To the Editor.—We read with interest the original article by Gandini et al1 “A retrospective study on human leukocyte antigen types and haplotypes in a South African population” recently published in Archives of Pathology & Laboratory Medicine. The study reports the human leukocyte antigen (HLA) types of 782 donors and the HLA antibody profiles of 200 recipients in the Johannesburg transplant program with the goal of “examining the feasibility of the reintroduction of HLA matching into the criteria of the Johannesburg program.” While we welcome the authors' report of the HLA antigens present in the South African donor population, we wish to highlight some issues with how the HLA data are described. Figure 1 lists molecular HLA alleles that do not exist in the international IMGT/HLA database (https://www.ebi.ac.uk/ipd/imgt/hla/allele.html). These include Figure 1, B as follows: “*96”; Figure 1, D: “*25,” “*52,” and Figure 1, E: “*07,” “*09,” “*14,” “*93,” and “*96.” Figure 1 also appears to confuse molecular “*” typing with antigen equivalents. For example, the B*15 allele group contains alleles that correspond to several serologically defined HLA antigens, including B62, B63, B75, B76, and B77. In Figure 1, C the authors list B*15 that could be correct, but they also list “*62” and “*63,” which are not molecular types, but distinct antigen equivalents in the B*15 allele family (ie, there is no HLA-B*62, rather the antigenic equivalent of HLA-B*15:01 is B62). In Figure 2, the allele haplotype analysis does not appear to account for linkage disequilibrium (LD). HLA loci are arranged in a specific order on the chromosome, which results in different recombination rates between the various loci. For example, the HLA-B and HLA-C loci tend to show stronger LD, so that linkage (haplotypes) can be more readily established. The authors report “2 allele” haplotypes for loci that have lower LD (HLA-A and HLA-DQB1). Haplotype analyses also typically require the use of specific software programs and reference databases. Finally, the virtual crossmatch analyses in Table 4 should be described relative to serologically defined HLA antigens. Virtual crossmatches use results from HLA single-antigen bead assays to create a list of HLA antigens to which a patient has developed antibodies. In Table 4, the authors list the percentage of patients who have antibodies to “B*15” and “DQB1*03.” As described above, the B*15 (B15) allele group includes 5 different antigens, and the DQB1*03 (DQ3) allele group includes 3 different antigens (DQ7, DQ8, DQ9). While patients certainly can have antibodies reactive to the parent (ie, B15 or DQ3) antigen, they may also have antibodies specific to only one of the “split” antigens, and would therefore be compatible with donors who express 1 of the other antigen specificities in the group.
The Johannesburg transplant matching criteria referred to in the article is not described, and we find it difficult to assess the authors' conclusion that “the pre-transplantation workup should not reinclude HLA matching,” based on the data presented. In the US, organ allocation is governed by a complex and evolving algorithm that encompasses multiple donor and recipient factors, including HLA matching (ie, zero mismatch organ offers) and the degree of recipient HLA allosensitization.2 Equitable organ allocation is certainly a complex topic, and the authors point out that South Africa does not have an automated organ allocation system that could be used to evaluate the effect of changing the allocation criteria. Although it is plausible that HLA matching may not be beneficial for optimal organ allocation or transplant outcomes in South Africa, this does not appear to be specifically addressed by the study design or the data presented.
The authors have no relevant financial interest in the products or companies described in this article.