Standardization of urinary tract infection diagnosis by ATR-FTIR/PLSDA

Introduction: The UTI diagnosis includes positivity for leukocyte esterase, nitrite, Gram staining, and urine culture (UC, reference method). UC can take 24h (bacteria) to 48h (Candida spp.) forcing the clinician to prescribe early antibiotic therapy, sometimes unnecessary. Therefore, it would be more practical to make this result faster. In this context, we proposed the elaboration of a predictive model of UTI using infrared spectroscopy (FTIR) and chemometrics aiming the substitution of the UC. Material and methods: Urine samples (n=232) were analyzed by UC (1 µL, CLED agar, 37ºC, 48h). Samples were classified as UTI(+) when have presented ≥ 1.105 colony-forming unit (CFU/mL). FTIR analysis was carried out by Attenuated Total Reflectance (ATR, 8 scans, 4000-650 cm-1) using dehydrated urine films (50 µL on 35 mm2 aluminum surfaces) in triplicates. Spectral dataset was normalized by amplitude, the average spectrum of each sample was calculated, and the average ATR-FTIR spectra dataset (Figure 1) was analyzed by PLS-DA with preprocessing (2nd derivative, 5 points, two orthogonal signal correction components) in calibration step (training samples, leave-one-out cross-validation, LOO-CV) and external validation step (test samples, EVS, 2:1).