Background: Metabolic syndrome (MetS), prediabetes (PreDM) and metabolic-associated fatty liver disease (MAFLD) share pathophysiological pathways concerning type 2 diabetes mellitus (T2DM) onset. The non-invasive assessment of MAFLD combined with PreDM and MetS features might provide further accuracy in predicting hyperglucaemic status with the description of singular clinical phenotypes.Methods: A retrospective cohort study was performed on 2799 patients recruited in the Vascular-Metabolic CUN cohort. The main outcome was the incidence of T2DM according to ADA criteria. MetS and PreDM were defined according to ATP III and ADA criteria, respectively. Hepatic steatosis index (HSI) was used to detect patients with MAFLD.Results: MetS and PreDM were more common in patients with MAFLD (35% vs. 8% and 34% vs. 18%, respectively). MAFLD showed clinical interaction with MetS and PreDM in the prediction of T2DM [MAFLD-MetS interaction HR= 4.48 (3.37-5.97) and MAFLD-PreDM interaction HR = 6.34 (4.67-8.62)]. These findings supported the description of five different liver status-linked phenotypes with increasing risk of T2DM: Metabolically healthy patients (1,5% of T2DM incidence), MAFLD (4,4% of T2DM incidence), MAFLD and MetS (10,6% of T2DM incidence), PreDM (11,1% of T2DM incidence) and MAFLD and PreDM (28,2% of T2DM incidence). These phenotypes provided independent capacity of prediction of T2DM incidence after adjustment for age, sex, tobacco and alcohol consumption, obesity and number of SMet features with a c-Harrell=0.84.Conclusions: MAFLD interplay with MetS and PreDM might help to discriminate patient risk of T2DM in the clinical setting through metabolic remodelling-based phenotypes.