Computational Approaches to Facilitate Epitope-Based HLA Matching in Solid Organ Transplantation

Abstract

Epitope-based HLA matching has been emerged over the last few years as an improved method for HLA matching in solid organ transplantation. The epitope-based matching concept has been incorporated in both the PIRCHE-II and the HLAMatchmaker algorithm to find the most suitable donor for a recipient. For these algorithms, high-resolution HLA genotype data of both donor and recipient is required. Since high-resolution HLA genotype data is often not available, we developed a computational method which allows epitope-based HLA matching from serological split level HLA typing relying on HLA haplotype frequencies. To validate this method, we simulated a donor-recipient population for which PIRCHE-II and eplet values were calculated when using both high-resolution HLA genotype data and serological split level HLA typing. The majority of the serological split level HLA-determined ln(PIRCHE-II)/ln(eplet) values did not or only slightly deviate from the reference group of high-resolution HLA-determined ln(PIRCHE-II)/ln(eplet) values. This deviation was slightly increased when HLA-C or HLA-DQ was omitted from the input and was substantially decreased when using two-field resolution HLA genotype data of the recipient and serological split level HLA typing of the donor. Thus, our data suggest that our computational approach is a powerful tool to estimate PIRCHE-II/eplet values when high-resolution HLA genotype data is not available.

Figure 1

Deviation of the observed ln(PIRCHE-II)/ln(eplet) values from reference ln(PIRCHE-II)/ln(eplet) values when using HLA haplotype frequency tables from 2007 or from 2011. For both ln(PIRCHE-II) (a) and ln(eplet) (b), the observed values do not deviate or only slightly deviate from the reference values. The use of haplotype HLA frequency tables from 2007 (gray line) or from 2011 (black line) resulted in similar deviations.

Figure 2

Comparison of the delta in ln(PIRCHE-II) and the delta in ln(eplet). (a) When using serological split level HLA typing of both donor and recipient, the majority of the donor-recipient couples had a comparable ln(PIRCHE-II) and ln(eplet) delta. For 13 donor-recipient couples, the observed ln(PIRCHE-II) or ln(eplet) values deviated substantially (>+1 or <−1) from the reference ln(PIRCHE-II) or ln(eplet) values. (b) When using two-field resolution HLA genotype data of the recipient and serological split level HLA typing of donor the deviation between the observed ln(PIRCHE-II)/ln(eplet) values and the reference ln(PIRCHE-II)/ln(eplet) values substantially diminished. The dashed squares indicate the delta<0.5 and delta<1 borders.

Figure 3

The reliability of the PIRCHE-II and eplet estimations in different settings. The percentage of typing with a delta of zero between the observed and the reference values was plotted for ln(PIRCHE-II) (a) and ln(eplet) (b). For both ln(PIRCHE-II) and ln(eplet), the percentage of typing with a delta of zero was diminished when HLA-C or HLA-DQ was omitted from the typing. The highest percentage was observed when using two-field HLA genotype data of the recipient and serological split level typing of the donor. The different percentiles observed in the different settings were plotted for ln(PIRCHE-II) (c) and for ln(eplet) (d). The dashed horizontal lines in (c) and (d) indicate a delta of zero.