Rapid and non-invasive screening of high-risk human papillomavirus using Fourier transform infrared spectroscopy and multivariate analysis

Abstract

Persistent high-risk human papillomavirus (HR-HPV) infection is leading cause for the occurrence of cervical cancer, and timely detection and early treatment of HR-HPV infection can effectively reduce incidence of cervical cancer. In this study, a rapid and non-invasive method for detecting HR-HPV was proposed by using Fourier transform infrared (FT-IR) spectra of cervical exfoliated cells combined with multivariate analysis. A total of 100 spectra were recorded from 50 HR-HPV positive patients and 50 normal subjects. The obvious difference in infrared spectrum between the two groups was mainly shown at 1042 cm−1 (mucin), 1246 cm−1 (amide III), 1396 cm-1 (proteins), 1543 cm−1 (amide II), 1651 cm−1 (amide I), 2361 cm−1 (CO2), 2928 cm−1 (lipids), and 3294 cm−1 (amide A). Then, a principal component analysis-linear discriminant analysis (PCA-LDA) diagnostic model was developed and applied on the FT-IR spectra of normal samples as well as HR-HPV positive patients, and satisfactory classification results were obtained. The diagnostic accuracy, specificity, and sensitivity were 98 %, 98 %, and 98 %, respectively. Furthermore, the area under the receiver operating characteristic (ROC) curve (AUC) was 0.997, which could further demonstrate the feasibility of the PCA-LDA model. Therefore, our exploratory work shows that the combination of FT-IR spectroscopy and PCA-LDA model has great potential for HR-HPV screening.