An immunoproteomic approach to identifying cancer-associated autoantibody biomarkers

Doctoral Thesis

2018

Permanent link to this Item
Authors
Journal Title
Link to Journal
Journal ISSN
Volume Title
Publisher
Publisher

University of Cape Town

License
Series
Abstract
Cancer is a heterogenous disease capable of forming and spreading in most tissues of the human body. Cancer screening and diagnosis can be performed through medical procedures, which are highly invasive, requiring an intensive infrastructure. It is therefore important to create cost-effective, non-invasive cancer diagnostic tools that also gives an indication of disease prognosis. With this in mind, the Blackburn lab previously created a cancer-testis antigen microarray (CT100plus) functionalised with tumour-associated and tumour-specific antigens, capable of detecting plasma- or serum-derived autoantibodies in the picogram per millilitre (pg/ml) range. In this thesis, a newly established statistical pipeline was used to analyse colorectal cancer (CRC) patient-derived CT100plus data. Using the pipeline, the 10 antigens with the highest receiver operator characteristic (ROC)-derived area under the ROC curve (AUC)-values were identified as potential autoantibody-based biomarkers. The top 10 antigen biomarker candidates include CEACAM 1, COL6A1, GRWD1, MAGEA1, MAGEA5, MAGEA10, NY-CO-1, SGY-1, SPANXB1 and THEG. Using these biomarker candidates, distinct clusters of healthy controls (HCs) and CRC patients were observed using both unsupervised hierarchical clustering and principle component analysis (PCA) analysis. Combinatorial ROC analysis indicates that CEACAM1 and GRWD1 as the top autoantigen combination for CRC diagnosis, together producing sensitivity-, and specificity-, and AUC-values of 1.00, 0.77 and 0.94, respectively. Furthermore, other top autoantigens, including COL6A1, THEG and CEACAM7, a homologue of CEACAM1, were also identified in this thesis by affinity purification-mass spectrometry (AP-MS) for patients from the same cohort, providing supporting evidence that these antigens are associated with CRC. The CT100plus microarray content was enzymatically modified to include citrullinated proteins, with the subsequent assessment of CRC patient autoantibody response. Significantly (p-value ≤ 0.05; adjusted p-value ≤ 0.05) higher signal intensities were detected in CRC patients versus HCs for citrullinated CDK7, MAGEB1, MAGEB5, MAGEB6 and SYCP1, whereas no significant (adjusted p-value > 0.05) difference in autoantibody signal was detected for these autoantigens on the noncitrullinated microarray for the same patient cohort. ROC analyses of these antigens resulted in 2 an AUC-, sensitivity- and specificity-values of 0.91, 0.87 and 0.89, respectively. Together, this thesisshowsfor the first time that cancer patients elicit an autoantibody response to citrullinated proteins, resulting in potential novel CRC biomarkers. A novel AP-MS assay was developed to detect autoantibody responses to autologous native CRC tissue proteins. Using the optimised methodology, proteins or homologues of proteins that were significantly (> cut-off value) detected on the CT100plus microarray for the same 5 patients were re-identified by the orthogonal AP-MS method. Using the methodology, PAD2, an enzyme that catalyses the conversion of arginine to citrulline was also identified. In addition, citrullinated antigens associated with cancer were identified, including homologues of CDK7 and MAGEB supporting the conclusion that citrullinated homologues of these proteins induce an autoantibody response in CRC patients. Finally, serum and/or plasma samples of a cohort melanoma patients were analysed using the CT100plus microarray, and autoantibody signals were compared to those of healthy control (HC) samples. Using the established statistical pipeline, the 10 antigens with the highest ROC-derived AUC-values were identified as potential biomarkers. The top 10 biomarker autoantigen candidates for melanoma included CEACAM 1, DPPA2, FGFR2, ITGB1, MAGEA10, NANOG, PIM1, SPANXB1, THEG and XAGE1B. Using these biomarker candidates, distinct clusters of HCs and melanoma patients were identified in both unsupervised hierarchical clustering and PCA analysis. Combinatorial ROC analysis indicates that CEACAM1 and FGFR2 were identified as the top antigens for melanoma diagnosis, together producing sensitivity-, and specificity-, and AUCvalues of 0.96, 0.94 and 0.93, respectively. In conclusion, a statistical pipeline was established for microarray data, enabling the identification of potential antigens associated with cancer diagnosis, and the potential to determine disease prognosis. Using the established pipeline, cancer antigens associated with CRC and melanoma were identified. In addition, an AP-MS assay was developed for the identification of known and novel cancer antigen that can be added to the customisable CT100plus microarray.
Description

Reference:

Collections