NAFLD: Behind a statistical approach used for the inspection and association of omics and clinical data

Non-alcoholic fatty liver disease (NAFLD) prevalence and its related comorbidities are rapidly increasing and, if maintaining the current growth rate, they will reach epidemic proportions in the next coming few years. On the clinical point of view, several pathologic traits of NAFLD are shared with obesity, but others are closely linked to specific altered functions, proper of the symptomatic framework featuring the microbiota gut-brain axis of NAFLD. The altered homeostasis status revealed a burden of variables that we need to study to obtain a more complete clinical picture. In our advanced analysis workflow, the knowledge of gut microbiota members nearly inhabiting the gut niche, is granted by an approach that merges two distinct but converging omics i.e., faecal metabolomics coupled with metabarcoding taxa annotation. As a result, volatile organic compounds and 16S rRNA taxa signatures were used to ascertain the impact of dietary lifestyle and physical activity on a cohort of 109 NAFLD patients, randomly allocated to six life style intervention groups, where Mediterranean diet, aerobic and anaerobic activity were combined. Moreover, being us interested in inspecting other stratifications, the synergistic effect of diet and physical activity were also tested against a cohort subgroup that followed a treatment based on physical activity alone. The positive combined treatment evidently agrees with the decreased ‘controlled attenuation parameter’ (CAP) value, indicative of hepatic steatosis level. Additionally, a stringent statistical approach allowed the selection of those biochemical clinical parameters that we gathered into four factors, based on rotating factor analysis and namely: i) inflammatory/ sub-inflammatory, ii) a metabolic factor based on glucose, HbA1c, C peptide, insulin, (iii) a parenchymatous/hepatic factor and iv) a factor in common with cardiovascular pathologies (total and direct bilirubin and haemoglobin). Downstream, statistical correlation of VOC, taxa and clinical variables, allowed us to statistically weight each detected variables with a biological rationale