This is the first study to use a prospective, longitudinal approach to show decline in brain gray matter density shortly after chemotherapy and degree of recovery over time. No between-group gray matter differences were apparent at baseline (prior to chemotherapy, radiation, or anti-estrogen treatment), although declines were apparent from baseline to M1 in both cancer groups relative to controls. The within-group pattern of decline and partial recovery (i.e., the presence of regions where gray matter density fully recovered to baseline levels as well as regions where declines persisted from M1 to Y1) seen in chemotherapy-treated patients was not present in patients who did not receive chemotherapy or healthy controls. This suggests that these gray matter alterations are related to effects of chemotherapy, rather than solely reflecting host factors, the cancer disease process, or effects of other cancer treatments. Our results confirm findings from prior retrospective studies of changes in brain structure in long-term breast cancer survivors who were treated with chemotherapy [16
]. There is notable overlap in the present findings with frontal and temporal regions where gray matter differences were detected in prior retrospective studies, consistent with current results that some gray matter changes do not return to baseline over time.
The present results were consistent with our hypothesis that CTx+ patients would demonstrate decreased gray matter in the acute post-treatment phase (M1), with subsequent partial recovery (normalization) after 1 year (Y1), and provide evidence for a structural neuroanatomic basis for the cognitive problems most commonly reported during and after chemotherapy, including impairment in episodic and working memory. Medial temporal lobe structures are critical in episodic memory, whereas attentional and executive functions such as working memory are subserved by frontal brain systems. Cerebellar regions are also increasingly recognized as important in higher cognitive functions [34
]. The alterations in gray matter density observed in the CTx+ group are, therefore, consistent with the pattern of cognitive complaints and impairment found in neurocognitive studies. Such concerns tend to be most pronounced during and shortly after chemotherapy but remit in most patients over time, with only a subset reporting persistent cognitive problems. This temporal pattern is consistent with our finding that declines in gray matter density apparent at 1 month post-chemotherapy improve significantly over the following year, although they do not appear to normalize entirely. The presence of residual abnormalities is consistent with cognitive findings in long-term survivor cohorts [23
] and the limited available neuroimaging data, including our preliminary survivor study [28
]. Functional neuroimaging (functional MRI and positron emission tomography) studies have also shown abnormalities following chemotherapy corresponding to brain regions where we found decreased gray matter density, particularly in the frontal lobes and cerebellum [12
]. Our finding of decreased gray matter density in similar brain regions offers convergent data in support of a model of brain structural abnormality corresponding to functional changes after chemotherapy.
The specific etiology of these gray matter changes is unknown, although we and others have proposed possible mechanisms for chemotherapy-induced cognitive and brain changes, such as chemotherapy-induced DNA damage (directly or through increases in oxidative stress), individual variation in genes related to neural repair and/or plasticity, and chemotherapy-induced hormonal changes [1
]. A limited literature in animal models has demonstrated neuronal changes following chemotherapy, including alterations in cytoskeletal and calcium regulating proteins and pyramidal cell dendritic retraction, although not frank neuronal loss [19
]. However, these studies focused on cytosine arabinoside (Ara-C), an agent with a different mechanism of action than that of the drugs with which our patients were treated. The finding of recovery in gray matter density over time, while consistent with our a priori hypothesis, also requires further exploration with regard to etiology. VBM studies have demonstrated increases in gray matter volume in healthy controls after lithium treatment [22
], epilepsy patients following neurosurgery [41
], and anorexic patients after nutritional recovery [8
]. These increases were all seen after treatment and/or functional recovery, which is consistent with our hypothesized model of gray matter declines corresponding to chemotherapy-related cognitive impairment and later recovery relating to symptomatic improvement.
The findings of this study were statistically robust, as we used a stringent significance threshold for our imaging results. Some potential limitations must be considered, however. Group sizes are relatively small, although this is common in prospective studies of breast cancer patients, largely due to the challenges of recruiting and studying patients prior to adjuvant treatment. The cohort is racially and ethnically homogeneous (largely Caucasian, non-Hispanic), consistent with the rural northern New England population. In observational studies of this nature patients are not randomized to CTx+ and CTx– groups, with treatment practice patterns leading to between-group differences on disease-related variables. Our CTx– group tended to have earlier stage disease (predominantly stage 0–I) than the CTx+ group (mostly stage I–II), and a larger percentage of CTx– than CTx+ patients were on anti-estrogen treatment at M1, although comparable percentages were taking these medications at Y1. For both breast cancer groups, radiation and/or anti-estrogen treatment were administered to more than half of the participants. Although most patients were prescribed tamoxifen, other anti-estrogen agents were also used, potentially contributing to data variability. Several breast cancer patients were taking psychotropic medications (e.g., SSRIs for mood or menopause symptoms) at different points in the study. Although analyses covarying for medication use or excluding these individuals produced the same overall pattern as in the full sample, these agents may have independent effects which cannot be disambiguated from those of chemotherapy. This cohort was not powered to analyze other treatment effects fully or to clarify differential effects of chemotherapy regimens. In future studies it would be beneficial to include a larger, more diverse cohort, perhaps via multi-center collaborations, in order to assess individual contributions of specific cancer treatments and psychosocial variables.
Based on data from several groups, there is increased interest in abnormalities that may be present at baseline in breast cancer patients, as well as an emerging perspective that cancer or host factors may be associated with cognitive or anatomic changes over the course of treatment. The present finding of regional gray matter density reductions over time in patients who did not receive chemotherapy relative to controls suggests that the CTx– group also demonstrates gray matter changes related to cancer and/or its treatment, but that these are lesser in severity and spatial extent than those seen after chemotherapy. Further study will be needed to directly address the effect of anti-estrogen treatment on brain structure and function, as previous studies suggest this may be a contributing factor [11
In summary, we report breast cancer chemotherapy-related reduction in gray matter density and its partial recovery over time prospectively in a single cohort. This pattern of gray matter changes was not observed in breast cancer patients not treated with chemotherapy or healthy controls, and was not attributable to potential confounds such as disease stage, time since last cancer-related surgery, mood, or medications. Gray matter changes were most prominent in frontal, temporal, and cerebellar regions, consistent with the profile of cognitive dysfunction and complaints observed in a subset of breast cancer patients after chemotherapy in previous research. Prior cross-sectional studies using other imaging approaches have shown structural and functional abnormalities in similar brain regions; our findings extend these earlier studies by demonstrating the evolution of these structural changes in the first 2 years after breast cancer diagnosis and treatment. While further investigation is needed to elucidate the underlying neural mechanism(s), these findings provide important new information to help guide future investigation of the temporal course and regional specificity of these changes.