Diabetic nephropathy (DN) is a major cause of global kidney failure. While histological kidney biopsy is the gold standard for diagnosis, it primarily reveals tissue morphology. In contrast, near-infrared (NIR) microscopy offers a label-free method for detailed molecular characterization of kidney tissue. Hematoxylin and eosin-stained kidney tissue samples from 17 ob/ob mice with DN and 14 healthy mice were examined using Fourier transform-NIR microscopy. Four different spectra were obtained from both the mesangium and tubulus. NIR spectral analysis unveiled distinct differences in wavenumbers between DN-affected and healthy kidneys, notably in the carbohydrate and protein-associated region (5500-4200 cm−1). In the mesangium, DN tissue samples exhibited higher median values at 4235 cm−1, 4659 cm−1, 4844 cm−1, 4906 cm−1, and 5222 cm−1 compared to controls (P < 0.05, P < 0.01, P < 0.05, P < 0.05 and P < 0.001, respectively). In tubular spectra, higher median values were found at 4258 cm−1, 4659 cm−1, 5222 cm−1, and 5346 cm−1 in the DN group (P < 0.01, P < 0.05, P < 0.05 and P < 0.01, respectively). These spectral differences strongly correlated with metabolic, histologic, and urinary parameters, providing valuable DN progression insights. The classification model achieved a visible clustering between the control and DN group for both the mesangial and tubular measurements. NIR microscopy demonstrated significant spectral differences between DN and healthy kidney tissues in mice, hinting at its potential for providing chemical insights, aiding in more accurate diagnoses, and offering a foundation for further clinical exploration and potential therapeutic advancements in DN.
Read full abstract