Over time, donor and recipient profiles have changed substantially (1) modifying the risk of allograft rejection. Thus defining distinct alloimmune and non-alloimmune factors driving allograft rejection is greatly needed. For example, the proportion of sensitized [i.e., with circulating anti-human leukocyte antigen (HLA) antibodies] patients on kidney transplant waiting lists has gradually increased worldwide, because of both the implementation of highly sensitive immune assays to identify them and the increased proportion of retransplantations. In parallel, the number of expanded-criteria donors (ECD) or donors after circulatory death (DCD)—both groups that are often dominated by elderly people—may now exceed 50% in many transplant programmes. In such scenarios, it can be difficult to attain the excellent kidney transplantation outcomes observed for low-risk recipients and standard-criteria donors. However, low-risk donor and recipient is the usual pairing included in randomized controlled trials investigating new molecules and immunosuppression strategies. Given their real-world complexities, it would be useful to establish endpoints to identify clinically relevant and affordable improvements in outcome for distinct high-risk transplantation scenarios. This article presents evidence-based key determinants in immunological and non-immunological risk stratification, including but also extending beyond clinical research settings.

Technologies to assess alloimmune risk in transplant recipients have been developed and implemented in clinical practice, but further improvements to alloimmune risk stratification in kidney transplantation are needed. Such improvements would help to identify different subgroups of transplantation patients with distinct immune risks, which in turn would inform the development of clinical studies. Risk stratification is an essential first step toward personalized immunosuppression strategies for kidney transplant recipients.

Long-term immunosuppressive therapy may cause transplant recipients to experience various clinical events including cardiovascular disease, oncologic or metabolic complications, or opportunistic infections. Currently it is hard to individualize immunosuppressive therapy regimens to minimize the risk of such complications since data on individualization strategies remain limited and do not yet enable specific high-risk profiles to be identified (2). By contrast, the risk of allograft rejection (i.e., the immune-mediated destruction of transplanted organs), a major cause of graft loss, has been extensively investigated in large, retrospective population-based cohorts (3,4).

Alloantigens are unavoidably recognized by the kidney transplant recipient’s adaptive immune system. However, the innate immune system—which is triggered by damage-associated molecular patterns (DAMP) released in the circulation, because of ischemia-reperfusion injury (IRI) immediately after transplantation—is necessary to prime the adaptive alloimmune response. DAMP are strong stimulators of the immune system (5,6). Immunological dogma holds that rejection requires effectors of the adaptive immune system, namely alloreactive cytotoxic T cells and donor-reactive B cells, which produce destructive donor-specific antibodies (DSA). Notably, a key feature of the adaptive immune system compared with the innate immune system is that the former generates antigen-specific memory effectors (i.e., memory T and B cells), which respond rapidly when the same antigen is re-encountered. Importantly, although this vision of rejection as being largely dependent on the ability of the adaptive immune system to discriminate between alloantigens (i.e., a process named allorecognition) largely remains dominant, independent reports from basic-research and early clinical studies suggest that some innate effectors (including monocytes and natural killer cells) are also capable of allorecognition, leading to previously overlooked types of “innate” rejection episodes (7–9) and interfering with the adaptive immune mechanisms at stakes in “classical” rejection episodes (10).

Two main strategies are used worldwide for immune-risk stratification before kidney transplantation (11). First, evaluation of HLA disparity between recipient and donor, which quantifies the risk that a “naïve” transplant candidate will develop a de novo alloimmune response over time, by recognizing foreign alloantigens. Secondly, identification of preformed circulating IgG antibodies against HLA in the recipient’s serum, capable of lysing donor lymphocytes in a complement-dependent manner (“serological memory”); these antibodies are identified using a complement-dependent cell (CDC)-crossmatch assay. The latter approach aims to identify sensitized transplant candidates with preformed humoral alloimmunity, able to trigger complement cascade activation against the graft (i.e., preformed DSA responsible for rapid severe AMR and graft loss).

Advances in the characterization of donor/recipient HLA disparities at the molecular level, use of the flow crossmatch (FCXM) (in some centers), and novel and highly sensitive immunological tools to detect circulating IgG anti-HLA antibodies (whether complement binding or not), have substantially changed the landscape of immune-risk profiling.

Kidney transplants from donors who are elderly, ECD, ECD/DCD, or kidneys with pre-existing lesions have poorer prognoses than transplants from standard-criteria donors. Recipients of organs from high-risk donors tend to have poor renal function, with reduced medium-term graft survival (12,13). Such patients are also at high risk of delayed graft function (DGF) (14,15). These problems are not only related to the lower intrinsic quality of such organs but also to their highest level of immunogenicity (discussed more, below).

Immunosuppressive regimens administered to recipients of kidneys from ECD are adapted to avoid early acute rejection, which might worsen any pre-existing or ischemic injury of the graft; conversely, the goal of maintenance immunosuppression in such settings is to attenuate the long-term nephrotoxicity of calcineurin inhibitors. Indeed, reducing IRI in high-risk donors has been a major goal in kidney transplantation to minimize not only the risk of DGF but also to abrogate subsequent alloimmune activation favoring allograft rejection (16). Agents counteracting the effects of ischemia have been studied in selected populations, mainly by using donor/recipient risk indices to assess DGF risk (17). Interventional studies to prevent graft IRI have generally evaluated DGF occurrence as a qualitative phenomenon, although some also evaluated medium-term renal function. Given that IRI is a dynamic response to numerous molecular events, assessing DGF severity could help with the quantitative evaluation of the protective effects of anti-ischemic agents. Indeed, following discussions with the US Food and Drug Administration, this approach is in clinical investigation (ClinicialTrials.gov identifier: NCT02610296), with DGF severity (measured in terms of the number of dialysis sessions required in the first 30 days post transplantation, for participants starting dialysis on days 0–7) as the primary endpoint. However, DGF is unspecific and only partially relates to long-term graft function.

It is beyond the scope of the present paper to discuss ECD criteria and the definition of DGF as a potential endpoint in more detail, although it may also be a consequence of an early acute rejection episode; instead, we focus on the establishment of alloimmune risk in transplantation settings. Nevertheless, it should be noted that we consider DGF as a potentially meaningful endpoint for registration trials in transplantation.

Compared with other candidates, highly sensitized patients have reduced access to kidney transplantation and worse allograft outcomes, mainly due to their high risk for antibody-mediated rejection (AMR) (18).

The broadness of anti-HLA sensitization is evaluated using calculated panel-reactive antibody (cPRA) testing, which for each candidate estimate the percentage of donors against whom he is likely to show DSA, thus ultimately determining the proportion of unacceptable donors for a given transplant candidate. Candidates with very high cPRA values (>90%) have a reduced chance of finding a suitable kidney donor (19).

Several strategies enhance access to transplantation in highly sensitized candidates. The best option is the transplantation of a kidney from an HLA-compatible living donor (20,21) which, in the absence of an HLA-identical sibling volunteering for donation, may be achieved through large paired-donor exchange pools.

In the absence of a compatible donor, in the United States, desensitization protocols are commonly used (22). Standard-of-care desensitization regimens are based on a combination of off-label agent usage and techniques that aim to reduce antibody titers transiently, such as administration of intravenous immunoglobulin (IVIg), rituximab, and pre- and/or post-transplant apheresis with plasma exchange or immunoadsorption. However, such approaches are not widely followed in Europe, for reasons including evidence of inferior outcomes compared with HLA-compatible transplantation and a lack of robust data demonstrating the superiority of these high risk costly procedures (20,23,24).

In the absence of an HLA-compatible living donor, three strategies exist to increase access to transplantation for highly sensitized candidates on deceased-donor transplant waiting lists in Europe, which consider the degree of sensitization in algorithms for organ allocation. First, in the United States and some European countries including Spain and France, Kidney Allocation Systems prioritize candidates with very high cPRA values (percentages differ among countries but are ≥95%). This has increased the transplantation rate among highly sensitized candidates to levels similar to those for other candidates; however, in extremely sensitized patients (cPRA ≥99.9%), transplantation rates remain significantly lower than rates for less-sensitized patients (25). Secondly, the Eurotransplant International Foundation developed the Acceptable Mismatch (AM) Program for highly sensitized patients in the 1980s. Between 1989 and 2017, over 2,500 patients were listed on the Program, 57% of whom received a donor kidney (26). The 10-years graft survival rate among recipients listed on the Eurotransplant AM program was comparable to that for less sensitized recipients (26). The AM strategy is also used outside the Eurotransplant Program. For example, since 2005 France has operated a national AM policy (27). The EUROpe-wide Strategy to enhance Transplantation of highly sensitized patients based on Acceptable HLA Mismatches (EUROSTAM) project has developed and tested a tool to evaluate opportunities for sharing kidneys across different countries; the aim of this initiative is to increase HLA-compatible transplantation rates and thus, improve outcomes (28). There is also a third option: desensitization can be undertaken, sometimes in combination with specific allocation programs, to facilitate transplantation in sensitized recipients with preformed DSA (and/or positive crossmatch) (29,30).

Despite the creation of these programs to increase access to transplantation for highly sensitized candidates, a substantial number of people may not benefit (31), especially those with cPRA ≥98%, who often remain wait-listed for many years. These transplant candidates may need different strategies to increase their level of access to organs. In this regard, access to transplantation might be considered as a discrete endpoint among highly sensitized candidates enrolled in studies investigating whether new therapeutic approaches help to improve transplantation rates. Of note, fair evaluation of desensitization strategies based on access to transplantation requires that future studies enrol patients with homogeneous humoral immunological risk (discussed below). Clear distinction appears to be mandatory between candidates with positive lymphocytotoxicity test (LCT) cross match against their donor (who require therapeutic action pre-transplantation to reduce the titer of preformed DSA, to prevent hyperacute rejection) and candidates with lower DSA levels (positive FCXM and/or positive solid phase assay) and negative LCT, who can be transplanted without prior desensitization and only require adaptation of immunosuppression. In this regard, the transplantation rate alone is not a sufficient endpoint; only successful (e.g., rejection-free, good renal function) transplantations in the medium- or long-term (typically >10 years) should be considered. Highly sensitized transplant recipients are at high risk of developing AMR; they also have poor renal function and low graft survival rates (20,32). Although the relevance and impact of T cell-mediated rejection in these patients is lower compared with other transplantation groups, AMR with donor-specific anti-HLA antibodies exerts a detrimental effect on long-term graft survival (33,34). Hence, AMR could be a very suitable primary endpoint and surrogate for graft outcome in highly sensitized compared with HLA-compatible kidney transplant recipients.

The 2017 Banff conference described active AMR—which has several clinicopathological subtypes—as being indicative of ongoing disease activity. Active AMR is characterized by microvascular inflammation with or without graft remodeling; it is discussed further in the article by Becker et al. in the present Special Issue (35), and in Banff consensus publications (36,37).

The following is our proposal for alloimmune risk stratification in CDC-negative kidney transplantation.

• Transplantation performed with an HLA-identical donor carries a significantly lower immunological risk than transplantation from a donor of another HLA status.