The value of neck adipose tissue as a predictor for metabolic risk in health and type 2 diabetes

Upper-body adiposity is adversely associated with metabolic health whereas the opposite is observed for the lower-body. The neck is a unique upper-body fat depot in adult humans, housing thermogenic brown adipose tissue (BAT), which is increasingly recognised to influence whole-body metabolic health. Loss of BAT, concurrent with replacement by white adipose tissue (WAT), may contribute to metabolic disease, and specific accumulation of neck fat is seen in certain conditions accompanied by adverse metabolic consequences. Yet, few studies have investigated the relationships between neck fat mass (NFM) and cardiometabolic risk, and the influence of sex and metabolic status. Typically, neck circumference (NC) is used as a proxy for neck fat, without considering other determinants of NC, including variability in neck lean mass. In this study we develop and validate novel methods to quantify NFM using dual x-ray absorptiometry (DEXA) imaging, and subsequently investigate the associations of NFM with metabolic biomarkers across approximately 7000 subjects from the Oxford BioBank. NFM correlated with systemic insulin resistance (Homeostatic Model Assessment for Insulin Resistance; HOMA-IR), low-grade inflammation (plasma high-sensitivity C-Reactive Protein; hsCRP), and metabolic markers of adipose tissue function (plasma triglycerides and non-esterified fatty acids; NEFA). NFM was higher in men than women, higher in type 2 diabetes mellitus compared with non-diabetes, after adjustment for total body fat, and also associated with overall cardiovascular disease risk (calculated QRISK3 score). This study describes the development of methods for accurate determination of NFM at scale and suggests a specific relationship between NFM and adverse metabolic health.

Discovery of drug-omics associations in type 2 diabetes with generative deep-learning models

The application of multiple omics technologies in biomedical cohorts has the potential to reveal patient-level disease characteristics and individualized response to treatment. However, the scale and heterogeneous nature of multi-modal data makes integration and inference a non-trivial task. We developed a deep-learning-based framework, multi-omics variational autoencoders (MOVE), to integrate such data and applied it to a cohort of 789 people with newly diagnosed type 2 diabetes with deep multi-omics phenotyping from the DIRECT consortium. Using in silico perturbations, we identified drug-omics associations across the multi-modal datasets for the 20 most prevalent drugs given to people with type 2 diabetes with substantially higher sensitivity than univariate statistical tests. From these, we among others, identified novel associations between metformin and the gut microbiota as well as opposite molecular responses for the two statins, simvastatin and atorvastatin. We used the associations to quantify drug-drug similarities, assess the degree of polypharmacy and conclude that drug effects are distributed across the multi-omics modalities.