Age-Related Changes in BOLD Activation Pattern in Phonemic Fluency Paradigm: An Investigation of Activation, Functional Connectivity and Psychophysiological Interactions

Abstract

Healthy aging is associated with decline of cognitive functions. However, even before those declines become noticeable, the neural architecture underlying those mechanisms has undergone considerable restructuring and reorganization. During performance of a cognitive task, not only have the task-relevant networks demonstrated reorganization with aging, which occurs primarily by recruitment of additional areas to preserve performance, but the task-irrelevant network of the "default-mode" network (DMN), which is normally deactivated during task performance, has also consistently shown reduction of this deactivation with aging. Here, we revisited those age-related changes in task-relevant (i.e., language system) and task-irrelevant (i.e., DMN) systems with a language production paradigm in terms of task-induced activation/deactivation, functional connectivity, and context-dependent correlations between the two systems. Our task fMRI data demonstrated a late increase in cortical recruitment in terms of extent of activation, only observable in our older healthy adult group, when compared to the younger healthy adult group, with recruitment of the contralateral hemisphere, but also other regions from the network previously underutilized. Our middle-aged individuals, when compared to the younger healthy adult group, presented lower levels of activation intensity and connectivity strength, with no recruitment of additional regions, possibly reflecting an initial, uncompensated, network decline. In contrast, the DMN presented a gradual decrease in deactivation intensity and deactivation extent (i.e., low in the middle-aged, and lower in the old) and similar gradual reduction of functional connectivity within the network, with no compensation. The patterns of age-related changes in the task-relevant system and DMN are incongruent with the previously suggested notion of anti-correlation of the two systems. The context-dependent correlation by psycho-physiological interaction (PPI) analysis demonstrated an independence of these two systems, with the onset of task not influencing the correlation between the two systems. Our results suggest that the language network and the DMN may be non-dependent systems, potentially correlated through the re-allocation of cortical resources, and that aging may affect those two systems differently.

Keywords: BOLD fMRI; DMN; functional connectivity; normal aging; psychophysiological interactions; task positive network.

Figure 1

Phonemic verbal fluency task activation maps and quantitative change in activity. (Left) Activation maps in (A) young (≤ 30 years), (B) middle-aged (50–59 years), and (C) older adults (≥60 years) subjects. Weakening of task-positive and task-negative activation in the older subjects is evident. Cluster-wise corrected at p < 0.005, α = 0.05, minimum cluster size = 105 contiguous voxels. Maps are provided following neurological convention, left is on left. (Right, D) Task-positive (green) percent-change showing slow increase with age, and task-negative (blue) showing abrupt decline in the group of participants over 60 years. Statistical values are generated using a one-way ANOVA test (^*Significant at p < 0.05).

Figure 2

Phonemic verbal fluency task connectivity maps and quantitative change in functional connectivity between age groups. (Left) Connectivity maps from (A) left IFG seed in young (≤ 30 years), (B) left IFG seed in older adults (≥60 years) subjects; (C) pC/PCC seed in young (≤ 30 years), (D) pC/PCC seed in older adults (≥60 years) subjects, (E) mean connectivity score within the language network and the DMN for each group. No differences were observed in connectivity in either system.

Figure 3

(Top) Intensity and extent of activation over the task-positive network. Mean intensity (left) and cluster extent (right) within ROIs of the task-positive (language) network, showing areas actively implicated in the language function and their homologs. (Bottom) Functional connectivity strength and extent of connectivity over the task-positive network. Mean functional connectivity strength (left) and extent of connectivity (right) to left IFG for individual regions of the language network and their homologs. Statistical values are generated from ANOVAs between the three population groups for each of the ROIs. (^**Significant at p < 0.005, ^*Significant at 0.005 < p < 0.05; ^†trending).

Figure 4

(Top) Intensity and extent of activation over the task-negative network. Mean intensity (left) and cluster extent (right) within ROIs of the task-negative network (DMN): pC/PCC, bilateral IPL and medial PFC. (Bottom) Functional connectivity strength and extent of connectivity over the task-negative network. Average connectivity strength (left) and extent of connectivity (right) to the pC/PCC seed (TLRC [6, −60, 32]) within regions of the DMN. Connectivity strength (left) for regions of the DMN present an apparent step-wise reduction with age while extent of connectivity (right) is similar across groups for each region. Statistical values are generated from ANOVAs between the three population groups for each of the ROIs. (^**Significant at p < 0.005, ^*Significant at 0.005 < p < 0.05; ^†trending).

Figure 5

PPI maps representing the context-dependent interaction between two systems. A- PPI with seed placed in left IFG with DMN mask underlay (white). B- PPI with seed placed in right pC/PCC with left IFG mask underlay (white). Maps are presented with threshold p = 0.00001, with 165 contiguous voxels.

Figure 6

Statistical difference between the young group and the old group in terms of (A) PPI interaction with left IFG as seed region, (B) PPI interaction with right pC/PCC as seed region. Statistical maps are shown at p = 0.005.

Figure 7

Task activation and functional connectivity over the motor and auditory networks. (Top) Mean intensity (left) and cluster extent (right) within ROIs of the auditory and motor networks in a phonemic VF task in young (red), middle-aged (green) and older adults (blue). (Bottom) Mean functional connectivity strength (left) and extent of connectivity (right) to Broca's area (TLRC [−42, 3, 30]) for individual regions of the auditory and motor networks. Statistical values are generated using a one-way ANOVA test, uncorrected for multiple comparisons. (^*Significant at p < 0.05).