Prediction of viral loads for diagnosis of Hepatitis C infection in human plasma samples using Raman spectroscopy coupled with partial least squares regression analysis

• DOI: 10.1002/jrs.5108
• Publication Date: 2017-03-31T17:13:01Z
• Authors: H. Nawaz; N. Rashid; M. Saleem; M. Asif Hanif; M. Irfan Majeed; I. Amin; M. Iqbal; M. Rahman; Ola Ibrahim; S. M. Baig; M. Ahmed; F. Bonnier; H. J. Byrne
• Keywords: Partial Least Squares Regression; Principal Components Analysis; Raman spectroscopy; Hepatitis C Virus (HCV) infection; blood plasma; Principal Components Analysis; Partial Least Squares Regression; 01 natural sciences; 3. Good health; 620; 0104 chemical sciences; [CHIM.ANAL]Chemical Sciences/Analytical chemistry; [SDV.MHEP.MI]Life Sciences [q-bio]/Human health and pathology/Infectious diseases; Raman spectroscopy; [SDV.MP.VIR]Life Sciences [q-bio]/Microbiology and Parasitology/Virology; Medicine and Health Sciences; Hepatitis C Virus (HCV) infection; blood plasma

Raman spectroscopy has been used to identify the biochemical changes associated with the presence of the Hepatitis C virus (HCV) in infected human blood plasma samples as compared with healthy samples, as control. The aim of the study was to establish the Raman spectral markers of hepatitis infection, which could be used for diagnostic purposes. Moreover, multivariate data analysis techniques, including Principal Component Analysis (PCA), coupled with Linear Discriminant Analysis (LDA), and Partial Least Square Regression (PLSR) are employed to further demonstrate the diagnostic capability of the technique. The PLSR model is developed to predict the viral loads of the HCV infected plasma on the basis of the biochemical changes caused by the viral infection.Specific Raman spectral features are observed in the mean spectra of HCV plasma samples which are not observed in the control mean spectra. PCA differentiated the ‘normal’ and ‘HCV’ groups of the Raman spectra and PCA‐LDA was employed to increase the efficiency of prediction of the presence of HCV infection, resulting in a sensitivity and specificity 98.8% and 98.6%, with corresponding Positive Predictive Value of 99.2%, and Negative Predictive Value of 98%. PLSR modelling was found to be 99% accurate in predicting the actual viral loads of the HCV samples, as determined clinically using the Polymerase Chain Reaction (PCR) technique, on the basis of the Raman spectral changes caused by the virus during the process of the development of Hepatitis C. Copyright © 2017 John Wiley & Sons, Ltd.