Although these accomplishments are important, they fall short of what many persons see as the ultimate goal of transplantation: development of antigen-specific tolerance. This is generally defined as the absence of an immune response to a given antigen without the need for exogenous medication or treatment. Immunosuppression, in contrast, refers to a nonspecific state in which all immune responses are blunted, usually as a result of continuous administration of one or more medications. Tolerance is the more desirable state for two reasons. First, if immunologic nonresponsiveness is specific, the risk for opportunistic infections and cancer, two of the most common complications of transplantation, should be markedly reduced. In addition, many transplantation immunologists believe that if tolerance could be achieved, long-term graft survival would dramatically improve.

This belief underscores one of the most important current issues in transplantation: the disparity between short-term success and the high rate of late allograft loss. The past 20 years has seen great improvement in short-term survival of organ allografts: For example, 1-year graft survival for recipients of first cadaveric kidneys has increased from 45% to almost 90% [1]. These advances resulted from a combination of factors, including more powerful and less toxic immunosuppressive medications, better tissue matching, improved surgical techniques, and new treatments for opportunistic infections. In contrast, in 1975 the expected half-life of a cadaveric allograft that remained functional for 12 months was an additional 8.5 years, a duration that has not changed appreciably since then [1]. For most organs (especially the kidney, heart, and lung), the leading cause of late graft loss is chronic rejection, a condition for which neither specific prophylaxis nor treatment is available. Depending on the organ involved, the histologic appearance of chronic rejection may involve interstitial fibrosis, arteriosclerosis, or bronchiolitis obliterans [2]. Unfortunately, chronic rejection seems to be inexorably progressive and usually leads to graft failure.

Commercially available immunosuppressants (or immunosuppressants in the final stages of phase III trials) are very powerful. Their actions are directed primarily at T cells, the lymphocyte population responsible for initiating graft rejection. T-cell-mediated immune responses depend greatly on the expansion and differentiation of T cells and macrophages, both of which are under the coordinated regulation of soluble growth factors. The power of current immunosuppressive medications derives from their ability to block this step. Cyclosporine and tacrolimus (FK506) block cytokine (particularly interleukin-2) production by activated T cells [3, 4]. Sirolimus (rapamycin) inhibits signal transduction through selected cytokine receptors, such as the interleukin-2 receptor [5]. Among new agents, mycophenolate mofetil is distinct in that it does not target cytokine pathways but blocks the pathway of purine metabolism, on which lymphocytes are particularly dependent [6].

Why, then, has the risk for late graft loss not improved? One possibility is that some of the newest agents have not been used long enough to determine whether they affect the rate of late graft loss. One study suggests that tacrolimus may be substantially beneficial in this regard [7]. In addition, important nonimmunologic risk factors for chronic rejection exist, including ischemia and reperfusion injury, cytomegalovirus infection, and mismatching of donor and recipient for age and size [8]. However, I believe that one of the most important problems is that immunosuppression is not functionally equivalent to tolerance. Although tolerance has been defined in the past as the lack of an immune response to a given antigen, it is increasingly apparent that active immunoregulatory mechanisms are important in the development and maintenance of self-tolerance. Thus, tolerance is not the absence of an immune response but might be redefined as failure to mount a destructive immune response. To the extent that even the newest drugs are globally immunosuppressive, they inhibit graft rejection and (perhaps) the development of antigen-specific immunologic tolerance. This concept gives rise to two implications. First, more powerful immunosuppressive drugs may further improve short-term graft survival but are unlikely to lead to tolerance. In addition, current strategies of immunosuppression may, paradoxically, inhibit attempts to induce tolerance, greatly complicating the design of "tolerogenic" clinical trials. Clearly, new paradigms are needed.

Recent insights into the requirements for T-cell activation and the normal mechanisms of immunologic homeostasis and self-tolerance suggest several new approaches to achieve transplantation tolerance. These approaches are now in various stages of preclinical and clinical trials.

Tolerance can be grouped into two categories: central and peripheral. In the case of T cells, central tolerance refers to events in the thymus. During their development in the thymus, immature T cells that are potentially reactive to autoantigens are either physically deleted (by induction of apoptosis) or functionally inactivated (induction of anergy). Taking advantage of this process, Ildstad and Sachs [9] and Sykes and Sachs [10] pioneered central tolerance-inducing strategies (also called thymic tolerance) that involve the use of lethal or sublethal myeloablation followed by immune system reconstitution with a combination of donor and recipient bone marrow. The donor cells actually contribute to hematopoiesis and thus the animals harbor a mixture of recipient- and donor-derived hematopoietic cells, a state called chimerism. Because T cells become tolerant to self-antigens as they develop in the thymus, recreating the immune system in the presence of donor cells leads to specific tolerance of donor antigens. Studies in rodents and in much more stringent nonhuman primate models have proven encouraging. However, the need for chemotherapy or irradiation to ablate the recipient immune system has largely prevented the widespread adoption of these protocols. In response to this problem, Posselt and colleagues [11, 12] have shown that the introduction of donor antigens directly into the thymus in conjunction with more limited immunosuppression may lead to tolerance in rodents. However, this method has not been tested widely in primates, and it may have limited applicability in adult humans, in whom significant thymic involution has already occurred.

Peripheral tolerance refers to events occurring in the mature lymphocyte compartment (lymph nodes, spleen, and peripheral blood). Here, the mechanisms of tolerance are less well understood but probably involve some combination of deletion, anergy, and induction of suppressor cells. As in thymic tolerance, deletion refers to physical elimination of the cells and anergy refers to a state in which the cells are present but are intrinsically nonresponsive. Suppression is poorly understood but is an immunologically active state in which the response of an intrinsically normal T cell is suppressed by a soluble factor or through cell-to-cell contact. Although multiple mechanisms probably maintain peripheral tolerance, it has been suggested that self-tolerance can be induced and maintained in the peripheral compartment by depriving cells of selected accessory signals required for antigen-specific responses. For example, to respond to antigen, T cells require additional costimulatory signals delivered through surface receptors [13]. In the absence of co-stimulation, T-cell immune responses are aborted early, and the cells either become anergic or die through apoptosis. At present the best characterized co-stimulatory signals are those provided through the accessory molecule CD28 [14]. The importance of this pathway is emphasized by the fact that blockade of CD28-mediated signals can induce transplantation tolerance to vascularized organs in rodent models [15]. Similar efficacy has been demonstrated through blockade of other potential costimulatory pathways (such as those mediated by CD2 or CD40 ligand), cell-to-cell adhesion pathways (such as LFA-1), or accessory signaling pathways (such as CD4). Recent studies in rodents and primates [16, 17] suggest that combination blockade of CD28 and CD40 ligand may have additive or synergistic effects that prevent rejection. In each case, blockade has been achieved by using monoclonal antibodies or recombinant fusion proteins as soluble competitors. Incomplete understanding of the signals transduced through these pathways has slowed the development of pharmacologic inhibitors, although intensive efforts are currently under way.

One approach, which combines aspects of central and peripheral tolerance and which has achieved substantial attention in recent years, is the use of donor bone-marrow augmentation [18]. In this maneuver, donor bone marrow is injected into patients at the time of solid organ transplantation. The rationale for this approach derives from observations by Starzl and colleagues that long-term allograft survival in humans (and transplantation tolerance in animals) is associated with the persistence of donor-derived hematopoietic cells in the recipient [19]. Unlike in the myeloablation studies [9, 10], donor cells persisted at very low levels (a "microchimeric" state) in this study. It has been suggested that microchimerism promotes tolerance through central or peripheral mechanisms; however, other studies have not supported this hypothesis [20]. Perhaps it is tolerance, no matter how it is induced, that permits microchimerism by allowing small numbers of donor bone marrow-derived cells to engraft in the recipient. Although the cause-and-effect relation of this association has not been established, clinical trials are under way. Unfortunately, initial results from one study not only failed to show improvements in graft survival but suggested that patients receiving donor marrow have greater immunosuppression and are at higher risk for opportunistic infection [21].

Where does this leave us? I believe that future advances in tolerance induction will come from approaches that exploit the normal mechanisms that establish and maintain self-tolerance. These approaches may take the form of pharmacologic agents, biological agents, gene therapy, or some combination. Recent studies suggesting that expression of the Fas-ligand gene can make tissue sites immunologically privileged are an exciting example of how this knowledge and technology can be put to practical use [22]. One of the foremost difficulties will be clinical trial design. To the extent that current therapy provides excellent short-term outcomes, we are victims of our own success; large studies with long observation periods will be required to demonstrate that new strategies are superior. Alternatively, trials might focus on high-risk patients (such as those undergoing retransplantation or experiencing rejection); however, the likelihood of seeing success is concomitantly reduced. We also should consider the problem of how to individualize treatment because no treatment is uniformly successful. With the present medications, some patients are at low risk for rejection (and steroid therapy can perhaps be withdrawn), whereas others have a higher risk. Even with new forms of treatment, some patients will be probably be destined for acceptance of the graft and others will require drug therapy to prevent rejection. Developing the means to identify these two groups must be a high priority. Finally, we must consider the question of how and whether to include standard immunosuppressive drugs in trials designed to induce tolerance. Although current immunosuppression may actually prevent the development of true tolerance, is it ethical to deny patients the very real benefits of such drugs as cyclosporine for the unproven hope of a new approach? These are the challenges we face as transplantation immunology enters a new era.