We are grateful for these comments on the present limitations of Histocheck. This software is intended to be a first step towards intelligent mismatching when no HLA-identical donor is available. As stated in the manuscript, Histocheck is based exclusively on crystallographic data of HLA molecules, and on a certain set of data about functional similarities of amino-acid residues. These data have been combined in an elementary mathematical model that so far lack validation by clinical data. The small analysis by Dr Shaw and colleagues based on clinical data demonstrates that Histocheck's current scoring system has a limited predictive value and needs significant improvement. This information is very important, yet not unexpected. In a preliminary analysis recently performed with more than 1700 HLA class I mismatched transplant pairs from the Hematopoietic Stem Cell Transplant component of the 13th International Histocompatibility Workshop (Effie Petersdorf, Fred Hutchinson Cancer Research Center, Seattle, WA, USA), the Histocheck algorithm was not superior in predicting the severity of aGVHD compared to just counting the number of HLA class I mismatches.
In view of the restrictions of Histocheck's current algorithm, without doubt more sophisticated mathematical models need to be developed and to be tested on large-scale clinical data sets. Considering data showing on the one hand that a single amino-acid mismatch can make a big difference (eg Asp156Leu in HLA-B*4402 vs HLA-B*4403) and on the other hand that multiple amino-acid mismatches are well tolerated, it might appear that the assessment of the quality of HLA mismatches in stem cell transplantation is nearly impossible. However, we strongly believe that a mathematical model for the assessment of amino-acid mismatches can be developed if more experimental and clinical data are available for individual HLA alleles. An algorithm based on more data will probably work with individual weights on each HLA locus and individual weights on each amino-acid difference. However, one has to keep in mind that even in fully matched stem cell transplant pairs great differences with regard to GVHD and overall outcome are observed. This is mainly due to non-HLA variations presented as different peptides in HLA molecules as well as to inhomogenous clinical situations.
This is a preview of subscription content, access via your institution