Abstract 15752: Differential IgG Repertoire in Individuals With Chagas Cardiomyopathy

Chagas cardiomyopathy (CC) is caused by the infection of Trypanosoma cruzi ( T. cruzi ) and represents a major etiology for heart failure in Latin America. Trying fill gaps related to humoral response in CC, we used a PhIP-Seq technology to describe differential IgG profile between individuals with CC and individuals infected by T. cruzi , but without symptoms (Indeterminate - IND). We built a library comprising all T. cruzi proteome (12766 proteins) and incubated against 1293 serum samples: 396 T. cruzi seronegative samples (98 controls and 298 individuals with cardiomyopathy caused by other causes than Chagas disease) and 897 T. cruzi seropositive samples (684 CC and 213 IND). 1870 peptides were increased in T. cruzi seropositive IgG compared to T. cruzi seronegative samples. Peptides from Nucleoporin NUP149, a flagellar attachment zone protein, basal body protein, surface protein TolT and cytoskeleton associated protein were the peptides with highest IgG response. Using as little as 5 peptides, I derived a classifier model for T. cruzi infection with 97% accuracy. Then, I compared CC and IND samples using logistic regression models after splitting our sample into 2 cohort groups. From these, 316 peptides were significantly different between CC and IND: 289 peptides were increased in CC (27% of them belong to the trans-sialidase class) and 27 peptides were increased in IND (FDR>0.01). Aiming to develop a classifier to separate CCC from IND, I analyzed our data using machine learning approaches. LASSO returned the best model using 48 peptides, with 70% of accuracy, 74% of sensitivity and 69% of specificity. The peptides selected for the model belong to 38 proteins, 21% being unknow function proteins, 18% trans-sialidases, 8% flagellar associated proteins, and 60S ribosomal protein, Nucleoporins and R27-2 proteins classes representing 5% each. Ninety (90%) of selected peptides in LASSO were identified in the previous results using the logistic regression model. I showed for the first time that Chagas Cardiomyopathy individuals have a differential IgG profile against T. cruzi proteins compared to IND individuals. I described a differential IgG response against 84 sequences (62 increased in CCC and 22 increased in IND) that can be related to disease development.