Dynamic response antibodies SARS-CoV-2 human saliva studied using two-dimensional correlation (2DCOS) infrared spectral analysis coupled with receiver operation characteristics analysis

Abstract

COVID-19 has affected the entire world due to the rapid spread of SARS-CoV-2, mainly through airborne particles from saliva, which, being easily obtained, help monitor the progression of the disease. Fourier transform infrared (FTIR) spectra combined with chemometric analysis could increase the diagnostic efficiency of the disease. However, two-dimensional correlation spectroscopy (2DCOS) is superior to conventional spectra as it helps to resolve the minute overlapped peaks. In this work, we aimed to use 2DCOS and receiver operating characteristic (ROC) analyses to compare the immune response in saliva associated with COVID-19, which could be important in biomedical diagnosis. FTIR spectra of human saliva samples from male (575) and female (366) patients ranging from 20 to 85 ± 2 years of age were used for the study. Age groups were segregated as G1 (20–40 ± 2 years), G2 (45–60 ± 2 years), and G3 (65–85 ± 2 years). The results of the 2DCOS analysis showed biomolecular changes in response to SARS-CoV-2. 2DCOS analyses of the male G1 + (1579,1644) and -(1531,1598) cross peaks evidenced changes such as amide I > IgG. Female G1 cross peaks -(1504,1645), (1504,1545) and -(1391,1645) resulted in amide I > IgG > IgM. The asynchronous spectra in 1300–900 cm-1 of the G2 male group showed that IgM is more important in diagnosing infections than IgA. Female G2 asynchronous spectra -(1027,1242) and + (1068,1176) showed that IgA > IgM is produced against SARS-CoV-2. The G3 male group evidenced antibody changes in IgG > IgM. The absence of IgM in the female G3 population diagnoses a specifically targeted immunoglobulin associated with sex. Moreover, ROC analysis showed sensitivity (85–89 % men; 81–88 % women) and specificity (90–93 % men; 78–92 % women) for the samples studied. The general classification performance (F1 score) of the studied samples is high for the male (88–91 %) and female (80–90 %) populations. This high PPV (positive predictive value) and NPV (negative predictive value) verify our segregation of COVID-19 positive and negative sample groups. Therefore, 2DCOS with ROC analysis using FTIR spectra have the potential for a non-invasive approach to monitoring COVID-19.