Use of ATR-FTIR Spectroscopy Coupled with Chemometric Analysis to Diagnose HCV and Monitor Disease Progression

Abstract

The central hypothesis of this research thesis posits that the integration of Attenuated Total Reflectance Fourier Transform Infrared (ATR-FTIR) spectroscopy with advanced chemometric analysis (artificial intelligence) can effectively serve as a diagnostic tool for the detection of Hepatitis C Virus (HCV) infection and the monitoring of its disease progression. Through the implementation of this innovative approach, the study aims to identify specific molecular biomarkers associated with HCV, thus contributing to the development of a cost effective and efficient diagnostic tool. The focus of the investigation is on elucidating tentative pathological biomarkers through pilot studies, with the ultimate goal of establishing a foundation for a robust tailored diagnostic tool for HCV; which is highly endemic in Pakistan. To address the scientific question. The protocol for sera processing of this research work was established, optimized, and implemented. Freeze-dried sera were employed for spectral acquisition as the presence of moisture can negatively impact spectra recorded from biological fluids. Spectral acquisition was performed using a BRUKER FTIR model Alpha II Platinum FTIR. Background noise was subtracted, and a small amount of the freeze-dried sample was placed onto the ATR crystal. The spectra were recorded in the IR region (400 4000 cm-1) with a resolution of 4 cm-1 and 32 scans per sample. Subsequently, the data was processed using Unscrambler X version 10.5 to minimize scattering effects and background noise and enhance inter-sample differences. For improved discrimination between the classes, multivariate data classification algorithms were applied. The optimized sera processing protocol was used for research studies of this thesis and represents the first report on the use of freeze-dried sera for recording ATR-FTIR spectral signatures and the development of a multivariate model for: (1) qualitative diagnosis of HCV; (2) differential diagnosis of HCV with similar clinical presentation; (3) diagnosis of HCV and assessment of the non-cirrhotic/cirrhotic status of HCV-infected patients; (4) diagnosis of HCV-related hepatocellular carcinoma (HCC) and categorization of HCC into non-angio-invasive and angio-invasive HCC using solely HCV-infected sera. For the classification of HCV from healthy control sera samples for qualitative diagnosis, significant variation was observed in the spectral regions 3500–2800 cm-1 and 1800–900 cm-1. Principal component analysis (PCA) and linear discriminant analysis (LDA) were utilized as multivariate classification algorithms. The PCA scatter plot of the first two principal components showed 97% variation between the two sample classes. When the four principal components were projected to the LDA algorithm for the development of a PCA LDA diagnostic model, it discriminated the HCV and healthy classes with 100% accuracy. For the classification of HCV with similar clinical presentations (Dengue) for differential diagnosis, PCA and principal component regression (PCR) were employed as multivariate data classification algorithms. The outcomes revealed 96% inter-class variation as depicted by PCA and a classification accuracy of 99.2% using PCR, as indicated by the value of R2. For the detection of HCV and further assessment of the non-cirrhotic/cirrhotic status of HCV-infected patients, a PCA-LDA and support vector machine (SVM) model were computed, resulting in a diagnostic accuracy of 100% for the detection of HCV infection. To further classify the non-cirrhotic/cirrhotic status of a patient, a diagnostic accuracy of 90.91% ix for PCA-Quadratic discriminant analysis (QDA) and 100% for SVM was observed. Internal and external validation for SVM-based classifications resulted in 100% sensitivity and specificity. In the context of HCV-related HCC diagnosis, infrared (IR) spectroscopy was employed to differentiate between HCC and healthy individuals, as well as to categorize HCC into non-angio-invasive and angio-invasive types using HCV-infected sera. Analysis of IR spectral data revealed significant differences between the two groups in the spectral regions of 3500–2800 cm-1 and 1800–900 cm-1. The results of PCA-LDA and SVM models indicated high diagnostic accuracy, with 100% accuracy achieved for HCC diagnosis using PCA-LDA and SVM models. For the classification of non-angio-invasive and angio-invasive HCC, PCA-LDA achieved a diagnostic accuracy of 86.21%, while SVM showed a training accuracy of 98.28% and a cross-validation accuracy of 82.75%. An external validation of the SVM-based classification was performed, resulting in 100% sensitivity and specificity for the classification of freeze-dried sera samples in all categories. Building upon the obtained results and a thorough review of literature, it is postulated that ATR-FTIR spectroscopy coupled with chemometric analysis possesses significant promise not only in the diagnostic capacity for HCV but also in discerning patterns indicative of disease progression and distinguishing between diverse pathological conditions. This proposition is grounded in the empirical evidence derived from our investigations and aligns with the current body of scientific knowledge in the field.