Abstract
Purpose :
Perinatally HIV-infected children are at risk for cerebral injury and retinal structural abnormalities. As the retina shares developmental, physiological and anatomical features with the brain, these changes may have a shared pathogenesis. Retinal imaging is increasingly used to study pathology in the brain, but it has not yet been studied as a biomarker for cerebral injury in HIV-infection. This study aims to evaluate the relationship between retinal thickness and cerebral injury in perinatally HIV-infected children and matched healthy controls.
Methods :
We included patients (n=29) aged 8-18 years from the AMC outpatient clinic, and healthy controls (n=35) from the community, matched for age, gender, ethnicity and socioeconomic status to account for their potential effects on brain development. Spectral Domain OCT was used to measure retinal thickness (RT). Individual retinal layers were segmented (figure 1) using the Iowa reference algorithm. HIV-infected children had a lower total foveal RT, mostly due to a thinner ganglion cell layer and outer nuclear layer + inner segments. All participants underwent a 3.0 Tesla MRI scan, using T1-weighted scans to determine GM and WM volume, and diffusion tensor imaging to assess WM diffusivity. HIV-infected children had lower cerebral volume, lower fractional anisotropy (FA), and higher mean and radial diffusivity (MD, RD; figure 2).
Results :
The relation between RT and cerebral injury was assessed using multivariable linear regression analysis adjusted for age, gender and spherical equivalent, and additionally for intracranial volume with volumetric outcomes. RT was strongly associated with WM diffusivity in HIV-infected participants (figure 3a/b, table 1), but not in healthy controls. Contrastingly, RT was associated with cerebral volume in healthy controls only (figure 3c, table 2).
Conclusions :
The lack of relationship between RT and cerebral volume in perinatally HIV-infected children suggests interference by unmeasured factors (e.g. HIV-induced inflammation, early life circumstances). The different correlation patterns between RT, WM diffusivity and GM volume suggest that the pathogenic mechanism underlying reduced RT is similar to that of increased WM diffusivity, but different from the pathogenic mechanism leading to reduced GM volume.
This is an abstract that was submitted for the 2016 ARVO Annual Meeting, held in Seattle, Wash., May 1-5, 2016.