The following links list how search determinants are assigned to each HLA allele by the different registries; The Anthony Nolan, the NMDP, BMDW, Canadian Unrelated Bone Marrow Donor Registry, France Greffe de Moelle, BBMR and also the ZKRD approach to search determinants.
British Bone Marrow Registry (BBMR)
All valid HLA types, including serological specificities, alleles and ambiguous types are stored in a single table. Apart from the broad serological specificities each type is given a parent and this creates a hierarchy which is used for matching. An entry in the table can only have one parent, so for ambiguous types the parent of the whole string is usually the parent of the first allele in the string.
In general terms when types are being compared any 'children' of a given type are considered as matched.
Thus if a patient is typed as A24 then A*24, A*2402 and A*2402-05 are considered to be matched. Effectively the top level parent (in this case A24) can be considered as the search determinant.
There are also grades of matches within the branch of the hierarchy:
- Full match - Same single allele on both patient and donor
- Potential match - Serological type to DNA type, or single allele to string, or string to string where donor and patient potentially share at least one allele
- Minor mismatch - Single allele to string or string to string where donor and patient potentially share do not share an allele but have the same serological parent.
HLA types which are different splits of the same broad parent are considered to be complete mismatches. This also applies to all alleles which have been assigned to serological splits. Thus A24 and A23 are mismatched as are A*2401 and A*2301
This algorithm works well except for types such as B15 and B40 where the alleles are not named after the serological splits. In such case the algorithm breaks down because the hierarchy not only has to cope with splits of the broad parent has to have a branch for the generic allelic type to hold alleles or strings which have not yet been assigned to a serological split. For example B15 has branches for B62, B63, B70, B75, B76, B77 and B*15. There is a problem here with alleles whose parent is B*15 because they would be considered as mismatches for alleles with well defined serological types because B*15 is treated as a split of B15. However we should treat alleles whose parent is B*15 as potential matches of all other B15 alleles. To achieve this potential match some generic alleles are treated as special cases but it does make the algorithm inelegant and complicated.