OR6 The positive virtual crossmatch. Correlation between HLA DSA and flow cytometry crossmatch results

Abstract

Aim Virtual crossmatch (VXM) is a good tool to expedite and streamline pre-transplant recipient selection and facilitate national organ sharing. However, VXM can be a barrier to transplantation, based on donor specific antibodies (DSA) detected by single antigen bead (SAB) testing, especially when low MFI cutoffs are used to assign unacceptable antigens. Recent studies report good clinical outcomes in patients transplanted across DSA with a negative FCXM, emphasizing that the choice of VXM cutoffs should be based on careful correlation with FCXM. The purpose of this study was to assess the correlation between MFI of DSA and FCXM results. Methods The results of 2843 deceased donor FCXM performed at our center between 03/2015 and 03/2016 were analyzed and correlated with cumulative MFI of locus specific DSA. Donors were typed (HLA-A, B, C, DR, DQ) by SSP method. FCXM were performed as per Halifax protocol (Liwski et al., ASHI Q 2015) with spleen or lymph node derived donor cells. SAB testing (LABScreen) was performed with heat inactivated sera to prevent prozone. DSA were assigned based on 1000 (1K) MFI cutoff. All sera with DP antibodies were excluded. Results Of 1859 VXM −ve samples (no DSA), 56 were FCXM +ve (3% false positive rate). Of 318 VXM +ve (DSA; >1K MFI) samples, only 214 (67.3%) were FCXM +ve. The degree of concordance, VXM vs FCXM, was locus and MFI dependent (Fig. 1). HLA-A and B DSA predicted a +ve FCXM (T and/or B) in 80%, 93% and 100% of cases using 1K, 2K and 5K cutoffs respectively. HLA-DR DSA predicted positive B cell FCXM in 54%, 89% and 100% of cases with same MFI cutoffs. HLA-C and DQ DSA predicted positive FCXM reliably only when 10K cutoff was used (100% and 95% respectively; Fig. 1). Conclusions Concordance between VXM and FCXM depends on MFI cutoff and DSA locus. 1K cutoff is not useful for FCXM prediction regardless of DSA specificity. For HLA-A, B and DR DSA, 2–5K cutoff can be used to reliably predict FCXM, while for HLA-C and DQ only 10K cutoff is predictive. Download : Download high-res image (138KB) Download : Download full-size image