Application Of Quantitative Analysis Of Renal Histopathological Lesions In Patients With Diabetic Nephropathy

Chapter one:Artificial intelligence assist identification and pathological classification of kidney lesionsObjective:To assist identification of glomerular lesions and pathological classification in patients with diabetic nephropathy(DN)using artificial intelligence(ARPS system).Methods:The clinicopathological data of patients with DN were collected.The performance of ARPS identifying glomerular types and intrinsic cells in patients with DN was evaluated.The relationships between these structures and pathological classifications were evaluated.Results:Of 272 patients with DN,ARPS can identify glomerular types and intrinsic cells with accuracy above 0.84,except for segmental glomerulosclerosis and crescent.With pathological classification increased,the numbers of mesangial cells and the mesangial fractional area increased,the number of podocytes decreased(p<0.001),while the number of endothelial cells remains constant.Mesangial cells,podocytes and mesangial area related indicators predicted pathological classification with AUC above 0.7.Conclusion:ARPS can identify the glomerular lesions in patients with DN,and there were significant correlations between mesangial cells,podocytes,mesangial fractional area with pathological classifications.Chapter two:Measurement and analysis of renal cortex glomerular density in patients with diabetic nephropathyObjective:To measure glomerular density(GD)and analyze its possible influencing factors and prognostic effect in patients with DN.Methods:The area of renal cortex tissue was measured by Aperio’s image scope software.The total glomerular density(GD1)and non-sclerotic glomeruli density(GD2)was calculated.The structural and possible clinical factors for glomerular density were analyzed by several methods including artificial intelligence or multiple linear regression models.Kaplan-Meier curve was used to assess the impact of GD on renal survival rates.Results:Among patients with DN,GD1 increased with eGFR decreased,but GD2 decreased.GD was lowest in patients with DN than in those with others disease(P<0.001).Renal cortical volume,glomerular volume,glomerular sclerosis,and interstitial fibrosis were all associated with GD in DN.The strongest predictor for GD was eGFR(P<0.001).The increased or decreased GD1 were both related to poor renal survivals in patients with DN.Conclusion:The changes in GD were associated with the severity and renal survival rates of patients with DN,which were affected by several structural factors including renal cortical volume,glomerular volume,glomerular sclerosis,and interstitial fibrosis.Chapter three:The clinical acute kidney injury and histologic acute tubularinterstitial injury and prognosis in diabetic nephropathyObjective:To determine the association between the clinical AKI(cAKI)and histologic acute tubular-interstitial injury(hATI)and its relationships with prognosis in patients with DN.Methods:Patients with biopsy-proven DN were selected.The incidence of cAKI in patients with hATI were investigated.Kaplan-Meier curve and Cox regression model were used to analyze the effect of hATI and cAKI on the renal survival rates and prognosis.Results:Of 1414 patients with biopsy-proven DN,the incidence of cAKI was only 12.2%in patients with hATI.The incidence and degrees of cAKI were increased by the percentage involvement of hATI(P<0.001).hATI could significantly affect the renal survival rates(P<0.001).After adjusting for significant covariates,even in patients without cAKI,hATI was still an independent risk factor of progression to ESRD(HR,1.46;95%CI,1.05,2.02)and creatinine doubling(HR,1.68;95%CI,1.28,2.19).Conclusion:Our findings indicated that only 12.2%cases met the cAKI criteria among patients with DN accompanied with hATI.Even in patients without cAKI,hATI was still an independent risk factor of DN progression.