Abstract 3999: Human proteome arrays for auto-antibody identification in clinical cancer studies

Proceedings: AACR 104th Annual Meeting 2013; Apr 6-10, 2013; Washington, DC Early detection of cancer is a critical factor for successful treatment of cancer patients. Autoantibody signatures have value in the diagnosis and management of autoimmune disorders and may be valuable for detection of cancers. Previous groups have applied several techniques to identify cancer-specific autoimmune responses with varying degrees of success. A lack of high throughput methods to rapidly screen large numbers of patient samples and validate multiple antigens of interest is an important bottleneck in the development of autoantibodies as biomarkers for cancer diagnosis. We tested normal and cancer patient sera against protein arrays generated using a gene library capable of expressing full-length human proteins. Autoantibodies against proteins in these arrays were identified using electrochemiluminescence detection technology from MSD®, a proven technology for high-throughput, array-based measurements. This approach offers key advantages over existing array-based approaches by providing a high-throughput, sensitive, and specific assay platform, with large-scale protein arrays. This approach also enables a rapid transition from antigen discovery to clinical validation using small focused arrays on the same diagnostic platform. Protein arrays containing ∼7,000 unique (non-redundant) full-length human proteins were generated using a eukaryotic expression system. This set of proteins represents a significant portion (28-35%) of the human proteome. We estimate that greater than 80% of autoantibodies recognize unmodified epitopes on single protein antigens; the protein arrays, therefore, should be highly representative of the vast majority of autoantigens. These “proteome-scale” arrays were used to screen for cancer-specific autoantibodies in patient serum samples. Patient samples were initially screened as pools (n=5) followed by screening of individual patient samples. Initial screening was done using samples from patients with breast, lung, ovarian, prostate, colon, kidney, melanoma, and pancreatic cancers. Control samples included samples from normal patients as well as those with benign breast disease, bronchial dysplasia, benign ovarian disease, and benign prostate hyperplasia. Samples from patients with breast, lung, ovarian, and prostate cancers were sub-divided into additional groups based on clinical findings. In total we screened 22 different groups of patients. We validated this approach with the recovery of well-established tumor antigen p53 and additional antigens. Ovarian cancer yielded the highest percentage of positive autoantibody responses with 21% for p53, 8% for DEK, and 6% for GMPR2. Further evaluation of this approach using sets of molecularly classified patient samples may improve the discovery of diagnostic autoantigens in cancer, especially in combination with patient-specific protein arrays determined from tumor whole-genome sequencing data. Citation Format: John H. Kenten, Mikayla Higgins, Christopher Frye, Sudeep Kumar, John Joern, George B. Sigal, Eli N. Glezer, Jacob N. Wohlstadter. Human proteome arrays for autoantibody identification in clinical cancer studies. [abstract]. In: Proceedings of the 104th Annual Meeting of the American Association for Cancer Research; 2013 Apr 6-10; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2013;73(8 Suppl):Abstract nr 2495. doi:10.1158/1538-7445.AM2013-2495