Imaging was performed on a 3 Tesla (3T) whole-body system (3T and Excite; General Electric) using an 8-channel receive head coil. High-resolution structural images for anatomical confirmation were acquired using an inversion recovery prepared 3-dimensional fast spoiled pulse sequence (inversion time [TI], 450 ms; repetition time [TR], 7.9 ms; echo time [TE], 3.1 ms; flip angle, 12°; field of view [FOV], 25×25×16 cm3; matrix, 256×256×124). Images were reviewed to ensure absence of structural lesions. For the activation task, a block design was used with functional changes in cerebral blood flow and blood oxygen level-dependent signal determined within the visual cortex for a flickering black and white radial checkerboard (frequency, 8 Hz). Activation periods were 20 s in duration, whereas rest portions were 60 s in duration and consisted of an isoluminant gray screen with a center fixation square. A single-shot proximal inversion sequence that controlled for off-resonance effects by using quantitative imaging of perfusion with a single subtraction (TR, 2.5 s; TI1, 700 ms; TI2, 1500 ms; tag width, 20 cm; tag-slice gap, 1 cm), dual-echo gradient echo readout, and spiral acquisition of k-space (TE1, 9.4 ms; TE2, 30 ms; flip angle, 90°; FOV, 24 cm; matrix, 64×64) allowed for the alternate acquisition of “”tag”” and “”control”” images. The difference between images provided cerebral blood flow values, whereas the mean of the images yielded the blood oxygen level-dependent signal. An additional resting cerebral blood flow scan was acquired to quantify baseline values, using methods described elsewhere.
Images were coregistered to correct for subject movement. The visual cortex was defined as the region between the parietaloccipital sulci and was manually delineated on anatomical images. This region was resampled to match functional image scans and corrected for possible white matter involvement and partial volume loss. A general linear model used the stimulus function, acquired cardiac and respiratory data, and constant and linear terms as additional nuisance regressors to identify activated visual cortex voxels with a significance threshold (P=.05) that corrected for multiple comparisons. From these activated voxels, functional changes in cerebral blood flow and the blood oxygen level-dependent signal were determined for the task, using methods described elsewhere. The mean baseline cerebral blood flow was calculated by averaging all time points within activated visual cortex voxels.
Wilcoxon rank sum tests were used to assess the differences in fMRI measures between the HIV-positive and HIV-negative subjects. The association and interaction between fMRI measures and HIV status and age (in years) were investigated using a multiple-regression model. Each fMRI variable (baseline cerebral blood flow, functional changes in cerebral blood flow, and blood oxygen level-dependent signal) was log10-transformed to improve normality and homoscedasticity and backtransformed to the natural scale for plotting. An additive regression model was used with P values determined from the Wald test. The effects of age and HIV serostatus were determined for each fMRI outcome. The mean change in the fMRI outcome on the log-scale was transformed back to the response scale and expressed as proportional increase (positive) or decrease (negative) in the outcome, with the effect size computed as the regression effect on the logarithmic scale divided by the residual standard deviation. The age-equivalent effect of HIV infection was determined by dividing the HIV effect size by the yearly effect size of age. The age effect was expressed per 15 years of aging for comparison.