The Journal of Pain, Vol 15, No 3 (March), 2014: pp 293-303 Available online at www.jpain.org and www.sciencedirect.com
Classification of and Risk Factors for Estrogen Deprivation Pain Syndromes Related to Aromatase Inhibitor Treatments in Women With Breast Cancer: A Prospective Multicenter Cohort Study Francoise Laroche,* Jo€ el Coste,y Terkia Medkour,z Paul Henri Cottu,x Jean-Yves Pierga,x jj Jean-Pierre Lotz, Karine Beerblock,jj Christophe Tournigand,{ Xavier Decl eves,** Patricia de Cremoux,yy Didier Bouhassira,zz and Serge Perrotz,zz *Pain Clinic, Saint Antoine Hospital, Paris, France. y ^ tel Dieu Hospital, Paris, France. Biostatistics, Ho z ^ tel Dieu Hospital, Paris Descartes University, Paris, France. Pain Clinic and Internal Medicine Department, Ho x Medical Oncology Department, Institut Curie, Paris, France. jj Oncology Department, Tenon Hospital, Paris, France. { Oncology Department, Saint Antoine Hospital, Paris, France. ^ tel Dieu Hospital, Paris, France. **Pharmacology Department, Ho yy Molecular Oncology, APHP and Paris-Diderot University, Saint Louis Hospital, Paris, France. zz INSERM U 987, Paris, France.
Abstract: Aromatase inhibitors (AIs) are the first-line treatment in women with breast cancer for total estrogen depletion. Half the treated women may develop pain, and this condition may therefore be seen as a clinical model of pain related to estrogen deprivation. In this prospective multicenter study, we classified AI-related pain syndromes and identified their predictors. A 1-year, prospective, multicenter cohort study, with 6 visits, was carried out on 135 women with early-stage breast cancer and no pain at the start of AI treatment. At initial assessment, we investigated clinical (demographic and psychosocial, cancer characteristics and treatment, sleep, quality of life), biological (sex hormones, vitamin D, bone biomarkers, oxidative stress, immunologic and inflammatory markers), environmental, and genetic (polymorphism for pain mechanisms) risk factors for pain. During 1 year of follow-up, 77 women (57%) developed pain, leading to AI discontinuation in 12 cases. Five pain syndromes were identified: joint pain (36%), diffuse pain (22%), tendinitis (22%), neuropathic pain (9%), and mixed pain (11%), which are mostly persistent (57%), with diffuse and joint pains the most intense. Risk factors for the development of pain included higher levels of anxiety and impaired quality of life at the initial assessment, whereas cancer characteristics, genetic background, inflammation, and immunologic and hormonal status at baseline were not significant predictors. Perspective: This article presents a classification of AI–related pain syndromes induced by estrogen deprivation that were previously described as arthralgia, but not as neuropathic, diffuse, and mixed pain. This estrogen deprivation–related condition represents a clinical model of pain, and our study identified mostly psychological risk factors for pain development. ª 2014 by the American Pain Society Key words: Breast cancer, pain, estrogen deprivation, aromatase inhibitors, arthritis.
Received September 30, 2013; Accepted November 20, 2013. This study received financial support from the CNP foundation, the APICIL foundation and IUD-Institut Upsa de la Douleur. None of the authors have any conflicts of interest related to this study. Address reprint requests to Serge Perrot, MD, PhD, Internal Medicine ^ pital Ho ^ tel Dieu, 1 Place du Parvis Notre Department and Pain Clinic, Ho Dame, 75004 Paris, France. E-mail: [email protected] 1526-5900/$36.00 ª 2014 by the American Pain Society http://dx.doi.org/10.1016/j.jpain.2013.11.004
A
romatase inhibitors (AIs) are the first-line treatment for hormone-dependent breast cancers in postmenopausal women.1,17 The aim of AI treatment is to induce total estrogen deprivation and prevent breast cancer recurrence. In relation to this estrogen deprivation, data from clinical trials and prospective surveys indicate that up to 50% of patients may develop pain following AI treatment.1,5,13-15,20,21,30-32,34,35,37,38 Several symptoms are described, including joint pain, stiffness, myalgia, tendinitis, carpal tunnel syndrome, and bone pain.20,22,36,37 However, AI-related pain problems 293
294
The Journal of Pain
have not yet been systematically classified. Thus, AI-related pain syndrome represents a clinical model of pain related to estrogen deprivation that could shed light on the links between pain development and estrogen modifications. Studies have identified previous hormone replacement therapy, time since menopause, obesity,14,26,34,35 and poorer quality of life38 as risk factors for the development of AI-related pain. However, no study has yet considered psychological factors, such as anxiety, depression, personality traits, and coping strategies. Furthermore, no study has considered all these potential risk factors together in a single prospective analysis. The aims of this study were 1) to describe and classify all the AI-related pain symptoms developing within 1 year of estrogen deprivation related to the initiation of AI treatment in pain-free women and 2) to investigate extensive predictors of pain development in an estrogen deprivation clinical model of pain: clinical predictors (demographic, psychological) and biological predictors (inflammation, oxidative stress, autoimmune disorders, bone metabolism, genetic pain polymorphism, hormones). The subtyping of pain syndromes and its specific predictors should help to improve the management of AI-related pain and to reduce the number of women stopping their AI treatment.
Patients and Methods Design We carried out a 1-year observational multicenter prospective cohort study with recruitment at 4 medical oncology departments and extensive pain assessment at 1 pain clinic, all these centers being located in university hospitals in Paris, France.
Ethics Statement This study was carried out in accordance with the Helsinki Declaration. Informed consent was obtained from each subject before enrollment. Institutional review board and French data protection agency approvals (CCTIRS, CNIL) were obtained before subject enrollment.
Study Population Consecutive women treated for early breast cancer at 4 medical oncology departments were eligible to participate if free of pain and starting AI treatment from June 2009 to March 2011, and each participant was followed for 12 months. Exclusion criteria were pain conditions interfering with pain assessment, patient not able to follow the protocol, other treatment for cancer scheduled for the next 12 months, previous AI treatment, and estimated survival of less than 12 months.
Aromatase Inhibitors Pain-Related Syndromes tane). All other treatments, including analgesics, were permitted and their use was monitored.
Measurements Baseline and follow-up measurements were carried out under the supervision of a research nurse (T.M.) and rheumatologists specializing in pain medicine. Visits were scheduled for 1, 3, 6, and 12 months after the start of AI treatment.
Baseline Measurements The initial visit occurred just before the start of AI treatment, with a complete medical examination carried out by a rheumatologist (F.L. or S.P.).
Demographic and Cancer History Data The demographic data collected included age, body mass index (BMI), menopause duration, and history of hormone replacement therapy. Cancer history variables included pathologic subtype of breast cancer, type of surgery, history of chemotherapy, and history of radiotherapy.
Pain Assessment and Subclassification Pain intensity was assessed with a 100-mm visual analog scale (VAS),19 and we considered pain as clinically significant for VAS scores of at least 30 mm. Affective and sensorial domains of pain were assessed with the shortform McGill Pain Questionnaire, which has been validated for cancer and arthritis.19 The extent to which pain interfered with daily life was assessed with the Brief Pain Inventory,9 which has also been validated for use in populations with and without cancer.2 Pain duration was assessed, at each visit, in months since the start of pain. We included 4 subclasses of pain: joint pain, diffuse pain, neuropathic pain, and tendon pain. This classification was made in 3 steps: 1) the body diagram included in the Brief Pain Inventory for pain localization and classification of diffuse pain, 2) by specific questions when localized pain: first 7 items from the DN4 questionnaire4 to detect neuropathic component, and specific questions on pain location (around the joint, inside the joint), and 3) confirmation by clinical examination. We also defined 2 types of pain: transient pain, observed at 1 visit, and persistent pain, observed on at least 2 consecutive visits. On the basis of previous studies on AI-related pain, and according to the International Association for the Study of Pain pain taxonomy,28 we included patients developing pain during the study in the following categories: joint pain, diffuse pain, neuropathic pain, tendon pain, and mixed pain combining at least 2 of the previous categories of pain.
Psychological and Clinical Assessment
AI Treatments and Other Drugs Prescribed During the Study Period All postmenopausal women with hormonedependent breast cancer recruited for this study began oral treatment with AI (anastrozole, letrozole, exemes-
Anxiety and depression were assessed with the Hospital Anxiety and Depression Scale.3 Personality traits and pain catastrophizing were assessed with the Temperament and Character Inventory33 (TCI–French version) and the Pain Catastrophizing Scale.42
Laroche et al
The Journal of Pain 10,11
The TCI was developed by Robert Cloninger for assessment of the 7 factors of the psychobiological model of personality. The TCI is a 240-item true/false questionnaire measuring the 4 dimensions of temperament (novelty seeking, harm avoidance, reward dependence, and persistence) and 3 dimensions of character (self-directedness, cooperativeness, and selftranscendence); 25 facets are also measured. The validity and the reliability of the original American version of the TCI have been established in the general population of the United States.11 Overall breast cancer–related quality of life was assessed with the European Organization for Research and Treatment of Cancer (EORTC) Quality of Life Questionnaire– Breast Cancer Module 23 (QLQ-BR23).41 Fatigue severity was assessed with the 20-item Multidimensional Fatigue Inventory,40 sleep with a global VAS for sleep quality, and hot flushes with a Yes/No question.
Biological Factors Biological tests included determinations of inflammation markers (erythrocyte sedimentation rate, C-reactive protein), autoimmune responses (rheumatoid factor, antinuclear antibodies, anti-citrullinated protein antibodies), bone metabolism indicators (plasma calcium concentration, plasma C-terminal telopeptide, 25 OHvitamin D concentrations, and bone mineral density at the hip and lumbar column), circulating sex hormone concentrations (estradiol, estrone, delta-4androstenedione, testosterone, sex binding protein [SBP]), genetic polymorphisms (single-nucleotide polymorphisms) for selected pain mechanisms (COMT, OPRM1, CYP19A1), and oxidative stress markers (thiols and advanced oxidation protein products).
Outcome Measures for the 1-Year Period of Assessment Outcome measures were assessed at 1, 3, 6, and 12 months after the start of AI treatment. At each visit, a complete medical examination was carried out and patients were asked to complete questionnaires analyzing pain (VAS, Brief Pain Inventory, McGill Pain Questionnaire), psychological variables (Hospital Anxiety and Depression Scale, Beck Depression Inventory, EORTC QLQ-BR23, Pain Catastrophizing Scale), and fatigue (Multidimensional Fatigue Inventory–20-Item). Participants complaining of pain were also asked to complete the DN4 questionnaire,4 designed to screen for neuropathic pain. Participants were asked to indicate openly whether they were experiencing any additional symptoms. At each visit, as at the baseline visit, a blood sample was taken for the determination of inflammation markers, autoimmune responses, bone metabolism indicators, circulating sex hormones, and oxidative stress markers.
Data Analyses A power analysis indicated that a sample size of 135 would be sufficient to obtain an estimate of pain frequency with a precision of 8%, assuming a frequency
295
of pain of 20 to 50% in the population within the first year of AI treatment. We first carried out standard univariate (proportions, means, and standard deviations) and bivariate (chisquared and Wilcoxon tests) statistical analyses to describe the sample and determine whether there was an association between baseline predictor and outcome variables. We then used logistic regression models to identify predictors of the overall occurrence of pain and of the occurrence of specific types of pain during follow-up. Predictive models were constructed in stages. We first modeled genetic predictors to analyze genetic background. We then entered information about the disease itself: cancer diagnosis and treatment. In the third step, we entered the clinical, psychological, and biological variables measured at baseline. In the fourth step, we entered the type of AI (steroidal versus nonsteroidal). Odds ratios and 95% confidence intervals are presented, as appropriate. We compared the baseline characteristics of patients with persistent pain with those of patients with transient pain, and a logistic model was constructed to identify independent risk factors for pain persistence. Logistic repeated measures models were used to test the relationship between pain and hot flushes, or pain and sleep disorders. Repeated measures models were used to compare the impact of different pain patterns in terms of intensity, the interference of pain with daily life, and fatigue at the various time points during follow-up (1, 3, 6, and 12 months). We used multiple imputation techniques to derive estimates of missing scores at each time point. The most common reason for missing scores was the discontinuation of AI treatment due to pain. Statistical analyses were carried out with SAS, version 9.2 (SAS Institute, Cary, NC). All statistical tests were 2-tailed, with a significance level of .05.
Results We invited 179 women with breast cancer but no pain to participate in this study, just before they began AI treatment. In total, 135 women were included in the analyses (Fig 1).
Pain and Pain Syndromes Related to AIs Pain Frequency During the study period, 77 (57%) of the 135 patients developed pain, with a Kaplan-Meier estimate of pain occurrence at 1 year of .59 (95% confidence interval = .51–.68). AI treatment was stopped because of pain in 12 cases. Table 1 shows the baseline characteristics and differences between women who developed pain and women who remained pain-free during the 12 months of study.
Five Types of Pain Syndromes We identified 5 primary types of pain syndromes (Fig 2): joint pain (the first pain to emerge) in 36%,
296
The Journal of Pain
Aromatase Inhibitors Pain-Related Syndromes
Figure 1. Flow chart for the recruitment of women with breast cancer free of pain, starting AI treatment, for the evaluation of AIrelated pain over a 1-year period. diffuse pain and tendinitis (both emerging at 3 months) both in 22%, neuropathic pain in 9% (emerging at 6 months on average), and mixed pain (mostly a combination of joint and tendinitis pain) in 11% (Table 2). Overall pain intensity did not differ significantly between the pain types (Table 2), but pain intensity was significantly higher in diffuse pain syndrome than in tendinitis and neuropathic pain.
Analgesics Prescribed for Patients With Pain During the Study Period During the study period, patients with pain were prescribed various analgesics and nonsteroidal antiinflammatory drugs: acetaminophen was prescribed to 48 women, oral nonsteroidal anti-inflammatory drugs to 9, topical nonsteroidal anti-inflammatory drugs to 7, and weak opioids (eg, tramadol, codeine) to 24.
Transient and Persistent Pain We also classified pain syndromes into 2 categories (Table 3): transient pain, for pain observed at only 1 visit (n = 26), and persistent pain, for pain observed on at least 2 consecutive visits (n = 34), in 60 of the 77 women with pain for whom it was possible to describe pain progression (for 17 women, it not possible to describe pain progression because they did not develop pain until the last visit or had to stop AIs because of severe pain). Diffuse pain was more frequently persistent (54.5%), whereas neuropathic pain was mostly transient (72.7%).
Pain Interference and Pain Domains of Patients With Each Type of Pain Daily pain interference scores were higher in the diffuse pain group than in the tendinitis group (Table 4). Indeed, the tendinitis group was chosen as the reference group because pain intensity and impact were lowest in this group.
Other Symptoms Occurring During Follow-Up The incidence of hot flushes increased significantly, from 8.3% at 1 month to 27.8% at 12 months, whereas the incidence of sleep disorders did not change significantly. Hot flushes and sleep disorders were not significantly related to pain during follow-up.
Biological Changes Associated With Pain Development During 1 Year of Follow-Up Pain development was not associated with any specific inflammatory, autoimmune, oxidative stress, or hormonal changes. Only low levels of sex binding protein were significantly (P < .01) associated with pain development during the study period, without correlation with a specific pain syndrome. One patient developed rheumatoid arthritis 3 months after the start of AI treatment.
Laroche et al
The Journal of Pain
297
Baseline Characteristics of All Patients (N = 135) and Comparison of Patients Developing AI-Related Pain (n = 77) and Patients Remaining Free of Pain (n = 57) Until the End of the Study Period
Table 1.
CHARACTERISTICS
TOTAL
PAIN
NO PAIN
Demographics (N) Age (y) (mean 6 StDev) BMI (StDev) Menopause duration (y) (StDev) HRT duration (y) (StDev) Type of surgery Tumorectomy Ganglion exeresis Previous chemotherapy Taxanes Anthracyclines Others Radiotherapy AI treatment allocation Anastrozole Letrozole Exemestane Psychological Factors Anxiety (HADS) Depression (HADS) Personality traits (TCI) Novelty seeking Harm avoidance Reward recompense Persistence Self-directedness Cooperativeness Transcendence Pain catastrophizing (PCS) Rumination Amplification Despair Quality of life (EORTC QLQ-BR23) Body image Sexual function Sexual enjoyment Future perspectives Systemic therapy side effects Breast symptoms Arm symptoms Upset by hair loss Sleep Not restorative (%) Fatigue (MFI20) Fatigue—general Fatigue—physical Fatigue—mental Fatigue—reduced activities Fatigue—motivation Genetics: SNPs (%) COMT: ValVal vs. ValMet 1 MetMet ORPM: AA vs. (AG 1 GG) CYP19: tandem repeats (7 1 8) vs. 10 – CYP 19A1 Hormone and binding proteins levels Estrone SBP Testosterone Delta4 androstenedione
135 61.5 6 7.1 25.28 6 5.32 11.65 6 7.60 8.03 6 5.50
77 61 6 7 25.91 6 6 11.57 6 8.44 7.53 6 5.60
57* 62.4 6 7.2 24.5 6 4.22 11.90 6 6.30 8.63 6 5.37
.17 .22 .50 .26
66 85 64 47 55 10 88
65 87 67 52 57 9 90
67 82 60 40 51 11 88
.83 .46 .35 .18 .47 .78 .73
32 67 .74
28.6 72 0
37 61 1.75
.32 .45 .27
7.18 6 3.78 3.61 6 3.02
7.65 6 3.65 3.56 6 2.70
6.53 6 3.92 3.75 6 3.40
.05 .81
8.02 6 2.54 8.55 6 4.55 10.06 6 2.25 3.26 6 1.37 19.16 6 3.78 20.37 6 2.83 6.27 6 3.43 14.30 6 11 5.89 6 4.44 2.94 6 2.61 5.45 6 4.83
8 6 2.41 8.77 6 4.51 10 6 2.30 3.50 6 1.31 18.75 6 3.80 20.40 6 2.61 6.61 6 3.73 14.65 6 10.50 5.95 6 4.30 3 6 2.60 5.70 6 4.80
8 6 2.74 8.26 6 4.66 10.10 6 2.24 2.94 6 1.4 19.72 6 3.77 20.31 6 3.14 5.89 6 2.95 13.86 6 11.80 5.84 6 4.71 2.87 6 2.71 5.14 6 5
.79 .41 .83 .02 .14 .58 .37 .56 .84 .62 .45
66 6 33 20 6 22.60 52 6 27.50 55.30 6 31.06 32 6 21.60 30 6 23 24.16 6 24.70 57.31 6 42.14
60.41 6 35.40 16.22 6 20.72 47 6 26.54 50.64 6 33 36 6 21.40 36 6 24 31 6 28 62 6 44.40
73.51 6 28.11 25.30 6 24.20 57.14 6 28.20 62 6 28 26.10 6 21 22.36 6 18.30 15.55 6 16 49.20 6 39
.04 .03 .23 .053 .0061 .0007 .003 .036
28 59.65 6 6.84 12.24 6 1.65 11.00 6 4.06 8.74 6 3.81 10.27 6 3.81 8.57 6 3.85
35 60.45 6 6 12.45 6 1.88 12.08 6 3.81 8.92 6 3.79 10.58 6 3.69 8.76 6 3.56
20 58.45 6 7.66 11.95 6 1.25 9.66 6 3.96 8.56 6 3.86 9.89 6 3.97 8.36 6 4.24
.24 .14 .13 .0006 .63 .34 .32
30 69 10
29 73 9.33
31.6 63 89
.78 .21 .82
33.57 6 20.27 72 6 32.36 .30 6 .17 1.08 6 .86
.9 .03 .16 .35
32.16 6 15.71 64.20 6 30 .28 6 .17 .97 6 .70
31 6 11.42 59 6 27 .26 6 .18 .90 6 .50
P VALUE
298
Aromatase Inhibitors Pain-Related Syndromes
The Journal of Pain
Table 1.
Continued CHARACTERISTICS
Inflammation (%) CRP
Classification of and Risk Factors for Estrogen Deprivation Pain Syndromes Related to Aromatase Inhibitor Treatments in Women With Breast Cancer: A Prospective Multicenter Cohort Study Francoise Laroche,* Jo€ el Coste,y Terkia Medkour,z Paul Henri Cottu,x Jean-Yves Pierga,x jj Jean-Pierre Lotz, Karine Beerblock,jj Christophe Tournigand,{ Xavier Decl eves,** Patricia de Cremoux,yy Didier Bouhassira,zz and Serge Perrotz,zz *Pain Clinic, Saint Antoine Hospital, Paris, France. y ^ tel Dieu Hospital, Paris, France. Biostatistics, Ho z ^ tel Dieu Hospital, Paris Descartes University, Paris, France. Pain Clinic and Internal Medicine Department, Ho x Medical Oncology Department, Institut Curie, Paris, France. jj Oncology Department, Tenon Hospital, Paris, France. { Oncology Department, Saint Antoine Hospital, Paris, France. ^ tel Dieu Hospital, Paris, France. **Pharmacology Department, Ho yy Molecular Oncology, APHP and Paris-Diderot University, Saint Louis Hospital, Paris, France. zz INSERM U 987, Paris, France.
Abstract: Aromatase inhibitors (AIs) are the first-line treatment in women with breast cancer for total estrogen depletion. Half the treated women may develop pain, and this condition may therefore be seen as a clinical model of pain related to estrogen deprivation. In this prospective multicenter study, we classified AI-related pain syndromes and identified their predictors. A 1-year, prospective, multicenter cohort study, with 6 visits, was carried out on 135 women with early-stage breast cancer and no pain at the start of AI treatment. At initial assessment, we investigated clinical (demographic and psychosocial, cancer characteristics and treatment, sleep, quality of life), biological (sex hormones, vitamin D, bone biomarkers, oxidative stress, immunologic and inflammatory markers), environmental, and genetic (polymorphism for pain mechanisms) risk factors for pain. During 1 year of follow-up, 77 women (57%) developed pain, leading to AI discontinuation in 12 cases. Five pain syndromes were identified: joint pain (36%), diffuse pain (22%), tendinitis (22%), neuropathic pain (9%), and mixed pain (11%), which are mostly persistent (57%), with diffuse and joint pains the most intense. Risk factors for the development of pain included higher levels of anxiety and impaired quality of life at the initial assessment, whereas cancer characteristics, genetic background, inflammation, and immunologic and hormonal status at baseline were not significant predictors. Perspective: This article presents a classification of AI–related pain syndromes induced by estrogen deprivation that were previously described as arthralgia, but not as neuropathic, diffuse, and mixed pain. This estrogen deprivation–related condition represents a clinical model of pain, and our study identified mostly psychological risk factors for pain development. ª 2014 by the American Pain Society Key words: Breast cancer, pain, estrogen deprivation, aromatase inhibitors, arthritis.
Received September 30, 2013; Accepted November 20, 2013. This study received financial support from the CNP foundation, the APICIL foundation and IUD-Institut Upsa de la Douleur. None of the authors have any conflicts of interest related to this study. Address reprint requests to Serge Perrot, MD, PhD, Internal Medicine ^ pital Ho ^ tel Dieu, 1 Place du Parvis Notre Department and Pain Clinic, Ho Dame, 75004 Paris, France. E-mail: [email protected] 1526-5900/$36.00 ª 2014 by the American Pain Society http://dx.doi.org/10.1016/j.jpain.2013.11.004
A
romatase inhibitors (AIs) are the first-line treatment for hormone-dependent breast cancers in postmenopausal women.1,17 The aim of AI treatment is to induce total estrogen deprivation and prevent breast cancer recurrence. In relation to this estrogen deprivation, data from clinical trials and prospective surveys indicate that up to 50% of patients may develop pain following AI treatment.1,5,13-15,20,21,30-32,34,35,37,38 Several symptoms are described, including joint pain, stiffness, myalgia, tendinitis, carpal tunnel syndrome, and bone pain.20,22,36,37 However, AI-related pain problems 293
294
The Journal of Pain
have not yet been systematically classified. Thus, AI-related pain syndrome represents a clinical model of pain related to estrogen deprivation that could shed light on the links between pain development and estrogen modifications. Studies have identified previous hormone replacement therapy, time since menopause, obesity,14,26,34,35 and poorer quality of life38 as risk factors for the development of AI-related pain. However, no study has yet considered psychological factors, such as anxiety, depression, personality traits, and coping strategies. Furthermore, no study has considered all these potential risk factors together in a single prospective analysis. The aims of this study were 1) to describe and classify all the AI-related pain symptoms developing within 1 year of estrogen deprivation related to the initiation of AI treatment in pain-free women and 2) to investigate extensive predictors of pain development in an estrogen deprivation clinical model of pain: clinical predictors (demographic, psychological) and biological predictors (inflammation, oxidative stress, autoimmune disorders, bone metabolism, genetic pain polymorphism, hormones). The subtyping of pain syndromes and its specific predictors should help to improve the management of AI-related pain and to reduce the number of women stopping their AI treatment.
Patients and Methods Design We carried out a 1-year observational multicenter prospective cohort study with recruitment at 4 medical oncology departments and extensive pain assessment at 1 pain clinic, all these centers being located in university hospitals in Paris, France.
Ethics Statement This study was carried out in accordance with the Helsinki Declaration. Informed consent was obtained from each subject before enrollment. Institutional review board and French data protection agency approvals (CCTIRS, CNIL) were obtained before subject enrollment.
Study Population Consecutive women treated for early breast cancer at 4 medical oncology departments were eligible to participate if free of pain and starting AI treatment from June 2009 to March 2011, and each participant was followed for 12 months. Exclusion criteria were pain conditions interfering with pain assessment, patient not able to follow the protocol, other treatment for cancer scheduled for the next 12 months, previous AI treatment, and estimated survival of less than 12 months.
Aromatase Inhibitors Pain-Related Syndromes tane). All other treatments, including analgesics, were permitted and their use was monitored.
Measurements Baseline and follow-up measurements were carried out under the supervision of a research nurse (T.M.) and rheumatologists specializing in pain medicine. Visits were scheduled for 1, 3, 6, and 12 months after the start of AI treatment.
Baseline Measurements The initial visit occurred just before the start of AI treatment, with a complete medical examination carried out by a rheumatologist (F.L. or S.P.).
Demographic and Cancer History Data The demographic data collected included age, body mass index (BMI), menopause duration, and history of hormone replacement therapy. Cancer history variables included pathologic subtype of breast cancer, type of surgery, history of chemotherapy, and history of radiotherapy.
Pain Assessment and Subclassification Pain intensity was assessed with a 100-mm visual analog scale (VAS),19 and we considered pain as clinically significant for VAS scores of at least 30 mm. Affective and sensorial domains of pain were assessed with the shortform McGill Pain Questionnaire, which has been validated for cancer and arthritis.19 The extent to which pain interfered with daily life was assessed with the Brief Pain Inventory,9 which has also been validated for use in populations with and without cancer.2 Pain duration was assessed, at each visit, in months since the start of pain. We included 4 subclasses of pain: joint pain, diffuse pain, neuropathic pain, and tendon pain. This classification was made in 3 steps: 1) the body diagram included in the Brief Pain Inventory for pain localization and classification of diffuse pain, 2) by specific questions when localized pain: first 7 items from the DN4 questionnaire4 to detect neuropathic component, and specific questions on pain location (around the joint, inside the joint), and 3) confirmation by clinical examination. We also defined 2 types of pain: transient pain, observed at 1 visit, and persistent pain, observed on at least 2 consecutive visits. On the basis of previous studies on AI-related pain, and according to the International Association for the Study of Pain pain taxonomy,28 we included patients developing pain during the study in the following categories: joint pain, diffuse pain, neuropathic pain, tendon pain, and mixed pain combining at least 2 of the previous categories of pain.
Psychological and Clinical Assessment
AI Treatments and Other Drugs Prescribed During the Study Period All postmenopausal women with hormonedependent breast cancer recruited for this study began oral treatment with AI (anastrozole, letrozole, exemes-
Anxiety and depression were assessed with the Hospital Anxiety and Depression Scale.3 Personality traits and pain catastrophizing were assessed with the Temperament and Character Inventory33 (TCI–French version) and the Pain Catastrophizing Scale.42
Laroche et al
The Journal of Pain 10,11
The TCI was developed by Robert Cloninger for assessment of the 7 factors of the psychobiological model of personality. The TCI is a 240-item true/false questionnaire measuring the 4 dimensions of temperament (novelty seeking, harm avoidance, reward dependence, and persistence) and 3 dimensions of character (self-directedness, cooperativeness, and selftranscendence); 25 facets are also measured. The validity and the reliability of the original American version of the TCI have been established in the general population of the United States.11 Overall breast cancer–related quality of life was assessed with the European Organization for Research and Treatment of Cancer (EORTC) Quality of Life Questionnaire– Breast Cancer Module 23 (QLQ-BR23).41 Fatigue severity was assessed with the 20-item Multidimensional Fatigue Inventory,40 sleep with a global VAS for sleep quality, and hot flushes with a Yes/No question.
Biological Factors Biological tests included determinations of inflammation markers (erythrocyte sedimentation rate, C-reactive protein), autoimmune responses (rheumatoid factor, antinuclear antibodies, anti-citrullinated protein antibodies), bone metabolism indicators (plasma calcium concentration, plasma C-terminal telopeptide, 25 OHvitamin D concentrations, and bone mineral density at the hip and lumbar column), circulating sex hormone concentrations (estradiol, estrone, delta-4androstenedione, testosterone, sex binding protein [SBP]), genetic polymorphisms (single-nucleotide polymorphisms) for selected pain mechanisms (COMT, OPRM1, CYP19A1), and oxidative stress markers (thiols and advanced oxidation protein products).
Outcome Measures for the 1-Year Period of Assessment Outcome measures were assessed at 1, 3, 6, and 12 months after the start of AI treatment. At each visit, a complete medical examination was carried out and patients were asked to complete questionnaires analyzing pain (VAS, Brief Pain Inventory, McGill Pain Questionnaire), psychological variables (Hospital Anxiety and Depression Scale, Beck Depression Inventory, EORTC QLQ-BR23, Pain Catastrophizing Scale), and fatigue (Multidimensional Fatigue Inventory–20-Item). Participants complaining of pain were also asked to complete the DN4 questionnaire,4 designed to screen for neuropathic pain. Participants were asked to indicate openly whether they were experiencing any additional symptoms. At each visit, as at the baseline visit, a blood sample was taken for the determination of inflammation markers, autoimmune responses, bone metabolism indicators, circulating sex hormones, and oxidative stress markers.
Data Analyses A power analysis indicated that a sample size of 135 would be sufficient to obtain an estimate of pain frequency with a precision of 8%, assuming a frequency
295
of pain of 20 to 50% in the population within the first year of AI treatment. We first carried out standard univariate (proportions, means, and standard deviations) and bivariate (chisquared and Wilcoxon tests) statistical analyses to describe the sample and determine whether there was an association between baseline predictor and outcome variables. We then used logistic regression models to identify predictors of the overall occurrence of pain and of the occurrence of specific types of pain during follow-up. Predictive models were constructed in stages. We first modeled genetic predictors to analyze genetic background. We then entered information about the disease itself: cancer diagnosis and treatment. In the third step, we entered the clinical, psychological, and biological variables measured at baseline. In the fourth step, we entered the type of AI (steroidal versus nonsteroidal). Odds ratios and 95% confidence intervals are presented, as appropriate. We compared the baseline characteristics of patients with persistent pain with those of patients with transient pain, and a logistic model was constructed to identify independent risk factors for pain persistence. Logistic repeated measures models were used to test the relationship between pain and hot flushes, or pain and sleep disorders. Repeated measures models were used to compare the impact of different pain patterns in terms of intensity, the interference of pain with daily life, and fatigue at the various time points during follow-up (1, 3, 6, and 12 months). We used multiple imputation techniques to derive estimates of missing scores at each time point. The most common reason for missing scores was the discontinuation of AI treatment due to pain. Statistical analyses were carried out with SAS, version 9.2 (SAS Institute, Cary, NC). All statistical tests were 2-tailed, with a significance level of .05.
Results We invited 179 women with breast cancer but no pain to participate in this study, just before they began AI treatment. In total, 135 women were included in the analyses (Fig 1).
Pain and Pain Syndromes Related to AIs Pain Frequency During the study period, 77 (57%) of the 135 patients developed pain, with a Kaplan-Meier estimate of pain occurrence at 1 year of .59 (95% confidence interval = .51–.68). AI treatment was stopped because of pain in 12 cases. Table 1 shows the baseline characteristics and differences between women who developed pain and women who remained pain-free during the 12 months of study.
Five Types of Pain Syndromes We identified 5 primary types of pain syndromes (Fig 2): joint pain (the first pain to emerge) in 36%,
296
The Journal of Pain
Aromatase Inhibitors Pain-Related Syndromes
Figure 1. Flow chart for the recruitment of women with breast cancer free of pain, starting AI treatment, for the evaluation of AIrelated pain over a 1-year period. diffuse pain and tendinitis (both emerging at 3 months) both in 22%, neuropathic pain in 9% (emerging at 6 months on average), and mixed pain (mostly a combination of joint and tendinitis pain) in 11% (Table 2). Overall pain intensity did not differ significantly between the pain types (Table 2), but pain intensity was significantly higher in diffuse pain syndrome than in tendinitis and neuropathic pain.
Analgesics Prescribed for Patients With Pain During the Study Period During the study period, patients with pain were prescribed various analgesics and nonsteroidal antiinflammatory drugs: acetaminophen was prescribed to 48 women, oral nonsteroidal anti-inflammatory drugs to 9, topical nonsteroidal anti-inflammatory drugs to 7, and weak opioids (eg, tramadol, codeine) to 24.
Transient and Persistent Pain We also classified pain syndromes into 2 categories (Table 3): transient pain, for pain observed at only 1 visit (n = 26), and persistent pain, for pain observed on at least 2 consecutive visits (n = 34), in 60 of the 77 women with pain for whom it was possible to describe pain progression (for 17 women, it not possible to describe pain progression because they did not develop pain until the last visit or had to stop AIs because of severe pain). Diffuse pain was more frequently persistent (54.5%), whereas neuropathic pain was mostly transient (72.7%).
Pain Interference and Pain Domains of Patients With Each Type of Pain Daily pain interference scores were higher in the diffuse pain group than in the tendinitis group (Table 4). Indeed, the tendinitis group was chosen as the reference group because pain intensity and impact were lowest in this group.
Other Symptoms Occurring During Follow-Up The incidence of hot flushes increased significantly, from 8.3% at 1 month to 27.8% at 12 months, whereas the incidence of sleep disorders did not change significantly. Hot flushes and sleep disorders were not significantly related to pain during follow-up.
Biological Changes Associated With Pain Development During 1 Year of Follow-Up Pain development was not associated with any specific inflammatory, autoimmune, oxidative stress, or hormonal changes. Only low levels of sex binding protein were significantly (P < .01) associated with pain development during the study period, without correlation with a specific pain syndrome. One patient developed rheumatoid arthritis 3 months after the start of AI treatment.
Laroche et al
The Journal of Pain
297
Baseline Characteristics of All Patients (N = 135) and Comparison of Patients Developing AI-Related Pain (n = 77) and Patients Remaining Free of Pain (n = 57) Until the End of the Study Period
Table 1.
CHARACTERISTICS
TOTAL
PAIN
NO PAIN
Demographics (N) Age (y) (mean 6 StDev) BMI (StDev) Menopause duration (y) (StDev) HRT duration (y) (StDev) Type of surgery Tumorectomy Ganglion exeresis Previous chemotherapy Taxanes Anthracyclines Others Radiotherapy AI treatment allocation Anastrozole Letrozole Exemestane Psychological Factors Anxiety (HADS) Depression (HADS) Personality traits (TCI) Novelty seeking Harm avoidance Reward recompense Persistence Self-directedness Cooperativeness Transcendence Pain catastrophizing (PCS) Rumination Amplification Despair Quality of life (EORTC QLQ-BR23) Body image Sexual function Sexual enjoyment Future perspectives Systemic therapy side effects Breast symptoms Arm symptoms Upset by hair loss Sleep Not restorative (%) Fatigue (MFI20) Fatigue—general Fatigue—physical Fatigue—mental Fatigue—reduced activities Fatigue—motivation Genetics: SNPs (%) COMT: ValVal vs. ValMet 1 MetMet ORPM: AA vs. (AG 1 GG) CYP19: tandem repeats (7 1 8) vs. 10 – CYP 19A1 Hormone and binding proteins levels Estrone SBP Testosterone Delta4 androstenedione
135 61.5 6 7.1 25.28 6 5.32 11.65 6 7.60 8.03 6 5.50
77 61 6 7 25.91 6 6 11.57 6 8.44 7.53 6 5.60
57* 62.4 6 7.2 24.5 6 4.22 11.90 6 6.30 8.63 6 5.37
.17 .22 .50 .26
66 85 64 47 55 10 88
65 87 67 52 57 9 90
67 82 60 40 51 11 88
.83 .46 .35 .18 .47 .78 .73
32 67 .74
28.6 72 0
37 61 1.75
.32 .45 .27
7.18 6 3.78 3.61 6 3.02
7.65 6 3.65 3.56 6 2.70
6.53 6 3.92 3.75 6 3.40
.05 .81
8.02 6 2.54 8.55 6 4.55 10.06 6 2.25 3.26 6 1.37 19.16 6 3.78 20.37 6 2.83 6.27 6 3.43 14.30 6 11 5.89 6 4.44 2.94 6 2.61 5.45 6 4.83
8 6 2.41 8.77 6 4.51 10 6 2.30 3.50 6 1.31 18.75 6 3.80 20.40 6 2.61 6.61 6 3.73 14.65 6 10.50 5.95 6 4.30 3 6 2.60 5.70 6 4.80
8 6 2.74 8.26 6 4.66 10.10 6 2.24 2.94 6 1.4 19.72 6 3.77 20.31 6 3.14 5.89 6 2.95 13.86 6 11.80 5.84 6 4.71 2.87 6 2.71 5.14 6 5
.79 .41 .83 .02 .14 .58 .37 .56 .84 .62 .45
66 6 33 20 6 22.60 52 6 27.50 55.30 6 31.06 32 6 21.60 30 6 23 24.16 6 24.70 57.31 6 42.14
60.41 6 35.40 16.22 6 20.72 47 6 26.54 50.64 6 33 36 6 21.40 36 6 24 31 6 28 62 6 44.40
73.51 6 28.11 25.30 6 24.20 57.14 6 28.20 62 6 28 26.10 6 21 22.36 6 18.30 15.55 6 16 49.20 6 39
.04 .03 .23 .053 .0061 .0007 .003 .036
28 59.65 6 6.84 12.24 6 1.65 11.00 6 4.06 8.74 6 3.81 10.27 6 3.81 8.57 6 3.85
35 60.45 6 6 12.45 6 1.88 12.08 6 3.81 8.92 6 3.79 10.58 6 3.69 8.76 6 3.56
20 58.45 6 7.66 11.95 6 1.25 9.66 6 3.96 8.56 6 3.86 9.89 6 3.97 8.36 6 4.24
.24 .14 .13 .0006 .63 .34 .32
30 69 10
29 73 9.33
31.6 63 89
.78 .21 .82
33.57 6 20.27 72 6 32.36 .30 6 .17 1.08 6 .86
.9 .03 .16 .35
32.16 6 15.71 64.20 6 30 .28 6 .17 .97 6 .70
31 6 11.42 59 6 27 .26 6 .18 .90 6 .50
P VALUE
298
Aromatase Inhibitors Pain-Related Syndromes
The Journal of Pain
Table 1.
Continued CHARACTERISTICS
Inflammation (%) CRP
