Salivary Glycopatterns as Potential Non-Invasive Biomarkers for Diagnosing and Reflecting Severity and Prognosis of Diabetic Nephropathy

Abstract

Discriminating between diabetic nephropathy (DN) and non-diabetic renal disease (NDRD) can help provide more specific treatments. However, there are no ideal biomarkers for their differentiation. Thus, the aim of this study was to identify biomarkers for diagnosing and predicting the progression of DN by investigating different salivary glycopatterns. Lectin microarrays were used to screen different glycopatterns in patients with DN or NDRD. The results were validated by lectin blotting. Logistic regression and artificial neural network analyses were used to construct diagnostic models and were validated in in another cohort. Pearson's correlation analysis, Cox regression, and Kaplan-Meier survival curves were used to analyse the correlation between lectins, and disease severity and progression. Liquid chromatography-tandem mass spectrometry (LC-MS/MS) and bioinformatics analyses were used to identify corresponding glycoproteins and predict their function. Both the logistic regression model and the artificial neural network model achieved high diagnostic accuracy. The levels of Aleuria aurantia lectin (AAL), Lycopersicon esculentum lectin (LEL), Lens culinaris lectin (LCA), Vicia villosa lectin (VVA), and Narcissus pseudonarcissus lectin (NPA) were significantly correlated with the clinical and pathological parameters related to DN severity. A high level of LCA and a low level of LEL were associated with a higher risk of progression to end-stage renal disease. Glycopatterns in the saliva could be a non-invasive tool for distinguishing between DN and NDRD. The AAL, LEL, LCA, VVA, and NPA levels could reflect the severity of DN, and the LEL and LCA levels could indicate the prognosis of DN.

Keywords: diabetic nephropathy; diagnosis; glycopatterns; non-invasive biomarkers; prognosis; saliva.

Figure 1

Schematic illustration of the present research. DN, diabetic nephropathy; NDRD, non-diabetic renal disease; MN, membranous nephropathy; IgAN, immunoglobulin A nephropathy; FSGS, focal segmental glomerulosclerosis.

Figure 2

Exhibition and confirmation of differentiation with DN and NDRD using lectins. (A) Comparison of all candidate lectins in salivary samples from DN and NDRD patients. The normalised fluorescence intensities of 37 lectins from DN and NDRD patients were compared based on fold-change and one-way ANOVA (*p < 0.05, **p < 0.01, and ***p < 0.001). Data are presented as the mean ± 95% CI. (B) Hierarchical clustering analysis of the three lectins with significant differentiation of NFIs between DN and NDRD. Glycan profiles of DN and NDRD patients were clustered (average linkage, correlation similarity). Samples are listed in columns, and lectins are listed in rows. The colour intensity of each square indicates the expression levels relative to other data. Red, high; green, low; black, medium. (C) The normalised glycopattern abundances of three lectins related to the two groups were subjected to principal component analysis (PCA). DN and NDRD samples were visualised by red and grey shadows, respectively. (D) Confirmation of salivary glycopatterns from DN and NDRD groups using LEL and VVA lectins was performed by lectin blotting. (E) Mean grey value of each apparent difference band was obtained using ImageJ. DN, diabetic nephropathy; NDRD, non-diabetic renal disease; NFIs, normalised fluorescence intensities; LEL, Lycopersicon esculentum lectin; VVA, Vicia villosa lectin.

Figure 3

Diagnostic accuracy of selected lectins and models was determined by ROC analysis with logistic regression and artificial neural network methods. ROC analysis for models constructed by logistic regression in the training (A) and validation (B) cohorts. ROC analysis for models constructed by artificial neural network in the training (C) and validation (D) cohorts. (E) A total of 37 candidate lectins for all patients are displayed in artificial neural network analysis. DN, diabetic nephropathy; NDRD, non-diabetic renal disease; ROC, receiver operating characteristic.

Figure 4

Kaplan–Meier analysis of dialysis-free survival in patients with diabetic nephropathy. Subjects were dichotomised based on the mean of the covariates: (A) 0.036 for LEL; (B) 0.060 for LCA. p-Values refer to log-rank tests. LEL, Lycopersicon esculentum lectin; LCA, Lens culinaris lectin.

Figure 5

Bioinformatics analysis of isolated glycoproteins from DN and NDRD. (A) Venn diagram of isolated proteins from DN and NDRD using LEL-coupled magnetic particle conjugates. (B) Venn diagram of isolated peptides from DN and NDRD using LEL-coupled magnetic particle conjugates. (C) Scatter plot of protein levels between DN and NDRD. y-Axis correspond to p-values (−log10) versus protein log2 fold-change (x-axis) in DN/NDRD. Colour indicates upregulation (orange) (fold-change > 2, p < 0.01) and downregulation (blue) (fold-change < 0.5, p < 0.01). Black represents the level of proteins without statistically significant difference between NDRD and DN. (D) Blast2GO was used to classify identified proteins into biological process, cellular component, and molecular function. DN, diabetic nephropathy; NDRD, non-diabetic renal disease; LEL, Lycopersicon esculentum lectin.

Figure 6

Bioinformatics analysis of differential glycoproteins isolated from DN and NDRD. Differential proteins were analysed using Gene Ontology (GO). DN, diabetic nephropathy; NDRD, non-diabetic renal disease. (A) Pie charts showing the biological processes of differential proteins between DN and NDRD. (B) Pie charts showing the cellular component of differential proteins between DN and NDRD. (C) Pie charts showing the molecular function of differential proteins between DN and NDRD. Next to their position are shown the associated term names on the chart. (D) STRING 9.0 was used to generalise and visualise the protein interaction network of differential proteins from DN. (E) STRING 9.0 was used to generalise and visualise the protein interaction network of differential proteins from NDRD. Line thickness represents the strength of the association between molecules. Networks with three or more protein interactions are shown. The confidence (score) required for protein association is high. The selected protein core complexes with important functions and proteins involved in the same biochemical reaction are marked with a red dotted line.