseline characteristics and changes in VAS, HAQ, and DASH scores between baseline and 3 months. A P value <.05 was considered to be statistically significant.
Characteristics of the 29 eligible patients are listed in Table 1. One patient was found to be ineligible for the parent study because she was not postmenopausal, and therefore she was excluded from all analyses. Of the 29 eligible patients, 25 underwent repeat evaluation at 3 months. Two patients discontinued AI therapy before the 3-month study visit because of musculoskeletal symptoms. One patient developed intolerable aches and pains, and the other had fatigue, headache, and myalgias. Of the other 2 patients who did not undergo ultrasonography at the 3-month time point, 1 had been noncompliant with study procedures and was taken off study and 1 developed fatigue, headaches, and an inability to concentrate and discontinued AI therapy.
|Characteristic||Baseline N = 29||3 Months N = 25||P|
|Median age (range), y||61 (47-83)||59 (47-83)|
|Race/ethnicity||Non-Hispanic White||25 (86%)||22 (88%)|
|Hispanic White||2 (7%)||1 (4%)|
|Black||1 (3%)||1 (4%)|
|Asian||1 (3%)||1 (4%)|
|BMI, kg/m2||<25||9 (31%)||8 (32%)|
|25-30||4 (14%)||4 (16%)|
|>30||16 (55%)||13 (52%)|
|Prior chemotherapy||Any||15 (52%)||13 (52%)|
|Taxane||12 (41%)||11 (44%)|
|Prior tamoxifen||10 (34%)||9 (36%)|
|Median HAQ score (range)||0 (0-0.625)||0 (0-0.750)||.97|
|Median VAS score (range)||2.8 (0-8.1)||2.6 (0-5.7)||.49|
|Median DASH score (range)||5.8 (0-44.8)||6.9 (0-55)||.95|
|Patient-reported joint pain||None||11 (38%)||7 (28%)||.41|
|Mild||8 (28%)||7 (28%)|
|Moderate/severe||10 (34%)||11 (44%)|
|Patient-reported neuropathy||None||13 (45%)||16 (64%)||.44|
|Mild||11 (38%)||4 (16%)|
|Moderate/severe||5 (17%)||5 (20%)|
Wrist Ultrasonography Findings
Abnormalities on the baseline high-resolution ultrasound of the wrist were common, especially in the joint recess (Table 2); a representative finding is shown in Figure 1, Top. Nearly half of the patients developed new findings on ultrasound after 3 months of AI therapy, also primarily located in the joint recess (Table 3) (representative finding shown in Fig. 1 Bottom). Abnormalities of the tendon itself were noted only rarely. Change in the mean median nerve area with 3 months of AI therapy was not statistically different (10.4 ± 2.2 mm at baseline vs 10.8 ± 2.5 mm at 3 months).
|Site||No. With Abnormalities Unilaterally||No. With Abnormalities Bilaterally||Total No. With Anormalities||Total No. With Abnormalities|
|Tendon sheath (flexors)||3 (10.3%)||1 (3.4%)||4 (13.8%)||14 (48.3%)|
|Fluid||3 (10.3%)||1 (3.4%)||4 (13.8%)|
|Tendon sheath (extensors)||6 (20.7%)||5 (17.2%)||11 (37.9%)|
|Fluid||6 (20.7%)||5 (17.2%)||11 (37.9%)|
|Synovitis||1 (3.4%)||0||1 (3.4%)|
|Hyperemia||1 (3.4%)||0||1 (3.4%)|
|Tendon (flexor + extensor)||1 (3.4%)||1 (3.4%)||2 (6.9%)||2 (6.9%)|
|Tendinosis||1 (3.4%)||1 (3.4%)||2 (6.9%)|
|Partial thickness tear||0||0||0|
|Full thickness tear||0||0||0|
|Joint recess (volar)||6 (20.7%)||7 (24.1%)||12 (41.4%)||24 (82.8%)|
|Fluid||4 (13.8%)||6 (20.7%)||10 (34.5%)|
|Synovitis||2 (6.9%)||3 (10.3%)||5 (17.2%)|
|Joint recess (dorsal)||12 (41.4%)||18 (62.1%)||24 (82.8%)|
|Fluid||5 (17.2%)||16 (55.2%)||21 (72.4%)|
|Synovitis||4 (13.8%)||9 (31.0%)||13 (44.8%)|
|Hyperemia||5 (17.2%)||2 (6.9%)||7 (24.1%)|
|Site||No. With Abnormalities Unilaterally||No. With Abnormalities Bilaterally||Total No. With Abnormalities||Total No. With Abnormalities|
|Tendon sheath (flexors)||1 (4%)||0||1 (4%)||5 (20%)|
|Fluid||1 (4%)||0||1 (4%)|
|Tendon sheath (extensors)||4 (16%)||0||4 (16%)|
|Fluid||4 (16%)||0||4 (16%)|
|Synovitis||1 (4%)||0||1 (4%)|
|Tendon (flexor + extensor)||0||0||0||0|
|Partial thickness tear||0||0||0|
|Full thickness tear||0||0||0|
|Joint recess (volar)||3 (12%)||3 (12%)||5 (20%)||12 (48%)|
|Fluid||3 (12%)||2 (8%)||5 (20%)|
|Synovitis||1 (4%)||2 (8%)||3 (12%)|
|Joint recess (dorsal)||7 (28%)||4 (16%)||11 (44%)|
|Fluid||1 (4%)||4 (16%)||5 (20%)|
|Synovitis||5 (20%)||2 (8%)||7 (28%)|
|Hyperemia||2 (8%)||0||2 (8%)|
Changes in Patient-Reported Symptoms
After 3 months of AI therapy, 72% of patients reported joint pain, with 44% of the total study population reporting moderate to severe pain (Table 1). Median scores on the HAQ and VAS for all patients were unchanged between baseline and 3 months. It is interesting to note that approximately half of the study patients reported pre-existing neuropathy, primarily of mild severity, and this symptom decreased during AI therapy. Because patients with CTS were excluded from participation, baseline neuropathy symptoms were likely the result of alternate etiologies.
Patients were followed on AI therapy until treatment discontinuation or for at least 1 year. Fifteen patients (51.7% of the eligible cohort) developed AIMSS after a median of 6.2 months of AI therapy (range, 1.2-18.6 months). Thirteen patients (44.8% of eligible cohort) discontinued AI therapy because of musculoskeletal symptoms after a median duration of 10.7 months of therapy (range, 0.9-21.2 months). The majority of patients subsequently switched to a different AI therapy. One additional patient discontinued AI therapy after the 3-month time point because of intractable diarrhea.
Associations Between Baseline Findings and AIMSS or Early Treatment Discontinuation
On univariate analysis, a higher body mass index (BMI) was found to be associated with a decreased risk of early treatment discontinuation (odds ratio [OR] across BMI groups, 0.16; 95% confidence interval [95% CI], 0.03-0.84 [P = .03]) (Table 4). There also was a trend toward prior tamoxifen use and a lower risk of early treatment discontinuation (OR, 0.05; 95% CI, 0.002-1.0 [P = .05]). No other significant associations were noted between baseline patient characteristics or questionnaire scores and either the development of AIMSS or early treatment discontinuation because of musculoskeletal symptoms (Table 4). There was a trend toward an association between the presence of tendon sheath abnormalities on ultrasound at baseline and the subsequent development of AIMSS (Fig. 2 Top). Ten of the 15 patients (67%) who developed AIMSS during 12 months of AI therapy had tendon sheath abnormalities at baseline, compared with 4 of 14 patients (29%) who did not develop AIMSS (P = .06). No statistically significant associations between baseline ultrasound findings and either the development of AIMSS (Fig. 2 Top) or early treatment discontinuation (Fig. 2 Bottom) were noted.
|Baseline Characteristic||Development of AIMSS||Early Treatment Discontinuation|
|OR (95% CI)||P||OR (95% CI)||P|
|Age (<55 y vs ≥55 y)||0.18 (0.01-3.1)||.24||0.23 (0.01-4.2)||.32|
|BMI (<25, 25-30, >30 kg/m2)||0.43 (0.11-1.7)||.22||0.16 (0.03-0.84)||.03|
|Prior tamoxifen (no vs yes)||0.21 (0.02-3.1)||.26||0.05 (0.002-1.0)||.05|
|Prior chemotherapy (no vs yes)||0.26 (0.04-1.9)||.19||0.19 (0.02-1.74)||.14|
|HAQ score||0.59 (0.31-1.1)||.11||0.73 (0.38-1.43)||.36|
|VAS score||1.26 (0.77-2.1)||.36||0.90 (0.52-1.55)||.70|
|DASH score||1.04 (0.92-1.2)||.52||1.09 (0.96-1.25)||.17|
|Change in measurement (baseline and 3 mo)|
|HAQ score||1.15 (0.57-2.3)||.70||1.39 (0.68-2.88)||.37|
|VAS score||1.56 (0.42-5.8)||.51||1.43 (0.46-4.48)||.54|
|DASH score||1.05 (0.91-1.2)||.52||1.09 (0.93-1.29)||.28|
|Change in measurement (baseline and 12 mo)|
|HAQ score||2.76 (0.3-26.6)||.38||6.02 (0.4-83.8)||.18|
|VAS score||2.11 (1.1-4.2)||.03||2.12 (1.1-4.1)||.02|
Associations Between Changes in Findings After 3 Months of AI Therapy and AIMSS or Early Treatment Discontinuation
No associations were noted between changes from baseline to after 3 months of AI therapy in HAQ, VAS, or DASH scores and the development of AIMSS or early treatment discontinuation (Table 4). Similar to the findings at baseline, we did not observe statistically significant associations between new findings on ultrasonography and either the development of AIMSS or early treatment discontinuation (Fig. 3).
Although a previous analysis suggested that the majority of AI-treated patients develop symptoms within 3 months of treatment initiation, the median time to the development of AIMSS in the cohort enrolled in the current substudy was 6.2 months. Patients enrolled in this substudy completed the HAQ and VAS questionnaires at multiple additional time points as part of the parent study. Therefore, we performed an unplanned analysis of the correlation between the development of AIMSS and early treatment discontinuation and changes in the HAQ and VAS scores from baseline to either 12 months or the time of early treatment discontinuation, whichever came first. We found a significant association between changes in the VAS score and the development of AIMSS (OR, 2.11; 95% CI, 1.1-4.2 [P = .03]) and early treatment discontinuation (OR, 2.12; 95% CI, 1.1-4.1 [P = .02]). A similar finding was not noted with changes in the HAQ score (Table 4).
In the current study, abnormalities of the tendon sheath and joint space at the bilateral wrist detected on high-resolution ultrasound were very common in women before the initiation of AI therapy. We observed a trend toward an association between tendon sheath abnormalities on ultrasonography before treatment initiation and the development of AIMSS. The identification of factors predictive of the increased likelihood of developing AIMSS would be useful for counseling and managing patients who are initiating treatment with adjuvant endocrine therapy. Although this study does not provide new insight into the mechanism of AIMSS, our results suggest that pre-existing abnormalities within the tendon sheath may predict which women are likely to experience AI-associated musculoskeletal pain.
The findings of the current study are similar to those of previous reports, the majority of which have been cross-sectional studies.16, 17, 22 A strength of the current study is the prospective, serial assessment of patients using both high-resolution ultrasound and patient self-report questionnaires.
A higher than anticipated number of patients discontinued therapy because of musculoskeletal symptoms, possibly because of chance in this small patient cohort, or possibly because they were enrolled in a clinical trial with a specific focus on musculoskeletal symptoms, which may have led to greater awareness of toxicity on the part of both the patient and physician. Similar to other reports, patients in this cohort who were previously treated with tamoxifen were less likely to discontinue AI therapy.13 However, contrary to other reports,12, 13 there was an association noted between obesity and a decreased risk of treatment discontinuation because of musculoskeletal symptoms.
A limitation of the current study is the lack of multiple, long-term ultrasound assessments during AI therapy. Patients were assessed after 3 months of therapy, but the median time to reported AIMSS in this cohort was approximately 6 months. Therefore, it is possible that we were unable to detect a statistically significant association between ultrasound-detected abnormalities and patient-reported symptoms because of insufficient time point assessments. This restriction to analysis of the data from the 3-month time point may also have led to the lack of a noted association between changes in the HAQ and VAS scores and the development of AIMSS or early treatment discontinuation, which is also supported by our finding of a statistically significant association between changes in the VAS score between baseline and 12 months and both the development of AIMSS and early treatment discontinuation. A second limitation of the current study is the relatively small subject cohort, which was further decreased by early treatment discontinuation by 5 patients, thereby reducing statistical power.
Another possible limitation is the use of ultrasonography as opposed to MRI. One potential criticism of ultrasonography is operator-dependent variability. In the current study, we attempted to decrease this variability as much as possible through the collection and storage of both static images and cine clips for each subject at each time point for later analysis by a single, blinded reviewer. Advantages of high-resolution ultrasound include lower cost, shorter evaluation time, and a lack of factors that can preclude assessment with MRI, such as claustrophobia and the presence of metal in the body.
In summary, the findings of the current pilot study suggest that the presence of abnormalities of the wrist tendon sheath before the initiation of AI therapy may be predictive of the subsequent development of AIMSS both at the wrist and at other anatomic sites. Future studies are warranted to confirm the trend noted in this pilot study, and to further elucidate the predictive role for imaging when making treatment decisions concerning adjuvant endocrine therapy.
CONFLICT OF INTEREST DISCLOSURES
Supported by Pharmacogenetics Research Network Grant U-01 GM61373 from the National Institute of General Medical Sciences, National Institutes of Health (NIH), Bethesda, Maryland (Indiana University) and Grant M01-RR000042 to the University of Michigan from the National Center for Research Resources (NCRR), a component of the NIH. The contents of this article are solely the responsibility of the authors and do not necessarily represent the official views of the NCRR or NIH. Grants also were received from Pfizer (to Dr. Hayes); Novartis Pharma AG (to Dr. Hayes); the Barbara Padnos Breast Cancer Research Fund, University of Michigan Comprehensive Cancer Center (to Dr. Henry); and the Expedition Inspiration Fund for Breast Cancer Research (to Dr. Henry). ClinicalTrials.gov number: NCT00228956. Drs. Hayes, Stearns, and Storniolo received research funding from Pfizer and Novartis. Drs. Henry and Hayes received research funding from AstraZeneca. Dr. Stearns received honoraria from AstraZeneca. Dr. Jacobson received other support (equipment) from Philips, SonoSite, and Terason; has acted as a consultant for BioClinica; and has received book royalties from Elsevier.
- 21Clinical global impressions. In: GuyW, ed. ECDEU Assessment Manual for Psychopharmacology. Rockville, MD: US National Institute of Health, Psychopharmacology Research Branch; 1976..