1 INTRODUCTION

Short-term outcomes for pediatric kidney transplant recipients have improved markedly over the last 50 years, with one-year graft survival increasing from 63%–80% in 1992 to over 95% in 2020.^{1, 2} However, long-term graft survival has proven a more challenging goal, with a current 10-year graft survival of 60.2%.³ Chronic rejection is the most common reason for late graft failure and has no clearly effective treatment,⁴ which makes early detection of graft inflammation a key necessary step to improving long-term outcomes. While monitoring renal allograft health has historically focused on serum creatinine trends, creatinine is known to be a poor biomarker for graft injury. Creatinine rise lags renal injury by 24–48 h and may not rise at all in cases of mild damage.⁵ Creatinine may be an especially poor marker in children who are transplanted with adult-sized kidneys; the mismatch between a small child's muscle mass and the large adult kidney suggests that significant damage must occur before the creatinine will increase.

The term ‘subclinical rejection’ was first coined by Rush et al.⁶ in 1994. Defined as histologic evidence of acute rejection without immediate functional deterioration, subclinical rejection at 3 months post-transplant has been reported in 44% of pediatric kidney transplant recipients, including Banff grade borderline, 1A, 1B, and 2A acute T-cell mediated rejection (TCR) on histology.⁷ Similarly, Buchmann et al.⁸ reported subclinical rejection in 41% of 148 children at 3 months post-transplant and 45% at 6 months post-transplant, approximately half of which were borderline changes. Subclinical rejection was associated with a ten-point lower estimated glomerular filtration rate at the end of the study compared to those with no pathology.

Early identification of graft inflammation is a key area of research in kidney transplantation. The ideal test to detect subclinical rejection must balance sensitivity, specificity, and clinical feasibility. The blend of these three test characteristics determines the test's usefulness to trigger a biopsy or avoid a biopsy. In this article, we will review current pediatric kidney transplant monitoring, including those available for clinical use and those still under study (Table 1).

2 SURVEILLANCE BIOPSY

Kidney biopsy is the gold standard for rejection diagnosis, making surveillance biopsy at pre-defined post-transplant intervals an attractive option for identifying subclinical rejection. A 2017 survey of United Network for Organ Sharing (UNOS) members reported that 17% of adult kidney transplant programs perform surveillance biopsies for all recipients.⁹ The optimal timing and interval for surveillance biopsy are unknown. Some experts suggest that surveillance biopsy after 1 year post-transplant is of low clinical utility¹⁰ while others report clinically actionable findings on 9%–30% of surveillance biopsies performed at 2 or 3 years for pediatric recipients.^{11, 12} The UNOS survey found that 3 and 12 months were the most common timepoints for surveillance biopsy among adult transplant programs,⁹ and the pediatric Improving Renal Outcomes Collaborative reported that surveillance biopsies were performed at 41.4% of the 29 centers in their study.¹³ Under modern immunosuppression practices, Landsberg et al.¹² identified TCR in 20.8% of pediatric kidney transplant recipients undergoing a surveillance biopsy at 3 months post-transplant and 42.9% 6 months post-transplant, the majority of which were borderline rejections, and Seifert et al.¹⁴ showed borderline rejection detected on surveillance biopsy prior to 6 months post-transplant was associated with a 5-year composite outcome of acute rejection or allograft failure (aHR 2.89 p < .01).

Adult and pediatric data suggest that intervention to treat histologic changes in the absence of clinical allograft dysfunction can improve long-term outcomes, though this remains controversial. A 1998 randomized trial of 72 adult kidney transplant recipients showed a lower incidence of acute rejection, less interstitial fibrosis, and tubular atrophy at 6 months post-transplant, and a higher estimated glomerular filtration rate at 24 months in the treated group compared to the control group.¹⁵ However, a replication of this study in 2007 showed a numerically improved eGFR in the treated group that did not reach statistical significance, which the authors attributed to modern immunosuppression medications and a lower incidence of subclinical rejection.¹⁶ In contrast, an observational study by Siefert et al.¹⁴ suggested that children diagnosed with subclinical borderline rejection who were treated with increased immunosuppression had a lower incidence of acute rejection or graft loss at 5 years post-transplant compared to those who were not treated, but only included 27 patients.

In addition to the diagnosis of rejection, there is increasing evidence for the use of biopsy for follow up of treated rejection episodes. As early as 1999, there was evidence showing that two-thirds of ‘clinically successful’ rejection treatments had residual inflammation on biopsy,¹⁷ and a 2022 cohort study of 163 adults with TCR similarly demonstrated that 14% of those with complete clinical response had only partial or no improvement in their inflammation histologically.¹⁸ In children, rates of persistent rejection are even higher; Landsberg et al.¹⁹ reported persistent TCR in 55% of follow-up biopsies at a mean of 1.7 months after the original diagnosis. Birk et al.²⁰ showed that serum creatinine at 1 month after rejection treatment is a poor marker of inflammation resolution and could not distinguish those with improved rejection histology (23.8%) from those with stable or worsening rejection (35.2%).

Biopsy is the gold standard for diagnosis of clinical and subclinical rejection, but it has disadvantages including its invasiveness and cost. A single-center study of pediatric transplant recipients reported complications in 1.9% of procedures, including fever, gross hematuria, and pain²¹; a higher incidence of major complications (8.3%) has been reported when adult kidneys are placed extraperitoneal or transperitoneal in infants.²² Biopsies incur additional costs for the procedure, radiology, pathology, and anesthesia.¹⁰ For the patient, the procedure may be anxiety-inducing and time-intensive, requiring a full day off of school or work and limitation of activities in the immediate post-biopsy period. These limitations have fueled interest in non-invasive methods for accurately identifying subclinical inflammation or earlier signs of allograft damage.

3 DONOR-SPECIFIC ANTIBODIES

Donor-specific antibodies (DSA) are one of the key criteria of antibody-mediated rejection (ABMR) diagnosis,²³ and their measurement is now standard of care in the post-transplant period.²⁴ The Sensitization in Transplant: Assessment of Risk (STAR) 2022 working group recommends testing for DSA when there is evidence of allograft dysfunction, when immunosuppression is minimized, when there is concern for nonadherence and “periodically” during stable graft function.²⁴ The Transplantation Society Antibody Consensus Group recommends screening nonsensitized primary transplant recipients at least once at 3–12 months post-transplant.²⁵ In a 2019–2020 survey of the pediatric Improving Renal Outcomes Collaborative, screening for DSA was performed, on average, every 3 months for the first 12 months post-transplant or when serum creatinine level was elevated. Ten out of 29 centers reported performing a biopsy with any new DSA detection, regardless of serum creatinine.¹³

It is well established that the presence of DSA is overall associated with worse long-term graft outcomes. Wiebe et al.²⁶ monitored 315 adult kidney recipients every 6 months between 1999 and 2008; 15% of patients developed new DSA at a mean of 4.6 ± 3.0 years post-transplant. Median 10-year graft survival for those with DSA was 57% compared to 96% in those without DSA. Comoli et al. followed 114 pediatric kidney recipients; 34% developed new DSA at a median of 24.6 months (IQR 3–115 months) post-transplant. Kidney biopsy was performed in 30 of 39 patients with DSA; nine patients did not undergo biopsy due to resolution of DSA, clinical stability, or patient refusal. ABMR was diagnosed among 21 of 39 patients with DSA. In patients with DSA but no rejection, the majority had interstitial fibrosis and tubular atrophy and/or mild peritubular capillaritis. The presence of new DSA was associated with a 5.4-fold increased risk of graft loss (95% CI 1.11–25.85, p = .02).²⁷

While DSA are associated with graft failure, some DSA may not have the same impact on outcomes.²⁶ Cooper et al.²⁸ reported a cohort of 244 adult kidney recipients, of whom 65 (27%) developed dnDSA. Graft survival at 2 years was 97.8% in the patients without DSA, 37% in the patients with DSA detected on for-cause testing, and 93% in patients with DSA detected on a screening test. In pediatric recipients, Engen et al.²⁹ reported that new DSA detected by a screening protocol in clinically stable patients was associated with rejection, microvascular inflammation, and C4d staining on a 2-year biopsy, but was not associated with eGFR decline or graft loss. In contrast, new DSA detected on for-cause testing was associated with a 2.8-fold increased risk of decline in graft function (95% CI 1.08–7.27, p = .034) and a 7.3-fold increased risk of graft loss (95% CI 1.37–39.23, p = .02).

In an effort to identify those DSA-positive patients at the highest risk for graft dysfunction, several expansions of DSA testing have been proposed. The STAR 2022 working group recommends high-resolution HLA testing of donor and recipient to more accurately identify whether an antibody is truly donor-specific (level 2 recommendation, quality of evidence D).²⁴ The HLA class against which the antibody is directed may be important, with HLA class II antibodies portending a higher risk of chronic allograft nephropathy than HLA class I antibodies,³⁰ and antibodies against HLA-DQ have been associated with a 76% graft survival at 3.5 years compared to 95.2% when antibodies were directed against other HLA loci (p = .012).³¹ Several studies have assessed new DSA for complement binding activity.³² In pediatric patients, Hayde et al.³³ reported DSA development in 56.3% of 48 children, 63% of which was C1q positive (C1q + DSA). C1q + DSA was associated with a higher incidence of rejection compared to C1q-DSA. In a larger cohort of 233 pediatric kidney recipients, 52 (22%) of whom developed C1q + DSA were treated with intensified immunosuppression, intravenous immunoglobulin, and rituximab which lead to resolution of C1q + DSA in 58.5% of treated patients. In these two pediatric studies, all patients with graft loss due to rejection had persistent C1q + DSA.³⁴ A 2022 meta-analysis of 26 studies estimated that patients with C1q + DSA had a 2.4 times increased risk of graft failure (95% CI 1.66–3.47, p < .00001) and a 3.13 times increased risk of death (95% CI 1.06–9.23, p = .04) compared to those with C1q-DSA.³⁵ While C1q testing is available for clinical practice, its overall utility remains controversial due to the strong association between C1q binding activity and MFI³⁶ or antibody titer,³⁷ and studies suggesting that C1q binding activity measurements can be affected by diluting or concentrating the serum samples.³⁸

DSA detection is critical to diagnosis of ABMR but its use as a screening test has certain limitations. DSA is thought to develop late in the process of graft inflammation, preceded by some level of TCR or other inflammatory process that stimulates endothelial expression of HLA class II antigens.²⁶ Therefore, the detection of de novo DSA is a late marker of such inflammation, and there is limited high-quality evidence for the management of DSA.³⁹ Given the strong association between DSA and poor graft outcomes, a marker that accurately identifies graft inflammation prior to the development of DSA, and the associated memory B cells, would be preferred. Additionally, a positive DSA test may signal a need for a biopsy, but a negative one cannot rule out the need for one in a clinically concerning patient.

4 NON-HLA ANTIBODIES

Evaluation, diagnosis, and management of antibody-mediated rejection have historically focused on donor-specific HLA antibodies, but the absence of DSA is described in over 50% of cases of biopsy-proven ABMR.⁴⁰ This has led to increasing concern about the potential impact of non-HLA antibodies on kidney allograft survival, the most common targets being angiotensin type 1 receptor (AT1R), MHC class I chain-related gene A (MICA), and endothelin-type A receptor (ETAR).

AT1R is a G-protein coupled receptor present in the kidney, among other cells. There are several polymorphisms, the most studied being A1166C, which does not affect amino acid sequence but does alter levels of expression and has been associated with cardiovascular and renal disease.⁴¹ It is hypothesized that antibodies to AT1R cause endothelial injury due to hyperactivation of the receptor, rather than by direct immune system recruitment.⁴² A 2022 meta-analysis of 15 adult studies found that AT1R antibodies were associated with a 1.94 times increased risk of ABMR and a 2.37 times increased risk of graft loss, though the majority of the included studies assessed AT1R antibodies levels only once pre-transplant.⁴³ A 2024 study by Pearl et al.⁴⁴ monitored AT1R antibody pre-transplant and at 6, 12, and 24 months post-transplant in 100 pediatric kidney transplant recipients. Forty-six patients (46%) were AT1R antibody positive, approximately half of whom had pre-transplant AT1R antibodies and half developed de novo antibodies. Developing de novo AT1R or endothelin antibodies was associated with an increased risk of allograft loss at 2 years but not at 5 years post-transplant. There were trends towards increased incidence of antibody mediated rejection, but these did not reach statistical significance. However, this study conflates AT1R antibodies and ETAR antibodies, making it difficult to assess the independent importance of AT1R. Further studies are needed to confirm these findings and better understand their short- versus long-term implications. There are currently no recommended screening protocols for AT1R, but further study is needed given the simple hypothesized intervention of an angiotensin receptor blocker.⁴⁵

ETAR is a G protein-coupled receptor located in the vascular smooth muscle, mesangial cells, pericytes of descending vasa recta bundle cells, and renal tubules that induces vasoconstriction.⁴⁶ In rat models, expression of ETAR is increased in ischemia.^{47, 48} Post-transplant ischemia–reperfusion injury is therefore hypothesized to increase ETAR expression and lead to autoantibody formation. Like AT1R antibodies, ETAR antibodies are thought to be pathogenic by activating the receptor, leading to a downstream proinflammatory cascade,⁴⁹ but they are also capable of activating complement.⁴⁶ There appears to be a strong correlation between the presence of ETAR and AT1R antibodies.⁵⁰ In the pediatric screening study by Pearl et al., 69% of patients with non-HLA antibodies were positive for both AT1R and ETAR, while only 5% of patients had ETAR antibodies alone.⁴⁴ In adult recipients, ETAR antibodies were present in 47.4% of patients pre-transplant and were associated with worse graft function during the first year post-transplant⁵¹; however, this study did not look for the coexistence of AT1R antibodies. Fitchner et al.⁵⁰ reported that ETAR antibodies are associated with ABMR in pediatric kidney transplant recipients with an ROC of 0.69, but this association has not been confirmed in larger studies. It currently remains unclear if ETAR antibodies are a primary cause of ABMR or a marker of vascular endothelial inflammation.

MHC class I chain-like gene A (MICA) are surface glycoproteins found on fibroblasts, endothelial, epithelial, and dendritic cells as well as activated CD4+ and CD8+ activated T cells.⁵² Exposure to a donor's MICA antigen can lead to recipient humoral response, but current crossmatch procedures using donor lymphocytes fail to detect MICA antibodies.⁵³ Studies of the association between pre- and post-transplant MICA antibodies and ABMR and graft survival had lead to conflicting results.^54-56 In a pediatric retrospective study by Fichtner et al.⁵⁰ focusing on post-transplant DSA and non-HLA antibodies, up to 80% of the 62 patients with MICA antibodies were also found to be positive for AT1R and ETAR antibodies. MICA antibodies' mean MFI value was significantly higher in the ABMR group compared to the TCR group, but AUROC analysis showed poor discriminative capacity of MICA antibodies (0.66, 95% CI 0.52–0.81, p = .028). A retrospective multicenter French study including 1356 adult kidney recipients showed an association between MICA antibodies and ABMR (HR 3.8 for pre-transplant antibodies and 9.9 for post-transplant antibodies). ABMR risk was even greater when patients also had anti-HLA DSA antibodies (HR 25.7 for pre-transplant antibodies and HR 82.7 for post-transplant antibodies). Anti-MICA presence was also associated with reduced graft survival, and this finding was confirmed in an independent cohort of 168 patients with ABMR. Interestingly, using the sequence-based molecular MICA genotyping, they found that patients who were MICA matched a better graft survival than MICA mismatched donor-recipient pairs,⁵⁷ findings similar to those previously published in a smaller cohort.⁵² Overall, the data on the role of MICA antibodies in antibody-mediated rejection remains limited, and more studies are needed to clarify the utility of MICA antibody testing.

5 DONOR-DERIVED CELL FREE DNA (dd-cfDNA)

Any allograft injury resulting in cell apoptosis or necrosis, such as ischemia–reperfusion injury, rejection, infection, or disease recurrence, should increase the absolute amount of graft's DNA released in bloodstream. Donor-derived cell-free DNA measurement is a non-invasive method which relies on the quantification of the portion of the total free DNA measured in recipient plasma that comes from the allograft. The genetic differences (single nucleotide polymorphisms, SNPs) are identified by computational methods without the need to genotype the donor or recipient, and next-generation sequencing or droplet digital PCR is used to discriminate the donor portion of total measured DNA. Mostly known on a commercial basis as a blood-based test (Allosure from CareDx and Prospera from Natera), its use in urine samples is also under study.

Two studies in adults showed that dd-cfDNA level was a better indicator of rejection than serum creatinine, with ABMR cases showing higher dd-cfDNA compared to TCR and non-rejection groups. In the DART study, a dd-cfDNA result above 1% was 84% specific but only 59% sensitive for a composite outcome of ‘any rejection’. The level of dd-cfDNA was significantly higher for ABMR than TCR, but the test could not differentiate TCR 1A from no rejection. Additionally, the vast majority of the biopsies were performed for clinical indication, so the study was not able to assess subclinical rejection.⁵⁸ The larger ADMIRAL study included 1092 adult patients with 5873 dd-cfDNA measurements; 203 patients underwent 219 biopsies, 110 of which were for-cause biopsies and 109 were surveillance biopsies. The median dd-cfDNA values of patients with ABMR and TCR, excluding borderline, were respectively 1.8% and 0.7%, compared to 0.23% in patients without signs of rejection on biopsy.⁵⁹ However, dd-cfDNA failed to differentiate between borderline TCMR and no rejection groups. Additionally, as most patients with dd-cfDNA measurements did not undergo kidney biopsy, there is no validation of true negative results. The classification of borderline rejection into the non-rejection groups is also problematic as it does not allow assessment of dd-cfDNA as a marker of very early or subclinical rejection; the ideal dd-cfDNA threshold should detect adequately early stages of rejection before rejection translates into more severe histopathologic features. In the Trifecta study of 300 mostly for-cause allograft biopsies from adult recipients using molecular microscope diagnosis, dd-cfDNA was significantly associated with both ABMR and TCR, though the association was stronger for ABMR (Spearman Correlation Coefficient 0.52), specifically peritubular capillaritis, than for TCR (Spearman Correlation Coefficient 0.22).⁶⁰ The authors noted that dd-cfDNA was more clearly associated with molecular diagnoses of rejection than histologic diagnoses, raising interesting questions about mild inflammation that may be missed on a histologic assessment.

Children have a more significant donor-recipient size mismatch and higher frequency of primary viral infections, which can both affect the amount of DNA released from the graft. Dandamudi et al.⁶¹ reported that pediatric recipients with a donor-to-recipient body surface area (BSA) ratio >1.5 took 8 months after transplantation for dd-cfDNA level to reach its baseline of 0.23%, whereas patients with a BSA ratio ≤1.5 reached dd-cfDNA nadir of 0.14% after 4 months post-transplant. Nie et al.⁶² reported that more than 60% of pediatric transplant recipients had dd-cfDNA over 1.0% at 1 month after transplant and this proportion decreased to 20.0% and 11.1% at 2 and 3 months, respectively. Thus, the 1.0% cut-off may lead to a high rate of false-positive in children early post-transplant.

Dandamudi et al.⁶¹ tested retrospectively 290 stored samples from 57 pediatric kidney recipients who had monthly blood tests for 12 months following transplantation as well as surveillance biopsies at 3-, 6- and 12-months and monthly EBV, CMV, and BKV PCR surveillance. No ABMR was diagnosed in the cohort. Median dd-cfDNA was 0.91% (IQR 0.54%–1.2%) in patients with TCR compared to the non-rejection group (median 0.22%, IQR 0.14%–0.45%). Using a cut-off of 1% to discriminate acute TCR versus no rejection, they found a specificity of 96% and sensitivity of 33%. When testing the same cohort with a positive threshold of 0.5%, they found 78% of specificity and sensitivity. The authors hypothesized that these differences compared to adult studies might be explained by the BSA ratio discrepancy and the use of surveillance biopsies.⁶¹ In contrast, Puliyanda et al.⁶³ studied 67 pediatric recipients undergoing dd-cfDNA surveillance and for-cause testing later after transplant (median test time 55.6 (22.2–79.8) months post-transplant). In this cohort, dd-cfDNA over 1% was diagnostic of rejection (n = 21) with a sensitivity of 86% and specificity of 100% but again, most rejection cases were represented by ABMR (n = 11) or mixed rejection(n = 8) and there were no surveillance biopsies to confirm lack of rejection in those without clinical dysfunction (Table 2).⁶³

Donor derived-cfDNA may also be a marker for rejection treatment response. Steggerda et al.⁶⁴ showed that dd-cfDNA levels decreased for all patients with rejection following treatment. However, all patients with ABMR or mixed rejection still had dd-cfDNA levels over 1.0% 8 months following the first positive result, possibly because 10 out of 16 of these patients had biopsy showing persistent rejection. Two patients were diagnosed with pure TCR on biopsy and received treatment accordingly but did not undergo a follow-up biopsy. Thus, it is difficult to conclude if dd-cfDNA reflects rejection resolution.

Current evidence for donor-derived cell-free DNA shows that it discriminates ABMR from non-rejection pathology with high negative predictive value, which could make it an interesting tool to rule-out rejection in cases of clinical suspicion. However, the strength of this evidence is limited by the lack of biopsy information in a significant portion of most study cohorts, the tendency of many studies to conflate TCR and ABMR, and the lack of evidence associating it with subclinical rejection. Like DSA, the higher sensitivity of dd-cfDNA for ABMR limits its ability to detect graft inflammation in an early subclinical stage. There is evidence that dd-cfDNA may be useful in pediatric patients when adjusted for size mismatch, but this needs further validation in independent cohorts with a broader range of rejection types and consistent biopsy of every participant. Recent studies have helped to improve understanding of dd-cfDNA kinetics over time and provide some potential normal values for children. Future studies will be needed to determine whether the threshold for defining a positive dd-cfDNA test should vary over time post-transplant or based on the pre-test probability for rejection.

6 TRANSCRIPTOMICS

Transcriptomics refers to the assessment of gene expression profiles by measuring messenger RNA (mRNA) via microarray or mRNA sequencing. Specific enzymes, receptors, and chemokines are known to be increased in rejection; thus, the genes encoding these proteins should hypothetically be over-expressed. Panels of gene expression have been studied and tested in both serum and urine.

In the blood, the largest studies include two gene expression panels called TruGraf (Eurofins and Transplant Genomics Inc.) and Allomap kidney (CareDx). The Transplant Genomics Collaborative Group studied a global gene expression using a 200 high-value probe-sets (TruGraf, Eurofins and Transplant Genomics Inc.) in 148 adults who underwent biopsy for surveillance or graft dysfunction. They showed that this panel could discriminate acute rejection from normal biopsy results (AUC 0.837) and acute dysfunction without rejection (AUC 0.893) with high accuracy.⁶⁵ However, this study did not differentiate TCR from ABMR, which is surprising as there is clear evidence of rejection type-specific transcript patterns.⁶⁶ Later, Marsh, et al.⁶⁷ demonstrated that this test was a reassuring asset in the follow-up of patients with stable graft function to discriminate subclinical rejection from normal histopathological findings with a negative predictive value of 90%, sensitivity of 74% and specificity of 73%. These findings are presently being validated among a pediatric cohort (NCT05335538).⁶⁸ Allomap Kidney (CareDX) is a 5-gene expression profiling test that has been tested in two different adult cohorts, where it showed a higher score in those with rejection compared to non-rejection and healthy participants. The AUC to discriminate rejection from quiescence reached 78% in two studies.^{69, 70} This panel had been tested in a small cohort of children who underwent surveillance and for-cause biopsy and showed to discriminate quiescence from rejection with an AUC of 0.89 (95% CI 0.80–0.98) and NPV of 97% considering rejection prevalence of 10%. Of note, children with a second transplant had consistently higher AlloMap scores.⁷¹

The CTOT-04 study assessed urinary cells' mRNA expression of a three gene panel, which included CD3ε, CXCL10, and 18S rRNA, in over 4300 adult kidney transplant recipients within 12 months post-transplant.⁷² Findings were compared to biopsy results and clinical outcomes. This panel was able to distinguish acute TCR versus non-rejection with an AUC of 0.85 in the initial cohort and 0.74 in an external validation cohort. A marked increase in levels was observed 20 days before biopsy in the rejection group (p < .001). Importantly, subclinical rejection cases were excluded from analysis; thus, it remains unclear whether this assay can truly detect rejection prior to its impact on allograft function. This panel also distinguished acute TCR from acute ABMR and borderline rejection (AUC = 0.78), but no further conclusions regarding ABMR could be drawn due to a small number of cases. Urinary tract infection did not result in elevated biomarker levels, nor did bacteremia or cytomegalovirus DNAemia. However, the diagnostic signature for rejection was also associated with BK virus infection.⁷² The currently ongoing VIRTUUS study aims to validate this same three gene panel in a pediatric cohort with a study completion date projected for May 2024.⁷³ Six hundred children are expected to be followed prospectively for 12 months with longitudinal urine testing and biopsies including for-cause and surveillance. Of note, early reports indicate that 17% of samples failed quality control,⁷³ a problem that highlights the methodological challenges of urinary mRNA, which tends to degrade without proper sample processing and storage even with high level of technical training.⁷⁴

7 PROTEOMICS

Rejection causes monocyte infiltration into the kidney tissue, leading to production of IFN-γ. IFN-γ stimulates renal tubular epithelial, mesangial, and infiltrating inflammatory cells to secrete the chemotactic cytokines (chemokines) CXCL9 and CXCL10, which in turn increase T cell recruitment to allograft via CXCR3 expressed on leukocytes.⁷⁵ Rabant et al.⁷⁶ showed that urinary CXCL9/Cr ratio was associated with tubulointerstitial inflammation and TCR in adult kidney transplant recipients with an AUC of 0.90, while CXCL10/Cr ratio was associated with peritubular capillaritis and ABMR with an AUC of 0.82 for mixed rejection and 0.7 for ABMR alone. CXCL10/Cr was found to be independently associated with graft loss (HR 2.2, 95% CI 1.4–2.2, p < .001), and patients who lost their graft because of ABMR had a higher ratio of CXCL10/Cr initially; thus, the ratio at the time of biopsy correlated with graft survival in a dose-dependent manner.

In pediatric studies, Blydt-Hansen et al.⁷⁷ showed a higher urinary CXCL10/Cr ratio in those with rejection compared to children with normal or indeterminate findings, and elevation of urinary CXCL10/Cr ratio anticipated rejection diagnosis up to 5–8 weeks prior to biopsy. In a second study pediatric recipients with clinical TCR had the highest ratio (23.3 ng/mmol) compared to normal histology and IFTA (1.4 ng/mmol) (p < .001), with subclinical rejection median ratio lying in between (4.4 ng/mmol).⁷⁸ The authors suggested two potential thresholds for interpretation of urinary CXCL10/Cr: the lower of 0.63 ng/mmol providing a false negative rate of 10.8% for ruling-out rejection, and the higher of 4.08 to reach a false-positive rate of 9.5% and maximal sensitivity.⁷⁷ The small number of cases of ABMR in pediatric studies has made it difficult to draw conclusions about its association with CXCL10. In one study, the CXCL10/Cr ratio of nine children diagnosed with pure ABMR was higher than those with normal histology, borderline tubulitis, or interstitial fibrosis and borderline tubulitis⁷⁸ but these findings will have to be tested in larger cohorts.

As CXCL10 and CXCL9 are inflammation surrogates, they have limited ability to discriminate rejection from other inflammatory pathologies such as BK viremia, BK viruria or urinary tract infection. In a study by Jackson et al.,⁷⁹ adult recipients with BK infection had elevated urinary CXCL10 levels that were indistinguishable from those with rejection. In a pediatric cohort, median urinary CXCL10/Cr was similar in children with BK infection with tubulointerstitial inflammation and Banff 1A rejection, but still higher than normal biopsy findings (p < .001). However, children with BK nephropathy without tubulointerstitial inflammation could not be distinguished from normal samples using urinary CXCL10/Cr.⁷⁷ Simultaneous acute rejection and BKVAN are common, with over 50% of adult patients showing histopathological findings of both, highlighting the need for a biopsy to orient clinical management.^{80, 81}

8 METABOLOMICS

Metabolomics refers to the quantification of small molecules (metabolites) considered as intermediate or end point products of diverse biological processes. Using mass spectrometry detection to measure hundreds of urine metabolites, patterns in particular metabolic pathways may be associated with tissue response phenotypes such as acute rejection,^82-86 IFTA, glomerulosclerosis, calcineurin inhibitor toxicity, and acute kidney injury.^87-89 Interestingly, many alloimmune and non-immune processes can be assessed simultaneously in the same provided samples.

Specific urine metabolites profiles have been associated with both TCR and ABMR in pediatric cohorts with a discriminant score (dscore) of high accuracy.^{83, 84, 86} Sidgel et al.⁸⁶ found that a panel of nine metabolites could discriminate urine samples of children showing acute or chronic alloimmune injury on biopsy from urine samples of stable patients with no rejection on biopsy with 95% accuracy (AUC 0.95). A slightly different 11 metabolite panel distinguished acute rejection, including ABMR and TCR (excluding borderline and subclinical rejection), from stable patients (AUC 0.985).⁸⁶ Similarly, Blydt-Hansen et al. reported two metabolite panels in pediatric kidney transplant recipients who underwent for-cause and surveillance biopsy, one of which identified TCR, including borderline, with an AUC of 0.88 (sensitivity and specificity 80%, NPV 97%, PPV 40%) and one that discriminated ABMR with an AUC of 0.84 (sensitivity 78%, specificity 82%, NPV 96%).^{83, 84} Although these findings are interesting, they need to be tested in independent prospective cohorts to further validate their possible clinical implementation.

9 TORQUE TENOVIRUS

Torque tenovirus (TTV) is a small, single-stranded DNA virus that appears to be part of the human virome and yet seems to cause no known clinical disease..⁹⁰ It is highly prevalent worldwide- epidemiologic studies have reported detectable levels in healthy populations ranging from 47% in the United Kingdom to 100% in Saudi Arabia⁹¹ – and is independent of age, sex, and socioeconomic status 74The virus is highly heterogeneous with at least 29 reported variants and an especially variable ORF1 region, but a highly conserved 5′ untranslated region.⁹⁰ Historically, studies of TTV have reported widely differing results depending on whether the PCR primers used target the N22 region of ORF1 or the 5′ untranslated region.^{92, 93} Current commercially available PCR tests amplify the conserved 5′ untranslated region.⁹⁴

TTV plasma loads are inversely correlated with the number and function of T cells⁹⁵ and levels are generally lower in healthy controls and higher in patients with immune system dysfunction from sepsis,⁹⁶ chemotherapy, and immunosuppressive medications.⁹⁷ Thus, TTV has been identified as a potential marker for overall immunocompetence after kidney transplantation. Among 169 adult kidney recipients, Strassl et al.⁹⁸ found that TTV was detectable in 80% of patients pre-transplant and 99% of patients by 3 months post-transplant. The TTV-POET trial explored all-cause infection and rejection outcomes in 683 adult kidney recipients monitored with monthly TTV via PCR from months 4–12 post-transplant. Patients with allograft rejection had lower levels of TTV at a median of 17 days prior to biopsy, and the risk of rejection decreased 25% with every log level increase in TTV load. The association between TTV and rejection had an area under the curve of 0.82 (95% CI 0.7–0.94, p < .001). In the same cohort, patients with infection had a higher TTV level at a median of 63 days prior to infection, with a 6% increase in infection risk for each log level increase in TTV but an area under the curve of 0.58 (95% CI 0.52–0.64, p = .039).⁹⁹ Van Rijn et al.¹⁰⁰ also showed that each log increase in TTV was associated with a lower risk of rejection (HR 0.74, 95% CI 0.71–0.76) but could not demonstrate any association with BK viremia or CMV DNAemia. A multicenter randomized controlled trial using TTV levels to guide tacrolimus dosing is currently ongoing in Europe.¹⁰¹

In children, TTV is present in 39% of children by age 4 months and 93% by 2 years of age with levels varying significantly between children and in serial samples from individual children.¹⁰² Similar to adults, TTV levels increase sharply post-transplant, peaking at approximately 3 months before declining to a steady state.¹⁰³ In a single-center study of 45 pediatric kidney recipients TTV was detected in 94.5%. Mean TTV viral load over the 12 months of the study did not vary with age, sex, primary disease, type of donation, or number of HLA mismatches. Of note, participants did not receive induction therapy and mean TTV load correlated with type and dosage of maintenance immunosuppression. There were no episodes of rejection in the study cohort, and there was no association between TTV level and infections or EBV, CMV, or BKV plasma loads. However, the study was limited by the small sample size and use of mean TTV level, rather than a time-varying TTV level.¹⁰⁴ Larger prospective studies are needed in pediatrics to assess: TTV association with rejection and/or infection, ideal target level or trend of TTV to minimize rejection and infection, and whether the interpretation of TTV in children varies by age, induction immunosuppression, and time post-transplant. As for other biomarker trials, it will be critical that such studies include surveillance biopsies in all patients, and inclusion of a diverse patient population with adequate power to detect infectious outcomes and all types of rejections.

10 CONCLUSION

Advancing the use of noninvasive immune monitoring in transplantation will depend on well-designed multicenter trials that meet three criteria. First, trials must include patients with stable graft function. The use of a convenience sample of patients undergoing biopsy due to graft dysfunction results in artificially high prevalence of rejection in the cohort and skews estimates of test parameters such as positive and negative predictive value. Second, biomarker trials should include routine comparisons to the gold standard biopsy for all study participants. Without biopsy, it is unknown if a ‘low risk’ biomarker result is truly negative. Third, biomarker studies need to show an association between the biomarker and a patient-relevant clinical endpoint, such as graft survival or a 50% decline in the estimated glomerular filtration rate. Studies have shown a poor correlation between some rejection episodes and long-term outcomes, making surrogate endpoints such as rejection and DSA detection inadequate. Finally, studies are needed to show that the biomarker can lead to specific changes in care that improve outcomes rather than simply identify high-risk patients.

CONFLICT OF INTEREST STATEMENT

The authors have no conflicts of interest to disclose.