This study examined the influence of prior treatment on the course of cognitive functioning in breast cancer survivors. Changes in cognitive functioning over time were compared in breast cancer survivors treated with chemotherapy plus radiotherapy, breast cancer survivors treated with radiotherapy only, and women with no history of cancer.
Stage 0-II breast cancer patients treated with chemotherapy plus radiotherapy (CT group; n = 62) or radiotherapy only (RT group; n = 67) completed neuropsychological assessments 6 months after completing treatment and again 36 months later. Women with no history of cancer (NC group; n = 184) were assessed over a similar interval.
A significant group × time effect was found for processing speed (P = .009) that reflected a tendency for the NC group but not the RT and CT groups to improve over time. There was also a significant group effect for executive functioning (P = .006) that reflected the NC group performing better than the CT and RT groups. Additional analyses found the administration of hormonal therapy was not associated with change over time in cognitive performance.
As survivors of breast cancer are living longer post-treatment, there is an increasing need to understand the long-term effects of adjuvant therapies. Numerous studies have examined the effects of cancer treatments on cognitive functioning among breast cancer survivors. Results of meta-analyses suggest adjuvant chemotherapy is associated with cognitive problems as measured by standardized neuropsychological tests.1, 2 However, most of the studies included in these analyses have been cross-sectional. Therefore, they do not provide information about changes over time in cognitive functioning among breast cancer survivors. Longitudinal studies are necessary to understand the course of post-treatment cognitive performance through survivorship. Of particular importance are studies that follow patients for extended periods of time after treatment completion.
Two longitudinal studies can be identified that followed breast cancer patients treated with chemotherapy for at least 18 months after treatment completion, with assessments conducted at standardized time points. Ahles and colleagues3 evaluated cognitive functioning in breast cancer patients before the start of chemotherapy and again at 1, 6, and 18 months post-treatment. These patients were compared with breast cancer patients who did not receive chemotherapy and women without cancer assessed at matched intervals. There was a significant group by time interaction for measures of verbal ability (P = .01) showing that patients who did not receive chemotherapy and women without cancer demonstrated improved performance across time, whereas patients treated with chemotherapy did not demonstrate improved performance until 6 months post-treatment. Additional analyses found that those in the no chemotherapy group who were prescribed tamoxifen performed worse on measures of processing speed (P = .016), verbal memory (P = .018), and verbal ability (P = .023) compared with the noncancer group. Mar Fan and colleagues4 assessed patients toward the end of chemotherapy (ie, after at least 3 cycles) and at 1-year and 2-year follow-ups and compared their performances to those of women without cancer. At the 2-year follow-up, the chemotherapy-treated patients performed worse on a measure of executive functioning than the women without cancer (P = .048). No differences were found based on whether the chemotherapy-treated patients received hormonal therapy.
The studies described above do not yield a consistent picture of the long-term course of cognitive functioning among breast cancer survivors. Furthermore, neither examined cognitive performance beyond 2 years post-treatment. The current study addresses these limitations by comparing changes in cognitive functioning over a 3-year period after treatment completion in 2 groups of patients (those who received chemotherapy plus radiotherapy and those who received radiotherapy without chemotherapy) and a comparable interval for women with no history of cancer. The aim was to determine whether there are differences in change over time in cognition among women who received chemotherapy plus radiotherapy (CT), radiotherapy without chemotherapy (RT), and noncancer controls (NC). We previously reported the results of cross-sectional comparisons conducted 6 months after treatment completion.5, 6 Findings showed that chemotherapy-treated patients performed worse than noncancer controls on measures of nonverbal memory and executive functioning.6 In addition, radiotherapy-treated patients performed worse than noncancer controls on measures of attention and executive functioning.6 Comparisons were not made between those treated with chemotherapy versus radiotherapy. Additional findings indicated there were no differences in cognitive functioning based on whether patients were receiving tamoxifen.6 In light of these and other findings,3, 4 we hypothesized that women who received chemotherapy would demonstrate worse cognitive functioning over a 36-month post-treatment follow-up period than women with no history of cancer. In light of our previous finding6 and inconsistent findings in the literature,3, 4, 7 no hypotheses were offered about differences in cognitive functioning in radiotherapy-treated patients relative to chemotherapy-treated patients and noncancer participants. The current study also explored the possibility that hormone therapy in combination with chemotherapy and radiotherapy might be associated with changes in cognitive functioning over time.
MATERIALS AND METHODS
Participants and Procedures
Data for the current report were drawn from a larger study of quality of life in women being treated for early stage breast cancer at the Moffitt Cancer Center and the University of Kentucky Chandler Medical Center.8 Patient eligibility criteria for the larger study were: diagnosed with stage 0-II breast cancer; scheduled to receive chemotherapy or radiotherapy; no other history of cancer besides basal cell skin carcinoma; no prior history of chemotherapy or radiotherapy; and no conditions in which fatigue is a prominent symptom (eg, acquired immunodeficiency syndrome, multiple sclerosis, fibromyalgia). Additional eligibility criteria for the current report were that the patients completed the cognitive assessments and had no recurrent breast cancer or new primary cancer. This study was approved by the University of South Florida and University of Kentucky institutional review boards, and informed consent was obtained from all participants.
Each patient who enrolled in the study (described below) was matched with a noncancer control (as described in detail elsewhere).8 Noncancer participants were women who were within 5 years of age and resided in the same Zip Code as their patient match. Additional eligibility criteria for the noncancer controls were: no history of cancer (besides basal cell skin carcinoma) or other potentially life-threatening diseases; and no conditions in which fatigue is a prominent symptom (as described above). Cognitive performance was assessed approximately 6 months after completion of radiotherapy (T1) and approximately 36 months later (T2), with noncancer participants assessed at comparable time intervals.
T1 cognitive assessments were completed by 449 participants (CT = 96, RT = 113, NC = 240). At T2, 133 participants declined participation or could not be contacted (CT = 33, RT = 44, NC = 56), and 3 were ineligible for T2 because of a recurrence (RT = 1) or second breast cancer diagnosis (CT = 1, RT = 1). Thus, this report is based on 313 participants (CT = 62, RT = 67, NC = 184) who had cognitive data at both T1 and T2.
Demographic characteristics and T1 cognitive domain scores (described below) of the 133 participants who withdrew or could not be contacted were compared with the 313 participants who completed T1 and T2. Among patients, the 129 patients with complete data had more years of education than the 77 patients who did not complete both assessments (P < .001). Also, patients who completed both assessments performed better on the verbal memory domain at T1 than those who did not complete both assessments (P = .011). There were no significant (P > .05) differences between the 56 controls who did not complete both assessments and the 184 controls who completed both assessments on any demographic variables. Controls who did not complete both assessments performed better at T1 on the attention cognitive domain than those with complete data (P = .041).
Demographic and clinical data
Demographic data were obtained through use of a standard self-report questionnaire. Variables assessed included age, menopausal status, race, marital status, annual household income, and educational level. Medical charts of patients were reviewed at the completion of study participation to obtain information about disease and treatment characteristics. Variables assessed included disease stage, type of breast surgery, and use of hormonal therapy at T1. Number of chemotherapy cycles and chemotherapy regimen (CT group) and number of radiation treatments and cumulative radiation doses were also recorded.
Cognitive performance was assessed using a battery of neuropsychological tests administered by doctoral students in psychology trained to conduct these evaluations. Tests were selected based on their reliability and validity, and the availability of published norms for each measure. Raw test scores were converted to standardized t scores (mean = 50, standard deviation = 10) based on published normative data. In addition to overall intellectual ability, 5 domains of cognitive functioning were assessed: verbal memory, attention, executive functioning, processing speed, and nonverbal memory.
Overall intellectual ability.
The National Adult Reading Test (NART)9 was used to estimate overall intellectual ability. The NART contains 50 irregular words that cannot be easily decoded phonetically. Previous work has shown that performance on the NART is highly correlated with general intelligence (factor g) as measured by the Wechsler Adult Intelligence Scale (WAIS)-Revised.9
This domain was comprised of 3 tests. For trial 1 of the Color Trails Test,10 participants are required to connect colored targets in numerical order as quickly as possible. For the Digit Span subtest of the WAIS-III,11 participants are read a list of an increasingly long series of number and asked to repeat the numbers. Next, the examiner reads sequences of numbers and the participant is required to repeat them in reverse order. The Spatial Span subtest of the WAIS-III11 is a visual analog of the Digit Span test.
This domain was comprised of 3 tests. For the Digit Symbol Coding subtest of the WAIS III,11 participants are required to match numbers with geometric symbols according to a specific key code in a speeded fashion. For trial 2 of the Color Trails Test,10 participants are required to connect colored number targets in numerical order in an alternating color as quickly as possible. For the Controlled Oral Word Association test (COWAT),12 participants are required to produce words beginning with a target letter over a series of three 1-minute trials.
This domain was comprised of 3 measures from the Visual Reproduction test of the Wechsler Memory Scales-III. In this test, cards with 5 novel geometric designs are presented 1 at a time for 10 seconds. Participants are then required to draw from memory the design they were just shown (immediate free recall). Delayed free recall and recognition recall were also measured after a 30-minute delay.
This domain was comprised of 2 measures from the Ruff 2 & 7 Test,13 which requires participants to read through 20 trials of numbers and letters and to mark through all occurrences of the numbers 2 and 7. Scores were derived for total speed and total accuracy in completing this task.
This domain was comprised of 3 subtests of the California Verbal Learning Test (CVLT).14 The CVLT consists of 5 learning trials of a 16-word list. Participants are read a word list and are then required to state as many words from the list as they can remember. After a 20-minute delay during which the participant is engaged in nonverbal tasks, the participant is asked to recall the list. The immediate recall, long-delayed recall, and recognition scores were included in this domain.
To examine longitudinal changes in cognitive performance across the 2 time points, repeated measures analyses of variance (ANOVAs) were conducted using SAS version 9.1 (SAS Institute, Cary, NC). The initial analysis consisted of a 3 (group: NC, RT, CT) by 2 (time: time 1, time 2) ANOVA for each cognitive domain in which the cognitive domain score represented the mean of the measures for that domain. Of principal interest was the group × time interaction effect, which assesses whether the groups changed at comparable rates over the follow-up period. If a group × time interaction was not significant, the corresponding group effect was examined.
Additional analyses were undertaken to determine whether prescription of hormonal therapy to patients at T1 was related to changes in cognitive performance. NC participants were removed from this analysis. For each of the cognitive domains, the interaction among hormone therapy (prescribed vs not prescribed), group (RT vs CT), and time (T1 vs T2) was evaluated.
A significance level of .01 (2-tailed) was used as the criterion for statistical significance for the group × time interactions and group effects. A .05 significance level (2-tailed) was the criterion for subsequent analyses in the case of significant group × time interaction or group effect. The sample size of 313 participants and P value of .01 allowed .80 power to detect a group × time interaction of d = 0.29 and a group effect of d = 0.44, small to medium-sized effects.
Because only 8 patients received chemotherapy in the absence of radiotherapy, those participants were excluded from the current analysis. Demographic characteristics of all participants are listed in Table 1. The CT group was younger than the RT and NC groups (P < .001), and the 2 patient groups had higher T1 NART scores than the NC group (P = .007). There were also differences in the T1 and T2 intervals, with the RT group having a longer interval than the CT or NC groups (P = .013). Thus, age, T1 NART scores, and time between assessments were used as covariates in all analyses.
Table 1. Demographic Characteristics of Study Participants
Clinical characteristics of the RT and CT groups are presented in Table 2. The CT group was more likely to have undergone mastectomy, to have more advanced disease, and to have a longer duration from cancer diagnosis to T1 than the RT group. The RT group was more likely to have received hormonal treatment than the CT group.
Table 2. Clinical Characteristics of Patient Participants
T score means and standard errors for each cognitive domain (adjusted for age, NART, and time from T1 to T2) for each group at T1 and T2 are presented in Table 3. There was a significant group × time interaction for the processing speed domain (P = .009) (see Fig. 1). Examining the effect of time for each group separately, there was a trend for the NC group to improve over time (P = .079), whereas there was no change over time in the CT group (P = .920) or RT group (P = .829). At T1, there were no differences between groups (P > .05). At T2, both the RT and CT groups performed worse than the NC group (P < .05).
Table 3. Adjusted Cognitive Domain T Score Means and Standard Errors by Groupa
T score means are adjusted for age, T1 National Adult Reading Test scores, and time from T1 to T2 assessments.
Significant group × time interactions were not observed for the verbal memory, executive functioning, attention, or nonverbal memory domains (P > .01). Examination of group effects for these domains yielded significant results only for the executive functioning domain (P = .006) (see Fig. 2). At both T1 and T2, the RT and CT groups performed worse than the NC group (P < .05). There were no differences between the RT and CT groups at either T1 or T2 (P > .05).
Analyses were undertaken to determine whether administration of hormonal therapy was related to changes in cognitive performance. There was no significant 3-way (hormone therapy × group × time) interaction, no significant 2-way (hormone therapy × group or hormone therapy × time) interaction, and no main effect for hormone therapy for any of the cognitive domains (P > .01). When patients who received a hormone therapy other than tamoxifen were excluded from analyses (n = 10), the same pattern of nonsignificant results was obtained (P > .01).
Findings provided limited support for the hypothesis that women treated with chemotherapy for breast cancer would demonstrate worse cognitive functioning over a 36-month post-treatment follow-up period relative to women with no history of cancer. For the processing speed domain, change in cognitive functioning varied as a function of group in the expected direction. Specifically, there was a trend toward performance improving over time in the NC group but not in the CT group. This pattern suggests that a practice effect (ie, improved performance because of prior test exposure) did not occur in women treated with chemotherapy. For the executive functioning domain, there was a main effect for treatment group, indicating the CT group performed worse than the NC group.
These findings are partially consistent with prior research. Although 1 longitudinal study found no differences between breast cancer patients receiving chemotherapy and noncancer participants 1 year post-treatment,15 other studies that assessed patients 6 months or longer post-treatment have observed differences in cognitive functioning between these 2 groups.3, 4, 16 Specifically, chemotherapy-treated patients performed worse than noncancer controls in the domains of processing speed,3 executive functioning,4, 16 and verbal ability.3 Although the current study did not include a verbal ability domain, a test of verbal fluency (ie, COWAT) was included in the executive functioning domain, for which a group effect was observed. Together, these results suggest that processing speed (the ability to quickly perform automatic cognitive tasks while under pressure to maintain concentration) and executive functioning (the ability to shift cognitive sets and solve novel problems) may be the domains most affected by chemotherapy.
Hypotheses were not offered about the performance of women who received radiotherapy without chemotherapy relative to the other 2 groups. Interestingly, results for the RT group were similar to those for the CT group. Regarding processing speed, NC participants improved over time, whereas RT patients, like CT patients, demonstrated no change. Regarding executive functioning, the NC participants performed better than RT patients as well as CT patients. In no instance did results show significant differences in performance between RT patients and CT patients. However, the study may not have been sufficiently powered to detect differences between the 2 patient groups.
These findings are also partially consistent with prior research. A longitudinal study that assessed cognitive functioning over the 6 months after treatment completion found no differences in rates of cognitive decline between breast patients who received chemotherapy and those who received radiotherapy only.16 In another study, breast cancer patients who received either chemotherapy followed by radiotherapy or radiotherapy only were assessed before radiotherapy and 1 year post-treatment.17 Before radiotherapy, rates of cognitive impairment were only marginally higher in the chemotherapy-treated patients (34% vs 24%, P = .06). One year after radiotherapy, differences among these patients were even less evident (19% vs 13%, P = .1). These results are consistent with our findings of no significant differences in cognitive functioning between the CT and RT groups.
The finding in the present study that the addition of hormone therapy was not related to cognitive performance is consistent with our prior research6 and results of a study in which women with breast cancer were followed for 2 years after treatment completion.4 However, other studies have reported findings suggesting a negative impact of hormone therapy on cognitive functioning, especially in the processing speed3, 18 and verbal memory3, 18, 19 domains. Some of these studies examined the effects of hormone therapy among women who received chemotherapy,18, 19 whereas others examined the effects of hormone therapy only among women who did not receive chemotherapy.3 It may be that tamoxifen in particular is associated with lower cognitive functioning,20 although this was not the case in the current study. Differences in designs and treatment groups across studies seriously limit the conclusions that can be drawn regarding how hormone therapy may interact with chemotherapy or radiotherapy to affect cognitive functioning.
Strengths of the current study include the inclusion of both radiotherapy and noncancer comparison groups, a 3-year follow-up period, and the relatively large sample size. Limitations include the homogeneity of the sample (mostly white, well-educated participants) and the lack of a pretreatment cognitive assessment. With patients assessed at only 2 time points post-treatment, acute versus late onset of cognitive dysfunction could not be evaluated, as has been done in prior research.21 In addition, patients who did not complete the second assessment had less education and worse performances on verbal memory at T1 compared with patients who completed both assessments, and noncancer participants who did not complete the second assessment had better performances on attention at T1. Thus, the education, verbal memory, and attention levels of the final sample did not reflect the larger group that completed the initial assessment. It is possible that the patients who were most vulnerable to cognitive dysfunction may not have completed the T2 assessment, suggesting results may be conservative estimates of the effects of cancer treatment on changes in cognitive functioning. Additional limitations include that many women had already been prescribed hormone therapy at T1, which limits the interpretation of the effects of hormone therapy on cognitive functioning over time. Also, data were not available regarding how many women receiving chemotherapy experienced premature menopause, which may be associated with changes in cognitive functioning.22 Finally, a limitation of using neuropsychological assessments is that it is not immediately clear how impairments may affect what an individual can and cannot do in everyday life. To better understand this, future investigators may consider using more ecological measures of cognitive functioning.
Results of the current study highlight the importance of including a radiotherapy comparison group. Interestingly, when cognitive functioning was worse in the CT group, differences were consistently noted with the NC group and not the RT group. Moreover, the RT group demonstrated worse cognitive performance than the NC group in many of the same domains as the CT group. Were the RT group not included in the current study, one might have concluded that worse performance in the processing speed and executive functioning domains was specific to chemotherapy administration. Because differences relative to the NC group were evident for both the RT and CT groups, the question arises as to what is driving these findings. Future research should investigate potential mechanisms such as residual symptoms (eg, pain23 and fatigue8) and persistent cytokine dysregulation24 that may be present in both chemotherapy-treated and radiotherapy-treated patients.
Findings of this study have important clinical implications. When educating patients about the longer-term impact of treatment on cognitive functioning, health care providers may wish to communicate that such effects may result from radiotherapy with or without chemotherapy. Although effects may persist for years after treatment and may possibly intensify, they tend to be domain specific rather than global. Patients who report that cognitive problems are interfering with their daily activities should receive a workup that includes referral to a neuropsychologist who can conduct a systematic evaluation of their cognitive functioning and provide recommendations to address any problems that might be identified.
This research was supported by National Institutes of Health grant R01CA82822.