Late effects of high-dose adjuvant chemotherapy on white and gray matter in breast cancer survivors: converging results from multimodal magnetic resonance imaging

Abstract

The neural substrate underlying cognitive impairments after chemotherapy is largely unknown. Here, we investigated very late (>9 years) effects of adjuvant high-dose chemotherapy on brain white and gray matter in primary breast cancer survivors (n = 17) with multimodal magnetic resonance imaging (MRI). A group of breast cancer survivors who did not receive chemotherapy was scanned for comparison (n = 15). Neuropsychological tests demonstrated cognitive impairments in the chemotherapy group. Diffusion tensor imaging (DTI) with tract-based spatial statistics showed that chemotherapy was associated with focal changes in DTI values indicative for reduced white matter integrity. Single voxel proton MR spectroscopy (1H-MRS) in the left centrum semiovale (white matter) showed a reduction of N-acetylasparate/creatine indicative of axonal injury. Voxel-based morphometry demonstrated a reduction of gray matter volume that overlapped with fMRI hypoactivation (as reported in a previous publication) in posterior parietal areas and colocalized with DTI abnormalities. Also, DTI correlated with 1H-MRS only in the chemotherapy group. These results converge to suggest that high-dose adjuvant chemotherapy for breast cancer is associated with long-term injury to white matter, presumably reflecting a combination of axonal degeneration and demyelination, and damage to gray matter with associated functional deficits. Hormonal treatment with tamoxifen may also have contributed to the observed effects, although results from other studies indicate that it is unlikely that tamoxifen is solely or largely responsible. Using this multimodality approach we provide for the first time insight into the neural substrate underlying cognitive impairments following systemic administration of cytotoxic agents many years after treatment.

Figure 1

Group differences in DTI values between the chemotherapy (CT) group and the no‐CT group. DTI data were analyzed with tract‐based spatial statistics (TBSS) on “skeletonized” white matter. Cluster‐based thresholding at P < 0.05 was applied, fully corrected for multiple comparisons. Areas of the white matter skeleton that show significant group differences are overlaid on a fractional anisotropy (FA) map, derived from the mean of the FA maps of all participants that have been normalized to standard MNI space. Significant clusters have been thickened for ease of visualization. Left panel shows regions where the CT group has a lower FA than the no‐CT group. Middle panel shows regions where the CT group has a higher mean diffusivity (MD) than the no‐CT group. Right panel shows regions where the CT group has a higher radial diffusivity (RD) than the no‐CT group. See text for description of affected white matter tracts. Color bars show range of corrected P values. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Figure 2

Group differences for DTI and fMRI. Increases in mean diffusivity (MD) and radial diffusivity (RD, upper and lower panel, respectively) for the chemotherapy (CT) versus the no‐CT group in white matter tracts (posterior thalamic/optic radiation) are colocalized with hypoactivation in bilateral posterior parietal cortex for the CT compared to the no‐CT group during a paired associates task [de Ruiter et al., 2011]. All images are overlaid on the MNI152‐template. For DTI, cluster‐based thresholding at P < 0.05 was applied, fully corrected for multiple comparisons. Significant MD and RD clusters have been thickened for ease of visualization. Color bars show range of uncorrected (fMRI) and corrected (MD and RD) P values. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Figure 3

Group differences for voxel‐based morphometry (VBM) and fMRI. Decreased regional gray matter volume in the chemotherapy (CT) group compared with the no‐CT group, as assessed with VBM, overlaps with fMRI hypoactivation during a paired associates task [de Ruiter et al., 2011] in left posterior parietal cortex. Right posterior parietal cortex only shows hypoactivation but no volume reduction. Also, volume reductions can be discerned in bilateral cerebellum (stronger in the left than in the right hemisphere). Color bars show range of corrected (VBM) and uncorrected (fMRI) P values. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Figure 4

Correlations of DTI values with single voxel proton MR spectroscopy (1H‐MRS) in the chemotherapy (CT) group. 1H‐MRS was performed in the left centrum semiovale (white matter). A: The approximate location of the voxel is shown in red on the MNI‐152 template (upper panel). Mean diffusivity (MD) values for skeletonized white matter were correlated with NAA and NAA/Cr. We also used tract‐based spatial statistics (TBSS) with cluster‐based thresholding at P < 0.05, fully corrected for multiple comparisons. Clusters showing a significant negative correlation with NAA/Cr are overlaid on the MNI‐152 template. Significant clusters have been thickened for ease of visualization. Color bar shows range of corrected P values. B: Correlations of mean DTI values of skeletonised white matter within the 1H‐MRS voxel. The scatter plots in the upper panels show correlations of MD, axial diffusivity and radial diffusivity with NAA/Cr. The scatter plots in the lower panels show correlations of MD, axial diffusivity and radial diffusivity with NAA (institutional units). *P < 0.05, **P < 0.01. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]