Abstract
BACKGROUND: The classical approach to the analysis of kidney biopsies is based on semi-quantitative scores of the amount of sclerosis, inflammatory infiltrate, fibrosis and vascular damage. However, advanced renal lesions may be accompanied by a paucity of clinical features and, conversely, important clinical abnormalities may be accompanied by minimal histopathological changes. The objective of this study is to correlate new, semiautomatic, quantitative features of kidney biopsies (e.g. fractal analysis) with clinical and hematological parameters using a cross-sectional design. METHODS: Whole slide images from sixty-seven biopsies of patients diagnosed for diabetic nephropathy, hypertensive nephropathy, focal segmental glomerulosclerosis (FSGS) or IgA nephropathy have been used. The images have been semi-automatically quantified in the ImageJ environment, in order to derive the glomerular density, the tubular density, the number of tubules per glomerulus and the fractal dimension of the tubular lumen in the cortex (an index of complexity of the tubular lumen). For each patient, hemato-chemical data have been retrieved, including the uric acid level and the creatinine-based eGFR. RESULTS: A linear relationship between eGFR and glomerular density was observed in hypertension and FSGS, but not in diabetic nephropathy. Conversely, the eGFR correlated with the tubular density across different glomerular conditions. Moreover, the tubular density was linearly correlated with uric acid levels in different pathological conditions. The fractal dimension of tubular lumen was correlated with the eGFR but only in hypertensive patients. Finally, blood pressure was not correlated to any of the morphological indices tested. CONCLUSIONS: We propose the use of the fractal dimension as a new morphological descriptor of the nephron integrity.