A common denominator of metabolic diseases, including type 2 diabetes Mellitus, dyslipidemia, and atherosclerosis, are elevated oxidative stress and chronic inflammation. These complex, multi-factorial diseases are caused by the detrimental interaction between the individual genetic background and multiple environmental stimuli. The cells, including the endothelial ones, acquire a preactivated phenotype and metabolic memory, exhibiting increased oxidative stress, inflammatory gene expression, endothelial vascular activation, and prothrombotic events, leading to vascular complications. There are different pathways involved in the pathogenesis of metabolic diseases, and increased knowledge suggests a role of the activation of the NF-kB pathway and NLRP3 inflammasome as key mediators of metabolic inflammation. Epigenetic-wide associated studies provide new insight into the role of microRNAs in the phenomenon of metabolic memory and the development consequences of vessel damage. In this review, we will focus on the microRNAs related to the control of anti-oxidative enzymes, as well as microRNAs related to the control of mitochondrial functions and inflammation. The objective is the search for new therapeutic targets to improve the functioning of mitochondria and reduce oxidative stress and inflammation, despite the acquired metabolic memory.

To assay new circulating markers related to macrovascular complications (MVC) in type 2 diabetes mellitus (T2DM), we carried out a descriptive cross-sectional study. We recruited 30 controls (CG), 34 patients with T2DM (DG), and 28 patients with T2DM and vascular complications (DG+C); among them, 22 presented MVC. Peripheral blood was used to determine redox status (superoxide dismutase, SOD; catalase, CAT; glutathione reductase, GRd; glutathione peroxidase, GPx; glucose-6-phosphate dehydrogenase, G6PD) and markers of oxidative damage (advanced oxidation protein products, AOPP; lipid peroxidation, LPO), nitrite levels in plasma (NOx). Inflammatory markers (IL-1β, IL-6, IL-10, IL-18, MCP-1, TNF-α) and the relative expression of c-miRNAs were analyzed. The real-time PCR results showed that the expressions of miR-155-5p, miR-21-5p, miR-146a-3p, and miR-210-3p were significantly higher in the DG group compared to the CG. The DG+C group presented statistically relevant differences with CG for four miRs: the increased expression of miR-484-5p, miR-21-5p, and miR-210-3p, and decreased expression of miR-126a-3p. Moreover, miR-126a-3p was significantly less expressed in DG+C compared to DG. The application of binary logistic regression analysis and construction of receiving operator characteristic curves (ROC) revealed two models with high predictive values for vascular complications presence: (1) HbAc1, creatinine, total cholesterol (TC), LPO, GPx, SOD, miR-126, miR-484 (Exp(B) = 0.926, chi² = 34.093, p < 0.001; AUC = 0.913). (2) HbAc1, creatinine, TC, IL-6, LPO, miR-126, miR-484 (Exp(B) = 0.958, Chi² = 33.863, p < 0.001; AUC = 0.938). Moreover, our data demonstrated that gender, TC, GPx, CAT, and miR-484 were associated with MVC and exhibited higher predictive values (Exp(B) = 0.528, p = 0.024, Chi² = 28.214, AUC = 0.904) than classical variables (Exp(B) 0.462, p = 0.007, Chi² = 18.814, AUC = 0.850). miR-126, miR-484, IL-6, SOD, CAT, and GPx participate in vascular damage development in the studied diabetic population and should be considered for future studies.

Osteoglycin (OGN) could be a biomarker of mild kidney function impairment in type 2 diabetes (T2D). Our study aimed to determine the association between serum OGN and impaired kidney function risk in T2D patients and to analyze its potential role as an estimator of kidney disturbances in this population. This cross-sectional study included 147 T2D patients (65 ± 8 years, 58.5% males), and 75 healthy controls (63 ± 10 years, 36% males). Circulating OGN levels were determined by ELISA. Linear regression modeling was performed to determine the variables influencing circulating OGN, and an ROC curve was plotted to assess the usefulness of OGN as an estimator of diabetic kidney disease risk. Circulating OGN was significantly increased in T2D patients compared to controls (18.41 (14.45–23.27) ng/mL vs. 8.74 (7.03–12.35) ng/mL; p < 0.001). We found a progressive increase in serum OGN according to the severity of kidney impairment in T2D patients (normal kidney function: 16.14 (12.13–20.48) ng/mL; mildly impaired kidney function: 19.15 (15.78–25.90) ng/mL; moderate impaired kidney function: 21.80 (15.06–29.22) ng/mL; p = 0.006). Circulating OGN was an independent estimator of mildly impaired kidney function risk in T2D patients. We suggest that serum OGN could act as an albuminuria-independent biomarker of incipient kidney dysfunction in T2D patients.

To evaluate new circulating markers related to macrovascular complications (MVC) in type 2 diabetes mellitus (T2DM), we recruited 30 controls (CG), 34 patients with T2DM (DG), and 28 patients with T2DM and vascular complications (DG+C), among them, 22 presented MVC. Pe-ripheral blood was used to determine redox status (superoxide dismutase, SOD; catalase, CAT; glutathione reductase, GRd; glutathione peroxidase, GPx; glucose-6-phosphate dehydrogenase, G6PD) and markers of oxidative damage (advanced oxidation protein products, AOPP; lipid pe-roxidation, LPO). Inflammatory markers (IL-1β, IL-6, IL-10, IL-18, MCP-1, TNF-α) and the relative expression of c-miRNAs were analyzed. The real-time PCR results showed that the expressions of miR-155-5p, miR-21-5p, miR-146a-3p, and miR-210-3p were significantly higher in the DG group compared to CG. DG+C group presented statistically relevant differences with CG for four miRs: the increased expression of miR-484-5p, miR-21-5p, and miR-210-3p, and decreased for miR-126a-3p. Moreover, miR-126a-3p was significantly less expressed in DG+C compared to DG. The application of binary logistic regression analysis and construction of receiving operator characteristic curves (ROC) revealed two models with high predictive value for vascular complications presence: 1) HbAc1, creatinine, total cholesterol (TC), LPO, GPx, SOD, miR-126, miR-484. 2) HbAc1, creatinine, TC, IL-6, LPO, miR-126, miR-484. Moreover, our data demonstrated that gender, TC, GPx, CAT, and miR-484 were associated with MVC and exhibited higher predictive values than classical variables. miR-126, miR-484, IL-6, SOD, CAT, and GPx participate in vascular damage development in the studied diabetic population and should be considered for future studies.

Continuous glucose monitoring enables Blood Glucose Level (BGL) forecasting, yet current deep learning models face a critical trade-off: general models lack individual accuracy, while personalized models suffer from a severe “cold-start” problem requiring extensive historical data. This study proposes an intermediate group-based forecasting paradigm utilizing time series clustering. We trained Long Short-Term Memory networks under a general, group-based, and personalized approach to forecast BGL at 60- and 120-min prediction horizons. Based on our previous research, patients were partitioned into six groups using K-means clustering applied to time series data. Results indicate that group-based models outperformed general baselines for patient groups with controlled BGL, enhancing clinical safety and reducing forecasting errors. Crucially, the group-based approach proved more robust than pure personalization, matching or exceeding personalized performance in 75% of overall cases. Most notably, personalized models failed to predict severe hypoglycemia in controlled patients due to a lack of these events in individual historical data, a vulnerability successfully bypassed by group-based models leveraging collective cluster histories. Ultimately, group-based forecasting emerges as a practical solution for Type 1 Diabetes management, providing the accuracy needed for critical event prediction while drastically reducing the data accumulation period required for rapid clinical deployment.