This paper discusses the application of Support Vector Machines (SVM) to classification of autoantibody profiles with a large number of antigens (attributes) and effectiveness gain produced by the proposed antigen reduction methods. As a result of research effort, a new biologically robust and meaningful method on selecting antigens with high expression level in cancer patients is introduced. The proposed antigen reduction method improves initial full-range SVM model and clearly proves the necessity for more careful attribute selection in classification tasks performed by SVM. The proposed method can also be used for novel biomarker discovery and can give a necessary insight into uncovered connections between already known and newly discovered CT (cancer testis) biomarkers.