Immunology and genetics have been among the most rapidly developing areas of science in the past few decades. It will be our purpose here to summarize basic information from these fields required for the work of the clinician who cares for recipients of transplanted organs. It is interesting to recall that the dawn of cellular immunology, as we now know it, and much of its early impetus to growth was generated by efforts to understand the behaviour of transplanted tissues.

The history of understanding of the fundamental processes of transplant rejection goes back further than is often appreciated. Much information, sometimes imperfectly interpreted, was gathered in the course of experiments in which samples of viable neoplastic tissue were transplanted to recipient animals which were immunogenetically different from those in which the tumours arose. This meant, as P. B. Medawar nicely put it, that investigators intending to ‘study cancer by transplantation were actually studying transplantation with cancer’. Thus, the transplanted neoplastic cells were subject to rejection in all but highly compatible recipients.

Out of such experiments with transplanted tumours and increasingly with normal tissues, especially skin grafts, came fundamental information about the effect on transplant survival of the genetic relationship between donor and host, as well as a realization that the destructive process of rejection is indeed immunological in nature. Furthermore, a dawning of understanding regarding the immunological mechanisms that bring about rejection was emerging at just about the time of the first clinical trials of kidney transplantation in Boston and in Paris in the early 1950s. It was also known at this time that whole-body irradiation and treatment with certain chemical agents, especially cortisone derivatives, could significantly reduce the severity of the rejection response. In the prevailing atmosphere, attempts at organ transplants, making use of the safest possible surgical techniques were felt to be reasonable. Kidneys were attached, for example, to vessels in the groin and placed in subcutaneous pockets while urinary drainage to a leg bag was achieved by cutaneous ureterostomy. In some cases, partly as a direct kind of histocompatibility test, the new kidney was maintained in a sterile box and was attached to arm vessels by percutaneous conduits. The intent here was to determine the early outcome of an allogeneic transplant before placing the organ within the body of the recipient.

Initially, animal experiments established very clearly that prior exposure to nucleated cell injections or organized tissue transplants from a given donor would leave a recipient primed for a greatly accelerated reaction to later transplants from the same donor. This more vigorous reaction was accompanied by a much more rapid and intense invasion of migratory recipient cells into the grafted tissue. Observations such at this, coupled with the finding of Mitchison that the immunity induced by transplant rejection could be transferred to another recipient by transferring cells from lymph nodes draining the graft led to a generally held view that the immune response to transplanted tissues was mediated by activated recipient cells exerting their destructive effects directly upon the graft. However, evidence then began to accrue from a number of sources that humoral factors could, at least under some circumstances, have a dramatic and decisive impact on transplanted cells and tissues. One particularly striking observation was that certain kidney transplants were seen to be destroyed within minutes of receiving blood from their new hosts in a reaction called ‘hyperacute’ rejection. The character of the pathological changes seen in such transplants after their removal made it unlikely that the process could have been mediated by recipient cells and the emergence of techniques (originally following the important work of Peter Gorer) that could detect humoral antibodies directed toward cell surface antigens made it possible to show that hyperacute rejection was attributable to pre-existing humoral antibody directed to at least some of the antigens present in the transplant. This stimulated further exploration of the participation of humoral antibody in rejection reactions in animals, where such reactions had been extremely difficult to find by methods such as the passive transfer of serum to a new recipient from a previously immunized subject. It also suggested the advisability of testing the reactivity of a recipient's serum to donor cells in the presence of complement as a preliminary ‘cross-match’ test before a transplant is performed.

All these points have subsequently received much attention, and a great deal of refinement in our knowledge of how cellular and humoral immunity are generated has taken place. Discussion continues, as we will describe below, regarding the most appropriate tests for humoral antibody to donor cells, which will predict with certainty the destruction of a transplanted organ from the donor of those cells. Some uncertainty remains as to the types of donor cells that can be considered representative of a later organ transplant, and even, in the minds of some, whether the presence of cytotoxic antibody to donor T lymphocytes, for example, is regularly associated with harmful effects on certain organ transplants, especially livers and hearts.

Knowledge derived both from laboratory experiments and clinical observations continues to expand in this complex field. Differences of opinion, often leading to new advances, are bound to be present from time to time, but, as mentioned above, few would dispute the contention that the process of rejection, a direct manifestation of immunogenetic factors, is of central importance to the outcome of any living cell or tissue transplant. Accordingly, these processes are of great importance to surgeons and other clinicians who seek to help patients by the transplantation of living tissue and organs of any kind.

Activation of the immune system requires that receptors on B cells or T cells recognize foreign substances. Those structures having the right size and configuration to fit within these receptors are called antigens. For the immune system as a whole, antigens may be soluble proteins or cell surface molecules, but only the cell surface antigens are capable of causing rejection of transplanted organs. Three types of surface antigens that are important in transplantation have been identified: (1) the major histocompatibility complex (MHC) antigens; (2) the minor histocompatibility antigens; and (3) the blood group antigens.

As their name implies, the major histocompatibility complex (MHC) antigens are the most important antigens causing graft rejection. Their presence was recognized from experiments using inbred strains of mice (see below) in which the products of one particular series of genes were found to be especially important in provoking graft rejection. The site within the genome which includes these genes is called the major histocompatibility complex. All species studied have been found to have MHC genes on a certain chromosome, and the antigens encoded within it have been well characterized in many of them.

Two types of MHC antigens have been identified and are now subdivided as class I and class II antigens. Over the years they have been given different names, as listed in Table 1 217. The class I MHC antigens are composed of two chains, one of which is very polymorphic (meaning that it varies from individual to individual) while the other, called &bgr;&sub2;-microglobulin, is similar for all members of the species. The class II MHC antigens are composed of two polymorphic chains, called &agr; and &bgr;. Class I antigens exist on almost all cells of the body, while class II antigens are less widely distributed, being expressed on macrophages, B cells, and other cells with ‘antigen-presenting’ function.

Despite the differences between the class I and class II MHC antigens, the two types of antigens probably have a similar tertiary configuration. In both cases, each chain has a short intracytoplasmic tail followed by a transmembrane portion with a relatively large extracellular portion which can be divided into regions called domains (Fig. 1) 672. The &bgr;&sub2;-microglobulin chain (molecular weight 12 000) is a constant chain with no polymorphism and is coded on a different chromosome to the MHC loci. It is necessary for the transport and expression of the class I MHC antigens on the cell surface.

At least three different loci within the MHC of humans encode polymorphic class I antigens. Three other loci encode class II antigens. The expression of MHC antigens is codominant (the genes on both chromosomes of a pair are expressed) and therefore up to 12 different MHC antigens are expressed on the cell surface (six encoded on the chromosome inherited from the father and six on the chromosome from the mother).

In humans the three class I loci are designated A, B, and C and the three class II loci are called DP, DQ, and DR. These six loci are organized on chromosome 6 of humans such that the class I loci are together, as are the class II loci (Fig. 2) 673.

One of the cardinal features of the MHC antigens is that there is substantial variation in the fine structure of each antigen between individual members of the species. This occurs because there are many different alleles within each species that can encode the MHC antigen determined by any particular MHC locus. For example, there are at least 20 different MHC class I antigens that can be encoded within the A locus. This variation in the MHC antigens between individuals is called polymorphism.

Structurally, most of the variation between different MHC antigens occurs in the outer two domains of the molecule. Recent crystallographic data have revealed that these areas together form a ‘cleft’ in the exterior of the antigen and that this cleft on the different MHC antigens therefore has many slightly different configurations (Fig. 3) 674.

The ability of the immune system to develop an antibody response against foreign MHC antigens has provided a tool with which the different antigens expressed by each individual can be characterized. The 20 or more class I antigens encoded in the A locus have been given numbers such as A2, A3, etc. Similarly, numbers have been assigned to antigens of the other loci. In practice, the available reagents are not adequate to characterize all of the HLA antigens, and the current clinical practice is to define serologically only the HLA A, B, C, and DR antigens. Since there are two antigens (from two chromosomes) for each of the loci, one person's phenotype might be described as: Equation 24

The process of determining the HLA phenotype is called tissue typing. It is often confused with cross-matching, a procedure that tests not for compatibility but for the presence of pre-existing antibodies in a recipient, which can react with a donor's MHC antigens. The two techniques use similar methods but the implications of finding an HLA-matched kidney and one with a negative cross-match are completely different.

In typical mendelian fashion, every person inherits one copy of the MHC from each parent. Each chromosome copy is called a haplotype. Unless the two parents express some of the same histocompatibility antigens, children will share half of their MHC antigens with each of their parents. This is called a onehaplotype match, and parents and children are often said to be ‘haplo-identical’.

The sharing of MHC antigens between siblings is more variable. Since each parent has two MHC haplotypes and will pass on only one to each child, four possible combinations of haplotypes may be inherited. Thus, for a particular child the chance that another sibling will have inherited the same two MHC haplotypes is 1 in 4, in which case the siblings are genotypic ‘HLA-identical’. The chance that the sibling will have inherited only one of the same MHC haplotypes is 2 in 4, in which case that sibling is ‘haplo-identical’. The chance that another sibling will have inherited two different MHC haplotypes is 1 in 4, in which case that sibling is HLA mismatched. These odds are altered slightly by the possibility of genetic recombination occurring within the MHC (the incidence of which is 1 per cent), such that it is possible, for example, for a sibling to share 1½ MHC haplotypes (Fig. 4) 675.

The presence of MHC antigens in every mammalian species and the strength of the immune response that they evoke suggests that they are important components of the immune system. However, despite their discovery in transplantation experiments and their importance in graft rejection, the MHC antigens probably do not exist in order to cause graft rejection since this would not offer survival benefit to the species. Despite this obvious point, it was not until the late 1970s that immunologists were able to assign a function to the MHC antigens beyond their role in transplantation.

It is now believed that the MHC molecules serve as focusing elements which direct the cellular elements of the immune system to the site where they can do the most good. If one considers a viral pathogen, it would be inefficient to occupy an entire T cell in neutralizing one, or even several, tiny viral particles. This can be accomplished by the small, soluble, and abundant antibodies which are produced and secreted by B cells. The bulky machinery of T cells would be better used to eliminate the larger cells that have been damaged by viral infection or which have undergone neoplastic transformation.

The MHC antigens accomplish their purpose by acting as antigen-presenting structures. Invading pathogens are broken down within cells into small peptide fragments and then carried to the cell surface in the polymorphic structural cleft of MHC molecules.

These peptides, sitting in the cleft of MHC antigens, are then exposed to the receptors of passing T cells. Since the cleft of each MHC antigen is different, it can only present some peptide fragments of some foreign pathogens. Therefore, it is to the advantage of an individual to express more than one kind of MHC antigen. If the cleft of the MHC antigen cannot present any peptides from a particular virus, the cleft of another MHC antigen might be able to do so. Furthermore, it is to the advantage of the species to have many different forms of each MHC antigen so that if one individual is unable to present peptides of the pathogen using any of its MHC antigens, another individual expressing different MHC antigens might be able to do so. Thus the antigen-presenting function of MHC antigens explains the multiple loci within the complex and the polymorphism of the antigens encoded there.

While the MHC antigens serve to present the peptides of pathogens on the surface of infected cells, the goal of restricting T cells so that they recognize these peptides only in this setting requires an additional feature. This restriction is imposed on the immune system by selection during T-cell maturation only of cells with receptors that are triggered by peptide fragments presented in association with MHC antigens. Part of the T-cell receptor must recognize the peptide itself and part must recognize surrounding determinants formed by the MHC molecule. Because of this dual recognition by T-cell receptors, cell-mediated responses are said to require ‘associative recognition’. The role that MHC antigens play in limiting the site of peptide recognition leads to their description as ‘restriction elements’.

Although MHC antigens do not exist to prevent organ transplantation, nonetheless they do act as the stimulators and the targets of the immune response in organ rejection. These antigens are particularly important in part because they elicit both T- and B-cell immune responses. Secondly, because the MHC antigens are so polymorphic, it is very unusual to find two unrelated individuals whose antigens are not different. Thirdly, it is a peculiar feature of transplantation immunity that the strength of cell-mediated immunity to allogeneic (those from another individual) MHC antigens is very strong. For example, the precursor frequency of T cells that respond to a foreign MHC antigen is 100- to 1000-fold higher than the frequency of T cells that respond to the peptides of a pathogenic virus presented with an MHC molecule. This surprising finding is discussed further below.

Minor histocompatibility antigens are simply defined as antigens that cause cell-mediated graft rejection but which are not major histocompatibility antigens. This definition by exclusion reflects the limited information available regarding the minor antigens, which are not, therefore, products of genes within the MHC. For many years they were pictured as similar in character but weaker in strength than the MHC molecules, but more recently this view has been replaced by a growing consensus that the minor antigens represent peptides of autologous proteins, which are presented on the surface of cells in association with major histocompatibility molecules. While such presentation of peptides probably occurs for many autologous proteins, minor histocompatibility antigens are formed when there is a genetically determined difference in a protein's structure between members of a species such that allelic variants among the peptide fragments can be presented in association with MHC antigens. As a hypothetical example, shown in Fig. 5 676, two different amino acid sequences for protein X (perhaps serum albumin) might exist such that a peptide of the protein from one individual, presented in association with an MHC molecule, will create a minor histocompatibility determinant which is foreign to the T cells of another individual.

The characterization of minor histocompatibility antigens described above suggests that there might be a large number of minor histocompatibility differences between any two members of a species. This does appear to be the case and this number has been estimated using inbred strains of mice and the principles of transplantation immunogenetics.

If one starts with two inbred mouse strains, ‘A’ and ‘B’, each homozygous throughout its genome, it is possible to generate F&sub1; offspring, called (A × B) F&sub1;, by breeding A with B mice. All F&sub1; mice will be heterozygous for all polymorphic loci within their genomes. In addition, one can generate F&sub2; mice by breeding mice from the F&sub1; generation together. At any given locus, 25 per cent of F&sub2; mice will be ‘AA’, 50 per cent will be ‘AB’, and 25 per cent will be ‘BB’ (Fig. 6) 677.

After breeding mice in the manner above, transplantation experiments can be performed between members of the generations. Such experiments have been carried out and the results summarized as the five ‘laws of transplantation’ (see Table 2 218). Although called ‘laws’, they were really experimental observations. They were described by Snell based on the work of C. C. Little and others at the Jackson Laboratory in Bar Harbor, Maine, United States.

C. C. Little recognized that these ‘laws of transplantation’ were the consequence of two fundamental principles:

(1)that graft rejection was determined by multiple histocompatibility loci with codominant expression of their products;

(2)that rejection would occur whenever the donor expressed products of histocompatibility loci which were not expressed by the recipient.

Little's insight identified the importance of histocompatibility loci in transplantation and initiated the experiments that identified what became known as the major histocompatibility complex of genes. He also recognized that the observation incorporated in the fifth ‘law of transplantation’ provided the means to determine the total number of histocompatibility loci.

To understand Little's reasoning, let us assume that there is only a single histocompatibility locus responsible for rejection, as shown in Fig. 6 677. Under these circumstances parental grafts from strain A to mice of the F&sub2; generation would be accepted by three-quarters of the offspring (those which were ‘AA’ or ‘AB’ at the important locus). If two histocompatibility loci were involved in graft rejection, then the probability of A strain graft acceptance by F&sub2; mice would be (3/4) × (3/4), or (3/4)&sub2;. If the number of loci involved was larger still, then the probability (p) of graft acceptance would be: Equation 25

where n is the number of histocompatibility loci. Since the fifth law of transplantation showed that P → F&sub2; grafts were usually rejected, Little concluded that n must be quite large. He also believed that it could be determined experimentally. In fact, the results of thousands of grafting experiments from parental to F&sub2; mice indicate that at least between 30 and 50 histocompatibility loci must exist. The major histocompatibility complex, encoding the MHC antigens discussed above, represents only one of these sites, and all the remaining loci encode minor histocompatibility antigens.

One of the important ways in which minor transplantation antigens differ from major antigens is the requirement that T cells recognize minor antigens only when their peptides are presented in association with major histocompatibility molecules. Minor antigens are therefore similar to external pathogens, such as viruses, and the ‘associative’ recognition of minor antigens represents the normal process of T-cell recognition. In contrast, the direct recognition of allogeneic major histocompatibility antigens, without a requirement for antigen processing and presentation in association with other MHC molecules, represents an unusual T-cell response.

The peptides that form minor histocompatibility antigens can be expressed in association with either class I or class II MHC antigens. When in association with class II antigens, they generally activate a subpopulation of T cells that expresses a surface antigen called CD4. These T-cells function primarily to provide ‘help’ for other T cell responses. On the other hand, when presented in association with class I MHC antigens, they generally activate a subpopulation of T cells that expresses an antigen called CD8. These cells usually have cytotoxic function. Thus the response to minor histocompatibility antigens usually requires the function of both of the two major subpopulations of T cells. As will be discussed below, this requirement is in contrast to the response to major histocompatibility antigens.

Another important feature of minor histocompatibility antigens, in contrast to the MHC antigens, is that there is generally no detectable antibody response to these antigens. This feature, although surprising at first, when it was assumed that minor antigens were cell surface structures, similar to but weaker than MHC antigens, proved not to be so when it was recognized that the minor antigens probably do not exist independently on the cell surface in three-dimensional form but only as peptides presented with MHC molecules. Antibody responses to peptide fragments presented by MHC antigens have been generally difficult to detect. Without antibodies to the minor antigens, we lack one of the important tools to identify and characterize them.

Another feature of the response to minor histocompatibility antigens, called immunodominance (which is not the same as genetic dominance), is that in the face of multiple minor antigen disparities, the immune response may select only a few of these disparities as targets. Therefore, in some experimental cases where grafts differing by multiple minor antigens are exchanged, rejection of a second graft of a type expressing only one of the original minor antigens may not show evidence of sensitization to that antigen. An explanation for this phenomenon is that when multiple peptides are available for presentation by MHC molecules, some of them may associate preferentially with the available MHC antigens, excluding the other peptides. Thus the requirement for associative recognition of minor antigens is probably responsible for the phenomenon of immunodominance.

One of many minor histocompatibility antigens has been studied extensively, namely the H-Y minor antigen, which can be detected by the rejection of skin grafts from male donors by female members of the same inbred strain of animal. As inbred females accept grafts from one another and inbred males accept grafts from females of the same strain, it appears that this graft rejection is due to expression of a minor antigen encoded on the Y chromosome. Some inbred strains of mice are capable of rejecting grafts exchanged between the sexes, while others are not. Therefore it is possible to investigate the requirements for a response to the H-Y histocompatibility antigen. Unfortunately this investigation has not provided simple answers. It appears that the ability to reject H-Y disparate grafts depends on multiple genes encoded both within the MHC and elsewhere.

The blood group antigens are carbohydrates and glycoproteins present on the surface of red blood cells and a few other cell types. They were identified and characterized because of their importance in blood transfusions, an early example of tissue transplantation.

Three major blood group antigens are commonly recognized, designated O, A, and B. In actual fact the situation is far more complicated than this simple designation implies, but for purposes of organ transplantation a simple picture is sufficient. In general, the different blood group antigens all arise from a single glycoprotein structure. The genes responsible for the several forms of this structure encode glycosylation enzymes which modify the core protein differently. Individuals of blood group O express the unmodified glycoprotein, those of blood group A express this glycoprotein with an additional external sugar, and those of blood group B express the core glycoprotein with a different additional external sugar. Individuals of blood group AB have the enzymes to add both additional sugar molecules. The inheritance of the major blood group antigens follows mendelian principles for a single locus with three allelic genes (O, A, B). The relationship between genotype and phenotype is shown in Table 3 219.

There is no T-cell mediated immune response to the blood group antigens, but there is an antibody response to these structures. The blood group antibodies are unusual in that they develop without prior exposure to foreign blood cells. Blood group antibodies are formed early in life, probably by exposure to cross-reactive determinants on common bacterial organisms. Since individuals do not form antibodies to structures that they express, individuals of blood group AB do not have antibodies to other blood group antigens while those of blood group O have both anti-A and anti-B antibodies. A and B individuals have antibodies to the antigen of each other.

The issue of blood group antigens is complicated further by the existence of hundreds of other antigens on the surfaces of blood cells, which may prevent successful transfusion of blood. For transplantation of solid organs, however, only the ABO antigens need be considered since they are expressed on essentially all organs, especially on vascular endothelium. Therefore, as far as blood group antigens are concerned, O blood group organ donors are universal donors, AB donors can be used only for AB recipients, and A or B donors are suitable for recipients of their own blood group or AB individuals.

In the practice of organ transplantation three modifications of these general principles are noteworthy. First, not all organs are equally susceptible to antibody-mediated rejection by blood group antibodies, and transplantation of the bone marrow and liver in particular are sometimes performed across blood group barriers. Secondly, there are two subgroups of blood group A, called A&sub1; and A&sub2;, and individuals of blood group O or B may not form antibodies which react with the A&sub2; determinant. Therefore it is sometimes possible to cross this blood group barrier in this special case (A&sub2; to O). Finally, removal of blood group antibodies from potential recipients by plasmapheresis before transplantation has occasionally allowed successful transplantation across blood group barriers. However, this approach is not widely applied, except for bone-marrow transplantation.

Although essentially all cells of an individual share the same complement of genes, not all cells in the body express all of the transplantation antigens at all times. It is likely, therefore, that the factors affecting antigen expression will have an important effect on graft rejection. When considering these issues, ‘constitutive’ expression refers to the baseline level of antigen expression by a given cell type, while ‘induced’ expression refers to expression that occurs in response to various stimuli.

Class I MHC antigens are expressed constitutively on essentially every type of cell in the body. Only a few cell types, including spermatazoa and red cells, have unusually low expression of these antigens, which may be important in protecting them from immunological responses.

Class II MHC antigens are not constitutively expressed on all cells and are normally found on those types of cells that specialize in the uptake, processing, and presentation of soluble foreign antigens. These include B lymphocytes, macrophages, and dendritic cells (including Langerhans cells in the skin and Kupffer cells in the liver). Endothelial cells of vessels also express class II antigens, as do thymic epithelial cells; this is important for their role in positive and negative T-cell selection.

Because of the differences in tissue distribution, not all types of organs and tissues have equal expression of transplantation antigens and hence are not necessarily equal in their susceptibility to rejection. For example, experimental tail skin grafts from mice have fewer Langerhans cells than trunk skin grafts and therefore have fewer cells expressing class II antigens. This difference leads to significant differences in the strength of rejection for the two types of grafts. Clinically, it is not clear that sufficient differences exist between the commonly used organs in the destiny of cells expressing class II antigens to affect the outcome of organ transplantation, but it would not be surprising to find that such differences do exist, say between the heart and the liver.

The constitutive expression of class I antigens is sufficiently high that variations in the level of expression do not have an appreciable effect on the immune response to them. Variations in class II antigen expression, on the other hand, probably do have important consequences in transplantation immunology. Not only can the level of class II expression on antigen-presenting cells be increased, but in addition, cell types that do not normally express class II antigens can be induced to do so in response to several stimuli. &ggr;-Interferon and interleukin-4 are especially important signals that include class II antigen expression.

In a normal immune response, the increased expression of class II antigens may amplify the immune response to antigens presented in association with class II molecules. In transplantation immunology, this effect has important implications regarding graft rejection. Since class II targets will only be expressed constitutively on certain cell types, many cells would avoid rejection aimed at these antigens unless induction occurs. Experimentally, class II induction can be prevented by antibodies against interferon and these antibodies can therefore be used to prolong graft survival in some cases of class II-only disparity. This experimental finding may explain the clinically recognized phenomenon that viral infections and other immune stimuli, such as vaccination, sometime induced rejection episodes. Perhaps the production of interferon or other lymphokines during infection induces the expression of class II antigens on the allograft, provoking previously dormant transplantation immunity.

The cell types that constitutively express class II antigens play a critical role in stimulating T cells. These cells are called antigen-presenting cells and they share the capacity to process foreign antigens, to present them in association with MHC antigens, and to secrete certain lymphokines which are required for T-cell activation. Thus it is more than just the expression of class II antigens that make these cells important.

The distribution of antigen-presenting cells in the body reflects the site where contact with foreign antigens will most likely occur or where immune responses can most powerfully be achieved. For example, the skin, intestinal tract, and lungs have large numbers of antigen-presenting cells and there are large numbers in liver, lymph nodes, and spleen. The function of the antigen-presenting cells in skin, where they are known as Langerhans cells, has been investigated in detail. Foreign antigens entering through the skin encounter abundant Langerhans cells at this site. These antigen-presenting cells, now known as ‘veiled’ cells because of their appearance, then carry the antigen from the skin to regional lymph nodes. At this site the antigen-presenting cells are known as ‘interdigitating cells’. Inserted into lymph nodes, these antigen-presenting cells are in a good position to present peptides of the foreign antigens to the many recipient T cells that circulate through each lymph node. This process increases the likelihood that individual T cells will encounter each foreign antigen without requiring that every T cell circulate to the remote corners of the skin.

The consequences of this mechanism for antigen recognition and presentation for graft rejection are several. First, sensitization to allogeneic antigens may often occur not in the graft itself but rather in peripheral lymph nodes. This has been demonstrated experimentally using tissue grafts to vascularized sites that had been surgically deprived of lymphatic drainage. These grafts were not rejected. Secondly, the critical function of antigen-presenting cells suggests that without them graft rejection might not occur at all. Several experimental systems have shown this to be true, giving rise to the notion that rejection might be prevented by eliminating ‘passenger leucocytes’. For example, under certain conditions tissue culture results in the selective loss of antigen-presenting cells from various endocrine tissues, including pancreatic islets, favouring the successful transplantation of these tissues without immunosuppression or with immunosuppression at reduced dosages. Unfortunately, the antigen-presenting cells of most organs include more than just leucocytes and it is difficult in clinical practice to eliminate all of the antigen-presenting cells from any organ.

Although prior removal of all antigen-presenting cells from donor organs is difficult to achieve, there is turnover of these cells in a donor organ after transplantation. Therefore, donor dendritic cells and macrophages are slowly replaced by cells arising from the recipient's bone marrow. In clinical transplantation, this exchange may contribute to a decline in the immunogenicity of a graft and of the frequency of rejection episodes over time.

So far this section has dealt primarily with the antigens that are the targets of the immune response. We will now consider features of the response itself. The presentation of this topic is confusing since there are two settings in which the mechanisms of graft rejection can be observed: the clinical and the experimental. While ideally the two sets of observations would be closely related by a clear understanding of the mechanisms involved, in fact that understanding is far from complete. Thus, not all features of graft rejection seen in one setting correlate clearly with observations in another.

Three main patterns of rejection are usually identified in clinical practice: (1) hyperacute rejection, (2) acute rejection, and (3) chronic rejection.

Hyperacute rejection is a form of rejection that occurs instantaneously after restoration of blood supply to the transplanted organ. Typically, the newly transplanted organ shows normal initial circulation followed by discoloration and then abrupt cessation of function as blood flow ceases throughout the organ. There is no known intervention that can do more than delay this process slightly.

In the mid-1960s investigators recognized that hyperacute rejection occurred when a recipient possessed antibodies in the serum before transplantation which were specific for donor antigens. With this information, it became possible to avoid hyperacute rejection by testing potential recipients for the existence of preformed antibodies, using a test called the cross-match, and thus to allow the selection of donors whose cells did not manifest any antigen against which the recipient had already formed antibodies.

As described above, preformed antibodies can arise from a number of causes. While blood group antibodies do not require prior exposure to blood cells of an incompatible donor, all other preformed antibodies to antigens of transplanted organs arise from prior exposure to foreign transplantation antigens. This can occur by blood transfusion, by pregnancy (which exposes the mother to the father's antigens expressed by the fetus), and by transplantation of organs. However, not every individual exposed to foreign antigens in these ways will necessarily form antibodies.

The correlation between the clinical pattern of hyperacute rejection and an immunological mechanism involving antibody-mediated rejection appears to be excellent. However, there is experimental evidence that an early rejection process resembling typical hyperacute rejection can occur in the absence of demonstrable antibody, presumably by cell-mediated mechanisms.

Since hyperacute rejection can usually be avoided by use of the cross-match, the most commonly encountered clinical pattern of rejection is acute rejection. In this case, organs that have functioned well for a week or so then demonstrate diminished function over a period of several days. Biopsies at this time typically show a lymphocyte infiltrate. Without treatment, graft destruction will almost always occur but, unlike hyperacute rejection, acute rejection can often be treated and reversed by forms of immunosuppression that are effective against cellular immunity.

Variations in the histological picture of acute rejection can be detected. In some cases there is a dense eosinophilic infiltrate, which correlates with especially severe rejection. In other cases there is striking vascular destruction, leading to the description ‘vascular rejection’, often thought to imply that there is an antibody-mediated component to the rejection mechanism. However, the evidence for the association of specific histological findings with particular immunological mechanisms remains imperfect.

Acute rejection episodes tend to occur as discrete events, once or sometimes several times after transplantation. Most of these episodes take place during the first few months after operation, but episodes of sudden organ dysfunction which can be reversed by increased immunosuppression may occur months or even years after transplantation. Because of their sudden onset, rapid pace, and reversibility, these late episodes are also described as acute rejection episodes. They may be triggered by reduction or cessation of immunosuppression, but often have no clear aetiology.

Chronic rejection refers to a clinical picture of slow deterioration in organ function over months or years. The deterioration is difficult to control by standard immunosuppression. Pathological examination of organs suffering chronic rejection often reveals a relatively sparse lymphocyte infiltrate and may show a characteristic ‘onion-skin’ appearance with concentric cellular thickening and obstruction of flow in the small arteries of the graft, a finding that correlates well with the presence of antibodies in the recipient which are specific for donor antigens. These features suggest that chronic rejection may be caused by antibody-mediated mechanisms resulting from antibodies induced in the recipient after transplantation. It is unlikely, however, that all cases of chronic rejection are caused by humoral mechanisms alone.

Characterization of the clinical picture of graft rejection using the terms described above is extremely useful in the management of patients. Particular patterns respond to particular forms of treatment and the ability to judge the prognosis for survival of a transplanted organ is important in determining the vigour of the treatment effort. However, these characterizations can become misleading when they are used to imply a particular mechanism of rejection. This suggests a better understanding of the processes involved than actually exists and implies a scientific basis for the treatment employed, which has not yet been achieved.

Numerous experiments demonstrate that antibodies can cause rejection which appears to be similar to the syndrome of hyperacute rejection recognized clinically. The antibodies must be present prior to the transplant and must be specific for antigens expressed on vascular endothelium. However, only certain vascularized organs are susceptible to immediate antibody-mediated rejection.

The pathophysiology of hyperacute rejection has been characterized in some detail. The initial step in the process involves binding of the preformed antibodies to the vascular endothelial cells of the new organ. This binding then initiates a series of interacting cascades. Complement fixation is involved, and activation of the classical complement system then activates the recipient endothelium, stimulates the clotting system, and triggers the kallikrein cascade. Within minutes there are substantial changes in the permeability and integrity of the endothelial lining, intense vascular constriction, and vascular thrombosis. These account for the clinical features and pathological findings of hyperacute rejection.

Some understanding of the mechanisms of hyperacute rejection has come from the many efforts to control it. Complement inhibitors, clotting inhibitors, and many other substances have been shown to delay the process. None of these interventions, however, is able to block the progression of hyperacute rejection in a clinically meaningful way. Once started, hyperacute rejection inevitably destroys the transplanted organ.

As a result of these findings, the principal effort in dealing with preformed antibody has been either to eliminate the preformed antibody prior to transplantation or to select donors or tissues which do not express the relevant target antigens. Elimination of preformed antibody by plasmapheresis or absorption, coupled with B-cell immunosuppression, has occasionally been successful in achieving successful transplantation across blood group barriers. In general, however, elimination of preformed antibody is an uncertain approach to transplantation when there is a positive cross-match. Some organs and tissues appear to resist rejection by antibody. Skin grafts do not suffer hyperacute rejection and can survive for long periods despite the presence of preformed antibody. The liver is quite resistant to antibody-mediated rejection, although liver transplants performed across a positive cross-match show inferior survival over the long term. Pancreatic islets, especially after periods of in-vitro culture, may also be resistant to antibody-mediated rejection.

Even when antibodies specific for donor antigens are not present before a transplant, they may be induced by exposure to foreign antigens after the transplant has been performed. This antibody formation by B cells requires the help of T cells. The antibody tends to be specific for MHC antigens and to be IgG in class.

It is difficult to determine the role of induced antibody in causing rejection. This is partly because induced antibody is not formed unless T cells are present, making it hard to separate the contribution of direct cell-mediated graft destruction. In addition, there are examples where organs survive well despite the presence of measurable quantities of induced antibody. There are also some cases where antibodies specific for donor antigens actually seem to prevent rejection by T cells (a phenomenon called enhancement). Nonetheless, there is evidence that organ rejection can be caused by induced antibody responses. Clinically, we believe that there is a strong correlation between pathological recognition of an ‘onion-skin’ appearance of small donor arteries, the presence of induced antibody in the recipient, and slow, progressive graft destruction. Experimentally, there are examples of early xenograft destruction, occurring too late for hyperacute rejection, which take place despite high levels of T-cell immunosuppression and which correlate well with the appearance of induced antibody.

In the absence of preformed antibody, the rejection of transplanted organs involves a cell-mediated mechanism that has three principal features. First, this rejection requires the function of T cells. This has been demonstrated by the indefinite survival of skin grafts on T-cell-deficient ‘nude’ mice. Secondly, the process shows immunological specificity and memory. This process has been demonstrated by showing that a second graft from the same donor is rejected more rapidly by a recipient than the first graft, while a second graft from an unrelated donor is not. Thirdly, the process is precisely targeted against the cells of the donor graft. This has been shown by skin graft experiments using ‘tetraparental mice’, formed by fusion of two embryos in the early stages of development. Mature tetraparental animals are composed of a mosaic of two different cell types interspersed throughout the body. For example, tetraparental mice originating from black and white parents show a speckled ‘salt and pepper’ coat colour. When skin from a mouse is transplanted to a recipient which is syngeneic with one parent but allogeneic to the other, only the allogeneic cells are rejected. The syngeneic cells survive the destruction of their allogeneic immediate neighbours.

Considerable effort has been devoted to determining the mechanisms by which T cells can cause graft destruction in a manner showing specificity, memory, and precise targeting. Several different experimental techniques have been used. First, there has been histological examination to determine the nature and distribution of T cells within a rejecting graft. Secondly, T cells within grafts have been removed and analysed for their type, function, and specificity. One way to achieve this has been to use ‘sponge-matrix’ allografts, which are inert sponges containing allogeneic target cells. These sponges can be removed from a recipient at any time after transplantation and the infiltrating cells squeezed out for analysis. Thirdly, T cells of different phenotypes have been transferred into T-cell-deficient animals to determine which types of cells are able to cause graft destruction. Fourthly, in-vitro assays of T-cell function, including proliferation interleukin-2 production, and cell-mediated cytotoxicity have been tested, seeking correlations between the development of these T-cell functions and the onset of graft destruction. All of these techniques for studying graft rejection have been refined by experiments using donors that differ from their recipients by only limited antigenic disparities and by using antibodies to T-cell antigens, such as the CD4 and CD8, which define subpopulations of the overall T-cell population.

Several conclusions have been generated from these studies. First, only a small portion of the T cells that invade an allograft are specific for the antigens of the donor. Secondly, only a small number of invading T cells are actually required for graft rejection. Thirdly, the T cells that invade a graft are generally of many types, with many functions, making it difficult to determine from this analysis the actual mechanism of graft rejection. Finally, there is an excellent correlation between the ability of the recipient to generate cytotoxic T cells in vitro against cells of a particular donor and the ability of that recipient to reject grafts from the same donor. These findings support the ‘cytotoxicity hypothesis’ that graft destruction is mainly caused by the development of cytotoxic T cells in a manner analogous to the in-vitro development of cell-mediated cytotoxicity.

The development of T-cell cytotoxicity in vitro requires more than one T-cell function. According to the standard model (Fig. 7) 678, the process begins with a random clustering of T cells around antigen-presenting cells based on the affinity of non-specific T-cell surface molecules for ligands present on these cells. One interaction is between a T-cell structure known as LFA-1 (for ‘leucocyte function antigen’) with a ligand called ICAM-1 (for ‘intercellular adhesion molecule’). While this interaction promotes binding, it is not in itself sufficient to cause T-cell activation. Activation requires the specific binding of the T-cell receptor with determinants on the MHC antigens of the antigen-presenting cells. These determinants may be formed by the association of peptides derived from foreign proteins with self-MHC molecules or, in the case of graft rejection, by the presence of donor MHC antigens which are recognized directly.

In addition to requiring engagement of the T-cell receptor, T-cell activation is also augmented by other accessory molecules. Especially prominent are the interactions of the CD4 antigen, present on one group of T cells, with a determinant expressed on all class II MHC antigens, and those of the CD8 antigen, present on another group of T cells, with a determinant expressed on all class I MHC antigens. These accessory molecules are responsible for the finding that CD4⫀ T cells have specificity for class II MHC antigens, while CD8⫀ T cells are specific for class I MHC antigens.

Finally, the standard model of T-cell activation also suggests that antigen-presenting cells secrete lymphokines, probably including interleukin-1 (IL-1), which bind to receptors on T cells to augment T-cell activation. One kind of T cell that is activated by these mechanisms is called a helper T cell. Once activated, helper cells begin to produce a lymphokine called interleukin-2 (IL-2), formerly described as T-cell growth factor, which contributes to the growth and maturation of activated cells. These activated T cells also express new receptors for IL-2, creating a positive feedback system which augments the helper response.

A second kind of T cell is a precursor of the cytotoxic T cell. It can also be activated by contact with antigen-presenting cells expressing foreign antigens. Proliferation and maturation of the cytotoxic T cell precursor into a functional cytotoxic T cell requires the IL-2 produced by helper cells. Once activated, these cytotoxic T cells can kill any cell expressing the same target antigens. Thus the development of T-cell cytotoxicity requires helper T-cell activity, precursors of cytotoxic cells, specific recognition of a foreign antigen, and the function of several non-specific cell surface molecules and soluble lymphokines.

The model of T-cell activation shown in Fig. 7 678 shows a helper T cell interacting with the same antigen-presenting cell as the precursor of the cytotoxic T cell. This ‘three cell cluster’ does appear to be an important requirement for T-cell activation in vivo. Apparently the IL-2 produced by a helper T cell can only provide help for cytotoxic T cells at a short distance. Thus lymphokines of this sort act like neurotransmitters rather than like hormones. In vitro the requirement for a three-cell interaction is not essential and excess IL–2 can be added to a culture, bypassing the requirement for the helper T cells in the generation of cytotoxic T cells.

The classic picture of T-cell activation usually shows the helper function being performed by CD4⫀ T cells and the cytotoxic function being performed by CD8⫀ T cells. This view has emerged because when both class I and class II antigenic disparities are present, CD4⫀ T cells do tend to perform the helper function, while CD8⫀ T cells tend to develop into cytotoxic T cells. Based on these findings, the designations ‘CD4⫀’ and ‘helper’ T cells are often used interchangeably as are ‘CD8⫀’ and ‘cytotoxic’ T cells. However, it is an important feature of alloreactivity that this association of phenotype and function is not absolute. CD4⫀ T cells can show cytotoxic function and CD8⫀ T cells can produce IL-2. Sometimes, these CD8⫀ helper cells are also able to develop cytotoxic activity, leading to the designation ‘helper-independent’ cytotoxic T cells.

Because of this overlap of T-cell functions, careful analysis has been required to identify the multiple avenues by which alloreactive cytotoxicity can develop. Currently at least three pathways are recognized by which the necessary ‘help’ can be generated:

(1)CD4⫀ T cells can recognize allogeneic class II MHC antigens directly;

(2)CD4⫀ T cells can recognize peptides of alloantigens in association with self-MHC antigens;

(3)CD8⫀ T cells can recognize allogeneic class I antigens directly.

In addition, three different populations of cytotoxic T cell precursors that can respond to alloantigens are now recognized:

(1)CD8⫀ cytotoxic cells can recognize class I MHC antigens;

(2)CD4⫀ cytotoxic cells can recognize class II antigens;

(3)CD8⫀ cytotoxic cells, defying the association of phenotype with antigen specificity, can recognize and kill targets expressing a class II MHC antigenic disparity.

The pathways for generating ‘help’ and cytotoxic T cells are shown in Table 4 220. In most clinical transplant situations the number of antigenic disparities between recipient and donor is large enough that all of these potential pathways for the development of cytotoxicity are available.

Although there is strong experimental evidence that cytotoxic T cells represent an important mechanism of graft destruction, there are several experimental situations in which rejection correlates poorly with the development of cytotoxicity in vivo. First, the rejection of grafts expressing only minor histocompatibility antigen disparities has been reported to occur even when cytotoxic T cells specific for the donor have not been found in the recipient. Secondly, vigorous cell-mediated rejection of tissues transplanted between species is always seen, even though there is sometimes weak cell-mediated cytotoxicity measured in vitro. These exceptional cases raise the possibility that other mechanisms of rejection may be brought into play. Delayed-type hypersensitivity is a T-cell response that might be involved. In this case, lymphokines from T cells may activate macrophages to cause tissue destruction. Alternatively, the development of natural killer cells or lymphokine-activated killer cells might destroy foreign grafts. Both of these mechanisms would appear to lack the means to achieve the precise target specificity that is a feature of graft rejection, and thus their role in graft destruction remains uncertain.

Since cell-mediated rejection is T-cell dependent, many forms of immunosuppression have been developed or selected because they interfere with T-cell function. For example, cyclosporin is a fungal product with a hydrophobic cyclic structure. It prevents the generation of interleukin-2 by T cells, blocking the generation of T-cell ‘help’. OKT3 is a monoclonal antibody which is specific for the CD3 antigen, present on all T cells, and intimately associated with the T-cell receptor. This antigen is involved in the transduction of signals to the interior of a cell after engagement of the external receptor. Administration of OKT3 to patients eliminates all T cells from the circulation and then allows the return of T cells which lack the expression of the CD3 antigen. These cells are not functional without their signalling mechanism.

Because cyclosporin and OKT3 block a large portion of the T-cell response, their use leaves the recipient broadly immunosuppressed. Therefore efforts have been made to use antibodies specific for subpopulations of T cells or for other structures involved in the immune response. These include antibodies for CD4 or CD8 antigens, antibodies directed at the IL-2 receptor, and antibodies directed at the intercellular adhesion molecule on the antigen-presenting cell. Some of these will probably find a role in clinical practice, although that role has not yet been defined.

In addition to new monoclonal antibodies, new chemical immunosuppressive agents such as FK-506, 15-deoxyspergualin, and rapamycin are also being tested. Like cyclosporin, these are also non-specific reagents; however, their site of action may differ from that of cyclosporin.

The scheme of T-cell activation leading to graft rejection described above starts with naive T cells that have never before encountered the donor antigens. While this process represents an important situation in graft rejection, it may not always reflect the situation in clinical transplantation. Just as B cells may have been activated before transplantation, so T cells may also have been sensitized by prior encounters with foreign MHC antigens. In addition, a majority of transplant recipients will experience an episode or more of graft rejection after their transplant, which will generate sensitized T cells. Thus the actual requirement in clinical transplantation may often involve the suppression of T-cell responses which have already been stimulated. The role of memory T cells, the factors that turn them from dormant to active cytotoxic T cells, and the difference between the requirements for immunosuppression before and after T-cell sensitization are all issues that require further study.

The response of T cells to the alloantigens of donor grafts is expected since T cells would be expected to respond to any foreign antigen. What is not expected is that the allogeneic response should prove to be so powerful. The precursor frequency of T cells that respond to a single determinant of an allogeneic MHC antigen is in the range of 2 per 100 T cells. In contrast, the precursor frequency of T cells responding to determinants formed by a foreign virus or other extrinsic antigen presented in association with a self-MHC antigen is about 1 in 10 000. Thus, the allogeneic response is enormously powerful relative to the response to a pathogenic virus and, in fact, the allogeneic T-cell response is the most powerful immune response measured. Why should T-cell alloreactivity be so strong when the immune system presumably evolved to protect us from viruses and other pathogens and not to protect us from receiving kidney transplants from cadaveric donors? A clear answer to this question is not yet available. However, much has been learned recently about the selection of T cells to form the mature T-cell repertoire, removing some of the mystery of alloreactivity.

T-cell development depends on the presence of the thymus, hence the designation ‘T’ cell. In addition, T cells that can react with self-antigens are eliminated in the thymus prior to maturation. It is now also recognized that the thymus plays a critical role in positively selecting the T cells that are allowed to mature, by selecting those cells with high affinity for modified forms of self-MHC antigens in addition to removing the cells that can react with unmodified self-MHC antigens. This dual process is presumably useful because mature T cells will only encounter foreign antigens when peptides of those antigens are presented in association with self-MHC molecules. Therefore, the mature T-cell repertoire is chosen for its affinity for modified self-MHC antigens.

In addition to this understanding of the selection of the T-cell repertoire, it is now also recognized that allogeneic MHC antigens present determinants that look to T cells like modified self-MHC antigens. In immunology, this feature is referred to by the shorthand notation that Equation 26

where ‘X’ is the peptide of a foreign protein which is presented in association with a self-MHC antigen. Thus the mature T-cell repertoire, selected for the ability to recognize Self + X, also includes T cells capable of recognizing allogeneic MHC antigens directly.

The similarity of modified self-MHC antigens to alloantigens is sufficient to explain the ability of T cells to recognize allo-MHC antigens without a requirement that these antigens be processed and presented as peptides in association with self-MHC antigens. However, the selection process does not, in itself, explain why the alloreactive repertoire should be more powerful than that for modified self-MHC antigens. Why should the number of T cells reacting with a particular allogeneic determinant be so much larger than the number of T cells reacting with a particular Self + X determinant?

Several theories have been proposed to explain the strength of alloreactivity. First, the genes that encode T-cell receptors might be maintained especially because they encode specificity for MHC antigens. This germline encoding of MHC specificity would increase the efficiency of the T-cell selection process, since the thymus will only select those T cells that are able to recognize modified forms of self-MHC antigens. Those T cells without any affinity for an MHC antigen would always be wasted. According to this hypothesis, the high precursor frequency of alloreactive T cells occurs because the receptor genes actually encode an alloreactive repertoire. Some T-cell receptor genes have been identified which do show affinity for particular MHC antigens of their species, providing some support for this hypothesis.

A second theory to explain the strength of alloreactivity suggests that the high precursor frequency of alloreactive T cells does not result from more T cells recognizing allogeneic MHC determinants, but rather from the greater number of allo-determinants expressed on the surface of an allogeneic antigen-presenting cell. For example, an immune response to the measles virus requires that peptides of that virus be presented in association with self-MHC antigens on a self-antigen-presenting cell. But probably only a few of the MHC antigens on a given antigen-presenting cell are engaged in the process of presenting measles peptides, while the other MHC antigens are occupied with presenting other peptides of autologous proteins, forming determinants that do not stimulate autologous T cells. In contrast, every one of the allogeneic MHC antigens on an allogeneic antigen-presenting cell will be able to stimulate a T-cell response since these allo-MHC antigens can each be recognized directly. According to this hypothesis, the greater strength of alloreactivity is, in a sense, an artefact of our assay system: there are more of a particular allogeneic determinant on allogeneic antigen-presenting cells than of modified self-determinants on self-antigen-presenting cells. Therefore alloreactive T cells will be stimulated more easily than those that recognize modified self and their number will appear to be larger by precursor frequency analysis.

A third hypothesis is based on the notion that all MHC antigens are engaged in presenting peptides all of the time, but that most often these peptides are from autologous proteins, forming determinants that are considered self-determinants by T cells. For example, some MHC antigens may present peptides from serum albumin while others might present peptides from haemoglobin. According to this hypothesis, the MHC antigens of an allogeneic antigen-presenting cell also present these same peptides in association with the allogeneic MHC molecules. However, the association of an albumin peptide with an allo-MHC antigen or a haemoglobin peptide with that MHC antigen would not form a self-determinant. Therefore, a single allogeneic MHC antigen on an allogeneic antigen-presenting cell could present multiple new foreign determinants to responding T cells. Thus, a higher precursor frequency of alloreactive T cells would result from the large number of new determinants created out of a single MHC difference.

These hypotheses need not be mutually exclusive, and it is quite likely that each is correct to some degree. However, the relative role of these potential mechanisms, or of others, has not yet been determined. Nonetheless, the ability to generate such plausible explanations for the enormous strength of alloreactivity has removed some of the mystery from this phenomenon.

One of the cardinal features of the immune response is that it does not occur to self-antigens under ordinary conditions. This is called ‘tolerance’ and it is critical to an individual's survival. Tolerance is also a source of fascination to those interested in transplantation since its existence suggests that it ought to be possible to recapitulate nature, instructing the immune system to accept additional antigens (those of the donor graft) as ‘self’, thus avoiding the need for non-specific immunosuppression.

Immunologists have long assumed that tolerance is determined primarily in the thymus where T cells first develop. Recently, confirmation of the thymus's important role has been achieved by analysing the expression of T-cell receptors. First, there is now evidence that T-cell receptors are not expressed prior to the entry of precursor T cells into the thymus. Since tolerance requires specificity (and therefore the expression of the receptor), T cells cannot learn tolerance prior to arrival in the thymus. Secondly, analysis of receptor expression in the thymus has shown that T cells with specificity for self-MHC antigens do not mature in that organ. Thus there must be a mechanism in thymic T-cell maturation whereby cells reactive with self-MHC antigens are eliminated.

The thymus is able to prevent the maturation of T cells reactive with self-MHC antigens because that organ expresses all self-MHC antigens. On the other hand, many have wondered how tolerance to all self minor histocompatibility antigens can be achieved in the thymus when some of the peptides forming these antigens are probably expressed only in particular tissues. Thus it has been assumed that an extrathymic mechanism to achieve tolerance also exists. Recent experiments demonstrating such an extrathymic mechanism have made use of transgenic technology. Mice have been generated that express foreign MHC antigens, inserted into their genome along with tissue-specific promoter genes. Even when such mice do not express the foreign MHC antigen in the thymus, and do not delete the alloreactive population of T cells during T-cell maturation, they nonetheless fail to reject cells that do express the foreign MHC antigen. They also fail to generate a T-cell response in vitro to that alloantigen. Thus this alloreactive population of T cells has been inactivated by some peripheral mechanism after thymic maturation.

Another set of experiments has revealed that mature T cells can encounter foreign antigens in ways that trigger the cell, but which do not lead to immune activation. Helper T-cell clones stimulated by immobilized antigen, in the absence of the additional signals usually provided by antigen-presenting cells, can undergo transformation into blast cells and secrete some, but not all, of their lymphokines. Subsequent stimulation of these clones by normal antigen-presenting cells finds them inert to their stimulating signals. Thus it appears that incomplete stimulation of T cells can achieve their down-regulation under some circumstances.

Given the importance of self-tolerance to the survival of an animal, it is not surprising that the immune system should have more than one mechanism to avoid self-responses. In addition, most biological systems with activation responses have a regulatory system to prevent even appropriate responses from getting out of control. Thus efforts have been made to recreate nature's mechanisms. Two experimental techniques have been most successful in achieving this goal. First, in a notable series of experiments in the early 1950s, Billingham, Brent, and Medawar, consistent with the hypothesis advanced by Burnet, were able to show that foreign antigens, introduced sufficiently early in the maturation of an animal, are treated as self-antigens, allowing the induction of ‘actively acquired immunological tolerance’. In mice the critical period extends for several days after birth, so that injection of allogeneic cells shortly postpartum induces ‘neonatal’ tolerance. Similar observations were made by the Czech biologist, Milan Hasek. Secondly, whole-body irradiation of adult rodents has been used to ablate the mature immune system, allowing reconstitution of the system by bone-marrow transplantation of cells expressing foreign antigens. This technique creates allogeneic bone-marrow chimeras, which are tolerant to both ‘self’ and to the foreign antigens. Although these two methods for achieving tolerance are useful experimentally, they require either too early or too drastic an intervention to be useful clinically. Therefore, efforts to achieve tolerance in ways that might be clinically applicable have been undertaken.

One approach aimed at specific downregulation of T-cell responses involves prior blood transfusion of allogeneic cells before organ transplantation. The use of this approach in clinical practice came about accidentally. Although some patients receiving blood transfusions before transplantation become sensitized to donor antigens, as might be expected, those recipients who had not become sensitized turned out to support subsequent graft survival better than expected. Experimental investigation suggests that this is partly due to a diminished T-cell response to the foreign antigens of the transplanted organ. The downregulation of T-cell responses as opposed to sensitization seems to depend on the intravenous route of antigen delivery and on the types of cells being transfused. Thus the finding of better graft survival after blood transfusion not only provided a clinical means of manipulating the immune system, but also stimulated investigation of means by which foreign antigens might be introduced in a non-stimulating way.

Part of the key to downregulating immune responses appears to be in avoiding the introduction of functional antigen-presenting cells. As discussed above, depletion of endocrine tissue antigen-presenting cells by tissue culture or by appropriate antibody treatment can achieve long-term graft survival. This technique may also achieve specific unresponsiveness to the antigens of the graft, such that subsequent grafts from the same donor, transplanted with the antigen-presenting cells still present, are not rejected. However, sufficiently strong stimulation of the immune system will often overcome this downregulation, causing rejection of the new and old grafts together. The survival of grafts depleted of antigen-presenting cells suggests again that antigens can be introduced into intact animals in a manner that downregulates rather than stimulates a T-cell response.

Clinically, depletion of antigen-presenting cells from donor organs has been difficult to accomplish. For example, it is not sufficient to remove only the leucocytes and lymph nodes from donor kidneys. Therefore, additional efforts have been made to inactivate donor antigen-presenting cells, rather than to remove them. One approach has been to use ultraviolet light to treat donor tissues. This has allowed successful tissue transplantation in some experimental situations where these cells are accessible to external illumination.

In addition to presenting antigens in a non-stimulating manner, efforts have also been made to achieve cellular chimerism, similar to that accomplished in bone-marrow chimeras, but using techniques that are less toxic than whole-body irradiation. Experimentally, lasting chimerism has been achieved by vigorous depletion of the mature immune system using several different monoclonal antibodies against T cells, low-dose irradiation of the recipient's bone marrow, and high-dose irradiation of the recipient's thymus. This is then followed by infusion of both donor and recipient bone marrow to reconstitute the immune system in the presence of foreign antigens. The technique has worked well in rodents and a few larger animals but has not yet been applied clinically. Some patients have been treated with antilymphocyte serum along with an infusion of donor bone marrow. However, lasting chimerism has not been achieved by this approach as yet.

The effort to achieve specific downregulation of T-cell responses and eventually of lasting tolerance represents a major area in which the immunological principles revealed by experimental methods and the achievement of their clinical application awaits completion. It seems that the tools are available to achieve specific immunosuppression and that appropriate modifications in techniques may well bring them to clinical practice. However, the potential for the introduction of such methods has been appreciated for several years. One reason for hesitating to embark on clinical trials of certain promising approaches, until they have been thoroughly explored in large animals trials, is the high level of success that is now being achieved with non-specific immunosuppressive agents. Nevertheless, the imperfections of our present approaches are clearly apparent and further progress is highly desirable.

This section has described some of the concepts in transplantation immunology, the science that underlies clinical organ transplantation. It is, of course, possible to perform clinical transplants quite successfully with little knowledge of these concepts. There are, however, some areas where a grasp of the relevant fundamentals of immunology can greatly facilitate one's understanding of clinical issues.

One of the hotly debated questions in clinical transplantation is how important it is to achieve matching of transplantation antigens between donors and recipients. Although the question is controversial, the data related to this issue are reasonably well established. In the first place, organs and tissues can be exchanged between identical twins, who express identical transplantation antigens, without any rejection and without immunosuppression. Secondly, among living-related kidney donors, roughly 95 per cent of kidneys derived from HLA-identical siblings survive 1 year, compared to about 90 per cent of those from one-haplotype-matched family members. In comparison, about 85 per cent of kidneys from unrelated cadaveric donors survive for 1 year at the best institutions. Thirdly, the survival of kidneys from cadaveric donors varies according to the degree of MHC antigen matching. Studies comparing the outcome depending on the number of matched HLA antigens have suggested that there is some benefit from matching for DR antigens, but that the clearest benefit (in the range of 10 per cent improved 1-year graft survival) occurs only when there is matching for the products of the HLA-A, B, and DR loci (six antigens in all). This benefit probably increases somewhat as more time elapses following transplantation.

These data support the notion that antigen matching ‘matters’. The controversy, however, arises between those who view the benefits of partial antigen matching as large, versus those who see them as relatively small in comparison to the disadvantages that occur when seeking better compatibility. Is it reasonable to subject living family members to surgery and the loss of a kidney to obtain better antigen matching? Is it better to gain a few percentage points in survival for a new patient whose name has recently been added to a long list of waiting recipients on the basis of a more favourable match, or to give that kidney to someone who has been waiting a year longer on dialysis? How much improvement in graft survival on the basis of compatibility will justify the cost of transporting kidneys over long distances and the negative effects of longer ischaemic times? These are difficult questions, which still lack precise answers and therefore generate substantial disagreement.

Antigen-matching is controversial and some are uncertain that any effort is necessary to achieve it. The cross-match, on the other hand, is a different issue entirely and is universally recognized to be important in kidney transplantation. A cross-match is performed by mixing serum from a prospective recipient with lymphocytes of a potential donor to determine whether the recipient has antibodies that are specific for the donor's antigens. If such antibodies are present, they bind to the donor's lymphocytes, causing lysis, when an exogenous source of serum complement, usually from rabbits, is added. A positive cross-match predicts with high accuracy that hyperacute rejection will occur if kidney transplantation is attempted between that donor and recipient, but there are exceptions in that not all antibodies in potential recipients are directed against HLA antigens. Not infrequently, autoantibodies are present as well as HLA antibodies and these may also be responsible for a positive cross-match, but in this instance the positive cross-match does not preclude transplantation.

Although simple in concept, there are practical difficulties with the cross-match which arise because the assay measures the presence of complement-fixing antibodies and not the initiation of hyperacute rejection. One problem is that not all antibodies fix complement equally well, but non-complement-fixing antibodies can cause hyperacute rejection. One approach to this problem is to amplify the assay by adding antibodies from another species that recognize and bind to the human antibodies of the recipient. If these recipient antibodies are bound to donor lymphocytes, the added antibodies will bind to them whether or not they are complement fixing, and these added antibodies, in turn, will fix complement to cause cell destruction. Another way to detect antibodies that do not fix complement is to use flow cytometry, which can detect antibody binding to cells even when they do not cause cell destruction. Both techniques increase the sensitivity of the cross-match assay, but both may be too sensitive, in some cases detecting antibodies that would not lead to hyperacute rejection. However, graft survival in highly sensitized recipients and retransplant patients are superior in those centres that use such sensitive procedures. Another problem with the cross-match is that the target cells (traditionally T lymphocytes from the donor) may not express all the target antigens of vascular endothelial cells that can elicit hyperacute rejection. Class II MHC antigens, in particular, are not usually expressed on human T cells, whereas antibodies against class II antigens can sometimes, but not always, cause hyperacute rejection. These antibodies can be detected using B lymphocytes as targets, a so-called ‘B-cell cross-match’. However, not all patients with a positive B-cell cross-match suffer hyperacute rejection. Obviously it is difficult to know exactly how stringent to be in the interpretation of cross-match results, and different centres handle this problem differently.

Potential recipients of kidney transplants who do not have any antibodies in their sera that react with foreign lymphocytes are said to be ‘unsensitized’. Those who have been exposed to foreign antigens may have antibodies against some or many different foreign antigens. These individuals are said to be ‘sensitized’. The level of sensitization of prospective recipients is tested by reacting their sera separately with lymphocytes from many different individuals, who are selected because they express a broad range of the human HLA antigens. If the donor's serum causes lysis of 45 per cent of the panel of different target lymphocytes they are said to have a 45 per cent PRA (panel reactive antibody). Some prospective recipients have antibodies that kill lymphocytes from every individual in the panel. They are said to be ‘highly sensitized’ or ‘100 per cent sensitized’. Obviously, for high sensitized patients it is extremely difficult to find donor kidneys which will not generate a positive cross-match, although it is not really ‘100 per cent’ impossible since a donor matched for all HLA antigens with the recipient should have a negative cross-match and, as mentioned above, some of these highly sensitized patients will have an autoantibody component.

The process of testing each individual against the panel is useful in predicting the difficulty of finding them a suitable kidney donor. It is also useful in identifying the specificity of the antibodies of the sensitized patient and therefore in determining which HLA antigens are most likely to generate a positive cross-match. This makes it possible to select kidneys that are most likely to be suitable for a highly sensitized recipient even before a cross-match, thereby increasing the efficiency of the screening and distribution process.

Many highly sensitized patients wait a long time for a negative cross-match, and some never receive transplants at all. As a result there has been great interest in the possibility of removing preformed antibodies from sensitized patients by plasmapheresis. Occasionally, successful transplants have been achieved in this manner, but the approach has generally been successful only in the case of ABO mismatched individuals. On the other hand, some highly sensitized patients show a gradual decline in both the percentage of ‘panel reactivity’ of their sera as well as the titre of antibody activity against individual cell samples from the panel, without any intervention. Some of these individuals have also developed antibodies against their own antibodies, which are referred to as anti-idiotypes. Why they develop in some but not other cases is not clear, but the possibility of using anti-idiotypes to remove the antibodies of highly sensitized patients remains intriguing.

Blood transfusions are one of the ways in which individuals may encounter foreign MHC antigens and thus become sensitized. Therefore it seemed reasonable in the early years of clinical transplantation to avoid transfusing prospective candidates for kidney transplantation whenever possible. Some patients were asked to struggle with very low haematocrits while waiting for an organ. Despite the sound rationale for this approach, data gathered during the mid 1970s indicated that patients who had received blood transfusions before transplantation and who were then given a cross-match-negative kidney transplant had better survival of their transplants than individuals who had never been transfused. Indeed, this issue was studied nearly 70 times between 1973 and 1984, almost always with the conclusion that blood transfusions were beneficial. Thus it became standard practice in many transplant centres to insist on a potential recipient receiving at least some blood transfusions prior to transplant, whether the potential recipient was anaemic or not.

Several explanations have been considered for the beneficial effect of blood transfusions. First, it is possible that viruses carried with blood transfusions cause chronic infections which are themselves immunosuppressive. Secondly, it may be that the introduction of multiple foreign MHC antigens before transplantation forces the eventual selection of a kidney donor that expresses the MHC antigens eliciting the weakest response from that particular recipient. This might occur because the antigens to which the recipient could react most strongly would have caused sensitization and therefore a subsequent positive cross-match when those antigens were expressed by a donor. In contrast, antigens to which the recipient was weakly reactive would not cause sensitization and hence would generate a subsequent negative cross-match. Thirdly, it is possible that the introduction of foreign MHC antigens intravenously may sometimes cause downregulation of the immune response to those antigens rather than stimulation, achieving a degree of specific tolerance. Each of these explanations is probably valid to some degree, but a great deal of scientific evidence suggests that downregulation by presentation of intravenous antigens is a factor in the better survival of transplants after transfusion. Although this finding encourages the hope of tolerance induction, it remains unclear exactly what mechanisms lead to downregulation in some cases and sensitization in others.

Although intentional transfusion of transplant candidates was the norm for many years, studies of the benefits derived from this approach, beginning in 1986, suggested that the advantage was becoming very small, and in some cases impossible to detect. This apparent change in an outcome that had been so repeatedly verified before was a startling event. The explanation for the shift probably lies in the improved results of transplantation for all recipients, whether transfused or not, derived from new immunosuppressive drugs and other technical advances. Thus, it was probably not that transfusions suddenly offered no benefit, but rather that the benefit was small compared to the strength of cyclosporin and OKT3 in preventing or reversing allograft rejection.

Even while the new data were being evaluated, a still more powerful influence caused a decline in the use of intentional blood transfusions before transplantation. The appearance of AIDS and the recognition that it could be transmitted by transfusion caused many patients to balk at transfusions offered in the absence of anaemia. Even when it became possible to assure patients that the risk of contracting AIDS was negligible, the new awareness by both patients and physicians of the risks of transfusion diminished enthusiasm for intentional transfusion. In recent years, the availability of recombinant erythropoietin for patients on dialysis has also decreased the need for blood transfusion to treat the anaemia associated with chronic renal failure. The number of patients coming to transplantation without everreceiving a blood transfusion has therefore increased during the past several years.

The shifting policies and standards regarding blood transfusions over the past 15 years is a fascinating story, revealing the way both scientific information and prevailing attitudes can alter clinical practice. The sequence reminds us that the most carefully considered rationale for our practices may not hold up with changing conditions and new insights. It is part of the special excitement of the field of transplantation that it is changing so rapidly, as both experimental and clinical studies provide new understanding.

Auchincloss H Jr, Sachs DH. Transplantation and graft injection. In: Paul WE, ed. Fundamental Immunology. New York: Raven Press, 1989: 889–922.

Barker CF, Billingham RE. The role of regional lymphatics in the skin homograft response. Transplantation 1967; 5: 962.

Bevan MJ. High determinant density may explain the phenomenon of alloreactivity. Immunol Today 1984; 5: 128–30.

Billingham RE, Brent L, Medawar PB. ‘Actively acquired tolerance’ of foreign cells. Nature 1953; 172: 603.

Bjorkman PJ, Saper MA, Samraovi B, Bennett WS, Strominger JL, Wiley DC. Structure of the human class I histocompatibility antigen, HLA-A2. Nature 1987; 329: 506–12.

Colvin RB. Immunopathology of renal allografts. In: Colvin RB, Bhan AK, McCluskey RT, eds. Diagnostic Immunopathology. New York: Raven Press, 1988: 151–97.

Cosimi AB, et al. Treatment of acute renal allograft rejection with OKT3 monoclonal antibody. Transplantation 1981; 32: 535–9.

Doherty PC, Zinkernagel RM. H-2 compatibility is required for T-cell mediated lysis of target cells infected with lymphocytic choriomeningitis virus. J Exp Med 1975; 141: 502.

Jeannet M, Pinn V, Flax M, Winn HJ, Russell PS. Humoral antibodies in renal allotransplantation in man. N Engl J Med 1970; 282: 111–17.

Jenkins MK, Schwartz RH. Antigen presentation by chemically modified splenocytes induces antigen-specific T cell unresponsiveness in vitro and in vivo. J Exp Med 1987; 165: 302.

Kahan BD. Cyclosporin. N Engl J Med 1989; 321: 1725–38.

Lafferty K, Prowse S, Simeonovic C, Warren HS. Immunobioloby of tissue transplantation. A return to the passenger leucocyte concept. In: Paul WE, Fathman CG, Metzgar H, eds. Annual Review of Immunology. Palo Alto, California: Annual Reviews, Inc., 1983: 143–73.

Mason DW, Morris PJ. Effector mechanisms in allograft rejection. Ann Rev Immunol 1986; 4: 119–45.

Merrill JP, Murray JE, Harrison JH, Guild WR. Successful homotransplantation of human kidney between identical twins. JAMA 1956; 160: 277–82.

Mitchison NA, O'Malley C. Three-cell-type clusters of T cells with antigen-presenting cells best explain the epitope linkage and noncognate requirements of the in vivo cytolytic response. Eur J Immunol 1987; 17: 1579–83.

Opelz G, Terasaki PI. Improvement of kidney-graft survival with increased numbers of blood transfusions. N Engl J Med 1978; 299: 799.

Opelz G, Terasaki PI. International study of histocompatibility in renal transplantation. Transplantation 1982; 33: 87.

Snell GD, Stimpfling JH. Genetics of tissue transplantation. In: Green E, ed. Biology of the Laboratory Mouse. 2nd edn. New York: McGraw-Hill, 1966: 457–91.

Williams GM, Hume D, Hudson R, Morris P, Kano K, Milgrom F. Hyperacute renal homograft rejection in man. N Engl J Med 1968; 279: 611–18.

Winn HJ. Antibody-mediated rejection. In: Williams GM, Burdick JF, Solez K, eds. Kidney Transplantation Rejection. New York: Marcel Dekker, 1986: 17–28.