Effects of aging on brain networks during swallowing: general linear model and independent component analyses

Abstract

Swallowing disorders occur more frequently in older adults. However, the effects of the aging process on neural activation when swallowing are unclear. We aimed to identify neural regions activated during swallowing and evaluate changes in neural activation and neural networks with aging. Using a general linear model (GLM) and independent component (IC) analyses, blood oxygen level-dependent (BOLD) signals were observed in the lateral precentral gyrus, postcentral gyrus, anterior insular cortices, supramarginal gyri, and medial frontal gyrus during swallowing. The right thalamus and anterior cingulate gyri were found to be active areas by GLM and IC analyses, respectively. In the correlational analyses, age was negatively correlated with BOLD signals of the lateral precentral gyri, postcentral gyri, and insular cortices in swallowing tasks. Additionally, correlation analyses between ICs of all participants and age revealed negative correlations in the right supramarginal gyrus, both anterior cingulate cortices, putamen, and cerebellum. In the network analysis, the BOLD signal positively correlated with age in the default mode network (DMN), and was negatively correlated in the lateral precentral gyri, postcentral gyri, and insular cortices. The amplitude of low-frequency fluctuations was significantly decreased in the DMN and increased in swallowing-related areas during swallowing tasks. These results suggest that aging has negative effects on the activation of swallowing-related regions and task-induced deactivation of the DMN. These changes may be used to detect early functional decline during swallowing.

Figure 1

BOLD signal in swallowing tasks in general linear model and independent component analyses. (A) In GLM analyses, significant BOLD signals were seen in both the lateral precentral gyri, postcentral gyri, anterior insular cortices, supramarginal gyri, pre-SMA, thalami, and visual cortices. (B) IC 17 showed the strongest positive correlation with BOLD signal (r = 0.64, p < 0.001) in the swallowing task in both the lateral precentral gyri, postcentral gyri, anterior insular cortices, supramarginal gyri, pre-SMA, and anterior cingulate gyri. GLM, general linear model; BOLD, blood oxygen level-dependent; SMA, supplementary motor area; IC, independent component; FWE, family-wise error.

Figure 2

BOLD signal in swallowing task contrasted with that in non-swallowing task. In the cognition-subtracted activation of swallowing versus non-swallowing tasks, the BOLD signal is predominantly seen in both the lateral precentral gyri, postcentral gyri, anterior insular cortices, supramarginal gyri, pre-SMA, and right thalamus. BOLD, blood oxygen level-dependent; SMA, supplementary motor area; FWE, family-wise error.

Figure 3

Negative correlation between age and BOLD signal during swallowing task. In the GLM, cognitively contrasted BOLD signals between swallowing and non-swallowing tasks showed a decline with healthy aging in both the lateral precentral gyri, postcentral gyri, and insular cortices (A) (r = 0.56, p < 0.001). Independent component 17, which was most positively correlated with BOLD signal (r = 0.64, p < 0.001) in the swallowing task, showed a clear negative correlation with age in both the lateral precentral gyri, postcentral gyri, and right supramarginal gyrus (B). BOLD, blood oxygen level-dependent; GLM, general linear model; FDR, false discovery rate.

Figure 4

ALFF results. In the DMN area, ALFF was significantly lower during swallowing tasks than during non-swallowing tasks. In contrast, ALFF in the medial thalamus, right anteromesial temporal lobe, and left inferior semilunar lobule of the cerebellum were significantly higher. Statistical significance was set at FDR-corrected p < 0.05. ALFF, amplitude of low-frequency fluctuations; DMN, default mode network; FDR, false discovery rate.

Figure 5

Network analysis of age and BOLD signal change during swallowing. During the 12-min tasks, the BOLD signals of the DMN areas were more positively correlated with age, whereas the BOLD signals of both lateral precentral gyri, postcentral gyri, and insular cortices declined with normal aging. The BOLD signal changes in the ICA and PCA showed similar patterns, although PCA more clearly demonstrated involvement of the DMN. ROI-based network analysis (right panel) suggests age-related decline of the BOLD signal in motor and insular areas, but strengthened networks in DMN areas such as the medial prefrontal, posterior cingulate cortices, dorsal superior frontal, angular gyri, and temporal pole. The color bar depicts t-values. Statistical significance was set at FDR-corrected p < 0.05. BOLD, blood oxygen level-dependent; DMN, default mode network; ICA, independent component analysis; PCA, principal component analysis; ROI, region of interest; FDR, false discovery rate.