Cognitive Complaints After Breast Cancer Treatments
Cognitive Complaints After Breast Cancer Treatments
Eligible women were aged 21–65 years; were newly diagnosed with stage 0, I, II, or IIIA breast cancer; had completed primary breast cancer treatments within the past 3 months; had not yet started endocrine therapy; were available for 12-month follow-up; and were English language proficient. Exclusions/ineligibility included current or past disorder/disease of the central nervous system or medical condition impacting cognitive functioning; head trauma history with prolonged loss of consciousness; epilepsy, dementia, or learning disability; current or past psychotic-spectrum disorder or current major affective disorder; current substance abuse/dependence; daily tobacco and alcohol use; whole brain irradiation or surgery; prior cancer diagnosis or chemotherapy treatment; active autoimmune disorder; insulin-dependent diabetes; uncontrolled allergic condition or asthma; chronic use of oral steroid medication; and hormone therapy (estrogen, progestin compounds) other than vaginal estrogen. Exclusions related to hormones and inflammatory conditions were required due to other MBS aims focused on the biology of cognitive dysfunction.
Study recruitment occurred primarily through rapid case ascertainment of stage-eligible patients identified through the Los Angeles County Surveillance Epidemiology and End Results (SEER) registry from selected collaborating physicians and hospitals. Prior publications include details of subject identification, screening, recruitment, and enrollment. Figure 1 shows the overall study design. Preliminary MBS data describing some baseline symptoms and inflammatory markers are reported elsewhere, and preliminary longitudinal data related to inflammatory markers, subjective complaints, and brain imaging are described in another recent report. The research was approved by the UCLA Institutional Review Board, and all participants provided written informed consent.
(Enlarge Image)
Figure 1.
Overview of the UCLA Mind Body Study, a prospective longitudinal cohort study designed to examine the impact of endocrine therapy for breast cancer on cognitive functioning. DX = breast cancer diagnosis; NP = neuropsychological; PET = positron emission tomography; rx = treatment.
Information was obtained from self-report and medical record abstraction. The Beck Depression Inventory–II (BDI-II) was used to assess depressive symptoms during the preceding two weeks, with higher scores indicating more severe symptoms. We also calculated two subscales of the BDI-II with 14 items representing the cognitive/affective component (eg, sadness, pessimism, guilt, loss of pleasure) and seven items focused on the somatic aspects of depression (eg, loss of energy, changes in sleep or appetite, loss of interest in sex).
Cognitive complaints were assessed with the Patient's Assessment of Own Functioning Inventory (PAOFI), a self-report instrument with prior evidence for correlation with NP test changes in cancer patient samples. The PAOFI contains 33 questions and is divided into four subscales: memory (10 questions), higher-level cognition (HLC) measuring executive function (nine questions), language and communication (nine questions), and motor-sensory perception (five questions). Each PAOFI complaint item was rated on a 6-point Likert scale from 1 ("almost always") to 6 ("almost never"). We scored the PAOFI by assigning a score of 1 to each item rated as "almost always," "very often," and "fairly often," whereas items rated "once in awhile," "very infrequently," or "almost never" were assigned a score of 0. Thus, the total score for the PAOFI ranges from 0 to 33, and each subscale from 0 to the total number of items in the subscale (ie, the memory subscale ranges from 0 to 10 and the HLC subscale ranges from 0 to 9). As normative data for the PAOFI were lacking, in this report we include comparative data from a healthy control population of women without breast cancer obtained from a concurrent study at the University of California, San Diego (UCSD). In that study, healthy volunteers were recruited to match a prospective, longitudinal study of cognitive function in women receiving breast cancer adjuvant chemotherapy (S. Ancoli-Israel, NCI R01 CA112035). Demographic data, body mass index (BMI), depression and PAOFI scores were available from the 63 healthy women enrolled in the UCSD study.
Fatigue was assessed using the Multidimensional Fatigue Symptom Inventory–Short Form (MFSI-SF), a 30-item questionnaire that includes five subscales: general, physical, emotional, mental fatigue, and vigor.
NP testing was conducted by a trained technician, closely supervised by a licensed clinical neuropsychologist using procedures detailed in an earlier report. The 120-minute test battery is described in detail in Supplementary Table 1. NP test scores were standardized to z scores, with positive scores indicating outcomes better than age-corrected normative performance scores with a mean of 0 and standard deviation of 1, and negative scores reflecting lower than normative performance (see Supplementary Table 1). These scores were then used to create NP test domains based upon prior factor-analytic studies of larger NP data sets, as well as groupings used in other studies with this population. We identified a priori variables most salient for the cognitive domain being studied. An estimate of full-scale IQ, the Wechsler Test of Adult Reading (WTAR), was administered only at baseline.
PAOFI total and subscale scores from the MBS sample were compared to data from the UCSD healthy female controls, using χ or t tests as appropriate, permitting definition of a normal range for each PAOFI subscale for comparison with the MBS sample. Only two PAOFI subscales (memory and HLC) showed statistically significant differences; these were examined in subsequent evaluations. Bivariate tables were constructed to identify demographic and treatment characteristics to include as predictors, as well as NP domains, adjusted for age and IQ, to include in multivariable models. MBS patients with PAOFI subscale scores >1 SD above the mean for the healthy controls (high complaints) were compared with patients whose scores were ≤1 SD above the mean (within-normal-limits levels of complaints) using χ or t tests. The PAOFI scores were log transformed [log(x+1)] due to nonnormality, in the multivariable linear regression analyses. Multivariable models included age and IQ, as well as statistically significant variables (P < .10) identified in the bivariate analyses. Final models were run with and without BDI-II scores to control for depressive symptoms, excluding the BDI-II item that asked about concentration from the total score. Additional multivariable models were run with the BDI-II somatic and cognitive/affective subscales instead of the BDI-II composite score. Finally, a stepwise linear regression was also performed to assess whether there was any redundancy in predictors selected for other reasons and to summarize the relative contribution of each factor to the overall model variance. For these regressions, the BDI-II somatic and cognitive/affective subscales were entered separately to discern any differential associations. All statistical tests were two-sided and all analyses were conducted in SAS 9.3 (Cary, NC).
Patients and Methods
Study Participants, Recruitment, and Design
Eligible women were aged 21–65 years; were newly diagnosed with stage 0, I, II, or IIIA breast cancer; had completed primary breast cancer treatments within the past 3 months; had not yet started endocrine therapy; were available for 12-month follow-up; and were English language proficient. Exclusions/ineligibility included current or past disorder/disease of the central nervous system or medical condition impacting cognitive functioning; head trauma history with prolonged loss of consciousness; epilepsy, dementia, or learning disability; current or past psychotic-spectrum disorder or current major affective disorder; current substance abuse/dependence; daily tobacco and alcohol use; whole brain irradiation or surgery; prior cancer diagnosis or chemotherapy treatment; active autoimmune disorder; insulin-dependent diabetes; uncontrolled allergic condition or asthma; chronic use of oral steroid medication; and hormone therapy (estrogen, progestin compounds) other than vaginal estrogen. Exclusions related to hormones and inflammatory conditions were required due to other MBS aims focused on the biology of cognitive dysfunction.
Study recruitment occurred primarily through rapid case ascertainment of stage-eligible patients identified through the Los Angeles County Surveillance Epidemiology and End Results (SEER) registry from selected collaborating physicians and hospitals. Prior publications include details of subject identification, screening, recruitment, and enrollment. Figure 1 shows the overall study design. Preliminary MBS data describing some baseline symptoms and inflammatory markers are reported elsewhere, and preliminary longitudinal data related to inflammatory markers, subjective complaints, and brain imaging are described in another recent report. The research was approved by the UCLA Institutional Review Board, and all participants provided written informed consent.
(Enlarge Image)
Figure 1.
Overview of the UCLA Mind Body Study, a prospective longitudinal cohort study designed to examine the impact of endocrine therapy for breast cancer on cognitive functioning. DX = breast cancer diagnosis; NP = neuropsychological; PET = positron emission tomography; rx = treatment.
Demographics, Clinical Information, and Symptoms
Information was obtained from self-report and medical record abstraction. The Beck Depression Inventory–II (BDI-II) was used to assess depressive symptoms during the preceding two weeks, with higher scores indicating more severe symptoms. We also calculated two subscales of the BDI-II with 14 items representing the cognitive/affective component (eg, sadness, pessimism, guilt, loss of pleasure) and seven items focused on the somatic aspects of depression (eg, loss of energy, changes in sleep or appetite, loss of interest in sex).
Cognitive complaints were assessed with the Patient's Assessment of Own Functioning Inventory (PAOFI), a self-report instrument with prior evidence for correlation with NP test changes in cancer patient samples. The PAOFI contains 33 questions and is divided into four subscales: memory (10 questions), higher-level cognition (HLC) measuring executive function (nine questions), language and communication (nine questions), and motor-sensory perception (five questions). Each PAOFI complaint item was rated on a 6-point Likert scale from 1 ("almost always") to 6 ("almost never"). We scored the PAOFI by assigning a score of 1 to each item rated as "almost always," "very often," and "fairly often," whereas items rated "once in awhile," "very infrequently," or "almost never" were assigned a score of 0. Thus, the total score for the PAOFI ranges from 0 to 33, and each subscale from 0 to the total number of items in the subscale (ie, the memory subscale ranges from 0 to 10 and the HLC subscale ranges from 0 to 9). As normative data for the PAOFI were lacking, in this report we include comparative data from a healthy control population of women without breast cancer obtained from a concurrent study at the University of California, San Diego (UCSD). In that study, healthy volunteers were recruited to match a prospective, longitudinal study of cognitive function in women receiving breast cancer adjuvant chemotherapy (S. Ancoli-Israel, NCI R01 CA112035). Demographic data, body mass index (BMI), depression and PAOFI scores were available from the 63 healthy women enrolled in the UCSD study.
Fatigue was assessed using the Multidimensional Fatigue Symptom Inventory–Short Form (MFSI-SF), a 30-item questionnaire that includes five subscales: general, physical, emotional, mental fatigue, and vigor.
Neuropsychological Assessments
NP testing was conducted by a trained technician, closely supervised by a licensed clinical neuropsychologist using procedures detailed in an earlier report. The 120-minute test battery is described in detail in Supplementary Table 1. NP test scores were standardized to z scores, with positive scores indicating outcomes better than age-corrected normative performance scores with a mean of 0 and standard deviation of 1, and negative scores reflecting lower than normative performance (see Supplementary Table 1). These scores were then used to create NP test domains based upon prior factor-analytic studies of larger NP data sets, as well as groupings used in other studies with this population. We identified a priori variables most salient for the cognitive domain being studied. An estimate of full-scale IQ, the Wechsler Test of Adult Reading (WTAR), was administered only at baseline.
Statistical Analyses
PAOFI total and subscale scores from the MBS sample were compared to data from the UCSD healthy female controls, using χ or t tests as appropriate, permitting definition of a normal range for each PAOFI subscale for comparison with the MBS sample. Only two PAOFI subscales (memory and HLC) showed statistically significant differences; these were examined in subsequent evaluations. Bivariate tables were constructed to identify demographic and treatment characteristics to include as predictors, as well as NP domains, adjusted for age and IQ, to include in multivariable models. MBS patients with PAOFI subscale scores >1 SD above the mean for the healthy controls (high complaints) were compared with patients whose scores were ≤1 SD above the mean (within-normal-limits levels of complaints) using χ or t tests. The PAOFI scores were log transformed [log(x+1)] due to nonnormality, in the multivariable linear regression analyses. Multivariable models included age and IQ, as well as statistically significant variables (P < .10) identified in the bivariate analyses. Final models were run with and without BDI-II scores to control for depressive symptoms, excluding the BDI-II item that asked about concentration from the total score. Additional multivariable models were run with the BDI-II somatic and cognitive/affective subscales instead of the BDI-II composite score. Finally, a stepwise linear regression was also performed to assess whether there was any redundancy in predictors selected for other reasons and to summarize the relative contribution of each factor to the overall model variance. For these regressions, the BDI-II somatic and cognitive/affective subscales were entered separately to discern any differential associations. All statistical tests were two-sided and all analyses were conducted in SAS 9.3 (Cary, NC).
Source...