Validation And Re-establishment Of The Differential Diagnostic Model Of Type2Diabetic Nephropathy

Background and objective:Renal diseases in diabetes include diabetic nephropathies (DN) and non-diabetic renal diseases (NDRD). Differential diagnosis of DN and NDRD is of great significance for clinical treatment and prognosis. This study aims to validate the previous diagnostic model (NDT equation) and analyze the changes of clinicopathological features of diabetes with kidney disease in order to establish new differential diagnostic equation.Methods:Retrieve the clinical data of335type2diabetic patients with kidney disease diagnosed by renal biopsy in the Department of Nephrology, PLA General Hospital from January2004to March2012, excluding DN superimposed on NDRD or with ambiguous diagnosis and finally enrolled200patients were divided into two groups according to pathological features,93in DN group and107in NDRD group. Clinical and laboratory data were compared between two groups and validation tests showed the diagnostic efficacy of the NDT equation. Based on the logistic regression model, a new diagnostic model was developed and evaluated by the area under the ROC curve. We then enrolled55patients with same inclusion criteria from April2012to December2012for validation of the new equation. PDN≥0.5were diagnosised as DN,PDN<0.5as NDRD.Results:Among200cases of patients with renal biopsy,46.5%of patients were DN and53.5%were NDRD.①We used pathology results as the golden standard, validation tests showed that the model had a sensitivity of58.1%, a specificity of94.4%, a positive predictive value of90.0%, a negative predictive value of72.1%and a total consistency rate of77.5%.②Compared to DN group from1993to2003, the DN group from2004to2012had higher incidence of hematuria (32.3%versus16.7%, P<0.01), longer diabetes duration (139.7±75.3versus87.6±54.3, P<0.01), better glycemic control (7.3±1.6versus8.4±2.0, P<0.05).③NDRD group from2004to2012included more pathological types than in the past. IgA nephropathy accounting for32.7%and membranous nephropathy ranked second accounting for18.7%of all NDRD. Obesity-associated glomerulonephritis, focal sclerosing glomerulonephritis (7.5%) and interstitial nephritis (4.7%) were most common in new pathological types.④Stepwise multivariate regression analysis identified diabetes duration (OR=1.022, P<0.001), systolic blood pressure (1.034,0.013), HbAlc (7.766,0.002), diabetic retinopathy (18.948, P<0.001), hematuria (0.070,0.004), hemoglobin (0.925, P<0.001) as significant predictors.⑤Based on the logistic regression model, a diagnostic model was constructed as follows:PDN=exp (0.864+0.022Dm+0.033Bp+2.050Gh-2.664Hu-0.078Hb+2.942Dr)/[1+exp (0.864+0.022Dm+0.033Bp+2.050Gh-2.664Hu-0.078Hb+2.942Dr)] The AUC was0.971and back-substitution validation tests showed better sensitivity (88.5%) and specificity (91.0%). Validation tests showed the new model had better total consistency rate (90.9%) than the NDT equation (81.8%).Conclusion:The clinical features of diabetic patients with kidney disease in China had great changes. The new established model may be better than NDT equation about the current clinical differential diagnosis of DN and NDRD.