Sensitization in Transplantation Assessment of Risk 2025 innate working group: The potential role of innate allorecognition in kidney allograft damage

Abstract

In solid organ transplantation, the alloimmune response is traditionally attributed to the action of alloreactive T cells that recognize mismatched human leukocyte antigens, as well as antibody formation and antibody-mediated rejection. However, recent evidence indicates that these paradigms of involvement of the adaptive immune system in organ transplant rejection do not explain all cases of graft inflammation and that innate cell allorecognition plays a role. This review, conducted by the innate team of the Sensitization in Transplantation Assessment of Risk workgroup, summarizes the concepts and empirical evidence supporting innate allorecognition. The focus is on natural killer cell activation via missing self and monocyte activation through the signal regulatory protein α-CD47 pathway and SIRPα gene polymorphisms. A consensus definition of genetic missing self is proposed, necessitating both donor and recipient human leukocyte antigen class I genotyping and evaluation of the recipient inhibitory killer-cell immunoglobulin-like receptor genotype. Although in vitro studies and preclinical validations corroborate the potential of innate allorecognition concepts, further research is required to establish clinical utility. This article delineated future research directions to bridge the gap between theoretical promise and practical application in clinical transplantation.

1. Introduction

Historically, the alloimmune response in solid organ transplantation has been attributed mainly to the action of alloreactive T cells recognizing mismatched human leukocyte antigens (HLAs). This explains why the vast majority of immunosuppressive drugs currently used in clinics were initially developed to target T cells. Although these drugs have contributed to the improvement of short-term graft survival, long-term graft survival has not improved accordingly. In addition, a strong correlation between pre-existence or de novo development of donor HLA-specific antibodies (DSAs) and inferior graft outcomes has been shown in several studies. However, recently, it has become evident that not all inflammatory phenotypes observed in human kidney transplantation can be explained by T cell activation or antibody-mediated allograft injury. Although microvascular inflammation (glomerulitis and peritubular capillaritis) is considered the hallmark of antibody-mediated injury via antibody-dependent cellular cytotoxicity and Fc gamma receptor 3A binding,^1–4 several studies indicate that cases with microvascular inflammation (MVI) are not always explained by detectable HLA-DSA and antibody-dependent cellular cytotoxicity.^5–14

To reflect these findings and to instigate further research on this topic, the Banff 2022 meeting proposed the creation of the “DSA-negative, C4d-negative MVI” category.¹⁵ Potential contributors to this phenotype include undetected DSA (eg, through antibody adsorption by the graft), alloreactive T cell-mediated responses, autoreactive or alloreactive non-HLA antibodies, ischemia-reperfusion injury, and primary (ie, DSA-independent) natural killer (NK) cell activation.^5,10,16,17

Interestingly, not only could MVI after kidney transplantation be heterogeneous in its cause. There is also evidence that tubulo-interstitial inflammation (the hallmark of T cell-mediated rejection [TCMR]) has a wide heterogeneity in the local immune cell composition.¹⁸ Some cases are primarily infiltrated by T cells, but other cases are mainly composed of monocytes/macrophages, and the latter population appears as important in relation to the severity of inflammation.¹⁹ The reasons for this heterogeneity are unknown but could be linked with the results of recent basic studies, indicating that innate myeloid cells are also capable of allorecognition.^20–22

In this manuscript, we discussed innate allorecognition by NK cells and myeloid cells in more depth and how it may impact graft rejection.

2. NK cells and missing self

2.1. The missing self theory

In 1986, Kärre et al²³ published a paper supporting the heretical notion that NK cells were inhibited by interacting with major histocompatibility complex (MHC) class I proteins. This seminal work demonstrated that syngeneic murine lymphoma cells selected for loss of MHC class I, which were therefore predicted to be more aggressive because they escape immunosurveillance by CD8⁺ T cells,²⁴ were less malignant, and that their elimination was due to NK cells. This work therefore predicted the existence of NK inhibitory receptors for MHC class I, which was later confirmed.^25,26 This investigation led to the concept of missing self-induced NK cell activation (Fig. 1).

Figure 1.

The missing self theory. Natural killer (NK) cells are educated on self-major histocompatibility complex (MHC) class I. Whenever a self-educated NK cell lacks ligation by the MHC molecule that resulted in the initial NK cell education, the NK cell will be activated. Created in BioRender. KIR, killer-cell immunoglobulin-like receptor.

Because allogeneic grafts express non-self MHC, it was logical to suspect that missing self could play a role in graft rejection. This hypothesis was rapidly proven true in the field of bone marrow transplantation by Bennet,²⁷ who demonstrated that an F1 hybrid mouse rejects parental bone marrow that lacks the expression of the MHC molecules of one of the 2 parental strains. However, this phenomenon, initially coined “hybrid resistance” (now commonly called missing self), remained ill-studied in rejection of solid organ grafts.²⁸

The first reason for the lack of interest among transplant immunologists in missing self is that hybrid resistance conflicts with Snell’s 3rd Transplantation Law,²⁹ which asserts that “skin from individuals of both parental strains transplanted onto members of the F1 hybrid generation is tolerated.” Indeed, due to the codominance of histocompatibility genes, the antigenic products of the different alleles of the 2 parental strains are part of the F1 hybrid’s alloantigenic makeup. The adaptive immune system of the F1 hybrid is thus unresponsive against the parental antigens. Indeed, the focus of interest of transplant physicians back in the 1980s was TCMR (which is characterized primarily by interstitial infiltration and epithelial injury). The dominant experimental model in transplant immunology was therefore skin grafting (the model used to establish Snell’s Transplantation Laws)²⁹ which presents evident advantages from a practical point of view and indeed adequately emulates TCMR. However, because the vessels of grafted skin are from the recipient origin (not from the donor as in transplanted organs), this skin graft model is inadequate to explore vascular rejection,³⁰ which, as a consequence, was largely overlooked at that time.³¹

Almost 2 decades after the initial studies by Bennett²⁷ in bone marrow transplantation, the first evidence that missing self can induce damage to the vasculature of grafted solid organs came from the experimental work by Madsen’s group³² who demonstrated the development of progressive arterial stenosis in a typical model of “hybrid resistance” (in which hearts from parental donors were transplanted into F1 hybrid recipients), and the contribution of NK cells to this vascular rejection.

More recently, starting from the opposite end of the cascade, Thaunat’s group hypothesized that the MVI lesions seen in renal graft biopsies in the absence of detectable DSA could be triggered by missing self-induced NK cell activation.¹⁴ They showed that this nonexceptional clinical situation (30%−50% of renal graft biopsies with MVI lesions) was indeed correlating with missing self¹⁴ (see below for details), a finding later confirmed in an independent and unselected observational cohort from Leuven.¹⁷ Finally, synergistic effects of DSA positivity with missing self on the risk for MVI were noted in both cohorts.^17,33

A third single-center cohort showed mixed results at first sight. In an initial side analysis that specifically focused on DSA-positive patients in a late phase posttransplant, no statistical association could be identified between missing self and MVI.³⁴ This can likely be explained by a lack of statistical power to disentangle the synergistic impact of missing self and antibody-dependent effects on NK cell activation and MVI in this highly selected and relatively small cohort, in contrast to the previous cohort studies.^17,33 In a larger, unselected cohort study in the same center, not restricted to DSA-positive patients, missing self was confirmed as an independent predictor of MVI, with similar effect size ratios as in the also unselected Leuven cohort.³⁵

To prove the causality of missing self in the generation of MVI lesions, the Thaunat group set up an in vitro model in which purified human NK cells were cocultured with glomerular endothelial cells and demonstrated that a single missing self molecule is sufficient to trigger the activation of the corresponding population of NK cells, resulting in endothelial damage.¹⁴ Finally, using a murine model of syngeneic cardiac transplantation, they demonstrated that the lack of MHC class I on graft vasculature was triggering the activation of primed NK cells, leading to the development of MVI lesions.¹⁴ Of note, the 2 experimental models formally excluded the possibility that the adaptive immune system of recipients (and in particular HLA and non-HLA DSA) could have contributed to the development of the MVI lesions.¹⁴

This conclusion that the adaptive immune system is not involved in the observed relationship between missing self and MVI is of particular interest. Indeed, because graft biopsies showing MVI lesions with or without DSA are indistinguishable at the transcriptomic level,^5,36 some authors have concluded that the 2 groups of patients with similar histo-molecular phenotypes should be merged and treated as being identical.⁵ Based on the previous discussion, it is clear that merging these patient groups could be a mistake. The observation that the DSA-positive and DSA-negative MVI share the same phenotype at the histologic and tissue transcriptome level reflects the fact that these types of rejection share the same final common pathway after NK cell activation. However, it is essential to recognize that the initial trigger may differ: DSA binds to Fc gamma receptors on NK cells in cases of antibody-mediated rejection (AMR), while NK cell activation through missing self occurs due to a lack of NK cell inhibition. Therefore, intravenous immunoglobulin and plasmapheresis, considered the gold standard therapy in AMR and aimed at reducing DSA titers, have no effect in patients with missing self-induced NK cell activation. Importantly, this situation also illustrates the limitations of biopsy-based transcript analysis,³⁷ which currently gains traction in the Banff community, despite its inability to determine the cause of MVI lesions.³⁸

2.2. Genetic prediction of missing self

Both the seminal work¹⁴ and the validation cohort studies^17,35 that have linked missing self to the presence of DSA-independent MVI on graft biopsies relied on simplified definitions of missing self based on (1) genetic analyses and (2) focusing solely on the compatibility between educated inhibitory killer-cell immunoglobulin-like receptors (KIRs) expressed on recipient NK cells (see below) and HLA class I of the donor.

Inhibitory KIRs have 2 (2D) or 3 (3D) extracellular immunoglobulin domains, a transmembrane domain, and a long (L) cytoplasmic domain containing 2 immunoreceptor tyrosine-based inhibitory motifs (Table).³⁹ Each inhibitory KIR binds a subgroup of HLA class I allotypes. The number of extracellular domains (2 or 3) gives them specificity for the HLA-C or HLA-A/-B allotypes, respectively.⁴⁰ KIR2DL1 recognizes HLA-C allotypes with 1 asparagine at position 77 and 1 lysine at position 80 (C2 allotypes), while KIR2DL2 and KIR2DL3 recognize those with a serine at position 77 and an asparagine at position 80 (C1 allotypes). KIR3DL1 recognizes the HLA-A and -B allotypes with the Bw4 motif (comprising 3 variables [77, 80, and 81] and 2 conserved residues [82 and 83]), and KIR3DL2 interacts with the HLA-A3 and A11 molecules. Although additional inhibitory KIR molecules have been described, the latter are not taken into consideration in the current definition of genetically predicted missing self. This is because KIR2DL4 recognizes the HLA-G (a nonclassical HLA-I with limited polymorphisms within the coding region), while the ligands of KIR2DL5 and KIR3DL3 are as yet unknown.

Table.

Killer-cell immunoglobulin-like receptors and their ligands.

Inhibitory receptors	Cellular ligands
KIR2DL1	HLA-C C2 allotype
KIR2DL2/DL3	HLA-C C1 allotype, HLA-B*46:01, HLA-B*73:01
KIR2DL4	HLA-G
KIR2DL5	Unknown
KIR3DL1	HLA-B Bw4 epitope, HLA-A*23, -A*24, -A*32
KIR3DL2	HLA-A*03, -A*11, HLA-F
KIR3DL3	Unknown

HLA, human leukocyte antigen; KIR, killer-cell immunoglobulin-like receptor.

Although an early study looking at KIR ligand mismatches between kidney transplant recipients and their donors showed an effect on long-term graft survival,⁴¹ merely focusing on KIR ligand mismatches is not sufficient to predict missing self.

To genetically predict the missing self, 3 steps are essential:

1. Since not all recipients have the genes encoding for all inhibitory KIRs, KIR genotyping of the recipient is the first step for identifying situations of missing self.

2. The second, essential step consists of determining which of the expressed recipient inhibitory KIRs are educated (ie, functional). Indeed, because the HLA locus is located on a distinct chromosome from KIR genes, HLA and KIR are inherited independently. NK cells, therefore, need to undergo a process of education in which autoreactive NK cells (because of the lack of expression of HLA class I ligands for their inhibitory KIRs) are rendered anergic (by chronic engagement of various activating NK cell receptors.⁴² Consequently, only NK cells expressing self-HLA class I-specific inhibitory KIRs are responsive when they encounter a cell not expressing its self-HLA class I ligand and should therefore be considered to determine missing self.

3. The third step for the definition of genetically predicted missing self consists in relating the list of recipients’ educated inhibitory KIR to HLA class I typing of the donor, which allows detecting all molecular conflicts where the recipient has an educated inhibitory KIR without expression of the corresponding ligand in the donor (ie, an HLA mismatch in the host-vs-graft direction). Of note, both in vitro models and population studies have demonstrated that the number of molecular conflicts (0-4 according to this method, see Table) correlates with the intensity of activating signal for the recipient’s NK cells.^14,17

2.3. Limits of the current definition of NK missing self

Although the genetic definition of missing self is informative at the population level and has been useful to identify a potential explanation for cases of MVI not explained by DSA, this approach has limitations when it is applied at the individual (patient) level:

1. The number of NK cells is highly variable between individuals, and the expression of inhibitory KIRs on NK cells is variegated (each receptor is expressed by a random subset of NK cells, and each cell can express between 0 and 4 different inhibitory receptors chosen in a relatively random fashion from the available gene repertoire). As a result, individuals may possess remarkable diversity in unique NK cell phenotypes.⁴³ In the transplant setting, the number of NK cells expressing the optimal combination of inhibitory KIRs to detect the molecular conflict with the graft is extremely variable between individuals with the exact same genetically predicted missing self.¹⁴

2. NK cells, even when expressing the relevant educated inhibitory KIR(s), do not react against endothelial cells lacking the expression of their ligand(s) unless they have been primed before. In the murine model, priming of NK cells was induced by danger signals, such as by induction of ischemia/reperfusion injury or by injecting CpG oligodeoxynucleotides (a TLR9 agonist mimicking viral infection).¹⁴ In line with these observations, viral (in particular cytomegalovirus) infections and longer cold ischemia time were statistically significantly associated with DSA-independent MVI lesions in the 2 cohorts.^14,17 Accordingly, 2 recipients with the same genetically predicted missing self and even the same number of reactive NK cells could have different histologic findings. Such results might underpin the observation that while 30% of donor-recipient pairs have a genetically predicted missing self, only a fraction of them develop MVI lesions.¹⁴ In this regard, genetically predicted missing self can be considered as a risk factor for DSA-independent MVI, similar to high low-density lipoprotein cholesterol being a risk factor for myocardial infarction.

3. Recent progress in the understanding of the biology of inhibitory KIRs suggests that the strength of the signal delivered to NK cells is influenced by polymorphisms of both KIR molecules and their ligands (eg, through effects on ligand affinity or signal transduction).^39,44 Similarly, KIR allelic polymorphism impacts KIR protein expression. This implies that there are situations in which NK cells could receive insufficient stimulation despite the apparent a lack of genetically predicted missing self. Furthermore, beyond inhibitory KIRs, other inhibitory receptors on the NK cell surface bind to HLA molecules and could be involved in the missing self. This includes the CD94/NK group 2A complex,⁴⁵ which binds to the complexes consisting of HLA-E and peptide derived from HLA class I signal sequences.⁴⁶

4. In theory, recipient NK cells could also be activated by graft endothelium in the absence of a deficit in inhibitory signals, in case of an excess of activating signals.⁴⁵ Several lines of evidence suggest that this situation is dominant in NK cell reactivity against tumor cells.⁴⁷ It is worth noting that NK cell education is also applicable to activating KIR, but in the opposite direction to inhibitory KIR, whereby the presence of a strong cognate ligand for an activating receptor leads to downregulation of NK cell responsiveness. These points may partly explain why up to 30% of patients with MVI in graft biopsies but without detectable DSA (neither anti-HLA nor non-HLA antibodies) have no genetically predicted missing self.¹⁴

2.4. Future studies on NK cell missing self allorecognition

Although there is ample evidence on the importance of NK cells in both antibody-dependent and independent acute and chronic kidney injury, literature on the impact of missing self-mediated NK cell allorecognition in transplant outcomes can sometimes be seen as conflicting.^{14,17,28,33,41,48–50} This, however, largely reflects the methodological heterogeneity of relevant studies, including KIR genotyping, assessment of NK cell education, and definition of clinical outcome measures (eg, rejection phenotypes). Conversely, as discussed above, it is also clear that the presence of a missing self is neither sufficient nor necessary for the development of microvascular injury.

To confirm the causal role and evaluate the impact of NK cell missing self allorecognition on transplant rejection and clinical outcomes, forthcoming studies should adopt consistent definitions and focus on key objectives:

1. Conduct recipient KIR genotyping and donor-recipient HLA genotyping to accurately define missing self according to the potential consensus definition (see above), which accounts for NK cell education in the recipient.

2. Provide a precise definition of spatial patterns of graft injury/rejection, such as MVI and tubulointerstitial inflammation.

3. Document the presence of other mechanisms of NK cell activation, since synergy between missing self and anti-HLA DSA on NK cell activation and rejection outcomes has been described.^17,33

4. Validate findings from association studies in external and more heterogeneous transplant cohorts, as current clinical data come from single-center European cohorts with relatively homogeneous Caucasian populations and, in general, lower-risk transplantations.

5. Evaluate whether there is added value in evaluating missing self in specific clinical contexts, for example, for risk stratification (at time of transplantation) or diagnostic purposes (at time of posttransplant detection of MVI or other inflammation patterns).

6. Leverage omics approaches—such as single-cell RNA sequencing, spatial transcriptomics, and proteomics analyses—to dissect the phenotypic and functional diversity of intragraft NK cells. These technologies can help identify and characterize NK cell subsets potentially involved in MVI and vascular injury.

However, while these approaches are powerful, they primarily reveal correlations. To advance the field, such insights must be integrated with experimental models capable of establishing causality and unraveling the molecular mechanisms driving innate allorecognition. This translational strategy will deepen our understanding of innate immune processes in allograft pathology and ultimately support the development of novel therapeutic approaches.

Beyond this immediate objective, many key questions remain to be addressed to better define the larger concept of NK cell allorecognition in the context of solid organ transplantation. Future studies needed include:

• Investigate the impact of KIR polymorphism, copy number variation, KIR protein expression levels (and number of expressed receptors), and KIR ligand interaction strength on NK cell effector function in the setting of transplantation.

• Investigate the relative contributions of missing self vs other mechanisms of NK cell activation to clinical and histologic phenotypes. In particular, the balance between inhibitory vs activating NK cell KIR receptors and NK cell phenotypic diversity on NK cell activation and effector function at steady state and under conditions of graft stress.

• Investigate whether therapeutic strategies targeting NK cells lead to improved clinical outcomes (eg, abrogation of MVI).

3. Monocyte involvement in allograft rejection

Recipient monocytes and monocyte-derived cells (macrophages and dendritic cells) have long been suspected to contribute to allograft rejection in experimental animals and humans. In humans, macrophages are a prominent, if not the major, cell population detected in biopsies of renal allografts showing acute rejection.⁵¹ Macrophage and dendritic cell density in human kidney allograft tissue strongly correlates with graft fibrosis and poor long-term allograft survival.^52–55 In line with this, profoundly T cell-depleted renal transplant patients often present with early allograft dysfunction (within the first month) before any significant T cell reconstitution has taken place, and in these patients, the biopsy is dominated by a monocytic infiltrate.⁵⁶ Recent transcriptional and spatial profiling of human renal allograft biopsies demonstrated a specific association between recipient-derived FCGR3A+ monocytes, FCGR3A+ NK cells, and the severity of intragraft inflammation.¹⁹ Interestingly, activated FCGR3A+ monocytes overexpressed both CD47 and leukocyte immunoglobulin-like receptor (LILR) genes and showed increased paracrine signaling pathways possibly involved in the promotion of T cell infiltration in TCMR cases. In contrast, in AMR, there was stronger enrichment of FCGR3A+ NK cells in the neighborhood of FCGR3A+ monocytes, indicating a differential impact of these monocytes in different contexts. Finally, a recent comprehensive meta-analysis of transcriptomic data from multiple solid organ transplants has highlighted that differential enrichment in myeloid cell subsets (monocytes, macrophages, and dendritic cells) underpins pan-organ molecular markers relevant to acute rejection.⁵⁷ The role of CD47 in innate allorecognition will be discussed in more detail below. LILRs are broadly expressed on the monocytic lineage and NK cells and can interact with (isoforms of) HLA class I.^58,59 Although it is known that LILRs can have both inhibiting and activating effects, their role in innate allorecognition, or their regulation thereof, is not yet clearly defined, warranting further research. Recent data underscore the potentially significant role of LILRs in immune risk following kidney transplantation. For instance, single-nucleotide polymorphisms in LILRB3 have been associated with poor posttransplant outcomes in African American kidney transplant recipients.⁶⁰ Moreover, growing evidence highlights the contribution of ischemia-reperfusion injury and infections to innate immune responses in monocytes and macrophages—phenomena collectively referred to as “trained immunity.” This further emphasizes the importance of these cells in posttransplant inflammation, not solely restricted to allorecognition, as recently reviewed.⁶¹

3.1. Innate allorecognition and SIRPα polymorphism

The pioneering work by Lakkis’ group over the last 15 years has established that the murine innate immune system can distinguish between self and allogeneic non-self, independently of the adaptive immune system and innate lymphoid cells (including NK cells).^{20–22,62–64} In contrast with recipient’s NK cells, which sense the absence of self-MHC class I molecules on the surface of graft cells,¹⁴ monocytes discriminate between self and allogeneic non-self by both MHC-dependent and independent mechanisms.^21,62 Upon allorecognition, monocytes differentiate to antigen-presenting dendritic cells²⁰ and activated macrophages, endowed with direct allocytotoxic functions.⁶³

A key non-MHC ligand-receptor pair that mediates innate allorecognition in the mouse is signal regulatory protein alpha (SIRPα)-CD47.²¹ SIRPα is a polymorphic immunoglobulin super-family protein exclusively expressed in the immune system on myeloid cells, but can also be induced on kidney stromal cells.⁶⁵ Amino acid polymorphism in the immunoglobulin V (IgV) extracellular domain of mouse SIRPα modulates binding to its monomorphic ligand CD47 on monocytes.^21,66,67 Mismatch between donor and recipient SIRPα triggers monocyte activation by disturbing the balance between activating and inhibitory signals mediated by CD47 and SIRPα, respectively²¹ (Fig. 2). If self is encountered, the interaction between CD47 and self-SIRPα is balanced—the interaction at both sides has equal intensity, and the equal signals cancel each other out. In contrast, transplanting an allograft expressing a mismatched (non-self) SIRPα that has higher affinity for CD47 than the recipient’s self SIRPα upsets the balance in favor of the stimulatory signal, causing the recipient’s monocyte differentiation into mature interleukin-12-producing dendritic cells that drive T cell proliferation, interferon gamma production, and rejection.^20,21

Figure 2.

The CD47–SIRPα pathway. In physiological situations, there is a balance between reciprocal CD47–SIRPα ligation, resulting in an inhibitory effect. Whenever a higher affinity CD47–SIRPα interaction occurs (eg, in transplantation), monocytes can be activated to differentiate into mature, interleukin-12-producing dendritic cells. Created in BioRender. SIRPα, signal regulatory protein α.

The human orthologs of mouse SIRPα and mouse CD47 are human SIRPα and human CD47. As in mice, human SIRPα is polymorphic, while CD47 is monomorphic. Recent analysis of ~5000 publicly available human genomes for SIRPα single-nucleotide polymorphisms that translate to amino acid polymorphisms in the CD47-binding IgV domain of the SIRPα protein identified 13 haplotypes that account for ~90% of variation in the human population,⁷⁹ consistent with a smaller published data set.⁶⁶ Absence or presence of a 3-nucleotide deletion that eliminates proline at position 99 and causes an aspartate-to-proline switch at position 100 of the IgV domain FG loop that interacts with CD47 divided the top haplotypes into 2 groups: v1-like and v2-like.^68,69 A polymerase chain reaction genotyping method devised by Jar-How Lee (One Lambda, Inc) that determines whether an individual is homozygous for A (v1-like) alleles (low affinity for CD47), homozygous for B (v2-like) alleles (high affinity for CD47), or heterozygous has been validated by several groups.⁶⁸ Genotype distribution in the general US population is approximately 40% v1v1, 45% v1v2, and 15% v2v2.

3.2. Human studies of SIRPα mismatch

Two published studies have found an association between SIRPα mismatch and outcomes in kidney and hematopoietic stem cell transplantation. Garcia-Sanchez et al⁶⁹ found in a small cohort of HLA-identical, living donor renal transplant recipients (n = 55) a trend toward greater 10-year graft failure (estimated glomerular filtration rate <30 mL/min/1.73 m², return to dialysis-dependence, or retransplantation) in SIRPα-mismatched patients (P = .06 vs matched patients). In an allogeneic hematopoietic stem cell transplantation study (n = 350), Saliba et al⁶⁸ observed an association between chronic graft-versus-host disease and SIRPα mismatch (P = .03). Next to the small sample sets and lack of statistical power, another main limitation of these studies is that analysis of monocytes (or monocyte signatures) in patients was not performed. A recent large-scale study involving 455 kidney transplant recipients, supported by an independent validation cohort of 258 patients, found that SIRPα mismatch is significantly associated with a higher incidence of acute rejection and graft fibrosis during the first year posttransplant.⁷⁹ Moreover, recipients with the A allele who received kidneys from B allele donors experienced reduced long-term graft survival. These patients also displayed an activated monocyte phenotype. Collectively, these clinical findings further substantiate the preclinical research, highlighting the potential impact of SIRPα compatibility on transplant outcomes.⁷⁹

3.3. Recommendations on how to address future SIRPα-CD47 studies in human studies

Additional independent large epidemiologic studies are needed to substantiate the effect of CD47-SIRPα in the human solid organ transplantation setting and establish whether this non-HLA mismatch has the potential to risk-stratify renal allograft recipients with respect to the development of graft fibrosis and long-term allograft loss. These future studies will have to include deep phenotyping of peripheral blood and graft-infiltrating monocytes to further support the association between the genetic mismatch and the events taking place within the graft.

4. The precision medicine pipeline: From preclinical models to analytical validity, clinical validity, and clinical utility

Both the missing self as a cause of NK cell activation in allogeneic organ transplantation and monocyte activation through SIRPα mismatches are potentially of clinical interest. However, even if further validated in additional human cohort studies, epidemiologic/statistical validation of sound hypotheses does not translate directly into clinical utility.

Insights into innate allorecognition have broad potential across the spectrum of precision medicine—from immune risk stratification and enhanced diagnostics to improved prognostication and even the prediction of therapeutic outcomes. To move toward clinical utility, the following aspects need to be addressed in further validation studies:

• Which clinical scenario would benefit from information on missing self or SIRPα mismatches (context of use)?

• How relevant is the measurement of missing self or SIRPα mismatches for clinical decisions?

• When would the measurement of missing self or SIRPα mismatches be best evaluated? At baseline for immune risk stratification, at the time of posttransplant rejection/inflammation, or to guide therapeutic decisions?

• What is the quantifiable added value of data on missing self or SIRPα mismatches on top of currently available standard-of-care information?

• Would information on missing self or SIRPα mismatches influence therapeutic decisions?

• Does having information on missing self or SIRPα mismatches improve outcomes?

The latter questions are becoming increasingly relevant. Although currently used immunosuppressive drugs like corticosteroids, mycophenolate, calcineurin inhibitors, mammalian target of rapamycin inhibitors, etc, impact innate immune cells through various mechanisms,⁷⁰ better-targeted therapies are entering the field. Recent data indicate that NK cell depletion by anti-CD38 therapies could become attractive for treating HLA-DSA-positive AMR.^71–73 Whether cases of MVI that are explained by missing self could equally benefit from such therapies seems a reasonable assumption based on data from the murine preclinical model,¹⁴ but still needs to be validated in the clinic.⁷⁴ In such a context, it remains unstudied whether evaluation of the missing self could be a predictor of therapy response or whether evaluation of the inflammatory cell composition could be more relevant as a companion diagnostic for clinical decision-making.

All these questions become very important and should be answered by appropriately designed intervention studies and clinical trials. In parallel, analytical validity and quality control are necessary before assays/testing for missing self or SIRPα mismatches are implemented in clinical-grade diagnostic settings. Finally, the extensive and ongoing research in oncology harnessing NK cells⁷⁵ and macrophages⁷⁶ to enhance antitumor responses is expected to also inform our understanding of pathobiology and potential therapeutic targets in allorecognition and transplantation. For example, 2 recent studies in tumor immunology identified missing self-induced NK cell activation⁷⁷ and SIRPα/CD47-mediated modulation of monocyte activation⁷⁸ as key molecular pathways in innate immune responses against tumors. These mechanisms can be viewed as the mirror image of what transplant immunologists aim to achieve—namely, the prevention or modulation of innate immune activation in the context of allografts.

5. Conclusion

In conclusion, in recent years we have seen a great body of research supporting the role of innate immune cells (NK cells and monocytes) not just in the downstream immune activation upon allorecognition by the adaptive immune system, but also as key players involved in the upstream mechanisms of allorecognition. Although the knowledge on this topic is advancing rapidly, more data are needed before these concepts can be translated into routine clinical practice. Most importantly, it is believed that clear definitions of, for example, missing self, and larger multicenter cohort studies will shed more light on the true clinical relevance of these concepts, particularly in relation to established immune risk stratification (eg, about HLA mismatching and HLA antibody profiling), which is currently focused on the adaptive immune system.

Abbreviations:

AMR

antibody-mediated rejection

DSA

donor HLA-specific antibody

HLA

human leucocyte antigen

IgV

immunoglobulin V

KIR

killer-cell immunoglobulin-like receptor

LILR

leukocyte immunoglobulin-like receptor

MHC

major histocompatibility complex

MVI

microvascular inflammation

NK

natural killer

SIRPα

signal regulatory protein alpha

TCMR

T cell-mediated rejection

Data availability

This review paper does not include original research data.