Despite advances in immunosuppressive drugs, antibody-mediated rejection (AMR) remains a critical problem in solid organ transplantation: 30% of transplant recipients of all organs develop AMR at some point after transplantation (1, 2). While current standard of care (SOC) immunosuppressive regimens target T cell-mediated rejection (TCMR), there is currently no reliable and durable treatment for AMR. As AMR is mediated by donor-specific antibody (DSA), the most widely-used approaches aim to rapidly remove DSA and to eliminate antibody-producing cells (e.g., B cells and plasma cells) but none durably modulate antibody production. Such approaches include plasmapheresis (PP), intravenous immunoglobulin (IVIg), anti-CD20mAb, increased maintenance immunosuppression, and/or lymphocyte-depleting agents (3–6). IVIg and PP are considered first-line treatments for AMR despite lack of evidence of efficacy or safety (4, 7). Furthermore, no mechanistic evidence suggests that approaches targeting antibodies alone can adequately control DSA production in the long term.

Rituximab, an anti-CD20 mAb, is the third most commonly used agent for AMR treatment and its use in clinical practice has increased as initial or rescue therapy (4, 8–10). As with antibody-targeting approaches, evidence for the efficacy of rituximab in treating AMR is limited (11). A single-center study of rituximab with PP for treatment of AMR reported two-year graft survival rates at 90% in the dual treatment group compared to 60% in the PP alone group (12). In contrast, multi-center clinical trials evaluating rituximab and standard of care (SOC: plasma exchange, IVIg, and corticosteroids) or rituximab and IVIg found no difference in graft survival, renal function, or adverse events (13, 14).

As such, there are currently no FDA-approved treatment for AMR. A recent meeting of international experts noted that there was no conclusive evidence to support any specific AMR therapies, highlighting the need to develop and rigorously evaluate AMR therapies (15).

Many biologics are available to target different components of the humoral response (16). These agents target initial B cell help from T cells (costimulation blockade) (17, 18), B cell survival (BAFF) (19), plasma cells (proteasome inhibitor, anti-CD38mAb) (17, 18, 20–22), antibodies (i.e., Imlifidase or anti-FcRn mAb) (23–25) and complement activation or split products (i.e., Eculizumab, C1-INH) (26–28), and have been tested in the clinic or in pre-clinical models for a variety of indications. However, optimal outcomes may not be achieved by just targeting a single component of the humoral response, due to compensatory pathways and mechanisms (20). We hypothesized that combination therapies offer a more promising approach to modulate the post-transplant humoral immune response.

Approaches to AMR treatment are often similar to those used for desensitization therapies prior to HLA-incompatible transplants, given the similar pathophysiologic mechanisms. Indeed, desensitization therapies aim to reduce the immune barrier by removing antibodies and targeting antibody-producing cells. Accordingly, therapies that successfully achieve this goal in the context of desensitization provide promising candidates for the treatment of AMR. Our group has previously tested several dual desensitization strategies of costimulation blockade with proteasome inhibition in sensitized NHP (17, 18, 29, 30). The combination of belatacept and carfilzomib effectively reduced plasma cell and follicular helper T cell (Tfh) populations, which reduced the incidence of early AMR and prolonged kidney allograft survival (29). Therefore, we hypothesized that proteasome inhibition and costimulation blockade with carfilzomib and belatacept would reduce antibody-mediated graft damage and prolonging graft survival in the context of AMR.

This study presents the use of carfilzomib/belatacept therapy in a NHP model of AMR following kidney allotransplantation. Following development of graft dysfunction and corresponding increase in DSA, animals were treated with corticosteroids, carfilzomib, and belatacept, with or without the C3 complement inhibitor Compstatin, and compared to controls treated with corticosteroids alone (Figure 1). Treatment of AMR with carfilzomib and belatacept improved kidney allograft function, prolonged graft survival (Figure 1) and suppressed the production of DSA (Figure 2) in accordance with our previous study in which we established this regimen for pretransplant desensitization (29). The contribution of complement inhibition in this regimen needs to be further evaluated. Animals experiencing chronic AMR (as defined by onset > 6 months post-transplant) were less responsive to AMR treatment, with less reduction in DSA titers (Figure 3).

Our group has previously shown that dual desensitization with carfilzomib and belatacept effectively lowers DSA, plasma cells (PC), T follicular helper cells (Tfh), and B cell follicles in the germinal center (GC) (29). Prevention of cognate T-B cell interactions through costimulatory blockade inhibits GC reactions and prevents repopulation of PCs from alloreactive memory B cell (Bmem) reservoirs. Similar combination regimens such as carfilzomib with lulizumab or tocilizumab, and daratumumab and plerixafor have similarly blunted the post-transplant humoral response (21, 22, 30). Given similarities in desensitization and AMR treatments, we applied this dual targeting strategy to AMR. In particular, this study examines chronic AMR, which likely occurs in a unique immunological context compared to acute AMR. Acute AMR likely arises in the context of DSA whose formation has been initiated prior to treatment, produced by both PC and germinal center B cells. Furthermore, the plasma cell repertoire implicated in early AMR is likely comprised of short-lived PCs and plasmablasts compared to long-lived plasma cells in chronic AMR. Thus, the impact of costimulation blockade and proteasome inhibition is likely to be different in early acute vs. late chronic AMR as costimulatory/coinhibitory signals are implicated in GC activation and contraction, and proteasome inhibition has been shown to preferentially act upon plasmablasts (39). Accordingly, we observed decreased efficacy of carfilzomib and belatacept in the chronic AMR group.

The combination of proteasome inhibitor and costimulation blockade has been used in humans in treatment of acute post-kidney transplant AMR (40). Bortezomib/belatacept treatment was given after PP, IVIG, and steroids failed to improve graft function. This regimen successfully resolved the episodes of acute AMR in six kidney transplant recipients, and suppressed DSA for up to 30 months (40). While this combination did not reverse or permanently halt the process of AMR in any patient, it may constitute an improvement over current approaches to AMR treatment.

One major barrier to the development of effective AMR treatments is the lack of a clinically relevant and reproducible animal model. There is no existing model that consistently produces an AMR phenotype. Furthermore, since organ transplantation promotes intense responses from both T and B cells, we cannot selectively elicit AMR or TCMR in this outbred, large animal model. The timescale of rejection is also an important consideration. Clinically, chronic AMR is a major source of graft loss in kidney transplant recipients but in animal models the development of a chronic AMR phenotype is difficult to achieve practically, as it requires long-term humoral activation and production of DSA while avoiding graft loss from acute AMR and TCMR, infection, and animal welfare concerns. Furthermore, long-term maintenance of post-transplant NHPs is costly and difficult to justify without reproducibility of a chronic AMR phenotype. Accordingly, we used sensitized maximally-MHC mismatched NHP to accelerate the development of AMR—a process that takes years in human recipients. Maximal mismatching of kidney grafts is not a clinical goal since it results in greater risk of rejection and poorer graft survival; however, this approach experimentally may trigger a robust and accelerated AMR response, resulting in a stringent and practical model of AMR. Given this stringency, we hypothesize that patients undergoing HLA-compatible transplants may actually have better responses to AMR treatment compared to our NHPs. To address this limitation, we have recently developed a naturally sensitized multiparous female primate model (32). Additionally, all animals received desensitization therapy prior to transplant to avoid early acute AMR. The present study includes NHP that developed biopsy-proven AMR on late and chronic timescales while in other immunosuppression studies, thus providing a clinically relevant and resource-sparing approach to preclinical AMR models.

The inclusion of animals maintained on different immunosuppressive regimens is also a limitation. While the NHP were impartially assigned to control and treatment groups, we cannot account for the potential impact of varying immunosuppression on subsequent AMR treatment. Previous studies have shown that CNI-based regimens do not adequately control the humoral response and thus animals maintained on costimulation blockade-based regimens may respond better to carfilzomib and belatacept AMR therapy (41).

Despite these limitations, the present study provides a useful and resource-efficient preclinical model to develop AMR treatments. Belatacept and carfilzomib dual therapy for AMR resulted in significant prolongation of post-transplant survival via reduction of DSA production, but failed to prevent rebound under conventional CNI-based immunosuppression. Given the published clinical experiences and our preclinical data, carfilzomib and belatacept offer a promising approach to AMR treatment that may improve on SOC antibody-targeting strategies.