Psycho-Oncology Psycho-Oncology (2010) Published online in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/pon.1860 Cognitive limitations associated with tamoxifen and aromatase inhibitors in employed breast cancer survivors Lynn M. Breckenridge1, Gina L. Bruns2, Briana L. Todd1 and Michael Feuerstein1,3 1 Uniformed Services University of the Health Sciences, Bethesda, MD, USA American University, Washington, DC, USA 3 Georgetown University Medical Center, Washington, DC, USA 2 * Correspondence to: Department of Medical and Clinical Psychology and Preventive Medicine and Biometrics, Uniformed Services University of the Health Sciences, 4301 Jones Bridge Road, Bethesda, MD 20814, USA. E-mail: mfeuerstein@usuhs.mil Received: 23 April 2010 Revised: 27 July 2010 Accepted: 26 August 2010 Abstract Objectives: Previous research has suggested that endocrine therapy is associated with cognitive limitations in breast cancer survivors (BCS); this study examined the relationship in employed BCS, an average of three years post-primary treatment. Methods: 77 BCS with past or current exposure to tamoxifen or aromatase inhibitors and 56 BCS with no history of endocrine therapy completed self-report measures of cognitive function, anxiety, depression, and fatigue as well as an online neurocognitive battery. Results: Exposure to endocrine therapy was not related to scores on the objective measures, but moderately related to perceived attentional problems at work (b 5 0.20; CI0.95 5 2.75, 0.25) and perceived cognitive functioning in overall life (b 5 0.17; CI0.95 5 0.33, 11.47) in excess of what could be explained by symptom burden measures. No differences were reported between groups on symptom burden measures. Symptoms of physical fatigue, depression, and anxiety were positively associated with self-report of general cognitive limitations (R2 change range: 0.28–0.37), and symptoms of depression and anxiety were positively associated with perceived cognitive limitations at work (R2 change range: 0.21–0.28). Discussion: Symptoms of depression, anxiety, and fatigue should be screened for and treated in BCS, as an approach to mitigating perceived cognitive limitations. However, healthcare providers should be aware that cognitive limitations exist in excess of what can be associated with symptom burden, and may be related to endocrine therapy and other cancer treatments. Copyright r 2010 John Wiley & Sons, Ltd. Keywords: breast cancer; adjuvant therapy; oncology; tamoxifen; aromatase inhibitors; cognitive function Introduction Most patients with breast cancer are offered adjuvant chemotherapy followed by adjuvant endocrine therapy for patients whose tumors are estrogen receptor positive [1,2]. Advanced treatments such as endocrine therapy result in substantial improvements in overall survival for women with breast cancer [3,4]. Despite this benefit, research indicates that the leading reason for non-adherence to adjuvant therapy among breast cancer survivors (BCS) is side effects of the medication [5]. Additionally, as the rate of survivorship has increased, there have been increasing reports of cognitive dysfunction and chronic fatigue [6,7]. A relationship between chemotherapy (cyclophosphamide/methotrexate/5-fluorouracil, 5-fluorouracil/ epidoxorubicin/cyclophosphamide (FEC), cyclophosphamide/thiotepa/carboplatin with FEC, cyclophosphamide/doxyrubicin/5-fluorouacil, and other regimens) Copyright r 2010 John Wiley & Sons, Ltd. and neurocognitive dysfunction has been reported in a number of studies [6,8–10]. For example, one investigation [11] reported that 25% of BCS had scores suggestive of cognitive decrements on a task of verbal learning. Further, in observational studies of BCS with a history of adjuvant chemotherapy, 17–50% are reported to have experienced cognitive limitations one to two years after treatment [12,13]. Recent evidence, however, suggests that there may not be a direct association between cancer treatments and cognitive deficits. Quesnel et al. [14] reported that in an observational study, some cognitive deficits existed in BCS regardless of whether or not they received either chemotherapy or radiotherapy. Additionally, in some breast cancer patients attention deficits exist prior to the initiation of treatment [15]. These results suggest that cognitive deficits seen in BCS may not be directly accounted for by exposure to cancer treatment. L. M. Breckenridge et al. Selective estrogen receptor modulators (SERMs) such as tamoxifen are frequently prescribed to patients with breast cancer and estrogen receptor positive (ER1) tumors. According to Castellon et al. [7], BCS who received tamoxifen exhibited decrements in verbal learning, language, visuospatial functioning, and visual memory, in excess of what was seen in patients who had chemotherapy without tamoxifen between two and five years after initial diagnosis and treatment. Further, crosssectional studies suggest that tamoxifen is associated with difficulties in memory, fluency, visuospatial ability, processing speed, and semantic memory [16,17]. However, Paganini-Hill and Clark [18] reported no significant effect of past or present tamoxifen use on neurocognitive battery scores (i.e., clock drawing, box copying, and narrative writing tasks) in patients less than five years from diagnosis. Hermelink et al. [19] reported that within the first year after diagnosis, BCS who had or had not had adjuvant endocrine therapy did not differ in scores on 12 cognitive tests (i.e., verbal memory, word fluency, digit span forward, digit span backward, and Trail Making Test A and B). Aromatase inhibitors (AIs) are becoming more frequently prescribed for patients with breast cancer [1]. Some studies have observed that AIs are associated with cognitive deficits, as indicated by lower verbal learning, visual memory, and working memory scores in postmenopausal women with early-stage breast cancer, a minimum of three months after initiation of adjuvant endocrine therapy [20]. However, one substudy of a randomized control trial (RCT) reported that AI use is not related to cognitive performance [21] and another RCT substudy [22] reported that after five years of AI therapy, BCS displayed fewer deficits than BCS taking tamoxifen. Another potential mechanism by which endocrine therapy drugs may be associated with cognitive deficits is through mood modulation. Lack of estrogen or failure of estrogen to bind with its receptors in the brain may be associated with increased depression, anxiety, and mood lability [23–25]. Between 7 and 46% of women with breast cancer report clinically significant levels of depression or anxiety within the first six months after diagnosis [26]. Depression and anxiety, as well as pain, distress, and fatigue have been demonstrated to have a negative relationship with cognitive performance in a number of studies of BCS [13,27]. However, Bender et al. reported that even after accounting for anxiety, depression, and fatigue in BCS on adjuvant therapy perceived cognitive limitations still existed and was predictive of poorer performance-based verbal learning and memory [13]. Studies show that 65–88% of BCS continue to work post-diagnosis [28–31]. Many BCS report that maintaining employment is important for their Copyright r 2010 John Wiley & Sons, Ltd. quality of life, including physical and mental health, as well as financial well-being [31]. However, a recent meta-analysis reported that BCS were more likely to be unemployed than healthy control participants [30]. It is often assumed that many women who stop working after a breast cancer diagnosis do so because of unspecified health-related reasons [31]. BCS are more likely than women with no history of cancer to have symptoms of depression, anxiety, and fatigue [32,33], and fatigue is more highly associated with workplace limitations in BCS than in women with no history of cancer [33], which may account for the lower employment rates. The current study was designed to assess whether occupationally active BCS with a history of endocrine therapy differ from occupationally active BCS with no history of endocrine therapy exposure on measures of perceived and performance-based cognitive function, as well as symptoms of fatigue, 7anxiety, and depression. We also assessed how symptoms of fatigue, depression, and anxiety are associated with perceived and performancebased cognitive function in occupationally active BCS. This information is important to BCS in making informed treatment and occupational decisions, to clinicians and occupational specialists in screening and treatment of occupationally active BCS with symptoms of cognitive dysfunction, fatigue, or distress, and to employers in understanding the needs and limitations of their BCS employees. Methods This study was part of a larger study of cognitive limitations and perceived work output in working BCS [34], which was approved by the Uniformed Services University of the Health Sciences (USUHS) Institutional Review Board. Data from 77 BCS who had adjuvant endocrine therapy (tamoxifen or AI) and 56 BCS who had not had adjuvant endocrine therapy were examined. Participants Participants were recruited by advertisements and fliers placed at cancer clinics and centers, as well as support groups and primary care centers across the United States. Newspaper ads, hospital bulletin boards, and websites were also utilized. Exclusion criteria included a diagnosis of dementia, brain injury, adult attention deficit hyperactivity disorder (ADHD), epilepsy, drug or alcohol abuse, or metastatic cancer. Potential participants who met the inclusion criteria were randomized to complete either the self-report measures or the neuropsychological test first. Psycho-Oncology (2010) DOI: 10.1002/pon Cognitive limitations associated with tamoxifen Procedure The participants supplied demographic information including health and work history, and completed a series of self-report measures on cognitive limitations, depression, fatigue, and work limitations. The health questionnaire included information about history of cancer, treatments, and current medications, as well as other factors that could impact mental status and well-being such as menopausal status as well as alcohol and caffeine use. The work history questionnaire asked for information about duration and type of work currently and prior to diagnosis. Upon completion of the study, participants were provided with a choice of either a cancer survivorship workbook or a stress reduction workbook. All participants were provided with an online list of resources and agencies specializing in psychooncology, and were given the option of receiving the results of the study upon its completion. While the participants provided sensitive information, all data were de-identified prior to analysis. Measures Hospital Anxiety and Depression Scale (HADS) [35] is a self-assessment scale for measuring depression and anxiety in a general medical population. The HADS consists of 14 items on two subscales, one measuring anxiety (A-scale) and the other measuring depression (D-scale). The scales are scored separately and have independent validity. The HADS has been demonstrated as a valid assessment of depression and anxiety in cancer patients [36] and has been used extensively for this purpose [37,38]. Multidimensional Fatigue Symptom InventoryShort Form (MFSI-SF) [39] is a 30-item self-report measure of fatigue. It assesses five symptom domains: general fatigue, physical fatigue, emotional fatigue, mental fatigue, and vigor. The MFSI has been validated with the breast cancer population and has been able to detect differences in fatigue in breast cancer patients related to different cycles of anthracycline-based chemotherapy, as well as detect differences in levels of chemotherapy-induced inflammatory mediators [40]. The MFSI-SF also was reported to have excellent reliability (a coefficients from 0.87 to 0.96) [39]. In an effort to ensure divergent validity (e.g., to avoid the redundancy of emotional and mental fatigue that is also being captured by the depression measure), the physical fatigue subscale of the MFSI-SF was used for our analyses. Visual Analog Scale of Pain (VASP) is a measure that consists of a single bar-line measure in which patients point to an area representing how much pain they feel. In studies of trauma patients [41] and emergency room patients [26,42], it was reported that change in VAS pain scores was Copyright r 2010 John Wiley & Sons, Ltd. strongly associated with patients’ verbal assessments of change in pain. Functional Assessment of Cancer Therapy Cognitive Scale Version Two (FACT-Cog) [43] is a 50question subjective measure designed to assess cognitive limitations and their effect on quality of general functioning in cancer survivors. The scale measures the frequency of positive and negative cognitive functioning events over the past seven days, based on self-report. The measure utilizes a five-point Likert-type scale (ranging from 0 5 never to 4 5 several times a day) to assess several different aspects of cognitive function. Lower scores are indicative of poorer perception of functioning. The FACT-Cog assesses a broad range of cognitive domains (not just those specifically related to work) and provides a multidimensional view of the cognitive deficits often experienced by patients with cancer [44]. The Perceived Cognitive Impairment (PCI) and Impact of Perceived Cognitive Impairments on Quality of Life (PCIQOL) subscales were used in this study as assessments of perceived cognitive performance and its impact on functioning and quality of life. Cognitive Symptom Checklist-Modified (CSC) [45] was developed for use as a patient checklist to assist in orienting physicians to patients’ cognitive difficulties. Unlike the FACT-Cog, which measures general functioning, the modified CSC is a specific measure pertaining to ability to function in an occupational setting, particularly in areas of work that require specific cognitive functions [46]. These areas of functioning include attention/concentration, memory, visual processes, and executive function. In a previous study [45], the number of items on the CSC was reduced from 100 to 59 based on a factor analysis (varimax rotation) that revealed a three-factor solution (working memory, executive functioning, and attention), followed by reduction of items to only those with a factor loading of 0.4 or higher on one of the three factors. This 59-item version of the CSC was used as a measure of perceived cognitive limitations encountered by BCS in their daily occupational duties, particularly in the domains of memory, attention, and executive function. Higher scores indicate lower functioning. CNS Vital Signs (CNSVS) is a remotely administered neurocognitive battery that objectively measures memory, psychomotor speed, reaction time, complex attention, and cognitive flexibility. The battery is composed of several well-established neuropsychological tests, such as finger tapping, symbol digit coding, the Stroop test, and the continuous performance test. The CNSVS takes approximately 30 min to complete. The test has been standardized with a normative sample and has been used to detect mild and moderate cognitive limitations in numerous neuropsychiatric patient groups including patients with mild and Psycho-Oncology (2010) DOI: 10.1002/pon L. M. Breckenridge et al. severe brain injury, early dementia, post-concussion syndrome, ADHD, and depression [47]. For this study, we utilized scaled scores of visual memory, verbal memory, and executive functioning to assess impairment in the domains of cognitive function that are most often reported in BCS [13]. Demographics, Work-Related, and Treatment Factors. Current job was measured by a self-report item in which participants chose between managerial, non-managerial, or self-employed. Income was recorded as a categorical variable with $10 000 to $20 000 increments, from $0–$10 000 per year up to $100 0001 per year. In the current study, history of radiation therapy or chemotherapy was selfreported as a ‘yes’ or ‘no’ rather than as a measurement of duration of treatment or time since treatment. Mental Status at Test Administration. Distraction, or whether or not the participant felt distracted during the test administration, was measured as a subjective binomial (yes/no) variable. Nicotine use prior to testing was assessed based on the information from the Behavioral Risk Factor Surveillance System Questionnaire. Caffeine use assessment was derived from a single item on the Caffeine Consumption Questionnaire, a measure commonly used to assess caffeine consumption, including use of various over-the counter substances that include caffeine. decision was made to combine groups of endocrine therapy users. One-way, univariate analyses of variance were conducted to determine whether there were between-group differences in anxiety, fatigue, and depression based on exposure to adjuvant endocrine therapy. A multivariate analysis of variance was conducted to detect between-group differences in perceived and performance-based cognitive function, after accounting for anxiety, fatigue, and depressive symptoms. Because of the large number of potential confounding factors relative to the sample size, a modified variable reduction technique was employed [48,49]. We conducted univariate regressions for 21 potential contributing variables, in relation to each of the outcome measures (cognitive impairment measures) for the purpose of eliminating potential confounders that were not likely to be significant in the current study. Any variable that reached a significance of po0.1 in relation to any outcome measure was retained for entry into the final multivariate regressions. After the final regression model was determined, a hierarchical multiple linear regression was computed for each of the outcome variables. Variables were entered by block based on expected significance (least to most). All analyses were conducted using the Statistical Package for Social Sciences (SPSS) version 16.0. Data analysis Power Analysis. Power analyses were run using nQuery Software. Assuming a two-tailed test with the current sample size (n 5 122) and the standard deviations from previous literature, we calculated that our study was adequately powered (0.80) to detect a difference as small as 2.5 points on the MFSI-SF, 1.8 points on the HADS-D, and 1.6 on the HADS-A. For the multiple regressions, separate power analyses were conducted for perceived and performance-based cognitive function measures. For self-report measures of cognitive function, we computed adequate power (0.80) to detect a moderate effect size of f2 5 0.07 with an a-level of 0.05. For the performance-based measures, we computed adequate power to detect a moderate effect size of f2 5 0.06. Prior to hypothesis testing, one-way, univariate analyses of variance were conducted to detect differences in observed and reported cognitive limitations between users of tamoxifen and aromatase inhibitors, and between past and present users. No differences were noted; however, the calculated and observed power of the analyses (0.48–0.63) indicated the possibility of a Type II error. Because power analyses indicated inadequate power if groups were split by past and present use or by history of tamoxifen, aromatase inhibitors, or both drugs, and based on results of previous studies, the Copyright r 2010 John Wiley & Sons, Ltd. Results Case definition Seventy-seven BCS who had previously used or were currently using adjuvant endocrine therapy (tamoxifen or AI) and 56 BCS who had never had adjuvant endocrine therapy were included. Of the adjuvant endocrine therapy (SERM/AI) group, 57% (n 5 44) reported use of tamoxifen (but not an AI) at some point during their treatment, 19% (n 5 15) reported use of an AI (but not tamoxifen) during their treatment, and 22% (n 5 17) reported use of both tamoxifen and an AI at some point during their treatment. Twenty-seven (35%) were currently taking tamoxifen, and 27 (35%) were currently using an AI. Demographics Table 1 presents demographic information for BCS exposed to adjuvant endocrine therapy (n 5 77) and BCS who had never been exposed to adjuvant endocrine therapy (n 5 56). Analyses of variance and chi-square analyses indicated no significant differences between the two groups on any demographic variables. The sample was primarily Caucasian (87.3%) and highly educated, with approximately 76% of participants possessing an Psycho-Oncology (2010) DOI: 10.1002/pon Cognitive limitations associated with tamoxifen Table 1. Participant characteristics SERM/AI group (n 5 77) Age p40 years old 41–50 years old 51–65 years old Mean (SD) Race Caucasian African American Asian American/Pacific Islander Other Ethnicity Hispanic Non-Hispanic Education Less than High School High School Graduate Some College Associates/Bachelors Some Graduate School Graduate Degree Marital status Single Cohabitating Married Divorced Widowed Current job characteristics Managerial Non-Managerial Self-Employed Primary occupation Clerical Sales Management/Administration Professional/Technical/Science Service Worker Years at current job 1 year or less 2–10 years 11–19 years 201 years Mean (SD) Annual income $10–19 000 $20–39 000 $40–59 000 $60–79 000 $80–99 000 $100 0001 No SERM/AI group (n 5 56) n % n % 21 20 28 44.93 (9.99) 30.4% 29.0% 40.6% 16 20 14 44.80 (8.88) 32.0% 40.0% 28.0% 68 6 3 0 88.3% 7.8% 3.9% 0.0% 48 2 4 2 85.7% 3.6% 7.1% 3.6% 2 69 2.8% 97.2% 2 46 4.2% 95.8% 1 6 8 27 5 30 1.3% 7.9% 10.4% 35.1% 6.5% 39.0% 0 1 14 14 6 21 0.0% 1.8% 25.0% 25.0% 10.7% 37.5% 12 5 54 5 1 15.6% 6.5% 70.1% 6.5% 1.3% 8 0 41 7 0 14.3% 0.0% 73.2% 12.5% 0.0% 20 53 4 25.97% 68.83% 5.19% 19 29 7 34.5% 52.7% 12.7% 7 6 22 37 3 9.3% 8.0% 29.3% 49.3% 4.0% 7 1 25 21 1 12.7% 1.8% 45.5% 38.2% 1.8% 10 48 10 7 7.48 (7.23) 13.3% 64.0% 13.3% 9.3% 11 36 10 4 7.13 (6.68) 21.6% 70.6% 19.6% 7.8% 2 4 7 16 9 39 2.6% 5.2% 9.1% 20.8% 11.7% 50.6% 1 1 8 6 10 30 1.8% 1.8% 14.3% 10.7% 17.9% 53.6% Note: Not all participants responded to all questions. No demographics were significant by group. associates degree or higher. The mean age for both the SERM/AI and non-SERM/AI group was almost 45 years old. Table 1 also presents the job characteristics of participants in the study. Analyses showed no significant differences between groups in work characteristics. Over 50% of participants reported an annual household income of over $100 000. Copyright r 2010 John Wiley & Sons, Ltd. For both groups, the majority of participants were in non-managerial jobs, approximately 69% in the SERM/AI group, and 53% in the nonSERM/AI group. For both groups, the majority of participants reported that they had worked at their job for an average of between two and ten years, with a median of six years at their current job. Psycho-Oncology (2010) DOI: 10.1002/pon L. M. Breckenridge et al. Cancer and treatment A summary of cancer and treatment characteristics is included in Table 2. There was no significant difference in time since treatment for the two groups, with an overall mean of 3.08 (SD 5 2.37; range for both groups 1–10) years since most recent primary treatment. The majority of participants had had multiple treatments for cancer. Menopausal status was not statistically significant between the two groups. Additionally, chemotherapy exposure did not differ significantly between the groups. Adjuvant endocrine therapy and symptom burden As shown in Table 3, there were no significant differences between groups in symptoms of fatigue, depression, or anxiety. We were, however, able to detect a significant difference between groups on the Cognitive Symptom Checklist-Attention subscale, and the FACT-Cog Perceived Cognitive Impairments subscale, with BCS exposed to adjuvant endocrine therapy reporting significantly greater cognitive limitations (po0.05). Perceived cognitive function Tables 4 and 5 display the results from the multivariate linear regressions for the perceived cognitive function outcome variables. In each regression, possible covariates were entered first (income, current job, and smoking today, followed by fatigue, depression, and anxiety), followed by the variable of interest (use of a SERM and/or AI). No demographic, medical/treatment history, or substance use variables were found to be significant for any measure of perceived cognitive function. Symptom Burden Factors in Relation to Cognitive Function. On the Cognitive Symptom Checklist, symptom burden factors accounted for 20.7–28.2% of the variance in the model for each subscale. Symptom burden factors are defined as the symptoms that may be confounds for cognitive function and include depression, anxiety, and Table 3. Fatigue, distress, and cognitive symptoms in BCS exposed or not exposed to adjuvant endocrine therapy MFSI fatigue HADS depression HADS anxiety CSC- memory CSC- attention CSC-executive function Fact-cog perceived cognitive Impairment (PCI) Fact-cog PCI quality of functioning CNSVS- visual memory CNSVS- verbal memory CNSVS- executive function SERM/AI (n 5 72) Mean (SE) No SERM/AI (n 5 50) Mean (SE) 5.70 4.48 7.75 8.50 5.36 3.75 52.37 5.53(0.59) 4.66 (0.40) 7.77(0.39) 9.02 (0.41) 4.35 (0.35) 4.89 (0.38) 55.33 (1.14) (0.48) (0.33) (0.32) (0.34) (0.27) (0.32) (0.94) 11.49 (0.32) 10.99 (0.39) 46.26 (0.59) 52.82 (0.56) 45.07 (0.89) 46.92(0.72) 51.12(0.67) 45.15 (1.08) SERM/AI (n 5 77) No SERM/AI (n 5 56)  5 po0.05. Table 2. Diagnosis and treatment SERM/AI (n 5 77) n Tumor location Right breast 39 Left breast 33 Both breasts 4 Tumor stage I 25 II 35 III 15 Treatment (at any time since cancer diagnosis) Chemotherapy 63 Radiation therapy 62 Surgery 75 Herceptin (Trastuzumab) 12 Tamoxifen (only) 45 Both Tamoxifen and Aromatase Inhibitor 17 Aromatase inhibitor (only) 15 Current use of psychotropic drugs Psychotropic drug users 27 Antidepressant users 25 Current use of other hormonal therapy Other hormonal therapy for cancer 5 Other estrogen/progesterone therapy 0 No SERM/AI (n 5 56) % n % 51.3% 43.4% 5.3% 29 26 1 51.8% 46.4% 1.8% 32.9% 46.1% 19.7% 22 27 7 39.3% 48.2% 12.5% 81.8% 80.5% 97.4% 15.6% 58.4% 22.1% 19.5% 47 36 54 6 0 0 0 83.9% 64.3% 96.4% 10.7% 0.0% 0.0% 0.0% 35.1% 32.5% 18 18 32.1% 32.1% 6.5% 0% 2 1 3.6% 1.8% n Time since primary treatment 1 year 26 2 years 10 3 years 13 4 years 7 5 years 8 6 years 4 7 years 1 8 years 0 9 years 3 10 years 3 Mean (SD) 3.31 Menopausal status Premenopausal 27 Post-menopausal 14 Currently undergoing 35 % n % 34.7% 13.3% 17.3% 9.3% 10.7% 5.3% 1.3% 0.0% 4.0% 4.0% (2.51) 19 13 6 6 4 0 3 1 1 1 35.2% 24.1% 11.1% 11.1% 7.4% 0.0% 5.6% 1.9% 1.9% 1.9% 2.77 (2.28) 35.5% 18.4% 46.1% 13 24 19 23.2% 42.9% 33.9% Not all participants answered all questions. Copyright r 2010 John Wiley & Sons, Ltd. Psycho-Oncology (2010) DOI: 10.1002/pon Cognitive limitations associated with tamoxifen Table 4. Factors related to perceived cognitive function at work, (n 5 114) CSC-Memory b CSC-Attention b 95% CI Step 1: Demographic factors Income 0.134 Current job 0.163 Step 2: Substance-related factors Smoking today 0.157 CSC-Executive Function b 95% CI 95% CI ( 1.470, 0.250) ( 3.575, 0.264) R2 5 0.035 0.097 0.044 ( 0.802, 0.265) ( 1.461, 0.919) R2 5 0.009 0.167 0.026 ( 1.142, 0.075) ( 1.546, 1.170) R2 5 0.027 ( 0.258, 2.937) R2 5 0.059 R2 Change 5 0.024 0.181 ( 0.041, 1.932) R2 5 0.041 R2 Change 5 0.031 0.116 ( 0.437, 1.836) R2 5 0.040 R2 Change 5 0.013 Step 3: Symptom burden factors MFSI fatigue 0.172 HADS depression 0.251 HADS anxiety 0.241 0.087 0.241 0.253 0.119 0.330 0.230 Step 4: Use of endocrine therapy SERM/AI 0.073 0.198 0.033 ( 0.022, 0.450) (0.110, 0.818) (0.117, 0.843) R2 5 0.320 R2 Change 5 0.261 ( 0.085, 0.212) (0.045, 0.500) (0.075, 0.542) R2 5 0.247 R2 Change 5 0.207 ( 2.885, 1.079) R2 5 0.325 R2 Change 5 0.005 ( 0.062, 0.270) (0.182, 0.679) (0.067, 0.578) R2 5 0.321 R2 Change 5 0.282 ( 2.748, 0.254) R2 5 0.239 R2 Change 5 0.038 ( 1.112, 1.686) R2 5 0.322 R2 Change 5 0.001 po0.05 po0.01. Table 5. Factors related to perceived cognitive function, (n 5 114) FACT-COG PCI Step 1: Demographic factors Income Current job FACT-COG PCI QOL b 95% CI b 0.052 0.121 ( 1.800, 3.3152) ( 2.044, 9.005) 0.167 0.067 95% CI ( 0.077, 1.164) ( 0.894, 1.875) R2 5 0.014 Step 2: Substance-related factors Smoking today 0.162 Step 3: Symptom burden factors MFSI fatigue HADS depression HADS anxiety 0.225 0.267 0.188 R2 5 0.027 ( 8.547, 0.642) R2 5 0.040 R2 Change 5 0.025 ( 1.470, 0.122) ( 2.416, 0.395) ( 2.106, 0.029) 0.158 0.255 0.320 0.207 R2 5 0.315 R Change 5 0.275 0.168 ( 0.385, ( 0.661, ( 0.537, 0.071) 0.190) 0.054) R2 5 0.416 R Change 5 0.365 2 Step 4: Use of endocrine therapy SERM/AI ( 2.121, 0.183) R2 5 0.051 R2 Change 5 0.024 2 (0.330, 11.474) R2 5 0.342 R2 Change 5 0.027 0.043 ( 0.941, 1.703) R2 5 0.418 R2 Change 5 0.002 po0.05; po0.01. fatigue in this study. Depression and anxiety scores independently accounted for significant amounts of the variance on all subscales of the CSC. On the FACT-Cog, symptom burden accounted for 27.5–36.5% of variance in the model for both of the subscales. Depression, anxiety, and fatigue were all independently statistically significant factors for FACT-Cog measures of perceived cognitive function in everyday life. Endocrine Therapy. Past or present use of a SERM and/or AI was significantly associated with reports of decreased attention on the Cognitive Symptom Checklist (CSC-A; b 5 0.198, po0.05), Copyright r 2010 John Wiley & Sons, Ltd. and with greater perceived cognitive impairment on the Functional Assessment of Cancer TherapyCognitive (FACT-Cog PCI; b 5 0.168, po0.05). Use of a SERM and/or AI accounted for 3.8% of the variance in attention on the CSC-A (R2 Change 5 0.038). The overall model accounted for 23.9% of the variance in CSC-A scores (po0.01). On the FACT-Cog PCI, use of a SERM and/or AI accounted for 3% of the variance in perceived cognitive impairments (R2 Change 5 0.027, po0.05). The overall model for the FACTCog PCI accounted for 34.2% of the variance in perceived cognitive impairments (po0.01). Psycho-Oncology (2010) DOI: 10.1002/pon L. M. Breckenridge et al. Performance-based cognitive function In our analyses of performance-based cognitive function outcome variables, possible covariates identified in our variable reduction analyses (distractions during test, chemotherapy, and time since caffeine) were entered first, followed by our variable of interest (history of SERM/AI use). Compared to measures of perceived cognitive function, there were noticeably fewer significant variables on measures of performance-based cognitive function. Our analyses indicated that use of a SERM or AI was not a significant predictor of performance-based cognitive function in occupationally active BCS, three or more years after primary treatment. Discussion This study indicates that no differences existed between BCS who were exposed to adjuvant endocrine therapy and a group of BCS who had never been exposed to endocrine therapy on a performance-based measure of cognitive function. On the other hand, these groups significantly differed on measures of perceived attention difficulties in the workplace (CSC-A) and perceived cognitive dysfunction in everyday life (FACT-Cog PCI), with those exposed having higher levels of perceived cognitive dysfunction. This study also indicated that depression, anxiety, and fatigue were not significantly different based on exposure to adjuvant endocrine therapy. These symptoms were, however, positively associated with greater perceived cognitive dysfunction in both groups of BCS and accounted for 20.7–28.2% of the variance. Our study suggests that history of endocrine therapy was significantly associated with two measures of perceived cognitive function, but none of the performance-based measures of cognitive function. Similar results have been reported in other studies of BCS and perceived cognitive function [37,50]. However, several other studies have been able to detect performance-based cognitive deficits associated with endocrine therapy [7,20,27]. It has been suggested that computerized cognitive tests such as the one used in this study may be advantageous for sensitivity to subtle cognitive change in BCS [51], but that does not appear to be the case in our study. Although clinician-administered neurocognitive batteries are considered the gold standard, there is some concern that certain standardized neuropsychological measures (clinician-administered or computerized) may not be specific or sensitive to perceived deficits experienced by working BCS [33,52]. It is also possible that the participants in other [20,27] studies may have had more pronounced, easily detectable deficits because they were currently Copyright r 2010 John Wiley & Sons, Ltd. taking estrogen inhibitors at the time of the study and had not had adequate time to recover from cognitive deficits related to treatment. In comparison, the participants in the current study included 54 (41%) BCS actively taking tamoxifen and/or aromatase inhibitors and 78 (59%) BCS who were previously exposed to adjuvant endocrine therapy. All participants were at least one-year postprimary treatment with an average of three years post-primary treatment. The participants in the current study were all maintaining a daily level of cognitive activity via occupational tasks. Previous studies have suggested that maintaining cognitive activity can reduce such cognitive limitations as cognitive slowing [53,54], and can increase rate of cognitive recovery from neurotrauma [55]. While no baseline measures were obtained in this study, a possibility of adaptation or recovery in the period following treatment for breast cancer is suggested. Impairments associated with breast cancer treatment have been reported to dissipate in the years following diagnosis and treatment [8], most likely due to BCS’ learned use of compensatory strategies to overcome deficits or due to neuroplasticity effects [51,56]. In a workshop of BCS, use of cognitive compensatory strategies such as avoidance of concurrent multiple task situations, frequent list-making, avoidance of high-pressure work situations through planning of workload, and recognizing increased need for sleep to overcome cognitive impairments at work was consistently endorsed by working BCS [51]. A longitudinal study of cognitive activity and change in cognitive impairments would help clarify any link between cognitive demands in the workplace and recovery of cognitive function in BCS. Potential limitations A limitation of this study is that it was a secondary analysis based on previously collected data from a cross-sectional study. There was limited information on estrogen receptor positive/negative status, compliance to treatment regime, time since endocrine therapy treatment, and timing of endocrine therapy treatment regime. Additionally, there was limited information on timing of menopause and because the original study was not stratified based on menopausal status there was a slight difference in menopausal status by group. This is most likely because some endocrine therapy drugs are only indicated for post-menopausal women or because these drugs can induce a change in menopausal status. However, analyses indicated that the difference was not statistically significant and that menopausal status was not a significant factor on any cognitive measure. Another limitation of this study was that it was not adequately powered to separate past from present users of tamoxifen and AIs or separate Psycho-Oncology (2010) DOI: 10.1002/pon Cognitive limitations associated with tamoxifen BCS with a history of AIs from BCS with a history of tamoxifen use. While these groups are often combined in the literature, this seems to be due to difficulty in obtaining the large number of participants needed for adequate power to study groups separately. Previous studies [20,57] suggest that length of time on endocrine therapy trial is not related to cognitive decrements, but other studies suggest that current users of SERMs exhibit more cognitive deficits than those who had used but were no longer using the drugs [18]. Further, while many previous studies of cognitive decrements have not differentiated between BCS exposed to AIs, tamoxifen, or both, a few studies have raised concerns that there may be a difference in the deficits seen with tamoxifen versus the deficits seen with aromatase inhibitors [22,58]. Other adequately powered studies have reported no statistically significant differences between cognitive deficits associated with either SERMs or AIs ([19,27]; n 5 101 and n 5 94, respectively). Based on these findings, we assumed that there would be minimal differences in cognitive deficits associated with tamoxifen and AIs, if any. However, this is considered a limitation of the current study and a factor to be considered in future studies. While this study was reliant on self-report of medical history, including exposure to adjuvant endocrine therapy and other cancer-related treatments, other studies indicate that three years postdiagnosis, self-report of medical history and cancer treatment by BCS can provide accurate and consistent information when compared to medical records [59,60]. Also, in comparison with demographics from recent cancer statistics [4,61], women of color and women of lower socioeconomic status are under-represented in this study, restricting the generalizability of our findings and possibly influencing the finding that cognitive deficits were not objectively observed. Failure to detect decrements in objective performance measures may be due to the fact that without baseline measurements, it is difficult to estimate whether or not functioning changed following exposure to hormonal replacement therapy, especially in higher functioning samples. Further, it is possible that perceived cognitive dysfunction was reported in this sample because in a highly educated BCS population with Internet access, priming regarding possible side effects of treatment may play a role in the reporting of perceived deficits [62]. Although the study was internet-based, a recent study indicated that over 75% of cancer survivors and their family members across various demographics access the Internet for health-related information [63]. Selection bias in Internet studies may not be as substantial as it was once considered [64]. The focus of the current study was on working BCS; therefore, the results have limited generalizability to BCS who are not occupationally active. Copyright r 2010 John Wiley & Sons, Ltd. Implications This study adds to the growing body of evidence of perceived cognitive dysfunction associated with adjuvant endocrine therapy. The meaning of this between-group difference, however, remains to be clarified. Clinicians must be aware that a substantial number of BCS (16–50%, average 1–2 years but up to 10 years after chemotherapy treatment) report cognitive dysfunction, and that these perceived deficits impact BCS’s ability to function in occupational settings and in everyday life [51]. This study also suggests that fatigue, anxiety, and depression are strongly related to cognitive function in BCS. At this time it is unclear whether depression, anxiety, and fatigue are impacting BCS’ cognitive function or whether BCS who are having difficulties in cognitive functioning are becoming depressed, anxious, and/or fatigued. While this study was not designed to determine a causal relationship between cognitive impairments and distress, it is likely that the combination of these factors may have a negative synergistic effect on the individual. Further, it is important to note that symptoms of fatigue, anxiety, and depression are related to but not fully explanatory of cognitive dysfunction in BCS. These results should by no means be interpreted to state that BCS’ reports of perceived cognitive dysfunction are purely psychological in nature. Reports of cognitive dysfunction by BCS should be taken seriously and treated with cognitive rehabilitation strategies when appropriate. However, because treatments for cognitive impairments are limited, it is especially important for clinicians to be aware of any symptoms of distress that may be impacting patients’ functioning. Early identification and treatment of fatigue and distress may be crucial in BCS returning to work and may prevent further exacerbation of cognitive impairments. Acknowledgements The opinions and assertions contained herein are the private views of the authors and are not to be construed as being official or as reflecting the views of the Uniformed Services University of the Health Sciences or the Department of Defense. References 1. Hind D, Ward S, De Nigris E et al. Hormonal therapies for early breast cancer: systematic review and economic evaluation. Health Technol Assess 2007;11(26). 2. Visvanathan K, Chlebowski R, Hurley P et al. American Society of Clinical Oncology clinical practice guideline update on the use of pharmacologic interventions including tamoxifen, raloxifene, and aromatase inhibition for breast cancer risk reduction. J Clin Oncol 2009; 27(19):3235–3258. Psycho-Oncology (2010) DOI: 10.1002/pon L. M. Breckenridge et al. 3. Winer E, Hudis C, Burstein H et al. American Society of Clinical Oncology technology assessment on the use of aromatase inhibitors as adjuvant therapy for postmenopausal women with hormone receptor-positive breast cancer: status report 2004. J Clin Oncol 2005; 23(3):619–629. 4. Jemal A, Thun MJ, Ries LAG et al. Annual report to the nation on the status of cancer, 1975–2005, featuring trends in lung cancer, tobacco use, and tobacco control. J Natl Cancer Inst 2008;100(23):1672–1694. 5. Grunfeld EA, Hunter MS, Sikka P, Mittal S. Adherence beliefs among breast cancer patients taking tamoxifen. Patient Educ Counsel 2005;59(1):97–102. 6. Wefel J, Witgert M, Meyers C. Neuropsychological sequelae of non-central nervous system cancer and cancer therapy. Neuropsychol Rev 2008;18(2):121–131. 7. Castellon S, Ganz P, Bower J et al. Neurocognitive performance in breast cancer survivors exposed to adjuvant chemotherapy and tamoxifen. J Clin Exp Neuropsychol 2004;26(7):955–969. 8. Falleti M, Sanfilippo A, Maruff P, Weih L, Phillips K. The nature and severity of cognitive impairment associated with adjuvant chemotherapy in women with breast cancer: a meta-analysis of the literature. Brain Cogn 2005;59(1):60–70. 9. Schagen S, van Dam F, Muller M et al. Cognitive deficits after postoperative adjuvant chemotherapy for breast carcinoma. Cancer 1999;85(3):640–650. 10. Wieneke M, Dienst E. Neuropsycological assessment of cognitive functioning following chemotherapy for breast cancer. Psycho-Oncology 1995;4(1):61–66. 11. Von Ah D, Harvison K, Monahan P et al. Cognitive function in breast cancer survivors compared to healthy age- and education-matched women. Clin Neuropsychol 2009;23(4):661–674. 12. Fan HGM, Houede-Tchen N, Yi Q et al. Fatigue, menopausal symptoms, and cognitive function in women after adjuvant chemotherapy for breast cancer: 1- and 2- year follow-up of a prospective controlled study. J Clin Oncol 2005;23(31):8025–8032. 13. Bender CM, Pacella ML, Sereika SM et al. What do perceived cognitive problems reflect? J Support Oncol 2008;6(5):238–242. 14. Quesnel C, Savard J, Ivers H. Cognitive impairments associated with breast cancer treatments: results from a longitudinal study. Breast Cancer Res Treat 2009; 116(1):113–123. 15. Ahles TA, Saykin AJ, McDonald BC et al. Cognitive function in breast cancer patients prior to adjuvant treatment. Breast Cancer Res Treat 2008;110(1):143–152. 16. Palmer J, Trotter T, Joy A, Carlson L. Cognitive effects of tamoxifen in pre-menopausal women with breast cancer compared to healthy controls. J Cancer Survivor 2008;2(4):275–282. 17. Eberling J, Wu C, Tong-Turnbeaugh R, Jagust W. Estrogen- and tamoxifen-associated effects on brain structure and function. Neuroimage 2004;21(1):364–371. 18. Paganini-Hill A, Clark L. Preliminary assessment of cognitive function in breast cancer patients treated with tamoxifen. Breast Cancer Res Treat 2000;64(2):165–176. 19. Hermelink K, Henschel V, Untch M et al. Short-term effects of treatment-induced hormonal changes on cognitive function in breast cancer patients: results of a multicenter, prospective, longitudinal study. Cancer 2008;113(9):2431–2439. 20. Bender CM, Sereika SM, Brufsky AM et al. Memory impairments with adjuvant anastrozole versus tamoxifen in women with early-stage breast cancer. Menopause 2007;14(6):995–998. 21. Jenkins V, Ambroisine L, Atkins L et al. Effects of anastrozole on cognitive performance in postmenopausal Copyright r 2010 John Wiley & Sons, Ltd. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. women: A randomised, double-blind chemoprevention trial (IBIS II). Lancet 2008;9(10):953–961. Ribi KE, Phillips KA, Sun Z et al. Cognitive function in menopausal women receiving adjuvant letrozole or tamoxifen in the Breast International Group (BIG) 1–98 trial. J Clin Oncol 2009;27(15S):510. Archer J. Relationship between estrogen, serotonin, and depression. Menopause 1999;1:71–78. Kase N. Yes or no on estrogen replacement therapy? A formulation for clinicians. Clin Obstetr Gynecol 1976; 19:895. McEwen B. Estrogen actions throughout the brain. Recent Prog Horm Res 2002;57:357–384. Gallagher E, Bijur P, Latimer C, Silver W. Reliability and validity of a visual analog scale for acute abdominal pain in the ED. Am J Emerg Med 2002; 20(4):287–290. Jenkins V, Shilling V, Fallowfield L, Howell A, Hutton S. Does hormone therapy for the treatment of breast cancer have a detrimental effect on memory and cognition? A pilot study. Psycho-Oncology 2004; 13(1):61–66. Bradley CJ, Bednarek HL. Employment patterns of long-term cancer survivors. Psycho-Oncology 2002; 11(3):188–198. Bouknight RR, Bradley CJ, Luo Z. Correlates of return to work for breast cancer survivors. J Clin Oncol 2006; 24(3):345–353. de Boer A, Taskila T, Ojajarvi A, van Dijk F, Verbeek J. Cancer survivors and unemployment: a meta-analysis and meta-regression. J Am Med Assoc 2009; 301(7):753–762. Maunsell E, Drolet M, Brisson J et al. Work situation after breast cancer: results from a population-based study. J Natl Cancer Inst 2004;96(24):1813–1822. Feuerstein M, Harrington C, Hansen J. Patterns of symptom burden in cancer survivors. In Cancer Symptom Science, Cleeland C, Fisch M, Dunn A (eds). Cambridge University Press: Cambridge, MA, 2008. Hansen J, Feuerstein M, Calvio L, Olsen C. Work productivity in breast cancer survivors. J Occup Environ Res 2008;50(7):777–784. Calvio L, Peugeot M, Bruns GL, Todd BL, Feuerstein M. Measures of cognitive function and work in occupationally active breast cancer survivors. J Occup Environ Med 2010;52(2):219–227. Zigmond A, Snaith R. The hospital anxiety and depression scale. Acta Psychiatr Scand 1983; 67(6):361–370. Hopwood P, Howell A, Maguire P. Screening for psychiatric morbidity in patients with advanced breast cancer: validation of two self-report questionnaires. Br J Cancer 1991;64(2):353–356. Poppelreuter M, Weis J, Kulz A et al. Cognitive dysfunction and subjective complaints of cancer patients: a cross-sectional study in a cancer rehabilitation center. Eur J Cancer 2004;40(1):43–49. Spiegel D, Giese-Davis J. Depression and cancer: mechanisms and disease progression. Biol Psychiatry 2003;54(3):269–282. Stein KD, Jacobsen PB, Blanchard CM, Thors C. Further validation of the multidimensional fatigue symptom inventory-short form. J Pain Symptom Manage 2004;27(1):14–23. Mills PJ, Parker B, Dimsdale JE, Sadler GR, Ancoli-Israel S. The relationship between fatigue and quality of life and inflammation during anthracycline-based chemotherapy in breast cancer. Biol Psychol 2005;69(1):85–96. Psycho-Oncology (2010) DOI: 10.1002/pon Cognitive limitations associated with tamoxifen 41. Todd K, Funk K, Funk J, Bonacci R. Clinical significance of reported changes in pain severity. Ann Emerg Med 1996;27(4):485–489. 42. Kelly A. Does the clinically significant difference in the Visual Analog Scale Pain scores vary with gender, age, or cause of pain? Acad Emerg Med 1998; 5(11):1086–1090. 43. Cella DF, Tulsky DS, Gray G et al. The functional assessment of cancer therapy scale: development and validation of the general measure. J Clin Oncol 1993; 11(3):570–579. 44. Jacobs S, Jacobsen P, Booth-Jones M, Wagner L, Anasetti C. Evaluation of the functional assessment of cancer therapy cognitive scale with hematopoietic stem cell transplant patients. J Pain Symptom Manage 2007; 33(1):13–23. 45. Feuerstein M, Hansen JA, Calvio LC, Johnson L, Ronquillo JG. Work productivity in brain tumor survivors. J Occup Environ Med 2007;49(7):803–811. 46. O’Hara C, Harrell M, Bellingrath E, Lisicia K. Cognitive Symptom Checklists—Clinician’s Guide. Psychological Assessment Resources, Inc: Odessa, FL, 1993. 47. Gualtieri CT, Johnson LG. Reliability and validity of a computerized neurocognitive test battery, CNS Vital Signs. Arch Clin Neuropsychol 2006;21(7):623–643. 48. Hosmer D, Lemeshow S. Applied Logistic Regression. Wiley: New York, 2000. 49. Tabachnick B, Fidell L. Using Multivariate Statistics (3rd edn). American Psychological Associations: Washington, DC, 1996. 50. Klepstad P, Hilton P, Moen J et al. Self-reports are not related to objective assessments of cognitive function and sedation in patients with cancer pain admitted to a palliative care unit. Palliative Med 2002;16(6):513–519. 51. Tannock IF, Ahles TA, Ganz PA, Van Dam FS. Cognitive impairment associated with chemotherapy for cancer: report of a workshop. J Clin Oncol 2004; 22(11):2233–2239. 52. Calvio L, Feuerstein M, Hansen J, Luff G. Cognitive limitations in occupationally active malignant brain tumor survivors. Occup Med 2009;59(6):406–412. 53. Bieliauskas L, Langenecker S, Graver C. Cognitive changes in retirement among senior surgeons Copyright r 2010 John Wiley & Sons, Ltd. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. (CCRASS): results from the CCRASS study. J Vasc Surg 2008;48(3):769. Carlson MC, Saczynski JS, Rebok GW et al. Exploring the effects of an ‘everyday’ activity program on executive function and memory in older adults: experience corps. Gerontologist 2008;48(6):793–801. Noblett KL, Swain RA. Pretraining enhances recovery from visuospatial deficit following cerebellar dentate nucleus lesion. Behav Neurosci 2003;117(4):785–798. Johnson M. Plasticity in the developing brain: implications for clincal practice. Dev Disability 2009;15:94–101. Shilling V, Jenkins V, Fallowfield L, Howell T. The effects of hormone therapy on cognition in breast cancer. J Steroid Chem Mol Biol 2003;86:405–412. Schlider CM, Seynaeve C, Bez LV et al. Effects of tamoxifen and exemestane on cognitive functioning of postmenopausal patients with breast cancer: results from the neuropsychological side study of the tamoxifen and exemestane adjuvant multinational trial. J Clin Oncol 2010;28(8):1294–1300. Phillips KA, Milne RL, Buys S et al. Agreement between self-reported breast cancer treatment and medical records in a population-based breast cancer family registry. J Clin Oncol 2005;23(21):4679–4686. Liu Y, Diamant AL, Thind A, Maly RC. Validity of self-reports of breast cancer treatment in low-income, medically underserved women with breast cancer. Breast Cancer Res Treat 2010;119(3):745–751. Jemal A, Murray T, Ward E et al. Cancer statistics, 2005. CA Cancer J Clin 2005;55(1):10–30. Schagen SB, Das E, van Dam FSAM. The influence of priming and pre-existing knowledge of chemotherapy-associated cognitive complaints on the reporting of such complaints in breast cancer patients. Psycho-Oncology 2009;18:674–678. Simon C, Schramm S. Cancer and the computerized family: towards a clinical ethics of ‘indirect’ internet use. Med Health Care Philos 2008;11(3):337–341. Whitehead LC. Methodological and ethical issues in Internet-mediated research in the field of health: an integrated review of the literature. Soc Sci Med 2007; 65(4):782–791. Psycho-Oncology (2010) DOI: 10.1002/pon