Objective

Despite combined antiretroviral therapy, neuroinflammation may persist in persons living with HIV (PLWH) and contribute to cognitive impairment in this population. Positron emission tomography (PET) imaging targeting 18 kDa translocator protein (TSPO) has been used to localize neuroinflammation. We aimed to use TSPO-PET imaging to evaluate neuroinflammation in PLWH.

Design

Twenty-four virologically suppressed PLWH on combined antiretroviral therapy and 13 HIV-negative (HIV-) controls completed TSPO-PET imaging using the radiotracer [C]PBR28. Because of tracer complexity and differing procedures used in previous studies, we employed an expansive methodological approach, using binding potential (BP) and standard uptake value ratio and multiple different reference regions to estimate [C]PBR28 binding.

Methods

[C]PBR28 binding was measured in 30 cortical and subcortical regions and compared between PLWH and HIV- controls. Pearson correlation evaluated the association between [C]PBR28 binding and cognition and clinical measures of HIV.

Results

Analyses conducted using multiple reference regions and measures of tracer uptake revealed no significant differences between [C]PBR28 binding in PLWH compared with HIV- controls. In addition, [C]PBR28 binding in PLWH was not significantly associated with clinical measures of HIV or plasma biomarkers of inflammation. [C]PBR28 binding was not significantly elevated in cognitively impaired PLWH compared with unimpaired PLWH, but there were inverse relationships between cognitive performance (executive and global function) and [C]PBR28 binding in PLWH.

Conclusions

Our results suggest that neuroinflammation may play a role in cognitive deficits, but overall neuroinflammatory levels as measured by TSPO-PET imaging in PLWH are not significantly different from those seen in HIV- controls.

White matter hyperintensities (WMH) are nearly ubiquitous in the aging brain, and their topography and overall burden are associated with cognitive decline. Given their numerosity, accurate methods to automatically segment WMH are needed. Recent developments, including the availability of challenge data sets and improved deep learning algorithms, have led to a new promising deep-learning based automated segmentation model called TrUE-Net, which has yet to undergo rigorous independent validation. Here, we compare TrUE-Net to six established automated WMH segmentation tools, including a semi-manual method. We evaluated the techniques at both global and regional level to compare their ability to detect the established relationship between WMH burden and age. We found that TrUE-Net was highly reliable at identifying WMH regions with low false positive rates, when compared to semi-manual segmentation as the reference standard. TrUE-Net performed similarly or favorably when compared to the other automated techniques. Moreover, TrUE-Net was able to detect relationships between WMH and age to a similar degree as the reference standard semi-manual segmentation at both the global and regional level. These results support the use of TrUE-Net for identifying WMH at the global or regional level, including in large, combined datasets.