Transplantation/The HLA System and HLA Typing

The biological function of the MHC molecule

There are three classes of T cells. CD8 T cells are cytotoxic and are specialized to respond to intracellular sources of infection. CD4 TH2 cells stimulate B cells to produce antibodies against extracellular antigens. CD4 TH1 cells stimulate macrophages to destroy pathogens by phagocytosis.

There are two classes of MHC molecules. MHC class I molecules present peptides of intracellular origin to CD8 T cells. MHC class II molecules present peptides of extracellular origin to CD4 T cells. The MHC is called HLA in humans.

Peptides of intracellular origin are generated in the cytosol by the proteasome and are transported by the TAP transporter into the endoplasmic reticulum, where they bind to HLA class I molecules. Peptides of extracellular origin presented by HLA class II molecules are generated in acidified intracellular vesicles. HLA molecules bind a wide variety of peptides to be presented to T cells. Which peptides can be presented to T cells is determined by the polymorphism of the HLA molecules. The more different HLA molecules an individual has, the wider the variety of peptides this individual can present to T cells to stimulate an immune response. Nevertheless, there is a point at which having more HLA allotypes does not confer an additional advantage. In humans, the maximum number of HLA allotypes an individual can have is 6 class I and 12 class II allotypes. Based on the genetic polymorphism of class II genes, it is quite possible that the optimal number of class II isotopes is more like 9 or 10, as there has been no favorable selection of individuals that present 12 class II molecules.

There are other HLA genes, labeled HLA-E, HLA-F and HLA-G, known as HLA Ib antigens, or nonclassical HLA antigens. HLA class Ib molecules are much less polymorphic than HLA class I (Ia) and instead of playing a role in adaptive immunity, they regulate innate immune responses. Soluble HLA-G proteins, derived by an alternative splicing mechanism from the same gene HLA-G, may comprise a unique tolerogenic system for establishing local immune privilege during pregnancy. Along the same line, the expression of HLA-E and HLA-F in the placenta seems to indicate that they contribute to create the immune conditions to support fetal growth. HLA-C and Bw4-positive HLA-B molecules are the ligands of the KIR inhibitory receptors of NK cells, some HLA molecules binding to some KIR receptors and other HLA molecules binding to other KIR receptors. The absence of the corresponding HLA ligands can result in an allogenic NK-cell immune reaction.

Structure and genetics of the HLA molecule

Both class I and II HLA molecules have a similar molecular structure specialized in the presentation of peptides, with a groove to place the peptide being presented to T cells. The floor of this groove is a beta-pleated sheet, and the walls of the groove are alpha helixes. These structures are coded by exon 2 and 3 of class I genes and exon 2 of class II genes. Only genetic polymorphism in these exons has biological and clinical significance. Class I molecules are dimers with a polymorphic heavy chain coded by the HLA genes HLA-A, HLA-B and HLA-C, and a light chain formed by the monomorphic beta-2-microglobulin. Class II molecules are heterodimers formed by an alpha chain coded by genes HLA-DRA1, HLA-DQA1 and HLA-DPA1; and a beta chain coded by HLA-DRB1, HLA-DRB3/4/5, HLA-DQB1, HLA-DPB1. Except for HLA-DRA1, all these genes are polymorphic. For esoteric reasons HLA-DRB3, HLA-DRB4 and HLA-DRB5 are traditionally considered three different genes, but this only creates confusion and should really be considered the same gene. DRB1 and DRB3/4/5 molecules form heterodimers with DRA1, DQB1 with DQA1, and DPB1 with DPA1. All these genes are clustered together in chromosome 6 in the following order: DPB1, DPA1, DQB1, DQA1, DRB1, DRB3/4/5, DRA1, B, C, and A. This order is important to understand the linkage disequilibrium between these genes. Linkage disequilibrium is strong between B and C, and between DQB1, DQA1, DRB1, DRB3/4/5. Less strong, although still present, is the linkage disequilibrium between B-C blocks and A, and between B-C blocks and DR-DQ blocks.

Scattered among HLA genes, in the same section of chromosome 6, there are other genes, some of which are closely related functionally to HLA, such as TAP1, TAP2, that code for the transporter protein TAP; LMP2 and LMP7 that code for catalytic subunits that regulate the proteolytic activity of the proteasome; DMA, DMB and DOB codes for proteins that regulate the selection of peptides presented by class II HLA molecules. All these genes happen to be located between DPB1 And DQB1. In addition, between HLA-DRA1 and HLA-B there are a number of genes of diverse biological functions such as genes that code for some complement proteins (C4, Factor B, C2), some cytokines (LT-beta, TNF, LT), and the MICA protein. The function of MICA is still being investigated, but it is known to be very immunogenic, and anti-MICA antibodies may play an important role in transplantation. It goes without saying that all these genes are in linkage disequilibrium with HLA alleles and form part of HLA haplotypes.

The polymorphism of HLA alleles is mostly the result of gene conversion, that is, the transfer by recombination of a small fragment of DNA from one allele to another allele in the same locus. Only to a much lesser extend is a point-mutation the source of HLA polymorphism. This fact makes the polymorphism of HLA genes very informative, in the sense that sequence variations are not unique to a particular allele but common to many alleles, greatly facilitating the design of PCR primers and hybridization probes.

The HLA system in clinical practice

The HLA system and transfusion medicine

Even though HLA class I molecules are often said to be present on all nucleated cells, implying that non-nucleated cells, such as erythrocytes and platelets, do not express HLA molecules. As a matter of fact, HLA molecules are present on both erythrocytes and platelets. Platelet refractoriness is mainly caused by anti-HLA antibodies, and the treatment of platelet refractoriness is precisely to give HLA-matched platelets. As for erythrocytes, the antigens called Bg, are actually HLA molecules, which occasionally have been considered the target of antibodies responsible for post-transfusion haemolytic episodes. It must be clarified, however, that the presence of HLA on both erythrocytes and platelets is the result of adsorption of soluble HLA molecules from the plasma, rather than the result of the normal cellular antigen processing, for which non-nucleated cells are incapable.

Apart from 1) rare sporadic haemolytic transfusion reactions, and 2) platelet refractoriness, anti-HLA antibodies are important in transfusion medicine because they are the cause of: 3) most non-haemolytic febrile transfusion reactions, as the result of the destruction of donor leucocytes by recipient anti-HLA antibodies; and 4) non-cardiogenic pulmonary oedema (also called, transfusion related lung injury), produced by the interaction of anti-HLA antibodies from a plasma donor with recipient leucocytes.

In all cases anti-HLA antibodies are produced as a result of either pregnancy or the transfusion of leucocytes. Therefore, the most efficient way to prevent the problems associated with anti-HLA antibodies is to remove leucocytes from blood transfusion products by the use of special filters designed for that purpose.

A more serious problem, although extremely rare, is the condition of transfusion related graft-versus-host disease, which has a terrible prognosis causing death in many cases. This is caused by the destruction of the recipient bone marrow by T cells from the donor. Normally, in transfusion the recipient T cells destroy the donor T cells, preventing these from having any effect on the recipient. When the recipient does not allo-recognize the donor, failing to destroy the donor T cells, and the donor T cells do allo-recognize the recipient, then a fatal cellular immune response against the recipient’s haematopoietic system can be elicited with catastrophic consequences. This is due to the lack of symmetry in allo-recognition. If the recipient has more HLA allotypes than the donor, and all the allotypes present in the donor are also present in the recipient, but some allotypes in the recipient are absent in the donor, then the recipient fails to elicit an immune reaction against the donor, while the donor can and will elicit an immune reaction against the recipient. This is what happens if the donor is homozygous in one or more HLA loci, and the recipient is heterozygous, but carries the HLA alleles present in the donor. Such a situation is more likely to happen when donor and recipient are related. Immunosupression is also another cause for failure to elicit a T cell response by the recipient, leaving the donor T cells free to elicit a response against the recipient. Transfusion related graft-versus host disease is prevented by the use of gamma radiation of blood products, which blocks cell reproduction.

Lately there has been some discussion on the benefits of screening plasma donors for anti-HLA antibodies in order to prevent transfusion lung injury. Public health policy decisions are beyond scientific reasoning, but they must take seriously into account the allocation of resources, and the cost of preventing one case of transfusion-lung injury by screening hundreds of thousands or even millions of donors must be evaluated with precision.

The HLA system and bone-marrow transplantation

The success of bone-marrow transplantation greatly depends on matching the recipient’s HLA phenotype against the donor’s. There has been a long debate about which HLA loci are clinically relevant. As clinical outcome data accumulates, it becomes evident that all HLA loci are relevant, including A, B, C, DRB1, DRB3/4/5, DQB1, and DPB1. In addition, DQA1 and DPA1 are polymorphic and form different DQ and DP heterodimers with the same DQB1 or DPB1, and it is important to take that into account. Although all HLA loci may be relevant, not all are relevant to the same extent. It is well known that some HLA molecules are expressed on the cell surface more extensively than others. HLA-A, HLA-B, HLA-C, and HLA-DRB1 are expressed at a higher level, whereas HLA-DRB3/4/5, HLA-DQB1, and DPB1 are expressed at a lower level. Those loci corresponding to high-expression molecules play a major role in the immune interactions between donors and recipients, and those corresponding to low-expression molecules play a lesser role, with mismatches in these loci being more tolerable. Nevertheless, it is the cumulative effect of multiple mismatches that seems to be critical, and clinical outcome studies should not isolate the effect of one HLA locus, but analyze the cumulative effect of mismatches at all loci.

Being an HLA match is not a symmetric relation and both the graft-rejection and graft-versus-host directions must be considered and evaluated separately. A donor-recipient pair can be a match in the graft-rejection direction and a mismatch in the graft-versus-host direction, or vice versa. A major difference between solid-organ transplantation and bone-marrow transplantation is the immune direction that dominates the clinical course. Whereas in solid-organ transplantation the only concern is graft rejection, in bone-marrow transplantation graft-versus-host disease dominates the clinical picture in immune incompatibility. Graft rejection is still a concern in bone-marrow transplantation, but to a lesser degree, particularly in ablative protocols in which the recipient’s bone marrow is destroyed. Nevertheless, even though graft rejection is relatively rare in bone-marrow transplantation, the cost of managing such patients is so high that the risk must be taken into account earnestly.

Although it has not been studied properly, anti-HLA antibodies may play a role in bone-marrow graft rejection. This is particularly important in cord-blood transplantation, as the small number of cells transplanted can be easily destroyed by anti-HLA antibodies in the recipient.

The HLA system and solid-organ transplantation

In solid-organ transplantation, HLA-matching means absence of HLA antigens in the donor that could be the target of anti-HLA antibodies present in the recipient. The matching is not between the antigens in the recipient and those in the donor, like in bone-marrow transplantation. This does not mean that antigen matching is not beneficial, but it limits organ allocation in a way that is not acceptable in the current clinical setting. In fact, there is evidence that antigen matching, like in HLA-identical siblings, significantly improves long-term survival. The desperate clinical condition of patients and the limitations in the availability of organs gives priority to the problem of allocation and dictates the practice of matching recipient antibodies against donor antigens, instead of doing recipient-donor phenotypic matching.

This has brought the practice of cross-matching to the centre of attention in solid-organ histocompatibility. The new techniques of antibody identification using single-antigen luminofluorecence beads allow the precise characterization of antibody specificities. Although the standard of care is to rely on various forms of cross-matching the patient’s serum with donor cells, a more efficient and accurate determination of antibody-antigen matching can be achieve by taking advantage of the new techniques of identification of antibody specificities and DNA-based typing methods.

Disease association of HLA alleles

There are a number of diseases associated with certain HLA alleles, particularly autoimmune diseases and infectious diseases. The key principle in these disease associations is the fact that the presentation of peptides by the HLA molecule to T cells is the fundamental event in the immune response underlying the pathogenesis of these diseases. In so far as certain HLA alleles can present certain peptides and not others, and other HLA alleles can present other peptides and so on, the presence of specific HLA alleles can naturally predispose an individual to suffer from conditions caused by various immune reactions elicited against self antigens in the case of autoimmune diseases, and foreign antigens in the case of infectious diseases.

A restricted list of diseases clearly associated with HLA alleles is provided. See HLA/Disease association

Population genetics of the HLA system

Boundaries of HLA genetic diversity

HLA presents the highest degree of polymorphism of all human genetic systems. Prior knowledge of the extent of diversity is essential in the development and selection of molecular typing methods. Reliable allele frequencies are also important in allogeneic unrelated hematopoietic stem cell transplantation to determine the likelihood of finding closely HLA matched donors for each patient. We observe only a fraction of all the alleles so far described. The number of HLA alleles in the official catalogue increases at a regular rate. At the present time, only about 15% of the alleles in the high-expression loci (A, B, C, DRB1) and 30% in the low-expression loci (DRB3/4/5, DQB1, DPB1) are truly relevant in clinical typing. The number of official alleles will keep growing, but the number of significant and relevant HLA alleles will remain the same. The official names of HLA alleles can be found at: hla.alleles.org

The chances of an individual having an allele from what might be called the nebulous set of sporadic alleles are as low or even lower that the chances of having a totally new allele never reported before and absent in the official nomenclature.

Many of the alleles in the nebulous set are poorly documented, they have only been typed once and have no confirmation. They may very well have errors. Some alleles are known not to belong to the official list and one cannot help but wonder how many such alleles are in a similar situation. An example of such alleles are Cw*020204, which is really Cw*0210 and not a silent-mutation variant of Cw*0202. Another example is DQB1*060501, which is really DQB1*0609 and not a separate allele; and DRB1*1206 is identical to DRB1*120101.

In order for the official HLA nomenclature to be of value, each new allele sequence must be independently confirmed. In addition, the complete haplotype where the new allele has been identified must be fully characterized.

In spite of the fear only a few years ago that the ever growing list of HLA alleles will present an insurmountable barrier in finding eligible donors for bone-marrow transplantation, the difficulty does not lie in the high mutation rate of the HLA system producing endless numbers of alleles, but in the recombination rate of HLA genes. The enemy of the transplant coordinator is not the rare allele, but the rare association between HLA-B and HLA-C, it is the rare haplotype. From this point of view, more emphasis should be given to the study of HLA haplotypes and their distribution in various populations.

Measuring the genetic diversity

The genetic diversity of the HLA loci is responsible for the efficiency of the immune system in eliminating cells carrying foreign antigens. There is a need to develop a measure of this genetic diversity in order to assess how different populations are equipped to respond to foreign-antigen exposure, and to evaluate the contribution of each HLA locus From the point of view of Information Theory, developed by Claude Shannon in 1948, genetic diversity can be measured as the entropy at each HLA loci.

Calculating the entropy of a genetic system provides valuable conceptual clarifications in discussions regarding population genetics and human evolution. It allows the proposal of the following hypotheses:

Any change in the population genetic pool that leads to increased HLA entropy (new alleles or a more even distribution) provides a selective advantage.
There is a level of entropy beyond which the cost of increasing it by means of population genetic changes (mutations, migration, etc.) does not exceed the benefits.
When the entropy is lower than the efficiency level, the tendency will be for the frequency of rare alleles to increase.
When a population exceeds the level of efficient entropy by forced artificial changes such as repeated migrations, the entropy will tend to decrease by the loss of low-frequency HLA alleles.

The concept of HLA entropy is also relevant in clinical practice as it explains why it is difficult to find donor matches for patients of African ancestry, for example.

A preliminary evaluation of the entropy at different HLA loci in various populations gives the following values in an arbitrary scale ranging from 50 to 100.

Ethnicity	A	B	C	DRB1
Africans	75	83	66	73
Europeans	61	79	66	71
Hispanics	71	91	69	82
Asians	62	86	65	67

These values show the diversity in Hispanics due to ethnic admixture. There is an unusual relative diversity of HLA-A in Africans. It is possible that the higher general HLA genetic diversity observed in Africans is due to a higher level of antigen exposure requiring a higher polymorphism in the antigen-presenting system.

Boltzmann developed the concept of entropy in the field that has come to be known as statistical mechanics, and later developed into thermodynamics. This concept was applied by C.E. Shannon in 1948 to what he called 'A Mathematical Theory of Communication', a very valuable theoretical discipline that has been central in the technical developments of the information age, so much so that this discipline is now called 'Information Theory'. The concept of entropy plays a central role in information theory as a measure of information, choice and uncertainty. It is defined as follows:

eeee

where K is an arbitrary constant that defines the unit of measure.

The frequency of homozygosity has been used in the past as a measure of genetic diversity. The concept of entropy opens the door to the theoretical tools of information theory and provides a richer conceptual framework to discuss the diversity of a genetic system.

A catalogue of HLA alleles for clinical transplantation

The question concerning the genetic diversity of the HLA system is not only of anthropological interest. Being the primary genetic system responsible for the immune response against foreign antigens, the characterization of the HLA polymorphic variants donors and recipients are endowed with plays an essential role in clinical transplantation of both solid organs and bone marrow. How polymorphic the HLA system is has direct implications on the practice of transplantation medicine: 1) The number of probes in oligonucleotide-hybridization typing kits, or the number of primers in PCR-based typing methods, as well as how they are designed and combined, depends on the number of alleles under consideration. 2) The reactivity patterns produced by these typing methods are interpreted differently if a locus has 30 alleles or 500. 3) The advice of the laboratory consultant to the clinician in regard to the sensitivity and specificity of a particular allele assignment is also a function of the genetic diversity of the HLA system. 4) The donor-search strategy depends entirely on our knowledge of the population genetics of the HLA system. Although one must consider not just the diversity of the system, but also the linkage disequilibrium between loci, and the distribution of extended haplotypes in different ethnic reference populations. It is essential in clinical transplantation to evaluate the genetic diversity of the HLA system, and define the catalogue of HLA alleles relevant in clinical practice.

A distinction is made between 'well documented' alleles and 'clinically relevant' alleles. What is at issue is not only the technical expertise required to establish that a given DNA sequence has been identified with accuracy, but also whether the DNA in question is the result of an inconsequential sporadic random mutation that appears and disappears without making any impact on the genetic makeup of a given reference population. We consider that the minimum requirement for an identified allele to have any prospects of being part of a genetic system is to have been identified in at least three independent different individuals. When this minimum requirement is met, we label the allele as 'well documented'. At this point only its existence is ascertained, no claims are made about its relevance. We say an allele is 'clinically relevant' when its genetic frequency in a given reference population is at least 0.001. In creating a universal catalogue, an allele so characterized in any reference population would be included. It may come as a surprise to find that only about 15% of the WHO official alleles can be called 'clinically relevant' alleles. Only clinically relevant alleles should be considered in the design of HLA typing methods, in the interpretation of clinical typing results, and in the development of donor-recipient matching strategies.

The dynamic nature of a genetic system is such that its diversity is in constant movement and change. From a distant perspective it behaves as if there were a relatively small number of well established alleles, and a pool of a much larger not well defined number of alleles that appear and disappear. It is only the subset of well established alleles that account for the biological function of the genetic system, and it is only the polymorphism encountered in this small subset that should be considered of functional relevance. Obviously it is from the pool of sporadic alleles that new alleles with relevant functionality to face new genetic pressures come; but it is only after an allele comes out of that pool and becomes well established that the new functionality can be recognized. Only well established alleles are proper objects of study in population genetics. This classification of alleles as clinically relevant or not clinically relevant, and as well documented or poorly documented, presents HLA typing in the proper perspective. Not all alleles should be treated equally. Some deserve enough attention so that typing methods can be designed to include and exclude them. Other alleles do not deserve similar attention and might just as well be ignored.

The chances of detecting a totally new allele in a clinical laboratory are higher than the chances of finding not just one in particular, but any of the alleles not classified as well documented. There is no practical clinical difference between saying that an individual has a new allele and that he happens to have an allele previously considered poorly documented.

A catalogue of well-documented and clinically-relevant HLA alleles can be found at: Category:HLA CWD Alleles

Importance of the concept of haplotypes in full linkage disequilibrium.

The distinction must be made between ‘haplotypes in full linkage disequilibrium’ and haplotypes resulting from random recombination.

An extended haplotype (covering all HLA loci) is in full linkage disequilibrium if and only if the allele at each loci is in LD with the rest of the alleles at all other loci considered as a block. If a haplotype segment (one or more loci) is not in LD with another haplotype segment, the two haplotype segments can be put together in the same chromosome as a result of random recombination through the process of crossing-over. In this case, an extended haplotype is said to be in random recombination at crossing-over point X (e.g. C-A, that is, between HLA-C and HLA-A).

An extended haplotype is properly described when it is characterized as being either in 'full linkage disequilibrium' or in 'random recombination' at one or more crossing-over points.

When the various haplotype segments have high gene frequency, it does not matter whether the haplotype is in 'full linkage disequilibrium' or in 'random recombination' because the presence of the haplotype can be considered established in the reference population. Nevertheless, if a haplotype segment (one or more loci) has a very low gene frequency, only haplotypes in 'full linkage disequilibrium' can be considered established in the reference population.

A donor for a patient with a rare allele can only be expected to be found if the haplotype where the rare allele is located is in 'full linkage disequilibrium'.

Combinations of well-established alleles can be difficult to find if they are presented in unusual recombinations outside their usual haplotypes. It is not only that their respective frequencies must be multiplied reaching very low number, but also the frequencies used in the calculations must be evaluated carefully. If a well-established allele is presented in a given haplotype 80% of the time, then it would be misleading to use the gene frequency to calculate the probability of finding this allele in other haplotypes by random recombination. If the other 20% of the presentation of the allele in question is haplotypes by random recombination, then the frequency to use would be RGF = G * 0.2 , where 'G' is the gene frequency. This brings up the concept of 'residual gene frequency' (RGF) to indicate the probability of finding a given allele in random recombination. As a result of these calculations, unexpected combinations of fairly common alleles can be extremely difficult to find.

A haplotype is defined as a 'hybrid haplotype' if and only if a segment X of the haplotype has a high gene frequency in population P and low gene frequency in population Q, and another segment Y of the haplotype has low gene frequency in population P and high gene frequency in population Q. Hybrid haplotypes are the result of mating between members of different ethnic populations. The first result of inter-mating is the presence in the same subject of two haplotypes characteristic of different ethnic populations. The second result is the presence of two haplotype segments characteristic of different ethnic populations in the same haplotype. Hybrid haplotypes are the result of recombinations. Examples of cases identified in our laboratory are A*0206-B*0801-C*0701-DRB1*0301-DQB1*0201, and A*0201-B*1801-C*1203-DRB1*1402-DQB1*0301, the former having an A allele of Asian origin with the rest of the haplotype being Caucasian, and the latter having a DR-DQ block typically Amerindian with the rest being Caucasian.

The existence of hybrid haplotypes predicts the future difficulties of finding donors for transplantation. As different races intermingle, hybrid haplotypes will become more common. It is practically impossible to find a matching donor for a patient with a hybrid haplotype. A catalogue of HLA haplotypes and linkage-disequilibrium blocks can be found at: HLA/Linkage Disequilibrium

HLA Typing

The complexity of HLA genetic typing comes from: 1) the polymorphic variety of HLA genes, 2) the dynamic active changes of genetic material in the form of mutations and recombinations of genes, and 3) the diploid presentation of genes in pairs.

HLA Typing Methodologies

DNA analytical methods are superior to other detection methods, such as the use of antibodies, now obsolete, because of its lack of reliability and accuracy. In fact, the serologic behavior of alleles can be precisely predicted according to the DNA sequence of the allele. Therefore, HLA typing is based on the study of DNA sequences. This study can be carried out by the isolated or combined use of PCR primers (SSP), hybridization probes (SSO), or automated sequencing (SBT).

The design of SSO and SSP methods should be based on the detection and distinction of well-documented common alleles. There is no point in the design of methods for the detection of sporadic alleles. SSO methods are superior to SSP methods in that they can be easily automated, allowing the consistent processing of large number of samples.

SBT methods should be based on the study of exon 2 and 3 for class I, and exon 2 for class II. The study of other exons plays only an ancillary role to resolve ambiguities and identify null alleles. There are two types of ambiguity in the identification of HLA alleles. Ambiguity type I results when multiple alleles have the same reactivity pattern. Optimally designed methods will not have two well-documented common alleles having the same reactivity pattern. When this is not the case, these well-documented common alleles with the same reactivity pattern must be differentiated by complementary analytical methods. Ambiguity type II is the result of having to detect and identify two alleles in the same locus simultaneously. Usually HLA typing consists in the detection and identification of diploid combinations of alleles, not in the detection and identification of isolated alleles. It is possible for multiple pairs of haploid reactivity patterns to give the same diploid reactivity pattern. For this reason the ambiguity type II is usually referred to as ‘heterozygous ambiguities’. It is important to distinguish between these two types of ambiguities in the way HLA typing reports are designed.

Ideally the two alleles in each locus should be separated prior to sequencing. This can be achieved most of the time with a preliminary PCR amplification. Whenever only one allele is detected in a locus, the question should be raised whether the subject is homozygous in that locus, or there has been a failure to detect a possible second allele. Sequencing methods are particularly prone to show preferential detection of one allele over another that may go unnoticed.

A good approach in HLA typing is to design a two-step strategy with a first line of testing using a very sensitive method that is unlikely to miss any allele, using generic well tested PCR primers; and a second-line of testing using a very specific method that may miss alleles, but the ones it detects are identified without ambiguities. The point in the first step is not to miss any allele that is present, and the point of the second step is not to identify any allele that is not there. This strategy can be achieved by combining SSO and SBT. The role of SSP is relegated to resolving ambiguities.

The interpretation of HLA typing results in bone-marrow transplantation

HLA typing should not be considered as the independent allele assignment in different genetic loci. Linkage disequilibrium is strong enough between HLA-B and HLA-C on one hand; and between HLA-DRB1, HLA-DRB3/4/5 and HLA-DQB1 on the other, to consider HLA-B-C as one block and HLA-DRB1-DRB3/4/5-DQB1 as another block. Allele assignment for these blocks should be carried out as a single step. A weaker linkage disequilibrium exists between HLA-A and HLA-B-C, between HLA-B-C and HLA-DRB1-DRB3/4/5-DQB1; and between HLA-DRB1-DRB3/4/5-DQB1 and HLA-DPB1. Knowledge of the extent and nature of these associations between genes and gene blocks is essential in the interpretation of HLA typing results. Without this knowledge HLA typing results are devoid of meaning. The meaning of an HLA phenotype is conveyed by the inference of two haplotypes from the phenotype and the description of these haplotypes in terms of their position in the distribution of all haplotypes in a given reference population. Linkage disequilibrium data for HLA loci and blocks of loci allows the inference of HLA haplotypes from HLA phenotypes.

The following guideline is suitable for bone-marrow transplantation:

A distinction is made between “poorly documented” alleles and “well documented” alleles. A WELL-DOCUMENTED ALLELE has been identified in at least three independent unrelated individuals.
A distinction is made between “clinically relevant” alleles and alleles that are not clinically relevant. A CLINICALLY RELEVANT ALLELE has a gene frequency of at least 0.001 in any reference population, having been identified in that population at least thrice.
HLA typing results should be reported as DIPLOID COMBINATIONS of alleles, that is, pairs of alleles consistent with the reactivity pattern of a particular typing method. It is not good enough to provide a list of alleles assigned to a particular locus, which allele goes with which other allele must be explicitly stated.
A distinction must be made between high-resolution typing and low-resolution typing. In HIGH-RESOLUTION TYPING ambiguities of diploid combinations of pairs of clinically-relevant alleles must be resolved unless the ambiguities are the result of sequence differences in areas of the HLA molecule that are not immunologically functional (outside exons 2 and 3 in class I and outside exon 2 in class II ==M.R. Ostadali== .) All but one diploid combination of clinically-relevant-allele pairs must be ruled out. IN LOW-RESOLUTION TYPING more than one diploid combination of clinically-relevant-allele pairs can be left as ambiguities.
When both high-resolution and low-resolution typing is performed for a given subject, at least one high-resolution diploid combinations must be consistent with the low-resolution diploid combinations; otherwise results must be re-evaluated and confirmed by additional typing if necessary. It is possible in this way to rule out by low-resolution typing ambiguities encountered in high-resolution typing.
Ambiguities involving alleles that are not clinically relevant need not be resolved.
If all clinically-relevant allele pairs are ruled out and only a diploid combination involving at least one allele that is not clinically relevant is consistent with the reactivity pattern, then the results must be confirmed by another method.
All diploid combinations involving pairs of well-documented alleles must be reported; and a distinction must be made between diploid combinations of pairs of clinically-relevant alleles, and diploid combinations with at least one allele that is not clinically relevant.
Ambiguous diploid combinations involving poorly-documented alleles need not be reported.
A typing strategy to identify and rule out well-documented NULL ALLELES must be developed. This strategy needs not to include poorly-documented null alleles; but the identification of new alleles must always take into account the possibility that the new sequence identified could compromise gene expression. The strategy to identify null alleles should take advantage of our knowledge of the linkage-disequilibrium blocks and haplotypes null alleles appear in.
HLA-B and HLA-Cw typing must be checked for consistency with established linkage-disequilibrium HLA-B-Cw blocks. If there is no such consistency, then the typing must be confirmed by another testing method, or by family genotype analysis.
HLA-DRB1, DRB3/4/5 and DQB1 typing must be checked for consistency with established linkage-disequilibrium HLA-DRB1-DRB3/4/5-DQB1 blocks. If there is no such consistency, then the typing must be confirmed by another method, or by family genotype analysis.
In case of HOMOZYGOSITY, the possibility of allele drop out and preferential amplification must be ruled out by methodological validation. If the method in question has not been positively validated for the absence of allele drop out and preferential amplification, then homozygosity must be confirmed by another method, or by family genotype analysis. Heterozygosity must always be ruled out.
The HLA typing of family members must be consistent with genetic segregation. Any discrepancy of reported family relations with genetic segregation principles must be investigated and explained. This includes the explicit reporting of crossing-over instances.
Positive FAMILY GENOTYPE ANALYSIS requires the identification of four haplotypes in a family unit.
One-haplotype or two-haplotype HLA-identity among siblings can only be based on positive family genotype analysis that expands throughout the entire HLA region, including HLA-A-Cw-B-DRB1-DRB3/4/5-DQB1-DPB1.
In the absence of positive family genotype analysis, suspected HLA-identity between siblings by low-resolution typing must be confirmed by high-resolution typing, if a BMT clinical protocol requires HLA identity.
A new allele that has never been described before belongs to the same category as poorly-documented alleles. In both cases there is no expectation of finding an unrelated matched donor, and information must be provided regarding the closest similarities with other alleles that should be considered in donor searches.
Whenever a sequence is identified that does not match any well-documented allele, a search for matches with poorly documented alleles must be made before it is called new.
A GENOTYPE is a pair of haplotypes.
A PHENOTYPE is a given series of single diploid combinations at each loci. A phenotype is associated with the set of all possible genotypes consistent with it. For n heterozygous loci, the theoretical number of possible genotypes consistent with a given phenotype is 2(n-1). Due to the phenomenon of linkage disequilibrium, not all these theoretically possible genotypes are encountered.
GENOTYPE ANALYSIS is the evaluation of the set of all possible genotypes consistent with a given phenotype in terms of their distribution in one or several reference populations. In a strict sense, the interpretation of HLA typing results does not consist in allele assignment, but in haplotype assignment.

Errors and mistakes in HLA typing:

Thinking that because a sequence has been given an official name it is quite possible to find that sequence in clinical practice.
Thinking that, because of naming protocols, low 3rd-4th-digit numbers are common and high numbers are rare. (The naming of HLA alleles is so inconsistent that no conclusions can be derived from allele names. In spite of the tradition to convey meaning to allele names, they should be treated just as reference keys; reference keys, however, that unfortunately change over time.)
Confusing haploid ambiguities (type I) with diploid ambiguities (type II).
Failure to consider whether an allele that has been identified is a well-documented allele and a clinically-relevant allele. This failure will result in omitting the necessary confirmatory steps for unusual alleles.
Failure to take into account that linkage disequilibrium in HLA and omit the necessary confirmatory steps in the case of unusual inter-loci associations.
Forgetting that SSP interpretation tables are sometimes based on theoretical grounds relying on information about allele sequences, and not on empirical studies. Primer reactivity with extremely rare sequences not studied empirically cannot be taken for granted.
Readiness to change SSO or SSP reactions from positive to negative or vice versa just because the reading is close to the cut-off point, without evaluating the implications of the change from the point of view of population genetics and family genotype analysis.
Ignoring that whenever an unusual allele is detected by sequencing, there is always the possibility of a diploid combination ambiguity of a common allele with a new allele.

Detection and identification of anti-HLA antibodies

Traditionally, anti-HLA antibodies have been detected using panels of cells with well characterized phenotypes. This approach results in insurmountable problems of interpretation that give imprecise inaccurate results. The availability of single-antigen beads opens a new era in HLA antibody identification, making the use of panels of cells obsolete. The use of single-antigen beads, however, presents two problems. First, there is a need to evaluate the intensity of the antibody, which is measured better with cell panels. Second, there must be a complete representation of alleles of all serologic specificities. It is a mistake to rely on the first two digits of the name of the allele to predict its serologic behaviour, and, to be thorough, all well-documented common alleles must be represented in single-antigen bead array. Actually, the use of single-antigen beads has revealed that alleles previously thought to be in the same serologic group, do in fact exhibit different serologic reactivity, with public and private epitopes previously ignored.

Once the detection and identification of anti-HLA antibodies has been optimized, the reactivity between antibodies in the recipient and antigens in the donor can be accurately predicted without the need for crossmatches. The only pending question is the intensity of the antibodies detected, which traditionally has been evaluated by means of crossmatches. Once the intensity of reactivity of antibodies with single-antigen beads has been evaluated appropriately, HLA crossmatching can be considered obsolete.

HLA and transplantation: Matching donors and recipients

The interpretation of HLA typing consists in the comparison of a given haplotype to the distribution of all haplotypes in a given reference population. This comparison is the key to perform donor searches and to plan bone-marrow transplants. The proper interpretation of HLA phenotypes is in terms of the genotypes that account for the phenotypes; and the proper description of genotypes is in terms of where the two haplotypes in question fall in the distribution of all possible haplotypes in a reference population.

The difficulty of BMT is the difficulty of selecting a donor when no HLA-match is available. It can be said that the art of BMT is the art of HLA mismatching. It can be shown that relative comparisons between donors lead to non-transitive ambiguities and suboptimal donor selections. To solve these ambiguities HLA matching between donors and recipients must consist in a matching score function based on a geometric approach to HLA matching.

The logic of matching donors and recipients

We define ANTIGEN EXPRESSION VARIABLES in the following way: a donor is represented by ‘D’ and a recipient by ‘R’; the antigen in question is represented by a subscript, such as ‘DA’ for the expression of the A antigen in a donor; these are Boolean variables with a ‘true’ value indicating expression of the antigen and a ‘false’ value indicating lack of expression of the antigen. We now define two logical functions, the HvG crossmatch function for any antigen X as:

eeee

and the GvH crossmatch function for any antigen X as:

eee

The matching score

The concepts of HvG and the GvH matching scores are measures of how incompatible a mismatch is: the higher the score, the more incompatible the match is.

The HvG matching score:

eee

Where Min(x, y) is a function that returns the value of the argument that is lowest, i is the locus out of n loci to be considered in the match between donor D and recipient R; Ai and A'i are the two alleles corresponding for locus i in the donor if it appears as a subscript of D, or the recipient if it is a subscript of R.

The GvH matching score:

eee

In the GvH crossmatch the roles of donor and recipient a reversed and, since the D(X,Y) function is not symmetric, its arguments must be switched.

In transplantation the HvG matching score gives a prediction of graft rejection, and the GvH matching score a prediction of a graft-versus-host reaction.

Measuring histocompatibility in bone marrow transplantation

The problem in bone-marrow transplantation (BMT) is not matching, but mismatching a donor to a patient. When there is no HLA match, which of a series of mismatched candidates should be selected as the donor? The measurement of histocompatibility between two subjects must be based on:

Both host-versus-graft (HvG) and graft-versus-host (GvH) immune directions must be considered (I).
Each HLA locus must be evaluated separately (L).
A different measurement must be made for each HLA molecule in each locus in the immune target (the graft in HvG and the host in GvH); that will typically be two per locus, and one in homozygous cases, but it can be one, two or four in the case of DQ heterodimers (X). Which DQA1 alleles combine with which DQB1 alleles must be taken into account. It is postulated that the number of copies of an HLA molecule on the cell membrane are important in eliciting an immune response, and that the density of molecules is diluted with the number of combinations of DQA1 and DQB1 heterodimers (N).
Histocompatibility is measured based on the distance in a vector space where HLA molecules are represented as vectors, and where the coordinates are the sequence variables of these molecules (D). Each coordinate, representing an amino acid position, has a scale or weight depending on how critical that position is in the presentation of peptides and the recognition by T cells.
A distance matrix of allele distance based on an assigned weight system for each position is presented.
In addition, KIR-ligand matching must be considered for each of the main KIR receptors.
As the role of HLA antibodies in BMT is evaluated an antibody match must be included too.

Histocompatibility is defined as the relation R(I, L, X, N, D) between the variables mentioned above, and evaluated in reference to a donor and a recipient. This relation sets the foundation to select optimal donors in the absence of a perfect HLA match.

Antigen-antibody match

Anti-HLA antibodies can cause engraftment failure. It is important to know that, whenever there is any HLA mismatch at any locus:

the patient is evaluated for the presence of anti-HLA antibodies, and
the donor is evaluated to make sure he does not carry any HLA alleles with epitopes against which the antibodies in the patient show specific reactivity.

A donor-recipient pair is said to be an antigen-antibody-match if and only if the recipient does not show any anti-HLA antibodies with specificity against epitopes present in the donor.

KIR match

In addition, in so far as natural killer cells can elicit a cytolitic reaction against leukaemia cells, it is important to evaluate the KIR matching between donors and recipients. In order for donor natural killer cells to elicit a reaction against leukaemia cell in the recipient, the following three conditions must be met for each inhibitor KIR receptor:

The KIR receptor in question must be present in the donor.
The recipient must carry at least one HLA ligand to the KIR receptor.
The donor cannot carry any HLA ligand to the KIR receptor.

Conditions 2 and 3 imply an HLA mismatch. An HLA-mismatched donor is said to have KIR anti-leukaemia benefit if, in addition to being HLA-mismatched, it is also KIR-mismatched meeting the three previous requirements.

DQ heterodimers

Since DRA1 is monomorphic, the diversity of DR heterodimers depends solely on DRB1 polymorphism, an individual having one or two DR heterodimers depending on homozygosity.

DQ heterodimers, on the other hand, exhibit polymorphism in both alpha and beta chains. Given two DQA1 genes and two DQB1 genes, four different molecules can be put together, so matching for DQ may consist in the comparison not between pairs of molecules, but between sets of four molecules.

There are two groups of DQ molecules. The alpha chain of one group binds only to beta chains of the same group, and vice versa. One group includes the DQB1 alleles with DQw1 serologic specificity (DQB1*05xx and DQB1*06xx) and the corresponding DQA1 alleles found in the same haplotype (DQA1*01xx). The other group consists of all the other DQ haplotypes and associated DQA1 and DQB1 alleles.

The DQw1 alpha molecules are distinctly characterized by the amino-acid sequence FSKFGGFDPQGALRNMAVAKHNLNIMIKRY at positions 61-80. The DQw1 beta molecules are characterized by the amino-acid sequences EVAYRGI or EVAYRGI or EVAYRGI at positions 84-90. Whereas the non-DQw1 beta molecules have the sequence QLELRTT at the same positions.

We define the concept of DQ HETEROGENEITY as the characteristic of an individual having one DQw1 haplotype and a non-DQw1 haplotype. DQ HOMOGENEITY, on the other hand, is the characteristic of an individual having two DQw1 haplotypes or two non-DQw1 haplotypes. Subjects with DQ heterogeneity have two and only two DQ heterodimers. Subjects without DQ homogeneity have one DQ heterodimer if they are homozygous; and two if they are heterozygous for only one locus and four if they are heterozygous for both loci. So a given individual can have one, two or four DQ heterodimers.

Conclusion

Even though many factors are considered in the clinical practice of both bone-marrow transplantation and solid-organ transplantation, sometimes relegating HLA typing and matching to a secondary position, graft rejection and graft-versus-host disease are fundamentally dependent on the identification and characterization of the HLA phenotype in both donor and recipient, and the assessment of the degree of matching between them.

In so far as the genetic diversity is bound to increase due to the mixing of populations of different ethnic origins, it may be anticipated that in the long term the proportion of difficult HLA matches will increase. In these cases, the problem of selecting the optimal mismatched donor will require:

detailed knowledge of population genetics,
a functional assessment of mismatches between specific pairs of alleles, and
knowledge of the clinical implications of the cumulative effect of mismatches in multiple loci.

Additional Reading

Marrack P, Kappler J: The antigen specific, major histocompatibility complex restricted receptor on T cells. Adv Immunol 1986: 38:1-30.
Marrack P, Bender J, Jordan M, Rees W, Robertson J, Schaefer BC, Kappler J. Major histocompatibility complex proteins and TCRs: do they really go together like a horse and carriage? J Immunol 2001: 167(2): 617-21.
Marrack P, Kappler J. The T cell repertoire for antigen and MHC. Immunol Today 1988: 9(10): 308-15.
Saper MA, Bjorkman PJ, Wiley DC. Refined structure of the human histocompatibility antigen HLA A2 at 2.6A resolution. J Mol Biol 1991: 219(2): 277-319.
Brown JH, Jardetzky TS, Gorga JC et al. Three dimensional structure of the human class II histocompatibility antigen HLA DR1. Nature 1993: 364: 33-9.
McFarland BJ, Beeson C. Binding interactions between peptides and proteins of the class II major histocompatibility complex. Med Res Rev 2002: 22(2):168-203.
Lawlor DA, Zemmour J, Ennis PD, Parham P. Evolution of class I MHC genes and proteins: from natural selection to thymic selection. Annu Rev Immunol 1990: 8: 23-63.
Hughes AL, Ota T, Nei M. Positive Darwinian selection promotes charge profile diversity in the antigen binding cleft of class I major histocompatibility complex molecules. Mol Biol Evol 1990: 7(6): 515-24.
Gustafsson K, Wiman K, Emmoth E et al. Mutations and selection in the generation of class II histocompatibility antigen polymorphism. EMBO J 1984: 1655-61.
Hughes AL, Nei M. Maintenance of MHC polymorphism. Nature 1992: 355: 402-3.
Hughes AL, Nei M. Nucleotide substitution at major histocompatibility complex class II loci: evidence for overdominant selection. Proc Natil Acad Sci USA 1989: 86: 958-62.
Opelz G, Wujciak T, Dohler B, Scherer S, Mytilineos J. HLA compatibility and organ transplant survival. Collaborative Transplant Study. Rev Immunogenet 1999: 1(3): 334-42.
Erlich HA, Opelz G, Hansen J. HLA DNA typing and transplantation. Immunity 2001: 14(4): 347-56.
Hansen JA, Yamamoto K, Petersdorf E, Sasazuki T. The role of HLA matching in hematopoietic cell transplantation. Rev Immunogenet 1999: 1: 359.
Petersdorf EW, Anasetti C, Martin PJ, Hansen JA. Tissue typing in support of unrelated hematopoietic cell transplantation. Tissue Antigens 2003: 61(1): 1-11.
Muller JY, Halle L, Jaeger G. HLA-A and B antigens in AKA Pygmies. Tissue Antigens 1981: 17: 372-5.