1. Introduction
Routine graft monitoring in renal transplantation relies on non-invasive biomarkers such as serum creatinine, proteinuria, and HLA antibodies. More than 30 years ago, to explore the feasibility of histological monitoring, some centers started programs of protocol biopsies and observed that there were grafts with stable function depicting histological changes of rejection, leading to the definition of subclinical rejection (SCR) [
1,
2,
3]. During the cyclosporine era, SCR was prevalent (>30%) and it was proven that its treatment better preserves renal function [
2,
4]. However, a clinical trial addressing SCR treatment in patients on modern immunosuppression with tacrolimus/MMF yielded limited clinical benefit due to the low rate of SCR [
5]. For this reason, it has also been explored whether minor histological changes are associated with graft outcome. In a study including a large set of 6-month protocol biopsies (n = 957), it was shown that interstitial inflammation (i-score > 0) in otherwise normal protocol biopsies is associated with a significantly lower 15 y graft survival, comparable to SCR or interstitial fibrosis/tubular atrophy (IF/TA) with inflammation [
6].
Although the evaluation of renal biopsies based on the Banff classification for renal transplant pathology has been refined since 1991 [
7], some uncertainties persist, notably concerning the presence of borderline changes suspicious for T cell-mediated rejection (TCMR) and the incomplete phenotypes of antibody-mediated rejection (ABMR). To further characterize underlying mechanisms leading to different histological phenotypes, an analysis of the transcriptome has been incorporated [
8,
9,
10,
11]. It has been shown that molecular diagnostics allow the detection of transcript sets strongly associated with TCMR and have proven useful in differentiating borderline infiltrates likely to lead to the development of overt TCMR and/or graft fibrosis [
12]. Recently, we have shown that a rejection-associated gene expression score is present in 83% of protocol biopsies with SCR but only in 17% of protocol biopsies with borderline changes [
13]. Importantly, to distill information from RNA microarrays that evaluate thousands of genes, the Banff group has delineated gene sets related to TCMR, ABMR, tissue-repair injury, and other pathways implicated in graft dysfunction [
9,
10,
11,
12,
13,
14].
In studies of serial protocol biopsies performed in renal transplants, the prevalence of SCR is maximal during the initial three months, progressively decreases until the first year, and persists in a small number of patients after the first year. Risk factors associated with SCR are the number of human leukocyte antigen mismatches, the degree of sensitization, retransplantation, the presence of previous clinical acute rejection episodes, and the immunosuppressive regimen being lower in patients treated with tacrolimus and MMF [
3]. Our group and others have shown that reduced exposure to tacrolimus and/or MMF is associated with a higher incidence of subclinical inflammation in protocol biopsies performed during the first year [
15,
16]. Furthermore, while certain studies have linked SCR to HLA ABDR allelic mismatch [
17], disparities at the molecular level might offer more informative insights [
18]. Notably, in liver transplant recipients, reduced immunosuppression exposure and an increased number of HLA epitope mismatches between donor and recipient have been implicated in the molecular pathogenesis of subclinical liver allograft damage driven by an interferon gamma-orchestrated cellular immune response [
19].
In this current study, we employ microfluidic cards to scrutinize the transcriptome of a predefined set of genes related to different histological diagnoses (mainly TCMR and ABMR) previously described by the Banff group [
14]. We aim to determine whether transcripts increase or decrease in biopsies with rejection, reflecting changes in resident and/or infiltrating cells. To achieve this, we compare gene expression normal-protocol biopsies and for-cause biopsies that meet the Banff criteria for TCMR or ABMR. Subsequently, we quest these transcripts in a large set of early protocol biopsies to evaluate whether gene expression is associated with donor and recipient characteristics, including the intensity of immunosuppression and donor–recipient HLA mismatch at the allelic or molecular level.
3. Discussion
We conducted a prospective study on a set of 4-month protocol biopsies to evaluate whether validated rejection-associated transcripts are associated with tacrolimus exposure at the time of biopsy. The main findings of our study are that we confirm the discrimination capacity between normal and rejection biopsies of a large set of these genes and that the expression of 19 rejection-associated transcripts in early protocol biopsies is associated with tacrolimus exposure at the time of biopsy. Cluster analysis using this set of 19 genes identified a pool of patients with a higher proportion of inflammatory phenotypes, including TCMR, borderline lesions, and IFTA with inflammation. Interestingly, patients from this cluster had less exposure to tacrolimus and displayed a faster decline in renal function during follow-up. The low rate of de novo DSA development in our cohort (2.9%) limits further analysis of its association with subclinical inflammation. Thus, our results suggest that adjusted immunosuppression during the early months after transplantation favors a better control of the inflammatory response without deleterious effects on renal function in the mid-term.
Tacrolimus is the mainstay of immunosuppressive regimens for kidney transplantation since it prevents T cell activation and proliferation. Although tacrolimus reduces the acute rejection rate and improves short-term outcomes after kidney transplantation, it is associated with both acute and chronic nephrotoxicity and triggers serious side effects. Although monitoring of tacrolimus exposure relies on clinical practice for determining trough levels, there is no agreement on the target levels during the first year in renal transplant recipients. While the largest clinical trial supported the minimization of tacrolimus exposure [
20], one randomized clinical trial has shown that in case of steroid discontinuation and MMF reduction, maintaining TAC-C
0 > 7 ng/mL after the fourth month reduces the risk of acute rejection and appearance of de novo DSAs without increasing renal toxicity [
15]. Similarly, in low-immunological-risk renal transplants treated with TAC, reduced MMF, and low-dose steroids, TAC-C
0 levels are associated with subclinical inflammation in patients monitored by protocol biopsies [
16]. Additionally, it has been described that the effect of tacrolimus trough levels was modulated by the recipient’s baseline alloimmune risk, as defined by their class II HLA donor–recipient eplet mismatch [
18].
In the present study, we analyzed whether interstitial inflammation is associated with clinical characteristics of donors and recipients as well as with transplant-related variables. In our cohort, the presence of interstitial inflammation was associated with lower TAC-C
0 at the time of biopsy and with longer cold ischemia time, but it was not associated with mid-term renal function deterioration or the development of de novo DSA. Importantly, in our cohort, as in several others [
6,
16], few cases met the criteria for subclinical borderline rejection (11.7%) or TCMR/ABMR (5.8%). Since the presence of interstitial inflammation (i > 0) in otherwise normal biopsies has been associated with 15-year death-censored graft survival [
6] in a similar way to SCR, we chose this threshold for our analysis. Notably, in other studies including patients treated with a steroid-free regimen, the incidences of borderline rejection and TCMR were significantly more frequent (31% and 20.8%) [
21]. In this study, the authors did not find associations between TAC-C
0 and subclinical inflammation, but it should be noted that at the time of the 3-month protocol biopsy, the TAC-C
0 average was close to 10 ng/mL [
22]. In this study, SCR within the first post-transplant year is associated with a significantly greater hazard of subsequent clinical rejection and death-censored graft loss. On the contrary, other studies have shown that T cell-mediated inflammation detected in protocol biopsies mostly reflects the injury–repair response to implantation stresses and has little relationship with future events and outcomes [
23]. Acute kidney injury (AKI) after renal transplantation can also induce interstitial infiltration and tubulitis [
24] leading to a histological picture indistinguishable from that of TCMR. In this sense, in our cohort of protocol biopsies, we observed an association between interstitial inflammation and longer cold ischemia time. Thus, the presence of interstitial inflammation is uncommon in our cohort of low-immunological-risk kidney transplants maintaining steroids (27.7%) and it is associated with tacrolimus exposure and cold ischemia time, suggesting that both immune and non-immune factors may contribute to subclinical inflammation in well-functioning grafts.
The disagreement between different studies on the prevalence of subclinical inflammation and its association with later clinical outcomes is partly explained by the inclusion of different populations and different maintenance immunosuppression regimens. However, there is general agreement that conventional biopsy assessment is limited due to poor interobserver reproducibility of individual lesions [
11,
25]. To overcome these limitations, it has been proposed that molecular phenotyping be incorporated. The application of microarrays to transplant biopsies has been an ongoing effort by many groups and the interpretation of molecular changes aided by the understanding of their biological mechanisms led to the grouping of different transcripts [
10,
14,
26,
27]. To summarize information derived from RNA microarrays, which evaluate thousands of genes, in the last reports of the Banff meetings gene sets containing a few hundred genes related to TCMR, ABMR, tissue-repair injury and other pathways leading to graft dysfunction were described [
6,
26,
28,
29]. In the present study, we evaluated the panel of genes described in the Banff meeting in 2017 via RT-PCR [
14]. As expected, we confirm the discrimination capacity of a high number of these genes (111 out of 308 evaluated genes) to differentiate normal protocol biopsies from biopsies for cause with rejection. In the principal component analysis, we observed that TCMR-selective genes expressed in activated effector T cells (ADAMDEC1) and genes encoding different cytokines and their receptors mainly related to TCMR (CCL5, CXCL13 and CXLC9) were the most relevant in component 1, while genes encoding matrix proteins and solute transporters (COL1A1, SCL22A2 and SCL4A1) were the most relevant in component 2.
Regarding the derived gene set, we were interested in evaluating its relationship with tacrolimus exposure at the time of biopsy. We found that 19 of these 111 genes, mainly related to TCMR, were mildly correlated with TAC-C
0, suggesting that a higher tacrolimus exposure contributes to a better control of subclinical inflammation. Interestingly, in the small set of normal protocol biopsies, we observed a close correlation between TAC-C
0 and the expression of 2 out of these 19 genes (IKZF3 and CD2,
Figure 4) IKZF3, expressed mostly in the lymph and spleen, is found in several immune cell types, including B cells, NK cells, CD4+, and CD8+ T cells. It is expressed most strongly in B cells and studies of IKZF3 knockout mice indicate a critical role for IKZF3 in B-cell differentiation, maturation, proliferation, and T cell-dependent B-cell responses. IKZF3 is upregulated in pre-B cells, and it has been found to play a role in executing the transition from large pre-B cells to small pre-B cells during normal B-cell development. IKZF3 has also been found to play an important role in T cell regulation. It is expressed in interleukin-17-producing helper T cells and promotes differentiation through silencing of interleukin 2 production. Recently, it has been shown that IKZF3 is upregulated not only in ABMR but also in TCMR urinary cell specimens suggesting that B cells may play a more active role in TCMR than previously recognized, perhaps functioning as classical antigen-presenting cells [
30,
31,
32]. The CD2 family of costimulatory and adhesion molecules has also been shown to play a significant role in the execution of an alloimmune response since it is constitutively expressed by all T cells and upregulated upon antigen recognition. Importantly, CD2 is more highly expressed on effector memory T cells relative to central memory T cells and therefore more effectively targets those cells that are poised to rapidly exert effector function upon encounter with cognate antigen. In addition to its role in facilitating the adhesion of T cells to antigen-presenting cells during the immunological synapse, CD2 ligation results in the direct transmission of co-stimulatory signals to promote T cell activation and differentiation [
33,
34]. Importantly, in the evaluated set of biopsies, the expression of these 19 genes associated with TAC-C
0 and split our protocol biopsy group into two clusters one containing all but one normal protocol biopsies and the other containing all rejection biopsies. The large set of protocol biopsies was distributed in a similar proportion in both clusters. Patients with protocol biopsies grouped in cluster 2 received a lower exposure to tacrolimus, showed more frequently an inflammatory phenotype, and displayed a faster decline of renal function in the mid-term. Thus, our results suggest that more adjusted immunosuppression during the early months after transplantation favors a better control of the inflammatory response and better preserving renal function in the mid-term.
Our effort to detect associations between gene expression, tacrolimus exposure, and HLA compatibility at the allelic or molecular level did not show significant associations in the multivariate analysis. It should be remarked that HLA typing in this cohort was performed according to clinical practice and thus, high-resolution HLA typing was not performed and the availability of HLA typing for all loci (especially DQ) was limited. However, in this cohort of successfully immunosuppressed renal transplant recipients, the number of patients developing de novo DSA was very low (2.9%) and although patients from cluster 2 tended to develop de novo DSA more frequently, this association did not reach statistical significance. Additionally, the present study has other important limitations, since associations between tacrolimus exposure and histological findings or gene transcripts were based on a single determination of TAC-C
0 on the day of biopsy and a more refined evaluation of tacrolimus pharmacokinetics (e.g., area under the time–concentration curve) or pharmacodynamics (e.g., calcineurin activity) was not done. Finally, patients in group II (acute rejection) underwent biopsies later in comparison to patients in the other groups (protocol biopsies), and the impact of biopsy timing on gene expression has been widely acknowledged [
11,
35,
36].