Liver-related complications continue to pose a significant threat to the health of people living with HIV (PLWH),
despite the life-extending benefits of combination antiretroviral therapy (cART). Non-alcoholic fatty liver disease
(NAFLD) affects up to 50% of PLWH and can be caused by both HIV infection and cART. The risk factors for
NAFLD in PLWH are not fully understood, but hypertension, diabetes, hyperlipidemia, metabolic syndrome, and
gut microbiome changes are more common in this population.
This study aims to investigate the co-occurrence of NAFLD and HIV infection and their potential impact on brain
health. Our hypothesis is that PLWH with NAFLD are at higher risk of persistent blood brain barrier (BBB)
dysfunction, metabolite dysfunction, and microcirculation alterations compared to those without NAFLD. These
changes may result in more aggressive brain injury and poorer cognitive function, especially in older individuals.
To test this hypothesis, we will employ state-of-the-art, non-invasive MRI techniques to evaluate brain injury. Our
approach includes developing a novel deep neural network with compress sensing for high-quality reconstruction
of highly accelerated clinical chemical exchange saturation transfer (CEST), a molecular MRI technique used to
assess metabolic dysfunction. We will compare neuroimaging metrics of metabolite dysfunction, BBB
permeability, microcirculation, and neuroinflammation in PLWH with NAFLD versus those without NAFLD. We
will also investigate the association between these imaging metrics and blood biomarkers.
Additionally, we will assess whether NAFLD in HIV infection leads to reduced cognitive performance, mediated
by brain injury as measured by CEST and other imaging metrics. Finally, we will track longitudinal changes in
MRI metrics, blood biomarkers, and cognitive performance in PLWH with and without NAFLD and healthy
controls.
The significance of this research lies in several aspects. Firstly, proposed rapid and high-quality CEST imaging
at clinical 3T offers increased sensitivity to disease pathology compared to traditional methods. Secondly,
advanced MR techniques can help us better understand metabolic changes, microcirculation, and BBB
alterations caused by disease processes. Thirdly, by examining cross-sectional and longitudinal associations
between MRI metrics, blood markers, and cognitive scores, we can identify gaps in our knowledge of brain injury
and disease progression and their interaction with clinical variables. Fourthly, the use of deep learning models
in conjunction with multimodal features may aid in differentiating brain-related abnormalities of NAFLD from HIV,
as there is significant overlap in clinical and imaging manifestations. Lastly, the methods proposed in this study
may have implications for understanding other brain-related disorders.