INS NYC 2024 Program

Paper	Paper Session 15 Program Schedule 02/16/2024 03:30 pm - 04:55 pm Room: West Side Ballroom - Salon 3 Paper Session 15: Neuroanatomy, Neuroimaging, and Neurophysiology Final Abstract #1 Unique Contribution of Brain Age Gap (BAG) in Demographically Adjusted Neuropsychological Test Performance Kara Eversole, University of Florida, Gainesville, United States Cynthia Garvan, University of Florida, Gainesville, United States Catherine Price, University of Florida, Gainesville, United States Jared Tanner, University of Florida, Gainesville, United States Category: Neuroimaging Keyword 1: neuroimaging: structural Keyword 2: neuropsychological assessment Keyword 3: brain structure Objective: Machine-learning models predicting neurobiological age from neuroimaging data have been established as a biomarker of brain health, especially when considering the difference between predicted brain and chronological age (brain age gap [BAG]). In other studies, individuals with cognitive impairment have shown higher BAG values compared to cognitively normal controls. Little is known about how BAG associates with scores on cognitive domains in older adults. The current study examines the unique variance BAG explains in cognitive performance across domains, after accounting for demographic variables known to influence neuropsychological test performance. Participants and Methods: We performed a retrospective cross-sectional analysis of 137 older adults with chronic knee pain (Age = 69.12 ± 6.50; Education = 15.67 ± 2.79; 55% female; 92% non-Hispanic White, 8% non-Hispanic Black). For each participant, BAG was calculated for each participant by subtracting their chronological age from their predicted brain age, as calculated by a 2D Convolutional Neural Network. Higher BAG values reflect poorer brain integrity. Nested regressions were used to examine the unique contribution of BAG in explaining neuropsychological composite scores (i.e., reasoning, working memory, memory, motor, inhibition, language, and processing speed) after controlling for demographic variables (i.e., sex, race, education). Results: Controlling for demographic variables of non-interest via nested regression, brain age gap explained significant variance in processing speed (b = -.03, p = .01, sr² = .04), inhibition (b = -.03, p = .01, sr² = .05), and language composites (b = -.03, p = .01, sr² = .05). In the models examining working memory, reasoning, memory, and motor composites, BAG did not explain significant variance (p’s > .05). Regarding demographic variables, a main effect of sex was found for processing speed and memory, as women had higher composite scores compared to men (p’s > .05). A main effect of education was observed for processing speed, working memory, reasoning, and memory composites, as higher levels of education related to higher composite scores (p’s < .01). Lastly, compared to non-Hispanic White participants, non-Hispanic Black participants (n = 11) had lower processing speed, working memory, reasoning, language, and memory composite scores, and higher motor composite scores (p’s <.05). Conclusions: Brain age gap (BAG), a biomarker of brain disease and aging, explained variance in processing speed, inhibition, and language composites above and beyond demographic variables. Overall, these findings show a relationship between BAG and poorer performance on specific neuropsychological domains. Our results suggest that machine learning models of brain age can provide a metric that is sensitive to certain neuropsychological domains in older adults. Future directions include assessing if BAG predicts neuropsychological performance over time. Supported by NIH [RO1 NR014181].

Paper

02/16/2024
03:30 pm - 04:55 pm
Room: West Side Ballroom - Salon 3

Paper Session 15: Neuroanatomy, Neuroimaging, and Neurophysiology

Final Abstract #1

Unique Contribution of Brain Age Gap (BAG) in Demographically Adjusted Neuropsychological Test Performance

Kara Eversole, University of Florida, Gainesville, United States
Cynthia Garvan, University of Florida, Gainesville, United States
Catherine Price, University of Florida, Gainesville, United States
Jared Tanner, University of Florida, Gainesville, United States

Category: Neuroimaging

Keyword 1: neuroimaging: structural
Keyword 2: neuropsychological assessment
Keyword 3: brain structure

Objective:

Machine-learning models predicting neurobiological age from neuroimaging data have been established as a biomarker of brain health, especially when considering the difference between predicted brain and chronological age (brain age gap [BAG]). In other studies, individuals with cognitive impairment have shown higher BAG values compared to cognitively normal controls. Little is known about how BAG associates with scores on cognitive domains in older adults. The current study examines the unique variance BAG explains in cognitive performance across domains, after accounting for demographic variables known to influence neuropsychological test performance.

Participants and Methods:

We performed a retrospective cross-sectional analysis of 137 older adults with chronic knee pain (Age = 69.12 ± 6.50; Education = 15.67 ± 2.79; 55% female; 92% non-Hispanic White, 8% non-Hispanic Black). For each participant, BAG was calculated for each participant by subtracting their chronological age from their predicted brain age, as calculated by a 2D Convolutional Neural Network. Higher BAG values reflect poorer brain integrity. Nested regressions were used to examine the unique contribution of BAG in explaining neuropsychological composite scores (i.e., reasoning, working memory, memory, motor, inhibition, language, and processing speed) after controlling for demographic variables (i.e., sex, race, education).

Results:

Controlling for demographic variables of non-interest via nested regression, brain age gap explained significant variance in processing speed (b = -.03, p = .01, sr² = .04), inhibition (b = -.03, p = .01, sr² = .05), and language composites (b = -.03, p = .01, sr² = .05). In the models examining working memory, reasoning, memory, and motor composites, BAG did not explain significant variance (p’s > .05). Regarding demographic variables, a main effect of sex was found for processing speed and memory, as women had higher composite scores compared to men (p’s > .05). A main effect of education was observed for processing speed, working memory, reasoning, and memory composites, as higher levels of education related to higher composite scores (p’s < .01). Lastly, compared to non-Hispanic White participants, non-Hispanic Black participants (n = 11) had lower processing speed, working memory, reasoning, language, and memory composite scores, and higher motor composite scores (p’s <.05).

Conclusions:

Brain age gap (BAG), a biomarker of brain disease and aging, explained variance in processing speed, inhibition, and language composites above and beyond demographic variables. Overall, these findings show a relationship between BAG and poorer performance on specific neuropsychological domains. Our results suggest that machine learning models of brain age can provide a metric that is sensitive to certain neuropsychological domains in older adults. Future directions include assessing if BAG predicts neuropsychological performance over time. Supported by NIH [RO1 NR014181].