Lower cognitive reserve in the aging human immunodeficiency virus-infected brain

Abstract

More HIV-infected individuals are living longer; however, how their brain function is affected by aging is not well understood. One hundred twenty-two men (56 seronegative control [SN] subjects, 37 HIV subjects with normal cognition [HIV+NC], 29 with HIV-associated neurocognitive disorder [HAND]) performed neuropsychological tests and had acceptable functional magnetic resonance imaging scans at 3 Tesla during tasks with increasing attentional load. With older age, SN and HIV+NC subjects showed increased activation in the left posterior (reserve, "bottom-up") attention network for low attentional-load tasks, and further increased activation in the left posterior and anterior ("top-down") attention network on intermediate (HIV+NC only) and high attentional-load tasks. HAND subjects had only age-dependent decreases in activation. Age-dependent changes in brain activation differed between the 3 groups, primarily in the left frontal regions (despite similar brain atrophy). HIV and aging act synergistically or interactively to exacerbate brain activation abnormalities in different brain regions, suggestive of a neuroadaptive mechanism in the attention network to compensate for declined neural efficiency. While the SN and HIV+NC subjects compensated for their declining attention with age by using reserve and "top-down" attentional networks, older HAND subjects were unable to compensate which resulted in cognitive decline.

Copyright © 2013 Elsevier Inc. All rights reserved.

Figures

Fig. 1

Performance accuracy (% correct ± standard error) and reaction times during the functional magnetic resonance imaging ball tracking task for each subject group. All groups showed a progressive decline in performance accuracy (in %) with increasing task difficulty. Although the decline was not significant in the seronegative (SN) subjects, the HIV subjects with normal cognition (HIV+NC) performed significantly poorer on the task with high attentional load (tracking 4 balls), and the HIV-associated neurocognitive disorder (HAND) group performed significantly poorer on the tasks with medium (tracking 3 balls) and high attentional load (tracking 4 balls). Only the HAND subjects showed slower reaction times with the most difficult task (tracking 4 balls).

Fig. 2

Surface views of brain regions with age-dependent changes in blood oxygen level dependent signal during the visual attention (ball tracking) tasks in seronegative (SN) and HIV with normal cognition (HIV+NC) subject groups. Left panel: SN subjects show age-dependent increases in brain activation within the attention network, primarily in the left posterior cortex during low attentional load (2 balls), nonsignificant increase in activation during the medium attentional load (tracking 3 balls), and more extensive age-dependent increases in activation involving the left (L) frontal regions during the high attentional load (tracking 4 balls). Right panel: HIV+NC subjects show similar age-dependent increases in activation as the SN subjects in the L posterior attention network during the low attentional load (tracking 2 balls), but extensive age-related increases in activation throughout the attention network during the 3-ball and 4-ball tasks. Please see Table 2 for cluster maximum coordinates, size, corrected p values and t scores. Maps were created using t >2.0 and cluster size >750 voxels. Abbreviation: R, right.

Fig. 3

Images showing brain regions with significant age by HIV status (seronegative [SN] subjects vs. subjects with HIV and normal cognition [HIV+NC]) interaction on blood oxygen level-dependent (BOLD) signals for the 3-ball task (A, cluster corrected p < 0.0001) and the 4-ball task (B, cluster corrected p = 0.006); see also Table 3. The scatter plots validated the interactions found on SPM8, showing dependence of BOLD signals extracted (from regions of interest centered at the cluster maxima) on subject age. The HIV+NC subjects (black circles and regression lines) show significant increases in BOLD signals with age (3 balls: r = 0.46–0.59; 4 balls: r = 0.50–0.55), and the SN subjects (green circles and regression lines) show relatively flat or slightly declining slopes of the BOLD responses across the age span (3 balls: r = –0.06 to –0.27; 4 balls: r = –0.18 to 0.23). Abbreviations: BA, Brodmann area; L, left; R, right.

Fig. 4

Statistical parametric maps showing significant interactions between age and cognitive status on brain activation (tracking 4 balls) for the seronegative (SN) and HIV-associated neurocognitive disorder (HAND) groups (cluster corrected p = 0.009). Top row: magnetic resonance imaging scans showing brain regions with significant interactions in 3 subclusters (see also Table 3). Bottom row: the scatter plots show opposite slopes in HAND and SN subjects on their age-dependent changes in blood oxygen level-dependent (BOLD) signals in these 3 brain regions (HAND subjects: red circles and regression lines, r = –0.34 to –0.58; SN subjects: green circles and regression line, r = 0.17–0.38). Abbreviations: BA, Brodmann area; L, left.

Fig. 5

Subjects with HIV and normal cognition (HIV+NC) and HIV-associated neurocognitive disorder (HAND) subjects show opposite relationships on their age-dependent changes in blood oxygen level-dependent (BOLD) signals while tracking 3 balls (A) and 4 balls (B). Magnetic resonance imaging scans in each panel show brain regions with significant interactions in the subclusters (see also Table 3) and the corresponding scatter plots show opposite slopes in HAND and HIV+NC subjects on their age-dependent changes in BOLD signals in each of the subclusters (HAND: red circles and regression lines; HIV+NC subjects: black circles and regression lines). Abbreviations: BA, Brodmann area; L, left; R, right.

Fig. 6

Brain regions that show differences in load-dependent changes in blood oxygen level-dependent (BOLD) signals between subjects with HIV and normal cognition (HIV+NC) and control subjects include the fusiform gyrus (Brodmann area [BA] 20) and the superior temporal gyrus (BA 38). Though seronegative (SN) subjects showed load-dependent increases in activation in these regions, HIV+NC subjects showed load-dependent decreases in these brain regions. No age effects were seen in these regions, and the findings remained unchanged when age was added as a covariate. The surface maps are generated with t >1.67, cluster size >1000 voxels. Abbreviations: ANCOVA, analysis of covariance; FWE, family-wise error.

Fig. 7

Scatter plots showing the motor, learning, and memory Z-scores in each subject in relation to age. The Z-scores are normalized to a database of 342 healthy seronegative (SN) subjects evaluated using the same tests, and adjusted for age and education. All subjects showed age-dependent decline in motor function. Note steepest age-dependent decline in memory (r = –0.25; p = 0.20) and learning (r = –0.36; p = 0.06) in the HIV-associated neurocognitive disorder (HAND) subjects. Abbreviation: HIV+NC, subjects with HIV and normal cognition.

Fig. 8

Left, brain regions activated during tracking of 4 balls in relation to global cognitive Z-scores that showed significant group differences, covaried for age (cluster level p (family-wise error)-corrected <0.0001, k > 1000 voxels and T-threshold > 3.5). Percentage of blood oxygen level-dependent (BOLD) signals extracted from region of interest surrounding the 3 subcluster maxima are shown: right superior parietal lobule (Brodmann area [BA] 7: 36, –49, 67), right parahippocampal gyrus (BA 19: 39, –46, –2) and right medial frontal gyrus (BA 6: 15, 2, 27). While the HIV subjects showed decreased BOLD signals as they approached or were below global Z-score of –1 in all 3 regions, the SN healthy controls showed a U-shaped curve with greater BOLD signals in those who were able to compensate and perform better (+0.5 to +1 Z-score) or those who could not compensate (–0.5 to –1.0) for 2 of these subregions. Right, subclusters from the frontal brain regions where the HIV subjects showed a threshold effect on their brain activation during the tracking of 4 balls in relation to global Z-score (covaried for age, cluster level, p (family-wise error)-corrected p < 0.001, k = 2451 voxels, T > 3.0). HIV subjects that performed poorer than –1 Z-score, primarily HIV-associated neurocognitive disorder (HAND) subjects, showed progressively lesser activation in the right superior frontal gyrus (BA 6: –15, 38, 7 and 15, 14, 49) and the left anterior cingulate (BA 32: 15, –4, 67).