Cervical cancer (CC) is a significant cause of morbidity and mortality in women, often preceded by high-grade cervical intraepithelial lesions (HSIL). Although conventional cytology (Pap smear) is widely used for screening, its sensitivity limitations and high false-positive rate reinforce the need for complementary methods. This study investigated the feasibility of low-field 1H NMR spectroscopy combined with chemometric modeling to differentiate healthy individuals (CON) from patients with HSIL and CC. Principal Component Analysis (PCA) was applied to explore metabolic patterns and identify relevant spectral variables in group differentiation. PCA1 highlighted the separation between CC and the other groups, while PCA2 and PCA3 evidenced intermediate metabolic characteristics in HSIL, reinforcing its role as a transition stage. Three classification scenarios were evaluated using Data-Driven Soft Independent Modeling of Class Analogy (DD-SIMCA): (1) CON as target class, HSIL/CC as outclass classes; (2) HSIL as target class, CON as outclass class; and (3) CC as target class, CON as outclass class. Calibration was optimal (100 % SEN, SPE, ACC, MCC), and prediction showed higher efficacy in detecting CC (SPE = 100 %, MCC = 70 %), indicating that the model was more efficient in screening cervical cancer cases. Furthermore, low-field 1H NMR has demonstrated potential as a metabolomic screening tool. It is a promising alternative due to its greater accessibility, lower operational cost, and non-invasive nature, complementing traditional methods of early detection of tumors and cervical lesions.
Keywords: Cervical cancer; Chemometrics; DD-SIMCA; Metabolomic; low-field benchtop NMR.
Copyright © 2025 Elsevier B.V. All rights reserved.