Identification of histopathological and clinical spectrum of diabetic kidney disease based on an unsupervised hierarchical clustering analysis of elderly autopsy specimens

Abstract

Aim Diabetic kidney disease (DKD), a chronic kidney disease caused by diabetes and other comorbidities, is the leading cause of end‐stage renal disease. The pathogenesis of DKD is diverse and influenced by various causes, some but not all of which cause proteinuria. Some factors such as hypertension can modify DKD. Therefore, the spectrum of DKD is difficult to elucidate and remains unsolved. This study aims to classify and characterize DKD. Methods We examined autopsy specimens from type 2 diabetes mellitus (DM) (n = 44) and non‐DM (n = 21) groups. Results The frequency of interstitial fibrosis and tubular atrophy was higher in patients with proteinuric DKD than in those with non‐proteinuric DKD. The presence of polar vasculosis was associated with hypertension in DKD. In addition, an unsupervised hierarchical clustering analysis revealed the spectrum of renal histopathology findings for more‐proteinuric and less‐proteinuric DKD. With changes in the diagnostic criteria for hypertension and advances in antihypertensive drugs, the pathogenesis of DKD may be changing. Furthermore, a decision tree model suggested how diabetes, hypertension, and dyslipidemia interacted in predicting the characteristics of DKD. Conclusion Polar vasculosis is a good indicator of the presence of DM and hypertension. Furthermore, the histopathological and clinical spectrum of DKD were related to the interaction of diabetes, hypertension, and dyslipidemia. These histopathological and clinical results may help to show the range of patient characteristics when conducting clinical trials and could help to determine whether chronic kidney disease is caused by DM or some other cause.