Toward Optimal Selection of Cellular Products for Hematology/Oncology Patients Eliminating Adverse Effects

Abstract

Context: To provide critical cellular support to hem/onc patients undergoing chemotherapy, US hospitals consumed w675,000 single donor platelets (34.4% of the 2010 total transfused) and w2,100,000 red cells (15.1% of the 2010 total transfused). In the emerging paradigm of genomics and personalized care, the selection of cellular products for the treatment of oncology patients in accordance with their immunomolecular signature has the potential to reduce antibody-mediated destruction of cells and improve treatment outcomes. Turning this strategy into a logistical reality calls for three ingredients: the identification of patients at high risk of mounting an adverse immune reaction; the clinical decision to provide optimal cellular products; and real-time access to an adequate supply of such products. Objective: To clinically validate a novel process of classifying hematology patients and candidate donors on the basis of an immunomolecular attribute profile comprising platelet (HLA class I and HPA) as well as red cell antigen profiles for a total of w100markers (and the potential of scaling this process so as to permit the patient profiling and classification of 1,000 candidate donors per 24hr period). Design: Residual blood samples from MDACC blood bank were collected and analyzed by our novel process combining: the parallel sequencing for pools of samples at informative sites; haplotype recovery; antigen and epitope profile prediction; and classification by “machine learning”. Results: HLA class I and HPA profiles of samples from 296 hematologic and non-hematologic oncology patients, and from 160 candidate platelet donors were determined, and classified by HLA class I and HPA profiles; the process achieved 100% concordance with high-resolution HLA types available for 42 patients; in addition, we demonstrated the direct recovery of haplotype pairs and extended the method to red cell antigens including RH variants. Conclusions: Our new “virtual cell sorting” process offers an effective approach to the large-scale classification of patients and candidate donors, and optimal selection of cells, on the basis of immunomolecular signatures.