Moving Biomarkers toward Clinical Implementation in Kidney Transplantation

Abstract

Long-term kidney transplant outcomes remain suboptimal, delineating an unmet medical need. Although current immunosuppressive therapy in kidney transplant recipients is effective, dosing is conventionally adjusted empirically on the basis of time after transplant or altered in response to detection of kidney dysfunction, histologic evidence of allograft damage, or infection. Such strategies tend to detect allograft rejection after significant injury has already occurred, fail to detect chronic subclinical inflammation that can negatively affect graft survival, and ignore specific risks and immune mechanisms that differentially contribute to allograft damage among transplant recipients. Assays and biomarkers that reliably quantify and/or predict the risk of allograft injury have the potential to overcome these deficits and thereby, aid clinicians in optimizing immunosuppressive regimens. Herein, we review the data on candidate biomarkers that we contend have the highest potential to become clinically useful surrogates in kidney transplant recipients, including functional T cell assays, urinary gene and protein assays, peripheral blood cell gene expression profiles, and allograft gene expression profiles. We identify barriers to clinical biomarker adoption in the transplant field and suggest strategies for moving biomarker-based individualization of transplant care from a research hypothesis to clinical implementation.

Despite advancements in understanding immune responses induced in transplanted organs and despite diminution in acute rejection (AR) rates,¹ lifelong immunosuppression is required after kidney transplantation, and long-term allograft survival rates remain suboptimal.² The causes of late allograft loss are multiple and include late rejection as well as recipient death with a functioning graft.³ The prevailing immunosuppression strategies are center-based protocols, with potential for overimmunosuppression (predisposing to infection or drug toxicity) or underimmunosuppression (predisposing to immunologic graft injury) of individual transplant recipients. Currently used tactics used to guide immunosuppression choices and dosing are relatively rudimentary and include epidemiologic parameters (e.g., age and self-declared race), HLA mismatching, and donor-specific, anti-HLA antibody (DSA) screening. Identification and validation of biomarkers that correlate with and/or predict allograft injury and that could improve therapeutic decision making are priorities for the transplantation community.⁴ Herein, we review available data on immune monitoring assays that are moving toward clinical use in kidney transplantation, highlighting assays approved by the US Food and Drug Administration (FDA) and those that are validated in external cohorts but are not FDA approved (Table 1). We also discuss selected promising assays that require additional work before clinical implementation. Finally, we offer a framework monitoring strategy for consideration, and we outline likely barriers and potential solutions to clinical adoption of biomarker testing in transplantation.

Table 1.

Selected biomarkers for outcomes in kidney transplantation: summary of current status

Authors	Assay Name	Assay Type	Timing Post-Transplant	Outcome	Discovery Set: Se/sp/ppv/npv	Validation Set: Se/sp/ppv/npv	Training/Test Sample	Biomarker Lifecycle
FDA-approved assays
Patel and Terasaki19	CDC crossmatch	Microcytotoxicity assay	Pretransplant	Hyper-AR/early graft loss	0.75/0.97/0.80/0.97	Not applicable, no validation set	225	FDAA: Y; Comm: Y
Mahoney et al.149	Flow crossmatch	Flow cytometry	Pretransplant	Early graft loss (<2 mo)	0.71/0.74/0.33/0.93	Not applicable, no validation set	90	FDAA: Y; Comm: Y
Pei et al.150	Lumine	HLA beads; flow cytometry	Variable pre-/post-transplant	Anti-HLA Ab	Details not available in manuscript	Not applicable, no validation set	10	FDAA: Y; Comm: Ya
Ashokkumar et al.61	Pleximmune	T cytotoxic memory cell assay	Rejection episodes	Biopsy-proven AR	0.88/0.94/0.93/0.88	1.0/0.86/0.80/1.0	32/11	FDAA: Y; Comm: Y
He et al.63	Cylex-Immuknow	Lymphocyte ATP generation assay	Serial <30 mo	CD4-T cell function	Details not available in manuscript	Not applicable, no validation set	42	FDAA: Y; Comm: Y
Loupy et ald.32	C1q binding assay	Flow cytometric C1q bindingb	Baseline, 1 yr, rejection episode	TCMR/ABMR/graft loss	Details not available in manuscript	Not applicable, no validation set	1016	FDAA: Y; Comm: Yb
Selected externally validated assays in kidney transplantation (pending FDA approval)
Hricik et al.47	IFN-γ ELISPOT	Donor-reactive memory T cell	Preatransplant	De novo DSA and/or rejection	1.0/0.67/0.67/1.0	Not applicable, no validation set	21	FDAA: N; Comm: N
Hricik et al.70	Urine CXCL9	Urine ELISA	Serial <6 mo, rejection episode	TCMR	0.85/0.81/0.68/0.92	Not applicable, no validation set	258	FDAA: N; Comm: N
Suthanthiran et al.86	Urine three-gene signature	Urinary RNA by qPCR	Serial <12 mo, rejection episode	AR	Details not available in manuscript	Details not available in manuscript	485 (4300 urine samples)	FDAA: N; Comm: N
Roedder et al.119	KSORT	Peripheral blood RNA by qPCR	Serial <24 mo, rejection episode	Biopsy-proven AR	0.83/0.91/0.81/0.91	0.91/0.99/0.95/0.98	143/124	FDAA: N; Comm: N
Halloran et al.112	ENDAT	Graft biopsy RNA by microarray	Variable (1 wk to 31 yr)	ABMR	0.69/0.85/0.50/0.94	0.67/0.90/0.64/0.91	300/403	FDAA: N; Comm: N
O’Connell et al.18	GoCAR score	Graft biopsy RNA by microarray (3-mo protocol biopsies)	3 mo biopsy	Prediction of fibrosis/histologic progression	0.93/0.95/0.93/0.95	0.67/0.92/0.86/0.81	159/45	FDAA: N; Comm: N

Se, sensitivity; sp, specificity; ppv, positive predictive value; npv, negative predictive value; CDC, complement dependent cytotoxicity; FDAA, US FDA approved; Y, yes; Comm, commercialized assay available; TCMR, T cell–mediated rejection (Banff ≥1A); N, no; KSORT, kidney solid organ response test; ENDAT, endothelial-associated transcripts; GoCAR, Genomics of Chronic Allograft Rejection.

^aAssay approved for detection of DSAs (not for quantification).

^bC1q binding Bindazyme assay is FDA approved, but flow cytometric C1qScreen (One-Lambda) used in this study is not FDA approved.

Biomarker Overview

Clinically useful biomarkers may be prognostic, be predictive, and/or serve as surrogate end points,^5–7 and they may be within or outside the causal pathway of the disease process for which they are being used. In kidney transplantation, biomarkers can be used to predict/diagnose AR, immunologic/operational tolerance,^8,9 overimmunosuppression, nonadherence, opportunistic infections,^10,11 progressive, chronic graft injury,¹² and graft loss. We focus on biomarkers for AR, subclinical alloimmune injury, and graft loss.

An ideal immune biomarker should rapidly, accurately, inexpensively, and noninvasively identify subjects with or at risk for incipient allograft injury (e.g., rejection), discern the type of injury (e.g., antibody mediated versus cellular rejection), and differentiate rejection from other causes of graft damage (e.g., infection). Clinically useful assays have high sensitivity, specificity, and negative/positive predictive values and a diagnostic area under the receiver operator characteristic curve nearing 1.0.^7,13 Serial biomarker measurements should correlate with remission of the pathologic process, especially in response to therapeutic interventions. For most transplant trials, statistical analyses of biomarkers have been on the basis of comparison with the gold standard of histologic detection of rejection on an allograft biopsy. Because pathologic definitions of rejection are subject to intra-/interobserver and/or sampling error–related variability in biopsy readouts,^14,15 such variability must be considered when interpreting biomarker study results. Although graft loss is an undisputable outcome for any biomarker comparison, it is challenging to design prospective studies of sufficient length and power to adequately study graft loss.^16–18

Biomarker development occurs via a lifecycle (Figure 1) that includes discovery, internal single-center, and external multicenter validation, standardization, commercialization, and ultimately, adoption into clinical care. After entrance into the clinical arena, widespread use tends to generate new questions regarding assay utility, potentially spawning second-order, controlled trials.

Figure 1.

Biomarker development should proceed through a lifecyle that includes external validation. A proposed template depicting the various steps involved from biomarker discovery and validation to clinical application in transplantation.

Anti-HLA Antibody Testing by Solid-Phase Assays

As initially published by Patel and Terasaki,¹⁹ preexisting recipient serum antidonor HLA antibodies are associated with early rejection/graft loss (hyper-AR) after kidney transplantation. Accurate detection of these antibodies is essential; crossmatch testing by FDA-approved assays, including solid-phase assays (e.g., luminex based), is now routinely used clinically for pretransplant risk assessments.^19–24 The strengths and weaknesses of each approach have been reviewed elsewhere in detail,^25–34 but conclusions suggest that solid-phase assays (e.g., luminex based) have the highest sensitivity for predicting post-transplant antibody-mediated rejection (ABMR) and/or graft loss. Standardization of solid-phase assay techniques among laboratories is achievable,³⁵ although the assay is not designed (or FDA approved) to be quantitative. Although transplant centers routinely avoid transplantation with DSAs as detected by solid-phase assays with mean fluorescence intensity (MFI) of >10,000 (high risk of hyper-AR), the thresholds for defining positivity and the clinical implications of pretransplant DSAs with lower MFI remain controversial.³⁶

Accumulating evidence from multiple studies associates development of de novo post-transplant DSAs with an elevated risk of late graft loss,^34,37 particularly in the context of medication nonadherence.³⁸ To improve the prognostic utility of de novo DSAs for incipient graft injury, investigators have examined whether various DSA characteristics, including time of development post-transplant, specificity (class 1 versus 2 HLA), isotype (IgG subtypes), strength (MFI or titer), and function (e.g., complement fixing),^34,37 confer increased clinical risk. Reports suggest a higher risk for kidney allograft loss in subjects with serum DSAs that bind/activate complement as measured by a standard solid-phase testing assay that additionally detects C1q binding.^32–34 C1q-positive de novo DSA was associated with a shorter time to graft loss than C1q-negative de novo DSA or the absence of any DSA.³³ Although it was postulated that C1q positivity indicates antibodies preferentially capable of initiating complement-dependent allograft rejection, additional work suggests that C1q positivity is a consequence of higher serum DSA titers³³ rather than complement-activating activity per se.³⁹ Although solid-phase DSA testing is widely used, the clinical utility of C1q binding DSAs (among other innovations)^39,40 as a risk assessment tool and the timing/frequency of DSA testing for detection of de novo DSAs remain unclear. One barrier to implementing routine post-transplant DSA testing is the absence of evidence that available therapies can prevent/reverse incipient allograft injury/loss in DSA-positive transplant recipients.

Assessing Pretransplant Risk for Development of Post-Transplant DSAs

Building on the above-noted observations, research teams have attempted to identify pretransplant biomarkers that predict high likelihood of developing post-transplant DSAs.

Epitope mismatch analysis of donor and recipient HLA polymorphisms builds on current HLA typing to identify donor-recipient mismatches for both class 1 (triplets) and 2 (eplets) HLA at the molecular epitope level. The HLAMatchmaker software is an epitope analysis tool that integrates knowledge of HLA molecule three-dimensional structures⁴¹ with known correlations among sero- or genotyping results at HLA loci to identify polymorphic amino acid differences, which when located on exposed regions, are potential immunogens that stimulate antibody production.^42,43 Studies showed that high numbers of epitope mismatches between donor and recipient^44,45 are associated with an elevated risk of developing de novo DSAs, particularly in kidney transplant recipients nonadherent to immune suppressants⁴⁶ or recipients undergoing immunosuppression withdrawal.⁴⁷ One implication is that individuals with high epitope mismatches may require more immunosuppression to prevent de novo DSAs. Although epitope mismatch analysis requires high-resolution HLA genotyping, which incurs an additional expense, the software is freely available, making this a readily implementable risk assessment strategy that could be used by any transplant center today. Remaining issues requiring attention are multicenter validation of optimal thresholds for positivity and testing the hypothesis that differential treatment strategies on the basis of epitope mismatching will prevent DSA and graft loss in those at highest risk.

Anti-HLA alloantibodies are produced by antibody-producing plasma cells and long-lived memory B cells (Bmems), the latter of which differentiate into plasma cells on re-encountering antigen.⁴⁸ Donor-specific Bmems are detectable in humans independent of whether serum anti-HLA antibodies are demonstrable.^49–51 The B cell ELISPOT assay⁵² detects IgG-producing B cells, including Bmems.⁵³ Frequencies of circulating donor-HLA–reactive Bmems correlate with degree of sensitization and ABMR episodes.⁵⁰ Large European observational studies are ongoing to assess the value of quantifying HLA-specific Bmems in kidney transplantation (O. Bestard, personal communication). Commercialization efforts are underway and may become available to United States transplant centers in the near future.

Assays of T Cell Alloreactivity

Alloreactive T cells are essential mediators of allograft rejection,^54–57 spawning efforts to quantify alloreactive T cell immune responses as potential biomarkers of transplant outcome.

In vitro T cell responses to alloantigens are detectable by proliferative mixed lymphocyte reactions, in which recipient T cells are tested for reactivity to donor cells using various readouts.⁵⁷ Although weak proliferative responses (³H-thymidine incorporation) suggest overimmunosuppression and strong responses suggest elevated risk of allograft rejection, the proliferative mixed lymphocyte reaction has limited predictive value in clinical transplantation.^58,59 As an alternative, Ashokkumar et al.⁶⁰ showed that pretransplant mitogen stimulation of alloreactive T cells induces upregulated surface expression of T cell CD154. Higher expression levels associate with higher risk of liver transplantation rejection in children. Limited data from adult kidney transplant recipients suggest diagnostic potential for ongoing cellular rejection.⁶¹ Although T cell expression of CD154 is a commercially available and FDA-approved assay and can be performed by any clinical laboratory with a flow cytometer, multicenter validation of its diagnostic/prognostic biomarker utility in kidney transplantation remains to be determined.

Measuring ATP generation by mitogen-stimulated CD4 lymphocytes (Immuknow assay)⁶² is an FDA-approved biomarker that is potentially informative in transplant recipients. Results are reported as “within normal limits,” high (suggesting underimmunosuppression), or low (suggesting overimmunosuppression). Observational studies performed in small kidney transplant cohorts provide inconsistent conclusions^63–65 but implied that the assay may be best at detecting overimmunosuppression and an elevated risk of infection. The most informative study was a prospective trial of 202 liver transplant recipients⁶⁶ randomized to receive standard of care immunosuppression versus immunosuppression guided by ATP release assay results. One-year patient survival rates were higher and infection rates were lower in the group receiving ATP release biomarker-guided immunosuppression. Randomized, controlled trials in kidney transplantation have not been performed.

A portion of the alloreactive T cell repertoire derives from the memory pool, and memory donor–reactive T cells can negatively affect transplant outcomes.^67,68 Donor-reactive memory T cells are detectable by flow cytometry⁶⁷ and cytokine enzyme–linked immunosorbent spot (ELISPOT).⁶⁹ IFN-γ ELISPOT positivity before transplantation was shown to correlate with elevated risk of developing post-transplant acute cellular rejection (ACR) and/or poor graft function in multiple cohorts of renal recipients, particularly those who were not given induction therapy with T cell–depleting agents.^70–76 The ELISPOT procedure requires donor cells and takes 24–36 hours, which limits application in deceased donor recipients. An alternative T cell reactivity index (panel of reactive T cells [PRT]) is analogously performed using a pool of donor cells reflective of the organ donor pool; pretransplant PRT results correlate with an elevated risk of post-transplant allograft injury.^77–79 The complexities of performing ELISPOT assays require careful standardization⁸⁰ for reproducibility. Commercialization for donor-specific assays requires obtaining donor cells, but commercialization efforts for PRT assays are underway. Prospective ELISPOT-guided interventional trials are needed to determine whether targeting specific therapies to subjects exhibiting ELISPOT positivity will improve short- and long-term outcomes in kidney transplant recipients.

Urine Biomarkers

Suthanthiran and colleagues^81–86 first recognized the potential of studying urine as a window into kidney allograft inflammation. Multiple single-center and small observational studies showed that preselected immune transcripts in urinary cell RNA measured by quantitative PCR (qPCR) can noninvasively differentiate ACR from non-ACR in kidney transplant recipients.^81–86 Detailed summaries of the published findings are summarized elsewhere.^5,13 The most informative findings derive from a large, multicenter, prospective, National Institutes of Health (NIH)-funded, observational study that showed that an 18S-ribosomal RNA normalized 3-gene signature (CD3ε, IP-10, and 18S-rRNA) distinguished ACR from non-ACR with high accuracy. The biomarker was detectable before the clinical recognition of the rejection episode.^86,87 Standardization of urinary gene expression has been accomplished,⁸⁸ and optimal control genes have been identified,⁸⁹ making it feasible to adopt urinary gene expression profiling by qPCR in the clinic. Importantly, RNA must be produced rapidly from the collected urine (within 4 hours) to be informative.⁸⁸ Urine cell qPCR testing is not yet widely commercially available. Whether clinical outcomes are positively affected by urine qPCR-driven interventions in the absence of biopsies remains to be determined via prospective, randomized, multicenter studies.⁵

Among other testing strategies close to commercialization is urinary protein measurement of the chemoattractant chemokines CXCL9 and/or CXCL10.^47,90–92 Early cross-sectional and relatively small single-center studies showed strong correlations between urinary CXCL9 and T cell–mediated rejection,^90,91,93 with some indicating strong correlations between urinary CXCL10 and ABMR.^92,94,95 In an NIH-funded, prospective, multicenter trial, urinary CXCL9 (ELISA) outperformed CXCL9-mRNA levels (qPCR) for the diagnosis of ACR.⁹⁰ In a separate, prospective, randomized, tacrolimus-withdrawal study, elevations in urinary CXCL9 predicted incipient acute cellular rejection, providing the first evidence that serial monitoring of a urinary biomarker could be used to accurately guide clinical decision making.⁴⁷ Results reported at the American Transplant Congress 2016⁹⁶ suggest that serial urinary CXCL9 measurements can be informative with regard to efficacy of antirejection therapy. ELISAs can be performed by most clinical laboratories, making this approach readily adoptable in clinical care, although commercialized assays are yet to be routinely offered in the United States. Whether treatment decisions on the basis of urinary CXCL9 affect clinical outcomes in kidney transplantation remains to be tested.

Molecular Approaches

High-throughput technologies for broadly analyzing gene transcripts in blood or tissue, including microarray and RNA-sequencing (RNA-seq), can simultaneously map differential expression patterns of thousands of transcripts and associate them with particular outcomes or disease states. Microarrays^5,97 are reported as fold changes in reference to a standard. Falling costs make it possible that microarray-based approaches could be adapted to clinical care.

RNA-seq provides single-base resolution, absolute transcript quantification from total or fractionated RNA, detection of splice variants, a large dynamic range for detection of expression levels (8000-fold changes), and the ability to discover novel transcripts.⁹⁸ The enormous amount of data obtained from each RNA-seq experiment, the time involved in analysis, and the relative inability to analyze data regarding one or a few candidate transcripts alone could limit the use of this technology as a clinical biomarker assay.

Biopsy Transcript Assessment as Biomarkers

The biopsy transcriptome as interrogated by qPCR^99–101 or high-throughput techniques^102–107 confirms allograft upregulation of T cell–related/immune transcripts (e.g., chemokines, CD3, FOXP3, IFN-γ, Fas/FasL, perforin, and Granzyme B) compared with those without rejection. Beyond these observations, allograft transcriptome studies suggest that molecular phenotyping adds to the histologic diagnosis of ACR,^{103,104,108,109} in part by identifying histologically undifferentiated lesions,¹⁰⁵ improving prognostication,¹⁰⁵ and providing insights into the clinical importance of borderline ACR¹⁰⁶ and isolated “v-lesions.”¹¹⁰ Allograft transcriptomic data may be relevant for diagnosing and prognosticating ABMR in the absence of classic histologic criteria and/or C4d positivity.^{107,111–113}

As one example of how “molecular microscope” analyses of graft tissue improve prognostication,¹¹⁴ Halloran and colleagues¹⁶ developed a biopsy-derived molecular classifier that independently associated with an elevated risk of graft loss within 1 year. The molecular risk score had a higher hazard ratio for graft loss than histologic or clinical factors. These intriguing results have the potential to overcome variability in interpretation of kidney allograft pathology¹¹⁵ but will require prospective validation, standardization, and commercialization before they will be adopted into clinical care.

Data from the prospective, NIH-funded Genomics of Chronic Allograft Rejection Study provide additional insight.¹⁸ Microarray analysis was performed on 3-month surveillance biopsies to identify genes and pathways associated with 3- and 12-month Chronic Allograft Dysfunction Index scores. A biomarker comprised of a refined 13-gene set accurately predicted impending fibrosis at 12 months, predicted graft loss, independent of simultaneous ACR was validated in external cohorts, which included for-cause biopsies,^12,16 and was independently associated with histologic progression and graft loss.¹¹⁶

Together, these genomic analyses suggest that, if transcriptome assessments were performed on surveillance or for-cause biopsies, prognostic gene sets could risk stratify allografts for histologic progression or graft loss and thereby, guide treatment decisions. Ultimately, prospective studies will be required to test whether therapeutic interventions on the basis of biopsy-derived molecular signatures will improve outcomes compared with therapeutic interventions driven by results of standard histology.

Blood Transcriptome Assessments

Assessment of peripheral blood gene expression profiles circumvents the need for a biopsy and could serve as an attractive immune monitoring tool.¹¹⁷ High-throughput transcriptional profiling of total RNA from peripheral blood validated across three microarray platforms identified a specific five-gene set that correlated well with AR/no AR status on simultaneously procured allograft biopsies.¹¹⁸ The team subsequently developed and validated a 17-gene signature that accurately diagnosed AR in a multicenter adult and pediatric cohort.¹¹⁹ Other studies by Murphy et al.¹²⁰ identified a peripheral blood gene signature that showed diagnostic accuracy for subclinical rejection, including borderline patients, early post-transplant. Although these approaches could potentially be used to guide therapeutic decisions in the absence of biopsy tissue, multicenter validation and commercialization are required. Adoption of such assays into clinical care will require clinical experience and additional studies to support the hypothesis that interventions on the basis of peripheral blood signatures can positively influence patient outcomes.

Cutting Edge Biomarkers

Although not yet commercially available, several novel approaches reported since 2010 have potential to become clinically useful transplant biomarkers.

Quantification of donor-origin cellfree DNA (cfDNA) in recipient blood or urine (by filtration/secretion or injury) has potential diagnostic utility for detecting rejection.^121–124 Donor-origin cfDNA is likely released into recipient plasma during allograft injury and quantifiable as a fraction of total cfDNA from blood or urine.¹²⁵ The donor-recipient cfDNA distinction is made by identifying genetic differences between donor and recipient.^125–128 Small studies have shown associations between AR and the quantity of plasma and urinary donor cfDNA in cardiac¹²⁹ or kidney transplant recipients.^126,127 Available evidence suggests that elevations in cfDNA cannot distinguish AR from other causes of graft injury, and additional markers may need to be layered over this assay to improve specificity or accurately relay need for allograft biopsy.^125,126,130 Although cfDNA commercialization is underway (Allosure; CareDx, Brisbane, CA), multicenter, prospective, observational studies (ongoing as part of the NIH-funded Clinical Trials in Organ Transplantation 19 study; www.CTOT.org) and ultimately, interventional studies will be required to delineate the utility of this promising approach as a transplantation biomarker.

Urine proteomic profiling studies performed in the early 2000s identified several protein signals that differentiated subjects with AR from those with no AR without additional characterization of the proteins associated with these signals.^{90,91,131–135} Studies published since 2010 using for-cause biopsies have identified numerous candidate proteins differentially identified in urine of subjects with AR compared with those with other pathologies.^136–138 Urine proteomes of transplant recipients with AR contain upregulation of immune response–related proteins and acute-phase reactants, with downregulation of solute/ion transport proteins.¹³⁸ The overlap among proteins in various studies has been minimal.¹³² Standardization, validation, and commercialization remain barriers to moving proteomic approaches to the clinic.

Metabolomics involves the simultaneous profiling of dozens of small molecule metabolites (<1500 D) in tissue and biologic fluids,^139,140 and because the transcriptome or proteome can undergo additional regulatory alterations, the metabolome can reflect the downstream products of cellular processes. A composite urine metabolite-mRNA signature accurately differentiated cellular AR from non-AR in an adult cohort.¹⁴⁰ The metabolite signature alone was less accurate, suggesting that combinations of markers applied sequentially or simultaneously may ultimately be more effective for noninvasive diagnosis and surveillance. Multicenter validation and commercialization remain to be done.

Heritable genomic changes in either donor or recipient DNA single-nucleotide polymorphisms (SNPs) are stable and measurable. Multiple SNPs in various genes have been shown to correlate strongly with certain transplant outcomes,^141–144 and they have the potential to serve as risk stratifiers pretransplantation. Although these observations offer promise, small sample sizes, low odds ratios observed for outcomes between exposed and nonexposed, limited validation, and lack of evidence that immunosuppression choices on the basis of SNP risk stratification influence outcomes have thus far restricted the SNP analysis to research settings.

Sequencing of DNA sequences at the T cell receptor β-chain CDR3 region can identify a fingerprint of specifically donor-reactive T cell clones within recipients. In a small series, allograft tolerance was associated with the disappearance of these donor-reactive clones, whereas the lack of tolerance was associated with their persistence.¹⁴⁵ If the appearance of these T cells in the blood or urine precedes the development of AR and if reproducible in larger cohorts of patients on usual immunosuppression, then this assay could help monitor donor-reactive T cell immunity specifically and serve as a surveillance tool.

Potential Framework for Kidney Transplant Monitoring

A clinically useful immune monitoring strategy must build on the known mechanisms of allograft injury, the data indicating time dependence of alloimmune injury,^{1,18,108,109,146–148} and the recognition that subclinical inflammation contributes to chronic allograft damage.^109,147,148 We contend that multiple assays will need to be deployed pretransplant and recurrently post-transplant to optimally detect incipient injury and individualize therapy. As a framework (Figure 2), we envision comprehensive risk assessment via cellular assays, genomic technologies, and protein and metabolite profiling. SNP and epitope mismatch analysis, IFN-γ ELISPOT or PRT assay, and anti-HLA antibody assessment are likely to be informative pretransplantation and guide initial choices regarding immunosuppression. Noninvasive urinary gene expression, urine chemokine testing, and blood gene expression patterns are likely to replace or minimize biopsies to diagnose ongoing injury. Optimal post-transplant time points for other biomarkers need additional study. We envision that serial urinary CXCL9 or urinary RNA measurements will be required during the first few months post-transplantation (highest risk for AR), particularly during changes in immunosuppressive drug dosing. Peripheral blood expression profiling performed early (approximately 3 months) or during allograft dysfunction¹¹⁹ could indicate ongoing subclinical or clinical rejection and trigger a biopsy¹²⁰; a biopsy transcriptome assessment could then help prognosticate allograft injury with greater granularity than histology,¹⁸ guiding maintenance therapy and/or follow-up. Although DSA measurements could be performed in all subjects, intense monitoring schedules might be better restricted to those individuals deemed at highest risk (e.g., with high epitope mismatches^45,47 with positive donor-specific IFN-γ ELISPOT assays⁴⁷) and/or at risk for nonadherence⁴⁵ or who experienced prior AR episodes.¹⁴⁷

Figure 2.

Biomarker-based individualization of immunosuppression of kidney transplant recipients will require use of multiple testing strategies. The x axis depicts the time points relative to the transplantation (Tx) when these biomarkers could be applied. The dashed horizontal line separates the conventional clinical and laboratory tests that are used currently. The continuous horizontal line describes common clinical immunologic events encountered in Tx with respect to times after Tx; the widths of the T cell–mediated rejection (TCMR) and ABMR lines refer to frequencies of these events. PRA, panel of reactive antibody.

Perspective and Future Directions

Although significant effort by numerous research groups has been focused on developing biomarkers to detect kidney allograft rejection and/or predict graft loss, few assays have moved from the research arena to a clinically implementable testing strategy that could guide therapeutic decision making.

The road to clinical biomarker implementation in transplantation is fraught with barriers that include high costs of commercializing assays for relatively small market sizes and a field that is being driven by physician-scientists rather than commercial entities. Improved partnering approaches among scientists, academic institutions, and various pharmaceutical and/or biotechnology companies will be required to commercialize biomarkers and make them available to transplant clinicians, particularly for assays with limited intellectual property rights, because methods are in the public domain (e.g., ELISAs for chemokines).

In addition, the relative lack of information on how to clinically use biomarkers remains problematic. The transplant field needs controlled trials, in which subjects are randomized to receive standard of care treatment or individualized care on the basis of prespecified biomarker-directed strategies. End points should include graft function/survival, rates of infectious complications, and biopsy-related complications. The hypothesis to be tested is that biomarker-driven care will decrease the need for biopsies and limit infectious complications while improving graft function/survival. Such controlled trials would provide the best evidence for how to use a given marker, but commercialized, FDA-approved biomarkers can also be broadly tested by the transplant community without hard evidence. Such real world experience can be hypothesis generating and drive improved trial design to address how an assay could be optimally used in a given setting.

Overall, although work remains to be done, the past decade witnessed significant progress in biomarker development, testing, and clinical implementation in transplantation. We look forward to the next decade as we anticipate that the kidney transplant community will move beyond protocol-based immunosuppression and begin to individualize treatment of transplant patients on the basis of results of objective and reliable biomarkers.

Disclosures

None.