Comparison FT-IR spectral fingerprints of kidney Cancer in urine and serum: Clinical correlations and diagnostic potential.

The identification of non-invasive biomarkers that not only detect kidney cancer but also reflect clinically relevant tumor characteristics remains a major challenge in translational oncology. In this study, we aimed to identify specific FT-IR spectral wavenumbers in two biofluids: serum and urine and to directly correlate these spectroscopic markers with established clinicopathological parameters, including histological grade and tumor characteristics. By linking molecular-level spectral information with medical data, we sought to determine whether biochemical alterations detected in biofluids truly mirror histopathological findings and tumor biology. Second-derivative FT-IR spectra were collected from serum and urine samples obtained from patients with renal tumors and healthy controls. Multivariate exploratory analyses (PCA, UMAP), stability-based feature selection, and supervised classification with leave-one-patient-out cross-validation were performed independently for both biofluids. The most discriminatory and stable wavenumbers were identified and subsequently evaluated for their association with clinical variables. Urine spectra demonstrated stronger class separation and higher classification accuracy compared with serum, achieving up to 100% accuracy using SVM. Importantly, specific wavenumbers showed statistically significant correlations with histological grade in both serum (1160 cm-1, p = 0.0114) and urine (1300 cm-1, p = 0.0375), indicating that the detected spectral features reflect tumor aggressiveness rather than merely cancer presence. The most influential variables were located in the amide II and urea-related region (1530-1545, 1695 cm-1) for urine, and in phosphate, phospholipid and lipid carbonyl regions (844-980, 1160-1164, 1700-1780, 2984-2992 cm-1) for serum, consistent with locally excreted protein and urea changes in urine and systemic lipid-metabolism alterations in serum. These findings demonstrate that FT-IR spectroscopy of both serum and urine enables identification of spectral biomarkers that correlate with clinically meaningful pathological parameters. The integration of spectroscopic data with medical characteristics strengthens biological interpretability and supports the potential of FT-IR as a minimally invasive tool for kidney cancer detection and stratification.