I am looking for a program that uses 2-digit (low resolution) to impute a high-resolution 4-digit type. We call it an HLA converter, and it takes a 2-digit HLA type and race of a person, and finds their most likely 4-digit HLA type.
The one available in HLA Matchmaker is not suitable to me, because I have to convert a large database of patients. And I have trouble getting my boss to work with the Brazilian EpVix service.
Thanks!
Sorry, can´t help you there, Dulat - and I do not see the point in that either, to be honest...
I´d rather do four digit CT if a donor fits at two digit level. In my humble oppinion guessing HLA typings gets you nowhere. How come you have got a large data base on patients, anyway, instead of donors?
Thanks Julian,
The thing is I am going to do a retrospective analysis of both patients and donors, and I don't have access to their sera or anything, so I cannot type them in-house. All I can do is to rely on what transplant centers have in their records.
I was told (but couldn't find publications yet) that 2 to 4 digit conversion is reliable enough, at least for my purposes.
Dear Dulat,
Antje Lassahn just now wrote what I was about to write: We both do not see a way in which you can get reliable results by what we probably would call "the best guess"!
Do you want to analyze survival or relapse free survival (by the way: solid organs or stem cell Tx?) according to a donor-recipient match or missmatch? Maybe you´d better stick to your low resolution typing, because that is a real typing, not "the most probable" of results...
Hi Dulat,
The only way to get somewhat reliable high-resolution HLA results is by allelic association (haplotype analysis) of the low-resolution HLA data. In this regard, please find an abstract that we presented at the 2015 American Transplant Congress. I hope this helps.
Best regards,
Andres Jaramillo
Interesting data, Andres - but then, you´d certainly do a serological crossmatch anyway? And would you, in case of let´s say an HLA-B*40:01 prediction and a LSA-antibody specification of anti-HLA-B*40:02 allocate an organ?
Dear Dulat Bekbolsynov,
there is no way to go in this direction!
It is possible to go from high resolution to a low resolution typing, only.
Best regards,
Thomas
Hi Julian,
We have successfully used this tool to do virtual crossmatching to accept hearts and lungs (imports) in selected cases. We always followed with a flow crossmatch to confirm the virtual crossmatch. Accurate information about the donor's race helped a lot in these cases.
It is amazing what awesome stuff can be found here on ResearchGate. Thanks Julian and Andres! The website mentioned in your abstract is probably all I need to do that haplotype based conversion.
And just of note, Julian, your question about LSA reactivity against B*40:01 in recipient while the donor is probably B*40:02... I would look at shared eplets to add a layer of safety. I assume B*40:01 and B*40:02 will have difference in eplets, because they were serologically defined as different antigens (B60 and B61) before.
Thomas, I understand it is not as reliable as actual typing, but sometimes, say when there is a deceased donor, an option to conversion is declining a transplant offer. And previous studies, including that of Dr. Jaramillo, showed that you can predict 4-digit HLA types using haplotypes with about 90-95% accuracy, which to me would justify the risk of mistaking.
It´ll be very interesting to see where we´re going with the concepts of epitope matching and eplets.
That B40 thing was just an example - I am aware of the different serological specialities. However, if a patient refractory to platelet transfusion has an anti-B60-antibody, we would try to avoid a B61 positive donor because of the cross reactivity anyway.
Dulat , If you have any questions about using HaploStats you can send them to the [email protected] contact. We would be glad to work with you.
Instructions for processing small sets of HLA typing from an input file are outlined in the Research Documentation, but it takes a bit of formatting work to accomplish this. The imputations are from US based race and ethnic population HLA frequencies only.
Dear Dulat, I have no Programm. Scientific question: Is it corrct to work with assumptions where you do not know what the real data are? The end-user of a clinical Thesis is confronted with the Facts of daily life, the propablity to hit an unexpected exception somewhere in the world is one. Please make sure to outline the Transformation process properly.
Hi Carl-Ludwig, sorry about the delay. Of course it is not an optimal approach to impute all HLA types from 2-digit types reported by transplant centers and assume it is 100% correct. But I think this is the method of choice given the resources that I have for the project. Obviously, in clinical practice you shouldn't make assumptions like this, and our lab also argues that high resolution typing should be always done.
Thank you, agree. Sometimes limitations are frustrating.
- Badges
- Science topic