SEARCH

SEARCH BY CITATION

Keywords:

  • Co-stimulation;
  • kidney transplantation;
  • T cell activation

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Blocking T-cell Costimulation in Transplantation: A Shift in Emphasis from Tolerance Induction to Maintenance Immunosuppression
  5. Costimulation Blockade: Making the Move from Bench to Bedside, One Step Backward, One Forward
  6. Conclusions
  7. Acknowledgments
  8. References

Activated T cells orchestrate the immune response that results in graft rejection; therefore, a common goal among current immunosuppressive therapies is to block T-cell activation, proliferation and function. Current immunosuppressive regimens that inhibit T cells and immune cells have greatly reduced the incidence of acute rejection following solid-organ transplant. However, the expected improvements in long-term outcomes have not been realized. This may be related to the non-immune side effects of current maintenance immunosuppressants, which target ubiquitously expressed molecules. The focus in transplantation research is shifting in search of maintenance immunosuppressive regimens that might offer improved long-term outcomes by providing efficacy in prevention of acute rejection combined with reduced toxicities. An emerging therapeutic strategy involves an immunoselective maintenance immunosuppressant that inhibits full T-cell activation by blocking the interaction between costimulatory receptor–ligand pairs. This review describes costimulatory pathways and the development of molecules, which inhibit them in the context of transplantation research. Recent clinical data using the selective costimulation blocker, belatacept (LEA29Y), as a part of a CNI-free maintenance immunosuppressive regimen in renal transplantation is highlighted.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Blocking T-cell Costimulation in Transplantation: A Shift in Emphasis from Tolerance Induction to Maintenance Immunosuppression
  5. Costimulation Blockade: Making the Move from Bench to Bedside, One Step Backward, One Forward
  6. Conclusions
  7. Acknowledgments
  8. References

Despite dramatic reductions in acute rejection rates, and improvements in short-term graft survival of renal transplants, long-term survival rates have changed little during the last decade (1). This underscores the pressing need for new immunosuppressive strategies that focus on improving long-term net health outcomes, rather than emphasizing a one-dimensional aspect of short-term outcomes, such as early rejection rates.

Given the central role of T cells in transplant rejection, a unifying goal of immunosuppressive therapies is to block T-cell activation (Figure 1). To this end, current immunosuppressive regimens use a combination of therapies that provide more intense immunosuppression during the immediate post-transplantation period (induction phase). With time, most patients are transitioned to a multiple-drug cocktail (maintenance immunosuppressive phase) to maintain a state of low- or non-responsiveness to the allograft, while permitting protective immunity to environmental pathogens. Unfortunately, the immune and non-immune activities of current maintenance immunosuppressants contribute to toxicities that compromise the ability to achieve optimal long-term outcomes in renal transplantation. This fact drives the development of alternative maintenance therapies with greater selectivity and improved toxicity profiles.

image

Figure 1. The current immunosuppressive paradigm targets the T cell. APC = antigen-presenting cell; MHC = major histocompatibility complex; TCR = T-cell receptor; MMF = mycophenolate mofetil; MPA = mycophenolic acid; IL-2r = interleukin 2 receptor; NFAT = nuclear factor of activated T cells.

Download figure to PowerPoint

This review discusses research into the CD40 and CD28 costimulatory pathways, and specifically the development of molecules that target these pathways. Such agents may have immunosuppressive potential suitable for the transplantation setting.

Blocking T-cell Costimulation in Transplantation: A Shift in Emphasis from Tolerance Induction to Maintenance Immunosuppression

  1. Top of page
  2. Abstract
  3. Introduction
  4. Blocking T-cell Costimulation in Transplantation: A Shift in Emphasis from Tolerance Induction to Maintenance Immunosuppression
  5. Costimulation Blockade: Making the Move from Bench to Bedside, One Step Backward, One Forward
  6. Conclusions
  7. Acknowledgments
  8. References

Advances in immunologic research have resulted in the characterization of the pathways and molecules critical for T-cell activation. T cells require more than a single signal for their full activation. Signal 1 is delivered via the T-cell receptor (TCR); alloreactive T cells recognize polymorphic donor-derived proteins as peptides bound to self-major histocompatibility complex (MHC) molecules on the surface of autologous antigen presenting cells (indirect recognition). Additionally, allogeneic MHC molecules (presumably loaded with peptides) can be directly recognized on graft-derived cells. In either case, a costimulatory signal (Signal 2) is required for the induction of naïve T-cell activation (2,3). Indeed, naïve T cells receiving only Signal 1 are either rendered anergic (unresponsive to a subsequent specific antigenic challenge) or are programmed for apoptosis (4). Soluble factors, such as cytokines, also provide activating signals (sometimes referred to as Signal 3) to propagate and augment responses (Figure 2).

image

Figure 2. T-cell activation requires more than one signal.

Download figure to PowerPoint

Several families of costimulatory molecules have been identified, including the immunoglobulin (Ig) superfamily, the tumor necrosis factor:tumor necrosis factor receptor (TNF:TNFR) family and the integrin family (5). Costimulatory signals differ both in their ability to increase or decrease T-cell activation, and in their expression patterns (constitutive versus induced; early versus late). This enables fine bi-directional modulation of the T-cell response, with significant potential for overlap and/or redundancy.

The best-characterized, and perhaps most important, positive costimulatory signals are between CD40 and CD154 (CD40L) in the TNF:TNFR family and the interaction between CD80 (B7-1)/CD86 (B7-2) and CD28, members of the Ig superfamily. The mechanism by which these molecules exert their effects remains incompletely defined; however, inhibition of these pathways using transgenic knockout mice, monoclonal antibodies or fusion proteins, has demonstrated profound effects on the immune response (6).

Based on the concepts put forth by Lafferty, Schwartz and others (2–4), which suggest that deprivation of costimulatory signals during TCR signaling could lead to selective inactivation or death of antigen-specific T cells, a major focus of transplant researchers for many years has been investigation of this strategy to induce transplantation tolerance. Indeed, there have been numerous reports describing tolerance induction in rodent transplant models following blockade of various costimulatory pathways (7,8). While progress has been made in non-human primate models, identification of clinically applicable strategies that consistently induce tolerance in pre-clinical models has so far been elusive (9–12). The reasons for the apparent discrepancies between the results in rodents, primates and humans remain unclear, but may, in part, be due to the increased complexity of the human immune system and the more diverse and longer-term environmental exposure that humans experience compared with laboratory animals (13). In addition, animals used in pre-clinical transplantation studies do not suffer from the diseases that necessitate organ transplantation, adding another layer of complexity to the treatment paradigm (14). Nonetheless, rodent experimental models have played, and continue to play, a vital role in the process of identifying potential therapeutic targets/strategies in transplantation.

As the unfulfilled quest for tolerance in non-human primates continues, there has been a quiet but profound shift in the approach of the transplant community toward harnessing the therapeutic potential of targeting T-cell costimulation. Therapeutic development has recently focused on exploiting the relatively restricted distribution of costimulatory pathways within the immune system. Inhibition of costimulatory pathways may represent less toxic alternatives to currently available induction and/or maintenance immunosuppressive agents.

Costimulation Blockade: Making the Move from Bench to Bedside, One Step Backward, One Forward

  1. Top of page
  2. Abstract
  3. Introduction
  4. Blocking T-cell Costimulation in Transplantation: A Shift in Emphasis from Tolerance Induction to Maintenance Immunosuppression
  5. Costimulation Blockade: Making the Move from Bench to Bedside, One Step Backward, One Forward
  6. Conclusions
  7. Acknowledgments
  8. References

CD40:CD154—distribution and function

CD40, a member of the TNF receptor superfamily, is constitutively expressed on antigen-presenting cells (APCs), such as B cells, macrophages, dendritic cells (DCs) and thymic epithelia, and can be induced on endothelial cells and fibroblasts (15). CD40 plays a major role in B-cell activation and the maturation of DCs, which become 100-fold more potent initiators of T-cell responses following CD-40-dependent maturation. This is, in part, mediated by enhanced DC survival via nuclear factor (NF)-κB-responsive pathways; the upregulated expression of CD80, CD86 and adhesion molecules, such as ICAM-1; and increased production of IL-12 (16). The CD40 on B cells provides signals essential for proliferation, Ig production, isotype switching and memory B-cell development (15,17). CD40 stimulation also triggers macrophage effector function, causing the release of TNF-α and nitric oxide (17).

The ligand for CD40, CD154, is expressed both as a type II integral membrane protein and a soluble cleaved cytokine. Following TCR: antigen engagement, CD154 expression is rapidly induced on CD4+ T cells, as well as on some CD8+ T cells. The expression of CD154 on T cells is also influenced by signals from CD28 (Figure 3). More recently, it has been recognized that platelets express CD154 and that platelet-derived CD154 plays a major role in both inflammation and coagulation (discussed below).

image

Figure 3. T-cell costimulatory pathways.

Download figure to PowerPoint

Therapeutic targeting of CD40:CD154

Therapeutic targeting of the CD40 pathway has been of great interest to the transplant community for more than a decade. Remarkable effects on allograft survival and tolerance induction have been observed in rodent models using anti-CD154 antibodies (8–11,18,19). Anti-CD154 antibodies also prevent acute rejection and promote long-term allograft acceptance in non-human primates; however, withdrawal of therapy results in eventual graft rejection, demonstrating that immunologic tolerance is often not achieved when these antibodies are used alone (9–11,18,19). Despite positive pre-clinical experiences, clinical trials testing anti-CD154 antibodies as immunosuppressive agents in autoimmune disease and transplantation were terminated due to an unanticipated, elevated incidence of thrombo-embolic complications.

Subsequent work has shown that platelets are the main source of soluble CD154 (sCD154) in peripheral blood and that this molecule has a role in the control of thrombotic and inflammatory processes. Importantly, there is evidence that CD154 can stabilize arterial thrombi by an integrin glycoprotein (GP) IIb/IIIa-dependent (CD40-independent) mechanism (20). Others have suggested that sCD154 may activate platelets through CD40 ligation, although these CD40-dependent effects are relatively weak (21). Furthermore, it has been found that CD154−/− deficient mice have unstable thrombi, whereas this defect is not observed in CD40−/− mice (22). On balance, these studies suggest that CD154 may be involved in thrombus formation and platelet activation, primarily mediated via CD40-independent pathways.

Further studies in human renal transplantation raised additional concerns about the efficacy of anti-CD154 antibodies; data showed that five of seven patients treated with the anti-CD154 monoclonal antibody hu5c8 experienced rejection episodes (12). However, this initial trial tested hu5c8 in a low-dose steroid- and calcineurin inhibitor (CNI)-free protocol. It is possible that careful vigilance and liberal biopsies in the setting of an ambitious open-label trial, and/or increased inflammation caused by the pro-thrombotic properties of the drug may have contributed to these observations.

Alternative anti-CD154 antibodies have been developed, with the hope that targeting different epitopes on the CD154 molecule may not interfere with platelet function. Recent studies pairing the anti-CD154 antibody, IDEC-131, with a short course of sirolimus and a single donor-specific transfusion, have shown prolonged skin graft survival in primates and operational tolerance following primate renal allotransplantation (23,24). Another anti-CD154 antibody showed efficacy in preventing non-human primate renal transplant rejection, but again raised the specter of thrombotic side effects (25).

This insight into how CD154 functions to stabilize thrombi in a CD40-independent manner suggested that targeting this pathway via CD40 rather than CD154 might allow interruption or inhibition of the CD154/CD40-dependent interactions critical for T-cell responses, while leaving unaltered the CD154-integrin interactions necessary to regulate thrombus stability. The most obvious approach would be to develop antagonistic anti-CD40 monoclonal antibodies that block CD154 binding without delivery of signals via CD40. Several anti-CD40 antibodies have been developed, one of which has been evaluated, with promising results, in a primate renal allograft model (26,27).

While antibodies to CD40 with agonistic properties have been most extensively investigated as a strategy to augment immune responses (e.g. to enhance vaccines or boost anti-tumor activity), recent work in rodents demonstrated that some agonistic anti-CD40 antibodies may also attenuate immune responses (28–30). One such antibody, Chi220, a chimeric anti-human CD40 mAb, has been evaluated in non-human primate transplants models. Chi220 blocks CD154 binding, inhibits CD154-induced B-cell proliferation, but has weak agonist properties (31). As monotherapy, Chi220 promoted modest prolongation of renal and islet allograft survival in rhesus macaques, and was particularly effective when combined with the CD28 costimulation blocker, belatacept (LEA29Y). The relative contribution of blockade of ligand binding, the partial agonistic properties and complement and/or FcR-dependent cell depletion to Chi220's immunosuppressive activities are, as yet, unknown. Although Chi220 promotes transient B-cell depletion, the immunosuppressive effects of Chi220 in the islet allograft model could not be reproduced by B-cell depletion using the anti-CD20 monoclonal antibody, rituximab. Further clarification of these mechanisms will be an important area of focus for future research (27,31,32).

CD28:CD80/CD86 and CTLA-4—distribution and function

While many costimulatory pathways have been described, the interactions and functions of the CD28 and cytotoxic lymphocyte antigen-4 (CTLA-4; also known as CD152) receptors and their ligands, CD80 and CD86, remain the most thoroughly characterized (Figure 3). In humans, CD28 is constitutively expressed by a majority of CD4+ T cells and approximately 50% of CD8+ cells (33). CD28 delivers critical costimulatory signals, which synergize with TCR signals to lower the T-cell activation threshold. This results in enhanced proliferation, increased cell survival due to enhanced expression of anti-apoptotic molecules, preparation of the cellular bioenergetic machinery for the metabolic demands of the clonal expansion process and markedly increased cytokine secretion (34). CD28 signals promote T-cell differentiation into T helper (Th) cells, enhance B-cell help for antibody production and enhance proliferation of previously activated T cells (35). Recent studies have also shown that CD28 may signal to DCs through CD80 and CD86, inducing IL-6 production and preventing immunosuppressive tryptophan catabolism (36). Furthermore, CD28 controls both thymic development and peripheral homeostasis of regulatory T cells (Tregs), which constitute 5–15% of CD4+ T cells (37,38).

In contrast to CD28, CTLA-4, which shares ∼30% homology with CD28, is induced rapidly upon T-cell activation (Figure 3). CTLA-4 binds to the same ligands as CD28, namely CD80 and CD86, but with a 10- to 20-fold higher affinity for both than CD28. Unlike CD28, CTLA-4 delivers a potent inhibitory signal to T cells, attenuating CD28 and TCR signals. This inhibits IL-2 and IL-2R expression and arrests T cells at the G1 phase (35,39,40). CTLA-4 inhibits T-cell activation in both naïve and primed CD4+ and CD8+ T cells, although to a lesser extent in CD8+ T cells (40). In addition, CTLA-4 plays an important role in the function of Tregs. By ligating CD80 and CD86, CTLA-4 induces expression of the tryptophan-degrading enzyme, indoleamine 2, 3 dioxygenase (IDO), in DCs. This enzyme may allow suppression of antigen-driven proliferation of T cells in vitro, providing an additional mechanism of regulating T-cell responses (41).

CD80 and CD86 are thought to have overlapping functions—both bind to CD28 during activation of T cells—although evidence suggests that they may have functionally distinct roles in T-cell activation. In general, CD86 is constitutively expressed and is rapidly upregulated on APCs, whereas CD80 is usually induced following more prolonged stimulation (34,42). These differences suggest that CD86 may be important in mediating initial T-cell activation and CD80 may play more of a role in perpetuating the immune response (19). Recent studies suggest that, in some circumstances, CD86 binds preferentially to CD28, whereas CD80 primarily ligates CTLA-4 (43). These differences in affinity and avidity translate into a difference in the selective recruitment of CD28 or CTLA-4 to the immunologic synapse, with CD86 acting as the major ligand mediating CD28 recruitment to the synapse and CD80 playing the same role for CTLA-4 (43).

Therapeutic targeting of the CD28 pathway

Powerful effects of short-term blockade of the CD28 pathway were first observed in islet xenograft, and rat and mouse cardiac allograft models, triggering intense research. Studies in rodent models of transplant demonstrated that blockade of CD28:CD80/CD86-mediated T-cell costimulation prevented acute allograft rejection and in some models induced donor-specific tolerance, although this finding was model- and strain-dependent (44–47). CD80/CD86 double-knockout transgenic mice are unable to reject cardiac allografts, although they do reject skin and islet cell transplants (48,49). Anti-CD80 and -CD86 antibodies were able to delay renal allograft rejection in non-human primates even when used without CNIs and steroids (50–52). Combined blockade was most effective (51), possibly as CD86 signaling is implicated in early rejection and CD80 signals in later in the rejection process (42). Although not powered to evaluate efficacy, a Phase I study in renal transplant recipients provided evidence that antibodies against CD80 and CD86 (h1F1 and h3D1) were safe as part of a combination maintenance immunosuppressive regimen including cyclosporine, mycophenolate mofetil (MMF) and steroids (53). Further development was halted prior to any evaluation of their efficacy.

An alternative strategy involves targeting this receptor/ligand pair by blocking CD28. Recent reports show that short-term use of anti-CD28 monoclonal antibodies prevented acute and chronic rejection in rodent renal transplantation models (54,55). Formation of anti-donor MHC class II alloantibodies was largely prevented in tolerant recipients, and a subset of B7+ non-Tregs was generated, with tolerance sustained by IDO and inducible nitric oxide synthase (54).

The fusion protein, CTLA-4Ig, combining the extracellular binding domain of CTLA-4 with the Fc portion of IgG1, was developed as a drug candidate to block the interactions of CD28 with CD80 and CD86. Numerous studies have shown that CTLA-4Ig acts to inhibit immune responses, both in vitro and in vivo (56). In rodents, CTLA-4Ig inhibits T-cell-dependent antibody responses, humoral and cellular immunity, significantly prolongs transplanted organ survival, and slows disease progression in several animal models of autoimmune disease (57–59). CTLA-4Ig has also been shown to prevent the development and progression of chronic rejection in some animal models (58,59). Interestingly, even though CD28-mediated costimulation is thought to act primarily on naïve T cells, CTLA-4Ig has demonstrated significant clinical efficacy in the treatment of established T-cell-mediated autoimmune diseases, such as psoriasis and rheumatoid arthritis (60–62). Perhaps not surprisingly, while the results using CTLA-4Ig in non-human primate transplant experiments demonstrated some prolongation in allograft survival, the effects were much less pronounced than those seen in rodent models (9,63).

Rational development of belatacept (LEA29Y)

Despite initial disappointment with results of experiments using CTLA4-Ig in non-human primate transplant models, several factors led to continued efforts to develop variants of CTLA4-Ig with enhanced efficacy. These factors included: (1) the conceptual appeal of the target (i.e. the pivotal role of the CD28 pathway in T-cell responses, along with the restricted cell and tissue distribution of the receptors and ligands of the pathway); (2) the appreciation that optimal effects in allograft models required blockade of both CD80 and CD86, coupled with the realization that CTLA4-Ig showed less than optimal binding to CD86; and (3) the opportunity to manipulate CTLA4-Ig's binding properties afforded by the definition of the contact residues between CTLA4 and CD86 (64) (Figure 4). After screening a library of CTLA4-Ig variants with amino-acid substitutions in the binding domains, belatacept, which has two amino acids substituted versus CTLA4Ig, was selected for further development. This decision was based on data showing an approximately fourfold increase in binding avidity to CD86 and a twofold increase to CD80 and a ∼10-fold greater inhibition of T-cell activation in vitro as compared with CTLA4-Ig (65).

image

Figure 4. The rational development of belatacept.

Download figure to PowerPoint

Non-human primate studies demonstrated prevention of acute rejection and prolongation of renal allograft survival when using belatacept either as monotherapy or in combination with drugs that are typically used in human transplant immunosuppressive regimens—basiliximab, steroids and MMF (65). Belatacept also inhibited anti-donor antibody formation, thought to contribute to the development of chronic rejection and a major barrier to retransplantation (65).

A Phase II study involving ∼200 human recipients of de novo renal allografts compared the efficacy of belatacept with cyclosporine as the primary maintenance immunosuppressant in a combination regimen including basiliximab induction, steroids and MMF (66). Belatacept-based maintenance therapy, administered as a 30-minute intravenous infusion every 4 or 8 weeks, demonstrated equivalent efficacy in preventing biopsy-proven acute rejection at 6 months versus cyclosporine-based treatment (6–7% for belatacept versus 8% for cyclosporine). Additionally, belatacept-treated patients showed significant improvements in renal function and reductions in chronic allograft nephropathy compared with cyclosporine-treated patients at 1 year, a finding that may be predictive of improved long-term outcomes (66). Belatacept was safe and well tolerated in this Phase II study, with similar rates of graft loss, death, discontinuations and adverse events, including infections and malignancies, between groups. Interestingly, in the belatacept groups there were trends toward lower rates of typical CNI-related toxicities, such as hyperlipidemia, hypertension and post-transplant diabetes. Further, as belatacept interacts with the CD28 pathway, there were no reports of the thrombotic complications associated with therapeutic interventions in the CD40:CD154 pathway. Belatacept did not appear to affect the number or activity of Tregs (67) and clinical monitoring of lymphocytes in studies of belatacept in healthy human volunteers, rheumatoid arthritis patients (68) and renal transplant recipients did not reveal any depleting effects of this therapy (data on file). These findings suggest that belatacept acts by inhibiting initial T-cell activation, rather than by selective depletion or complement-mediated lysis (66).

Conclusions

  1. Top of page
  2. Abstract
  3. Introduction
  4. Blocking T-cell Costimulation in Transplantation: A Shift in Emphasis from Tolerance Induction to Maintenance Immunosuppression
  5. Costimulation Blockade: Making the Move from Bench to Bedside, One Step Backward, One Forward
  6. Conclusions
  7. Acknowledgments
  8. References

Current immunosuppressive therapies have greatly reduced the incidence of acute rejection following solid organ transplant, but have had little impact on chronic allograft dysfunction, late graft failure and death. Indeed, the non-selectivity of current maintenance immunosuppressants contributes to toxicities that compromise the ability to achieve optimal long-term outcomes in renal transplant. A greater understanding of the immune pathways that lead to rejection has helped foster the development of novel therapies that target critical T-cell costimulatory pathways. These strategies have the potential both to be effective and to provide a greater degree of immunoselectivity than current immunosuppressive agents.

The major shift seen in the past few years has been a change in emphasis from using costimulation blockers primarily as a tool for tolerance induction, toward exploitation of their potential as primary maintenance drugs in CNI-free regimens. While the initial clinical trials with anti-CD154 antibodies were met with unanticipated thromboembolic complications, efforts to target the CD40 pathway continue to hold promise. The results of Phase II studies in renal transplantation targeting the CD28 pathway using belatacept are very encouraging. As the actions of more recently discovered pathways are elucidated, including inducible costimulator (ICOS)/B7RP-1 and programmed death (PD)-1/PD-ligand (PD-L)1, further advances are expected.

It should be noted that, although exciting, the results seen with belatacept are preliminary and the implications of this approach in the longer term can only be borne out by larger studies and with longer-term observations. Long-term extension data from Phase II and the results of larger Phase III studies, including one in recipients of extended criteria donor kidneys, are eagerly anticipated to confirm these findings. However, in the near term, these agents may provide extraordinary clinical benefits through their role as part of less toxic maintenance therapy. With continued research, we hope that it may still be possible to use them someday as a component of the therapy that provides transplantation tolerance.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Blocking T-cell Costimulation in Transplantation: A Shift in Emphasis from Tolerance Induction to Maintenance Immunosuppression
  5. Costimulation Blockade: Making the Move from Bench to Bedside, One Step Backward, One Forward
  6. Conclusions
  7. Acknowledgments
  8. References

The authors thank Shula Sarner, Ph.D., for editorial assistance in manuscript preparation. Dr. Larsen and Dr. Pearson have received consulting fees from Bristol-Myers Squibb, Pfizer and Abbott and grant support from Bristol-Myers Squibb, Novartis and Abegenix and have assigned all future royalties relating to U.S. patent ‘Methods for inhibiting an immune response by blocking the GP39/CD40 and CTLA4/CD28/B7 pathways and compositions for use therewith’ (5,916,560), issued jointly to Bristol-Myers Squibb and Emory University, to Emory University. Dr. Adams and Dr. Knechtle have no commercial association or funding sources related to the submitted work. Dr. Kirk is funded by the Division of Intramural Research, National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Department of Health and Human Services.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Blocking T-cell Costimulation in Transplantation: A Shift in Emphasis from Tolerance Induction to Maintenance Immunosuppression
  5. Costimulation Blockade: Making the Move from Bench to Bedside, One Step Backward, One Forward
  6. Conclusions
  7. Acknowledgments
  8. References