Despite entering the era of combination antiretroviral therapy (CART), HIV-associated neurocognitive disorder (HAND) remains prevalent; however, less severe forms of HAND now predominate, and the most severe form, HIV-associated dementia, is rare In individuals treated with CART, the risk of HAND increases with age and in the presence of cardiovascular disease risk factors Latent HIV can persist in the brain even when systemic virological control is achieved with CART, thereby hampering efforts to eradicate HIV Animal models of CNS HIV infection — such as macaques infected with simian immunodeficiency virus — develop severe HAND, viral encephalitis and neuronal apoptosis, and are central to understanding the immunopathogenesis of HIV-induced CNS damage A growing body of work indicates that mild HAND can be modelled in immunocompetent mice infected with chimeric HIV (a model known as EcoHIV), and in chronically HIV-infected immunodeficient mice reconstituted with human immune systems To date, clinical trials of HAND therapies have been unsuccessful, but further trials for the treatment of HAND are forthcoming, including a trial of intranasal insulin

Vesicles shed from astrocytes after brain trauma trigger hepatic cytokine release to mobilize the peripheral immune response.

Nature Reviews Neurology 12, 234–248 (2016); 10.1038/nrneurol.2016.27 The initially published version of this article included an incorrect grant number. The correct number of the NIH grant for David J. Volsky is MH104145. This error has been corrected in the HTML and PDF versions of the article.

It remains unclear how different translocator protein (TSPO) ligands reflect the spatial extent of astrocyte or microglial activation in various neuroinflammatory conditions. Here, we use a reproducible lipopolysaccharide (LPS)-induced model of acute central nervous system inflammation to compare the binding performance of a new TSPO ligand (18)F-GE-180 with (11)C-(R)-PK11195. Using immunohistochemistry, we also explore the ability of the TSPO ligands to detect activated microglial cells and astrocytes. Lewis rats (n = 30) were microinjected with LPS (1 or 10 μg) or saline (1 μL) into the left striatum. The animals were imaged in vivo at 16 h after the injection using PET radiotracers (18)F-GE-180 or (11)C-(R)-PK11195 (n = 3 in each group) and were killed afterward for autoradiography of the brain. Immunohistochemical assessment of OX-42 and glial fibrillary acidic protein (GFAP) was performed to identify activated microglial cells and reactive astrocytes. In vivo PET imaging revealed an increase in the ipsilateral TSPO binding, compared with binding in the contralateral hemisphere, after the microinjection of 10 μg of LPS. No increase was observed with vehicle. By autoradiography, the TSPO radiotracer binding potential in the injected hemisphere was increased after striatal injection of 1 or 10 μg of LPS. However, the significant increase was observed only when using (18)F-GE-180. The area of CD11b-expressing microglial cells extended beyond that of enhanced GFAP staining and mapped more closely to the extent of (18)F-GE-180 binding than to (11)C-(R)-PK11195 binding. The signal from either PET ligand was significantly increased in regions of increased GFAP immunoreactivity and OX-42 colocalization, meaning that the presence of both activated microglia and astrocytes in a given area leads to increased binding of the TSPO radiotracers. (18)F-GE-180 is able to reveal sites of activated microglia in both gray and white matter. However, the signal is increased by the presence of activated astrocytes. Therefore, (18)F-GE-180 is a promising new fluorinated longer-half-life tracer that reveals the presence of activated microglia in a manner that is superior to (11)C-(R)-PK11195 due to the higher binding potential observed for this ligand.

Deep learning (DL), an emerging area of investigation in the fields of machine learning and artificial intelligence, has markedly advanced over the past years. DL techniques are being applied to assist medical professionals and researchers in improving clinical diagnosis, disease prediction and drug discovery. It is expected that DL will help to provide actionable knowledge from a variety of ‘big data’, including metabolomics data. In this review, we discuss the applicability of DL to metabolomics, while presenting and discussing several examples from recent research. We emphasize the use of DL in tackling bottlenecks in metabolomics data acquisition, processing, metabolite identification, as well as in metabolic phenotyping and biomarker discovery. Finally, we discuss how DL is used in genome-scale metabolic modelling and in interpretation of metabolomics data. The DL-based approaches discussed here may assist computational biologists with the integration, prediction and drawing of statistical inference about biological outcomes, based on metabolomics data.