*2.1. Study Design and Population*

We conducted a cross sectional study within the framework of the TOSCA.IT study (NCT00700856), a randomized controlled trial designed to evaluate the effects of sulfonylurea or pioglitazone, in add-on to metformin, on cardiovascular events in people with T2D. For the aim of the present study, we used data collected at baseline, prior to randomization to the study treatments.

The study participants were people with T2D, aged 50–75 years, on stable treatment with a full dose of metformin (2–3 g per day), and with a glycated hemoglobin (HbA1c) between 7% and 9%. Participants with incomplete data sets, those with alcohol intake exceeding 30 g/day if men and 20 g/day if women, or taking *n*-3 supplements were excluded from the analyses [28]; other exclusion criteria were severe hepatic dysfunction (plasma ALT values > 2.5 times the upper normal limit), serum creatinine > 1.5 mg/dL, history of congestive heart failure, (NYHA class I or higher), ulcer or gangrene of the lower extremities, cancer, substance abuse, and any health problem requiring special dietary treatments. Details of the study protocol have been published [29,30]. NASH was defined based on ION ≥ 50. To identify the association of the habitual diet with NASH, dietary patterns associated with ION ≥ 50 or <50 were derived using the K-means cluster analysis by which the sample population is classified into homogenous groups presenting different characteristics using a specific variable as the comparison criterion, in our case the ION. To perform this analysis, firstly the 248 food items were categorized into 59 food groups based on their similarity in term of nutrient composition. Then, the K-means clustering method was performed and the algorithm utilized to identify within each cluster the smallest variation. Two clusters were produced using a non-hierarchical K-means clustering method, with the random seed and 10 iterations to refine and optimize the classifications, and participants were grouped according to Euclidean distances. Two clusters were identified, one associated with an ION ≥ 50, the other associated with an ION < 50, respectively defined in the text and tables as cluster NASH or cluster NO-NASH. The anthropometric, metabolic and nutritional variables were compared in these two patterns. The study protocol has been approved by the Ethics Review Committee of the Coordinating Center and of each participating center. All participant provided written informed consent before entering the study.
