For screening to be successful, one should have an understanding of how underlying pathological processes can develop into overt graft dysfunction.

In case of ABMR, the screening marker itself seems to be implicated in the underlying pathological process [6, 12]. This is apparent with pretransplant DSA, considering the high risk of hyperacute rejection if transplantation proceeds despite a positive CDC-crossmatch. Modern practice precludes such transplantation with pretransplant listing of non-acceptable HLA antigens, or with measures such as paired kidney exchange programs or desensitization in the living donation setting. In contrast, CDC-crossmatch negative pretransplant DSA, which are identified through SPI only, are not necessarily a contraindication to transplantation in patients faced with no alternatives beyond dialysis [3]. However, these DSA still convey increased risk for ABMR and allograft loss according to a meta-analysis by Mohan et al. [18] Recent analysis of CTS data indicated that nearly 15% of recipients of deceased donor kidneys with crossmatch negative pretransplant DSA progressed to allograft failure within the first year post-transplant [19]. This figure was even higher in retransplant patients. In regards to dnDSA, a large meta-analysis by Sharma et al. [20] implicated development of dnDSA as a severe risk factor for notably cellular rejection, acute ABMR (aABMR), chronic ABMR (cABMR), and allograft loss. Moreover, CTS data showed that 20% of patients who developed dnDSA in the first post-transplant year progressed to allograft failure within the next five [19]. A recent randomized trial corroborated these results [21].

The genesis of DSA after transplantation is a complex process. B-cells can initiate and subsequently differentiated plasma cells (as well as B-cells) can maintain production of these antibodies as a result of sensitization of the adaptive immune system. Sensitization could be related to a period of underexposure, either due to non-adherence or iatrogenic reduction of immunosuppression [22–28]. Additionally, poor HLA matching [28–31] and previous episodes of T-cell mediated rejection (TCMR) [29, 30, 32–34] have been associated with DSA development. Other risk factors pertain to certain recipient characteristics such as age or ethnicity [35, 36]. The association of previous TCMR and dnDSA development is hypothesized to be explained by sensitization of the B-cell compartment through inflammation induced by T-cell alloimmunity, especially T-follicular helper cells [37–39]. The role of T-cells in a process which could ultimately lead to ABMR seems to question the dichotomous view on rejection (i.e., either TCMR or ABMR as separate entities). Perhaps a more contemporary view on rejection is that it is a heterogenous spectrum with different histological and clinical manifestations [40].

While the process of sensitization leading to DSA formation is complex and multifactorial, the risks DSA convey are clear. Still, this does not necessarily infer a causal relationship. Though the pathogenicity of HLA-DSA was extensively studied in recent years and a recent thorough literature review by Callemeyn et al. [40] attempted to untangle association from causation. This review assessed the possible causal relationship between HLA-DSA and microvascular inflammation (MVI), a histopathological hallmark of ABMR, through the Bradford-Hill criteria, which can be used as guide for causal inference in epidemiological research. These criteria include: strength of effect size and reproducibility, experimental evidence in vitro and in vivo, temporality between HLA-DSA appearance and graft injury, biological gradient, and coherence and analogy [40, 41]. Callemeyn et al; [40] illustrates that most criteria are met. However, more investigations are warranted to demonstrate a clear biological gradient between antibody titre and occurrence of ABMR or graft failure; [42]. Yet, recent studies by Viglietti et al; [43, 44] showed that treatment of ABMR through plasmapheresis (PP), intravenous immunoglobulins (IVIG) and rituximab is associated with a significant decrease in DSA MFI and capacity to bind C1q. Interestingly, these reductions in DSA properties were significantly associated with improved graft survival in patients with ABMR. However, treatment effects on more chronic or late ABMR are variable [15]. Furthermore, the histological presentation of ABMR including MVI is not always specific for antibody involvement, as other causes could appear clinicopathologically similar. Nonetheless, there seems to be clear preclinical and clinical evidence of a pathogenic relation between HLA-DSA and ABMR.

Despite this strong relationship, not all recipients with preformed DSA or dnDSA seem to progress to ABMR or graft failure [16, 29, 45]. Multiple mechanisms have been proposed to explain this variation in effect of HLA-DSA on graft outcomes. A recent comprehensive review has summarized HLA-DSA attributes and discussed mechanisms of HLA-DSA-induced effector functions in mediating allograft damage [46]. These effector functions may be Fc-dependent, such as the impact of antibody glycosylation status on complement activation and recruitment of cytotoxic NK-cells and macrophages [47, 48]. Regarding Fc-independent mechanisms, recent studies describe intracellular signalling downstream of HLA-antibody binding to endothelial cells that promote upregulation of adhesion molecules, proliferation and activation of endothelial cells, induction of dendritic cells and CD4⁺ T-cell maturation [46, 49, 50]. HLA-antibody ligation of the HLA-molecule of endothelial cells can also lead to anaphylatoxin production that can result in more monocyte recruitment. Recruitment is also mediated by the cellular expression of anaphylatoxin receptors on CD4⁺ and CD8⁺ T cells, and myeloid cells [51, 52]. Lastly, regulatory T and B-cell populations may play a pivotal role in suppressing the deleterious effects of DSA on the graft. Recent research indicates that these cell lines impart tolerogenic effects through impairment of the T-follicular helper cell – B-cell interaction and that these regulatory cells were significantly reduced in frequency in patients with DSA who developed ABMR, as compared to patients with DSA but absent ABMR [38, 39, 53].

Several studies have shown that high titre HLA-DSA reflected by high MFI levels, and inflammatory isotype switching toward IgG1 and IgG3, and thus their capacity to bind C1q or C3d are associated with significantly increased microvascular inflammation and C4d deposition [54, 55]. Although considered classically non-inflammatory, the IgG4 isotype has been associated with subclinical graft rejection, including ABMR, in several studies in kidney and other solid organ transplants [55, 56]. Subclinical ABMR was shown to lead to significantly more transplant glomerulopathy and accelerated graft loss when compared to subclinical TCMR [57]. In addition to subclinical ABMR, HLA-DSA have been shown to be significantly associated with kidney graft fibrosis and subsequent accelerated graft loss [58]. The relationship between HLA-DSA and graft fibrosis was independent of previous ABMR episodes. Thus, HLA-DSA, even detected at low strength/MFI, are associated with subclinical damage and fibrosis independent of clinical ABMR occurrence [58, 59].

Lastly, it must be mentioned that not all patients with MVI, indicative of injury attributed as being “antibody-mediated” by current Banff criteria, have detectable levels of HLA-DSA. The histopathologic entity of MVI without detectable HLA-DSA by definition suggests that factors other than HLA-DSA may mediate MVI, such as non-HLA antibodies [60–63]. Antibody-independent pathways may include NK-cell alloimmunity through a “missing-self” mechanism [64, 65] or direct allorecognition by monocytes [66, 67]. Other causes may not even be related directly to alloimmunity, such as recurrent complement-mediated renal disease, ischemia/reperfusion injury, or viral endothelial infection.

The above presented body of evidence illustrates that there are likely multiple individual pathways, not all of which are fully understood, that eventually lead to varying levels of microscopic injury that is currently defined as MVI and ABMR, which may need some clarification [6, 68]. Nonetheless, regardless of the incompletely understood natural history of ABMR and MVI, there is a large amount of preclinical and clinical evidence warranting strong support for the notion that anti-HLA-DSA are significantly associated with, and predictive of rejection and clinicians should be aware of this [5, 12, 15, 40].

For a valid screening strategy for disease, clinicians should also be able to identify a latent stage. In some cases, patients undergoing rejection present with clinical dysfunction of the graft as first sign. However, most have a latent phase with prevalent DSA prior to developing graft dysfunction. The first evidence of subclinical DSA as a marker for latent rejection came from preclinical studies in a non-human primate model with sequential protocol biopsies by Smith et al. [69, 70] They showed that development of dnDSA generally precedes graft dysfunction, as well as C4d-deposition or transplant glomerulopathy. Clinical studies of longitudinal protocol biopsies in stable renal transplant recipients with preformed DSA show substantial oscillations characterized by fluctuations in HLA-DSAs, C4d deposition and scores for glomerulitis and/or capillaritis in a dynamic and multidirectional fashion [71–73]. Seminal papers by Wiebe et al. [22, 34] have shown that this progressive subclinical injury can also be detected in patients with dnDSA several years after kidney transplantation. They found that of 64 patients who developed dnDSA, the majority was without graft dysfunction. Additionally, development of subclinical dnDSA was independently associated with transplant glomerulopathy (and thus chronic ABMR (cABMR)), decline in graft function, and allograft loss. Therefore, it is unlikely that chronic rejection is the result of a single spike of HLA-DSA or a single episode of ABMR. Instead, it represents a dynamic process that continues, unabated, at varying levels and eventually progresses towards chronic allograft injury, graft dysfunction and ultimately graft loss [12].

Development of latent rejection in patients with subclinical DSA has been observed in other types of organ transplants [74–76], as well as in more recent clinical studies in KTR, which show underlying rejection in roughly half of overall patients (Table 4). Bertrand et al. [77] recently analyzed 123 patients with subclinical dnDSA in a French multicenter cohort study and found that 41.5% of these patients had subclinical ABMR. Loupy et al. [57] showed in a large prospective cohort study of 1,001 patients with 1 year protocol biopsies that of 256 patients with subclinical DSA, 55% had ABMR. Of these cases, 78% were related to pretransplant DSA, further indicating that both pretransplant DSA and dnDSA can underlie a latent pathological process. Coemans et al. [78] recently studied longitudinal protocol and indication biopsies in a single-centre cohort of 1,000 Belgian patients. Of these, 108 had pretransplant DSA and 47 developed dnDSA. The prevalence of subclinical aABMR in protocol biopsies at 3, 12, 24 and 60 months post-transplant was 42.5%, 40.5%, 37.3% and 13.3% respectively in patients with HLA-DSA. Prevalence of transplant glomerulopathy increased over time and this was associated with previously diagnosed aABMR, further corroborating the notion that ABMR is a dynamic and continuous process [71, 72]. Schinstock et al. [30] retrospectively analyzed a single center cohort of patients with serial surveillance biopsies, but also included biopsies at graft dysfunction and upon subclinical dnDSA development. They found that of the 40 patients who were biopsied at the time of dnDSA development, 25%, 7.5% and 20% had underlying aABMR, cABMR, and TCMR respectively. Yamamoto et al. [79] reported on a Japanese cohort of 43 patients with subclinical dnDSA and found that 41.8% of patients had ABMR. Parajuli et al. [80] showed in an American retrospective single center cohort study with biopsies in case of dnDSA development or clinical indication that of 29 patients with subclinical dnDSA, 15 (51%) had underlying rejection. Of those rejections, 60% were ABMR, 20% mixed rejections, and 20% were TCMR. Waldecker et al. [81] retrospectively studied 84 German patients with indication biopsies or in case of dnDSA development from a single centre and found that out of 50 patients with subclinical dnDSA, 44% had ABMR, 15% had TCMR, 12% had mixed rejection and 15% had borderline rejection. Notably, only 14% of these patients had no histopathological signs of rejection at light microscopy. Eskandary et al. [82] retrospectively reported on the screening process for the BORTEJECT study, whereby 861 patients with stable grafts were cross-sectionally screened for presence of DSA [82]. Of 86 patients with subclinical DSA, 44 (51%) met the Banff criteria for ABMR. Lastly, Cornell et al. [83] analysed the results of a prospective trial on pretransplant desensitization with eculizumab in patients with a positive flow-crossmatch and compared the long-term outcomes to a historical matched cohort. The overall prevalence of subclinical ABMR at 3 months, 1 year and 2 years post-transplant was 41.8%, 37% and 20% respectively in a total of 78 patients.

While most studies only reported these biopsy results in terms of either positive or negative for ABMR, the studies by Schinstock et al. [30], Parajuli et al. [80] and Waldecker et al. [81] interestingly further show that subclinical dnDSA can also be a signal for underlying TCMR. Unfortunately, no biopsies were performed in a DSA-negative control group in these studies, making it difficult to ascertain the precise odds of dnDSA to signal TCMR risk. Nonetheless, as discussed above, the association between TCMR and dnDSA development has been described previously in multiple studies. The study by Loupy et al. [57] seems to contrast this suggested association, as they showed that significantly more patients without DSA developed subclinical TCMR compared to patients with DSA. However, the majority of patients with DSA in this study had preformed DSA, not dnDSA. This was also reflected in patients with subclinical TCMR, as only 5 out of 17 (29%) of these patients had dnDSA. This could imply that there is no association between preformed DSA and TCMR, while there might be one for dnDSA and TCMR. Others were also not able to relate pretransplant DSA and TCMR development [84, 85]. However, these studies did not take into account the possible presence of mixed rejection in these analyses, or pooled these type of rejections with patients with ABMR. The study by Coemans et al. [78] did adjust their analysis accordingly and found that while pretransplant DSA was not associated with isolated TCMR, it was associated with total TCMR including mixed rejections. Nevertheless, the results of these studies might be further evidence for the previously described view on rejection as a spectrum of different clinical and histological manifestations. These may occur in sequence, as TCMR can result in dnDSA development, which can lead to ABMR. In contrast, preformed DSA may not necessarily lead to development of isolated TCMR. Collectively, these results imply that a biopsy serves to diagnose latent rejection (ABMR, TCMR or mixed rejection) in around half of patients with subclinical DSA, which is in line with previous recommendations on performing an allograft biopsy these patients [15]. While the number of papers warrants strong support for this statement, the evidence is mainly observational. This is reflected in the grading.

Aside from the diagnostic purposes of an allograft biopsy in patients with subclinical DSA, it may also have prognostic value in predicting allograft loss. As stated before, not all patients who develop DSA seem to lose their graft or even show declining allograft function. Wiebe et al. [22] showed that 50% of patients still had a functioning allograft 10 years post-detection of the subclinical dnDSA. Therefore, further risk stratification regarding the pathogenicity of these subclinical dnDSA seems necessary. The Banff classified rejection diagnosis in these patients may provide further risk-stratification for detrimental transplant outcomes. Multiple studies showed that KTR with subclinical DSA and histological evidence of ABMR had significantly worse allograft survival rates and allograft functional decline than those without histopathological rejection at light microscopy [30, 57, 77, 78, 86]. Others show a similar trend, albeit not statistically significant [79, 82]. Moreover, Bertrand et al. [77] and Loupy et al. [57] showed that patients with subclinical dnDSA but without ABMR had excellent 8 years allograft survival (>90%) and stable graft function. This suggests that the rejection diagnosis could be prognostically more important for graft outcome than the dnDSA status itself in patients without graft dysfunction. This was further corroborated by Parajuli et al. [87] in a cohort of 587 patients without rejection at an initial protocol or indication biopsy, whereafter there was no difference in 5 years allograft loss rates between DSA-positive and DSA-negative patients. Though, dnDSA positivity in patients with a negative index biopsy was associated with subsequent rejection.

Beyond the prognostic value of the Banff classified rejection diagnosis in patients with subclinical DSA, the severity of the Banff recognized acute pathological lesions may perhaps offer further risk-stratification in patients with rejection. A prospective cohort study of 215 patients showed that while DSA were univariately associated with renal function decline, this was no longer statistically significant when analyzed with a multivariate model including MVI and tubulitis [88]. This suggests that the presence of histological markers may define a more severe phenotype in patients with DSA. Wiebe et al. [22] showed in multivariate analysis of a small subcohort of 23 subclinical patients with dnDSA that tubulitis was a strong predictor for allograft loss. Studies by De Kort et al. [89] and the iBox study by Loupy et al. [90] elegantly showed that increasing levels of MVI severity score in patients with DSA is independently associated with worse allograft survival. A more recent study on semi-supervised clustering through data-driven mathematic modeling by Vaulet et al. [91] further corroborated the prognostic value of inflammation activity as determinant of allograft loss within patients with DSA.

Histological chronicity markers also impact allograft survival. The presence of transplant glomerulopathy is implicated as independent risk factor for allograft attrition in multivariate analysis by multiple studies [22, 78, 90, 92]. Other chronicity markers such as interstitial fibrosis or tubular atrophy also seem to independently infer risk for allograft loss in patients with DSA [90]. More recent research showed that patients with increasing chronicity scores as determined by an aggregate of several chronicity markers have significantly worse prognosis in terms of allograft survival [93]. A follow-up study by Vaulet et al. [94] validated these results, again through a semi-supervised clustering approach, and found that clusters with higher levels of chronicity were associated with increasingly higher rates of allograft loss. Importantly, they assessed the impact of time since transplant of the biopsy in this study. Even though there was an association with clustering based on chronicity, clustering solely based on time of biopsy could not discriminate in allograft loss rate. This indicated that pattern of chronicity scores was an independent risk factor for poor allograft survival, regardless of the post-transplant time of the biopsy.

Even though activity and chronicity are viewed as separate entities for the sake of the analyses in these studies, it should not be forgotten that they are intertwined. The temporal association between aABMR and chronic lesions associated with cABMR such as transplant glomerulopathy, peritubular capillary basement membrane multilayering and transplant arteriopathy is well described in both preclinical models and clinical studies [57, 69, 70, 78, 95–97]. Previous research in patients with TCMR has shown that chronic scarring is a determinate for poor response to treatment [98, 99]. Haas et al. [100] has also previously shown that early intervention in patients with ABMR may prevent chronic lesions such as transplant glomerulopathy. Recently, Wu et al. [101] did not observe any effect of current treatment options in patients with chronic ABMR and transplant glomerulopathy further indicating that late stages are less responsive to therapy. It could thus be of interest to identify patients at an earlier stage of the ABMR disease process. If a latent phase with subclinical DSA is an earlier stage in the continuum of rejection, then a biopsy taken at this stage may theoretically show less chronicity and these patients could perhaps be more amendable to treatment. Unfortunately, very few studies investigated this. Parajuli et al. [102] found that the Banff sum chronicity and transplant glomerulopathy scores of patients with underlying ABMR at biopsy were significantly lower in subclinical ABMR compared to dysfunctioning allografts with ABMR. Additionally, Wiebe et al. [34] found that no patients with subclinical dnDSA had evidence of transplant glomerulopathy. However, more research is needed to confirm this hypothesis, as current data is insufficient to draw meaningful conclusions.

While this body of evidence seems to point out the additional prognostic value and possible clinical utility of an allograft biopsy in patients with (subclinical) DSA, it does not necessarily provide direction on when to perform this biopsy within the subclinical stage. The study by Schinstock et al. [30] clearly showed that within dnDSA-positive patients with a negative index biopsy, a follow-up biopsy 1 year later yields significantly more positive cases of ABMR. This appears to contrast aforementioned studies which suggested that an initial negative biopsy infers significantly less risk in patients who have developed dnDSA. However, the findings by Schinstock et al. [30] could be explained by the fact that not all their included biopsies at either dnDSA detection or follow-up were in fact in a subclinical setting. This may affect the a priori probability of finding underlying rejection at follow-up. Alternatively, the worse prognosis of the minority group who do eventually develop rejection may have been masked by the majority who remained without significant graft injury in other studies. Nevertheless, Schinstock et al. [30] could indicate that performing the biopsy too early could lead to false negative findings. Whereas no histopathological rejection at light or electron microscopy might be visible in these cases, there may in fact still be rejection at the molecular level. Previous research has shown the independent prognostic value of molecular ABMR gene transcripts for allograft attrition within patients with Banff classified ABMR [103]. The INTERCOMEX study has shown the added clinical value of these molecular gene expression transcripts for identifying rejection [104]. Two studies on molecular gene transcript classifiers further show that DSA-positive patients who present with high levels of these classifiers but show no histopathologic evidence of ABMR at light microscopy have more risk to develop histologic ABMR at subsequent biopsies compared to patients with low molecular ABMR gene transcript levels [105, 106]. Molecular analysis may thus offer additional prognostic value in case of a microscopically negative index biopsy, however these techniques are not yet available in most centers and require further validation. Some clinicians could perhaps consequently conclude that upon dnDSA detection, a certain amount of time should elapse before a biopsy is conducted, as this may decrease the chance of a negative biopsy result in those patients with molecular but not yet microscopic histological rejection. However, this may negatively impact potential efficacy of treatment in patients who would have more chronicity on later biopsies. Additionally, aforementioned studies on biopsy results upon subclinical dnDSA detection clearly show rejection in approximately 50% of cases. These would all be detected later by postponing a biopsy. An alternative strategy could entail a follow-up biopsy, if the index biopsy is negative. No study has been conducted which specifically addresses and compares the impact of these strategies on transplant outcomes. Therefore, more research on the optimal time of biopsy in patients with subclinical DSA is needed. Nevertheless, the additional prognostic value of renal biopsy information on both the Banff recognized rejection diagnosis and of the severity of the pathological lesions in patients with subclinical DSA seems clear. Therefore, despite the mostly single-center observational evidence, we strongly recommend to perform an allograft biopsy to further determine the pathogenicity and impact of developed subclinical dnDSA, if prognostication is desired.

A crucial element of a screening program is whether proper treatment exists and whether there is consensus on whom to treat. Optimization of maintenance therapy, which includes promoting adherence, reducing exposure to secondary risk factors such as hypertension and maintaining appropriate calcineurin inhibitor trough levels, has been recommended in previous consensus statements for the treatment of ABMR and TCMR [15]. Moreover, the consensus on managing modifiable risk in transplantation (COMMIT) workgroup addressed non-adherence and underexposure to immunosuppression as pivotal risk factors for poor transplant outcomes [140]. The importance of adequate exposure has also previously been demonstrated in patients with DSA. Multiple studies showed that DSA-positive patients with adequate exposure have better graft survival compared to DSA-positive patients who remain non-adherent or with iatrogenic underexposure to immunosuppression [22–24]. Development of dnDSA has been heavily correlated to underexposure to immunosuppression [22–28]. This risk factor for poor transplant outcomes can be addressed and this could be done irrespective of underlying histology, because dnDSA may still signal underexposure even if there is no microscopically visible rejection. However, the recently published OuTSMART trial, which analyzed the effects of optimization of maintenance therapy based on DSA monitoring results, seems to contradict these previous retrospective studies [21]. No significant difference was found in regards to graft survival between standard of care and optimization of maintenance therapy based on DSA monitoring. This randomized controlled trial (RCT) is qualitatively better evidence than observational research. However, it should be noted that the consenting participants in OuTSMART were already highly adherent at baseline. This is reflected by the low dnDSA incidence rate of 1.6% per year and may relate to the possibility of healthy survivor bias due to cross-sectional inclusion. Even though adherence improved significantly to even greater levels, it is uncertain whether it was to be expected that this should have resulted in improved graft survival. Nonetheless, this study does appear to show that broadening the immunosuppressive regimen does not have the expected effect on graft survival. Even the sensitivity analysis, which only included patients who were optimized to a triple therapy regime upon detection of dnDSA could not demonstrate survival benefit, though the confidence interval included both estimates of highly protective as well as highly hazardous effects. This could have been related to less allograft failure in DSA-positive patients than initially expected. Interestingly, total amount of biopsy-proven rejections was significantly lower in patients in the intervention arm, indicating that increased exposure does have immunological effect. Perhaps more benefit could be demonstrated if optimization of maintenance therapy is accompanied with biopsy-guided anti-rejection treatments as subclinical rejection was likely present in only 50% of subjects. More research in terms of broadening immunosuppressive regimen as a means of optimization of maintenance therapy is thus required for this to be recommended. Nevertheless, addressing non-adherence and secondary risk factors for progression are still important aspects of treatment, which we still strongly recommend in case of development of subclinical dnDSA. The ultimate goal is to optimize graft survival which includes taking into account competing mortality risk from infections, malignancies, and other toxicities.

When subclinical donor-specific antibodies emerge, it becomes crucial to detect potential non-adherence and optimize the maintenance immunosuppressive regimen, unless there are contraindications present. In case of signs of ongoing alloimmunity, the convention in many center is often to switch to triple therapy with tacrolimus, mycophenolate analogues and maintenance steroids balanced against toxic side effects. Unfortunately, there is a current lack of strong evidence for exposure targets in kidney transplant recipients with subclinical dnDSA. To give some clinical directions, target tacrolimus exposure could be extrapolated from trough levels to prevent (additional) DSA [25, 28] rejection [141], and to improve graft survival [142–144]. Collectively, these studies suggest that maintaining the tacrolimus trough level between 5 and 8 ng/mL, which is in line with international recommendations, might prevent alloimmunity and optimize survival, albeit two studies suggested a potential lower threshold of 4 ng/mL in patients with very low intrapatient variability [141, 142]. However, whether this target range is helpful once caABMR ensues remains unknown. A study by Sablik et al. [145] did not find any survival difference between a tacrolimus trough greater or lesser than 5.9 ng/mL. Interestingly, they did find that higher intrapatient variability was significantly associated with poorer survival in patients with caABMR, suggesting that adherence and time in therapeutic range are probably more important exposure variables than attained trough levels within current clinical practice. Even less evidence is available regarding optimal mycophenolate exposure. A small single study found a trough >1.3 mg/L to prevent DSA formation [146]. It is reasonable to hypothesize that a clear exposure-relationship curve between mycophenolate and antibody formation might exist, considering the almost linear relation between MPA exposure and SARS-CoV-2 antibody formation [147]. No evidence is available for reintroduction of low-dose steroids, it is however often assumed that the anti-inflammatory effects and the diminished chance of acute rejection from maintenance steroids might have beneficial effects in the long-term but need careful balancing against side-effects [148]. Some evidence has emerged regarding the effectiveness of conversion from a CNI based immunosuppressive regime to costimulation blockade with belatacept [149]. Perhaps optimization of maintenance therapy could entail such a strategy, as it would effectively eliminate occult non-adherence due to the necessity of intravenous administration. Additionally, belatacept’s immunological mode of action may be more fitted for patients who have already developed a dnDSA as it interrupts T-follicular helper cell—B-cell interaction and could thus decrease B-cell stimulation and further reduce the evolution of DSA formation [29, 150]. Some studies have shown effectiveness of belatacept on DSA levels and on the (lower) incidence of ABMR in sensitized patients [149–152]. Interestingly, DSA positivity was not associated with graft loss in a small cohort of patients converted to belatacept, though the presence of aABMR with MVI was independently associated with treatment failure [153]. It has to be noted however, that the incidence of TCMR was significantly increased, especially in patients converted within the first year post-transplant [154]. We therefore recommend more research to be conducted on the role of costimulation blockade as a means to optimize maintenance therapy in patients with subclinical DSA.

In regards to further treatment of patients with subclinical dnDSA before conducting a biopsy, evidence is lacking. Only one small cohort study has been identified, in which patients with subclinical DSA were treated with bortezomib, PP, IVIG and corticosteroids without performing a biopsy to confirm rejection [155]. This study showed that patients who achieved DSA clearance had more stable 2 year allograft function compared to those with persistent DSA. However, no control group was included and thus it cannot be concluded that improvement in outcome was due to treatment. Furthermore, irrespective of efficacy, subjecting all patients with subclinical dnDSA to such a strong and broadly targeting immunosuppressive regimen might be difficult to justify, considering that roughly half of this population have no underlying observable histological injury [30, 57, 78, 80–83]. In addition, transient spontaneous negativity of dnDSA has been observed in 24% of patients with subclinical dnDSA and complete clearance of dnDSA has been observed in around 10% of patients [29]. Lastly, identification of the Banff classified type of rejection through a biopsy will ensure that patients with underlying cell-mediated rejection are not unnecessarily subjected to therapy aimed at antibodies and vice versa. We therefore do not recommend additional preemptive treatment of patients with subclinical dnDSA, besides optimization of maintenance therapy, without performing an additional allograft biopsy.

Amongst dnDSA-positive patients with underlying rejection, those with subclinical TCMR may have the best evidence for gained benefit. Treatment of subclinical TCMR has been investigated in multiple studies (Table 5). A literature review by Mehta et al. [156] revealed that most available studies [157–161] at the time showed that subclinical acute TCMR (aTCMR) is associated with inferior outcomes. Choi et al. [162] observed significantly lower 10 years allograft survival in patients with untreated early subclinical TCMR vs. non-rejectors (62.3% vs. 96.2%). Consequently, ESOT advocates subclinical aTCMR to be considered as primary efficacy endpoints in clinical trials [163]. The first evidence of treatment came from a randomized trial by Rush et al. [161]. They showed that treatment of early subclinical TCMR detected in protocol biopsies leads to lower chronicity scores, less late rejections and more stable and lower creatinine levels at 2 years post-transplant than untreated patients. Another RCT by Kurtkoti et al. [160] showed similar results in regards to lower creatinine levels at 6 and 12 months. These older studies could be criticized for having been conducted before the tacrolimus era and thus being less applicable to current practice. A more recent randomized trial of early protocol biopsy and treatment of subclinical TCMR in patients with tacrolimus and mycophenolate analogues showed no benefit of treatment [164]. There was no difference in renal function at 6 months and chronic histology scores were in fact higher in the treatment arm. This study was, however, limited by the relatively low frequency of subclinical rejection at early protocol biopsy, as only 4.6% showed subclinical TCMR. Additionally, chronicity scores in the control arm appeared to improve from implantation to the 6 months biopsy in some patients with seemingly no additional intervention. This perhaps indicates other unknown factors may have influenced the results of this study and limits the potential conclusions that can be drawn from it. In terms of more recent observational research, Seifert et al. [165] analyzed protocol biopsies at 3 and/or 6 months in 120 pediatric patients. They showed that 13 treated patients with subclinical aTCMR still had a significantly increased risk of meeting the composite endpoint of death-censored allograft loss and acute rejection at 5 years post-transplant, compared to patients without rejection. However, choice of treatment modality of this low number of patients was at the discretion of the physician. In contrast, larger recent studies showed no significant difference in delta creatinine, odds of 50% eGFR loss, or allograft survival between subclinical TCMR patients treated standardly with pulse steroids and a control group without TCMR at protocol biopsy; [57, 166]. It should be noted that these studies were mainly performed in DSA-negative patients. Thus, less is known about treatment of DSA-positive subclinical TCMR cases, although there is a broad consensus about the detrimental long-term consequences on ongoing inflammation in renal allografts [163]. However, Cherukuri et al; [24] analyzed the effect of treatment with steroid pulses on patients with TCMR and/or DSA, although these were not specifically subclinical cases. Patients with underlying TCMR and no DSA had no significant risk of graft loss. However, TCMR with concurrent DSA was a significant risk factor for 4 years allograft attrition in multivariate analysis, even when treated. Crucially, this significant risk was attributable to non-adherence. Adherent and pulse steroid treated patients with DSA and TCMR had no increased risk of allograft loss compared to patients without DSA and rejection, whereas non-adherent, pulse steroid treated patients with DSA and TCMR had drastically lower graft survival rates. This seemingly indicates that DSA-positive patients with underlying TCMR may still be amendable to current treatment modalities, provided they are adherent. This further signals that strengthening adherence is an important treatment option and is recommended by us and others in patients with dnDSA [15, 140]. There are currently no guidelines on the treatment of subclinical TCMR [163]. A recent systematic review and meta-analysis by Ho et al. [167] showed through the included retrospective studies that most centers seem to treat subclinical TCMR (Banff 1a or higher) with pulse steroids and occasionally thymoglobulin. This is in line with two recent surveys, which show that more than 90% of North-American transplant centers have implemented pulse steroids or lymphocyte depleting antibodies as standard of care in these patients [168, 169]. Currently, ESOT is surveying this in Europe as well.

As a substantial amount (40%–50%) of patients with subclinical dnDSA will have signs of ABMR upon biopsy, it is important to review the evidence for treatment options in these patients. Recent consensus guidelines concluded that there is very little evidence for efficacy of current treatment protocols for ABMR in patients with dnDSA [15]. However, a retrospective study showed an incremental improvement in the treatment of ABMR; [170]. In addition, a small phase II prospective randomized trial with an IL-6 inhibitor has shown some promising results in chronic active ABMR (caABMR), and is currently being studied in a large multicenter phase III RCT [171, 172]. Additional evidence is emerging on the effectiveness of costimulation blockade, as discussed above, and anti-CD38 therapy in patients with aABMR and caABMR, the latter of which is currently being investigated in a phase II RCT in the form of felzartamab [149, 173]. In light of emerging data one may conclude that (early) acute ABMR with dnDSA (but without transplant glomerulopathy) could be more responsive to maintenance treatment optimization as well as PP and IVIG and eventually novel treatment regimens than patients with caABMR or cABMR, albeit all the treatment options have a low amount of supporting evidence. Active research in this area is ongoing and ABMR definition is becoming more precise [174]. Thus, there could potentially be benefit in finding and treating patients with early (subclinical) forms of ABMR before they present late with irreversible chronic lesions and clinical dysfunction. Some retrospective studies seem to support this hypothesis (Table 6). Parajuli et al. [102] showed similarly good post-biopsy allograft survival in patients with subclinical ABMR treated with IVIG and PP, as compared to protocol biopsied dnDSA-positive patients without rejection. Additionally, patients with treated subclinical ABMR had significantly better allograft survival than DSA-negative patients with indication biopsies or patients with treated clinical ABMR. Importantly, there was no difference in outcome between subclinical ABMR based on preformed DSA (type 1) vs. dnDSA (type 2). However, it must be noted that the post-biopsy follow-up time in patients with subclinical ABMR was relatively low at 31.0 ± 15.8 months. Orandi et al. [175] showed that patients with mostly type 1 subclinical ABMR treated by PP and in some situations rituximab or eculizumab had no significantly different rate of 5 years death-censored allograft loss compared to ABMR negative matched controls, whereas untreated patients had significantly more 5 years death-censored graft attrition rates compared to their control group. In addition, Yamamoto et al [79]. described some beneficial effects of PP and rituximab in 8 out of 18 (44%) of patients with subclinical type 2 ABMR whereby DSA levels reduced significantly or histological injury stabilized upon rebiopsy. In contrast, studies by Bertrand et al. [77] and Loupy et al. [57] found that allograft survival in treated subclinical ABMR patients was still significantly worse than patients without rejection. However, only 39% of patients with subclinical ABMR in the study by Loupy et al. [57] received specific treatment for subclinical ABMR and no analysis was performed comparing the treated and untreated group. It is apparent that more robust research on the effectiveness of treatment of subclinical ABMR is warranted. Nonetheless, the overall risk-benefit balance seems to be in favor of screening of DSA, which could result in early optimization of maintenance therapy. Moreover, further biopsy-guided treatment of subclinical TCMR and subclinical ABMR may be more effective than later treatment of clinical rejections, though evidence for this notion is more limited, as reflected in the grading of this recommendation.