Diabetic nephropathy (DN), as the one of most common complications of diabetes, is generally diagnosed based on a longstanding duration, albuminuria, and decreased kidney function. Some patients with the comorbidities of diabetes and other primary renal diseases have similar clinical features to DN, which is defined as non-diabetic renal disease (NDRD). It is necessary to distinguish between DN and NDRD, considering they differ in their pathological characteristics, treatment regimes, and prognosis. Renal biopsy provides a gold standard; however, it is difficult for this to be conducted in all patients. Therefore, it is necessary to discover non-invasive biomarkers that can distinguish between DN and NDRD. In this research, the urinary exosomes were isolated from the midstream morning urine based on ultracentrifugation combined with 0.22 μm membrane filtration. Data-independent acquisition-based quantitative proteomics were used to define the proteome profile of urinary exosomes from DN (n = 12) and NDRD (n = 15) patients diagnosed with renal biopsy and Type 2 diabetes mellitus (T2DM) patients without renal damage (n = 9), as well as healthy people (n = 12). In each sample, 3372 ± 722.1 proteins were identified on average. We isolated 371 urinary exosome proteins that were significantly and differentially expressed between DN and NDRD patients, and bioinformatic analysis revealed them to be mainly enriched in the immune and metabolic pathways. The use of least absolute shrinkage and selection operator (LASSO) logistic regression further identified phytanoyl-CoA dioxygenase domain containing 1 (PHYHD1) as the differential diagnostic biomarker, the efficacy of which was verified with another cohort including eight DN patients, five NDRD patients, seven T2DM patients, and nine healthy people. Additionally, a concentration above 1.203 μg/L was established for DN based on the ELISA method. Furthermore, of the 19 significantly different expressed urinary exosome proteins selected by using the protein-protein interaction network and LASSO logistic regression, 13 of them were significantly related to clinical indicators that could reflect the level of renal function and hyperglycemic management.
Read full abstract