A PARALLEL STUDY OF CLASS I SEROLOGY VS. SSP IN 107 BMT PATIENTS.
JM Westerkamp, ES Campbell, LH Hazy, MJ Heller, JC Inlow, PL Steller, PK Hennessy, PW Adams and CG Orosz, The Ohio State University Medical Center, Columbus, OH.
Because bone marrow transplant recipients are at high risk for graft rejection and/or graft vs. host disease, it is important to employ an HLA typing methodology that is both accurate and reliable. We performed a blinded study on 107 bone marrow transplant donors and recipients in order to determine which method of HLA typing, serologic or molecular, most accurately and reliably defined HLA Class I alleles. Serologic typings were performed using Class I specific beads and a variety of commercially available Class I serologic typing trays. Molecular typings were performed using Pel-Freez Class I SSP Unitrays. HLA-A or HLA-B homozygosity was confirmed by a secondary system (Biotest SSP). Agreement in this study is defined as identical resolution between both methods. Discrepant results are defined either as a further resolution of a broad allele, definition of a different split of a broad allele, or the identification of an additional allele. We observed that 75% (80/107) of the parallel typings in this study were in agreement, while 25% (27/107) of the typings were discrepant. Of the discrepant results , 9/27 samples were those in which SSP further defined a broad allele as identified by serology, 8/27 molecular results identified an additional allele not defined by serology, and 6/27 molecular results defined a different split of a broad allele than serology. These 6 samples were sent to a referee laboratory, and all 6 of the original SSP results were confirmed. Only 4/27 samples were SSP mis-typings identified by serology. All 4 SSP errors occurred during the learning curve period for DNA typing which was within the first 2 weeks of this 24 week study. When these 4 samples were repeated by SSP, the results agreed with serologic analysis. Thus, in this study, the true error rate for serologic typings was 23% (24/107) while the true error rate for SSP typings was 4% (4/107). We conclude that molecular techniques (in this case SSP) provide the most accurate and best resolution possible for bone marrow transplant donors and recipients in our transplant program.