Increased brain-predicted aging in treated HIV disease

Abstract

Objective: To establish whether HIV disease is associated with abnormal levels of age-related brain atrophy, by estimating apparent brain age using neuroimaging and exploring whether these estimates related to HIV status, age, cognitive performance, and HIV-related clinical parameters.

Methods: A large sample of virologically suppressed HIV-positive adults (n = 162, age 45-82 years) and highly comparable HIV-negative controls (n = 105) were recruited as part of the Comorbidity in Relation to AIDS (COBRA) collaboration. Using T1-weighted MRI scans, a machine-learning model of healthy brain aging was defined in an independent cohort (n = 2,001, aged 18-90 years). Neuroimaging data from HIV-positive and HIV-negative individuals were then used to estimate brain-predicted age; then brain-predicted age difference (brain-PAD = brain-predicted brain age - chronological age) scores were calculated. Neuropsychological and clinical assessments were also carried out.

Results: HIV-positive individuals had greater brain-PAD score (mean ± SD 2.15 ± 7.79 years) compared to HIV-negative individuals (-0.87 ± 8.40 years; b = 3.48, p < 0.01). Increased brain-PAD score was associated with decreased performance in multiple cognitive domains (information processing speed, executive function, memory) and general cognitive performance across all participants. Brain-PAD score was not associated with age, duration of HIV infection, or other HIV-related measures.

Conclusion: Increased apparent brain aging, predicted using neuroimaging, was observed in HIV-positive adults, despite effective viral suppression. Furthermore, the magnitude of increased apparent brain aging related to cognitive deficits. However, predicted brain age difference did not correlate with chronological age or duration of HIV infection, suggesting that HIV disease may accentuate rather than accelerate brain aging.

Figure 1. Study methods

Outline of machine learning brain age prediction methods used in the study. Data included 3 separate groups: healthy individuals (n = 2,001) comprised the training data, and HIV-positive individuals (n = 161) and HIV-negative controls (n = 102) comprised the test data, after quality control (n = 4 exclusions). (A) All data were preprocessed with statistical parametric mapping (SPM) to segment T1 images into gray matter (GM) and white matter (WM) images. These segmented images were then normalized to a custom template using DARTEL for nonlinear registration, before being resampled to Montreal Neurological Institute 152 (1.5 mm³) template space, using volumetric modulation and a 4-mm smoothing kernel. GM and WM images were then concatenated for each subject. (B) Machine learning age prediction used PRoNTo. (a) Representation of all data in a linear kernel form as a similarity matrix of the dot products between pairs of vectorized and concatenated volume images. (b) Supervised learning stage. Data from the training set were run through a Gaussian processes regression model to define the correspondence between brain volume maps and chronological age. Model accuracy was assessed on predictions made during a 10-fold cross-validation procedure. (c) Test set prediction. The coefficients from the model trained on the healthy sample were used to generate predicted age values from the data in the HIV-positive individuals and HIV-negative controls. Brain-predicted age difference (brain-PAD) scores were defined by subtracting chronological age from predicted age. (C) Statistical analysis based on brain-PAD scores as an index of apparent brain aging.

Figure 2. Brain-predicted age in the training dataset

Scatterplot of chronological age (x-axis) and predicted brain age (y-axis), based on results of 10-fold cross-validation of the Gaussian processes regression model in the training dataset (n = 2,001). Dashed line (black) represents the line of identity (y = x), where predicted age = chronological age.

Figure 3. Predicted age differences in HIV infection

(A) Grouped data plot of brain-predicted age difference (brain-PAD) in HIV-positive individuals (red triangles) and HIV-negative controls (blue spots). Solid black lines indicate group mean brain-PAD values. (B) Scatterplot of chronological age in years (x-axis) against predicted brain age (y-axis) generated using structural neuroimaging. Points indicate HIV-positive individuals (red triangles) and HIV-negative controls (blue spots) and lines are regression lines for each group (HIV-positive = red; HIV-negative = blue), with 95% confidence intervals displayed. Dashed gray line is the line of identity (y = x).