Urine CXCL10 as a biomarker in kidney transplantation

Abstract

Purpose of review

Urine CXCL10 is a promising biomarker for posttransplant renal allograft monitoring but is currently not widely used for clinical management.

Recent findings

Large retrospective studies and data from a prospective randomized trial as well as a prospective cohort study demonstrate that low urine CXCL10 levels are associated with a low risk of rejection and can exclude BK polyomavirus replication with high certainty. Urine CXCL10 can either be used as part of a multiparameter based risk assessment tool, or as an individual biomarker taking relevant confounders into account. A novel Luminex-based CXCL10 assay has been validated in a multicenter study, and proved to be robust, reproducible, and accurate.

Summary

Urine CXCL10 is a well characterized inflammation biomarker, which can be used to guide performance of surveillance biopsies. Wide implementation into clinical practice depends on the availability of inexpensive, thoroughly validated assays with approval from regulatory authorities.

INTRODUCTION

During the last decades kidney graft survival has mainly improved due to better short-term outcomes including relatively low clinical rejection rates of 10–20% within the first year after transplantation with currently used immunosuppressive therapies [1,2]. On the other hand, efforts are still needed to improve the long-term outcome of kidney allograft recipients. To date, development of immune-mediated allograft damage is still the leading cause of graft failure [3,4,5]. The problem is that compared to the relatively small number of clinical rejection episodes, up to 50% of patients experience subclinical rejection within the first year posttransplant, which can today only be detected by an invasive procedure like performance of surveillance biopsies at regular intervals [6,7]. We learned from previous studies that these so-called smoldering rejection processes are clinically relevant [8,9]. However, there is still a matter of debate nowadays at which timepoints surveillance biopsies should be performed and whether modern noninvasive monitoring strategies like the use of a biomarker-driven routine follow-up could help to target the performance of surveillance biopsies because of their invasive nature and risks (i.e., bleeding complications and sampling errors which may obscure the diagnostic usefulness).

Furthermore, infection complications are frequent and BK polyomavirus (BKPyV) replication an important issue after kidney transplantation [10,11]. The problem is that the currently used immunosuppressive therapies are not specific and suppress both the allo-reactive as wells as pathogen-specific defense mechanisms. So far, the immunosuppressive burden of kidney transplant patient is still largely driven by center-specific therapy protocols and the amount of immunosuppression is mainly adjusted by infectious complications (i.e., over-immunosuppression) and the occurrence of clinical rejection as a histology result, obtained by for cause biopsies due to graft dysfunction (i.e., under-immunosuppression). With other words, so far there is a lack of precision medicine in kidney transplantation. This affects the health status of each individual patient and impacts the healthcare system.

Thus, from a clinical perspective, the leading goals of novel biomarker-driven noninvasive monitoring posttransplant are to identify patients at risk for (sub)clinical rejection and thus allow a more targeted use of surveillance biopsies, to monitor rejection processes and predict response to treatment, but also to detect over-immunosuppressed patients. Ideally, novel noninvasive biomarkers could help to tailor immunosuppressive therapies to the individual needs of the patients [12]. 

Box 1.

no caption available

URINE CXCL10 IN REJECTION

As the product of the renal ultrafiltrate, urine is in ‘close contact’ with the inflammatory process taking place during rejection within the allograft, making it a promising source for the detection of early inflammation biomarkers. One of the most promising urine biomarkers for the detection of early allograft rejection is the chemokine C–X–C motif ligand 10 (CXLC10). CXCL10 is secreted by infiltrating inflammatory cells and renal tubular cells and is involved in leukocyte recruitment and mediating the CD4 Th1 response [e.g., up-regulation of pro-inflammatory cytokine production like interferon (IFN)-gamma, interleukin (IL)-2 and tumor necrosis factor (TNF)-alpha] [13,14]. In acute allograft rejection, CXCL10 is highly expressed in infiltrating leukocytes and renal tubular cells [15,16]. There is a large body of evidence, mainly from observational studies of prospectively collected cohorts that urine CXCL10 is an important early noninvasive diagnostic biomarker for (sub)clinical allograft rejection in adult and pediatric populations [17–30]. Nevertheless, urine CXCL10 is an allograft inflammation biomarker which has important confounders. The most important ones in kidney transplantation are bacterial urinary tract infection (UTI) and BKPyV infection, as it is not a specific biomarker for alloimmune-mediated inflammation. This has been thoroughly investigated in different studies [21,23,31,32^▪▪,33^▪▪], including two studies using a multiparametric model taking meaningful confounding factors into account [32^▪▪,33^▪▪].

Besides the diagnostic value, previous studies also showed that urine CXCL10 is predictive for long-term outcomes [34,35]. Moreover, a recent study including 141 patients with 182 clinically indicated allograft biopsies performed >12 months posttransplant demonstrated that high urine CXCL10 was associated with worse outcome even in histologically quiescent patients (i.e., with histology including inflammatory processes not reaching current Banff classification criteria of acute rejection) [36].

The recently published results of the first single-center randomized clinical trial using a biomarker-based monitoring strategy in kidney transplantation could not show a beneficial effect on one-year outcomes in a study population of 241 patients (i.e., n = 120 stratified into an intervention and n = 121 stratified into a control arm, respectively) with broad inclusion criteria reflecting a real-life population, and with high compliance with the study protocol including one-year surveillance biopsies in > 80% of patients [37^▪]. Nevertheless, the study revealed precious observations. First, persisting low urine CXCL10 levels reflected a low risk for (sub)clinical rejection. On the other hand, persisting elevated indicated ongoing inflammation due to allograft rejection, UTI or BKPyV infection. Overall, the study supports the use of urine CXCL10 to assess the inflammatory status of the renal allograft for individual patients. Furthermore, the study is a pioneering source for future trial designs aiming to explore the clinical utility of noninvasive biomarkers, like for example definition of the study population size [37^▪]. Nevertheless, to pick out three limitations of the study, we first need to mention that the study was not powered to detect clinically relevant differences when the more stringent Banff 2019 [38] compared to the Banff 2015 [39] classification was applied. Second, the assumed effect size of 50% reduction in the primary outcome was ambitious and according to very recent reports likely difficult to reach [40,41]. Third, the incidence of the primary outcome was largely driven by the rejection category of borderline changes, which changed from Banff 2015 [39] to Banff 2019 [38] classification. The clinical relevance of borderline changes is still debated and thus might be a problematic trial endpoint [42]. Thus, further studies including multicenter studies are warranted to characterize the clinical utility of urine CXCL10 monitoring in kidney transplantation. Behind this background, the results of the ongoing randomized controlled multicenter trial (ClinicalTrials.gov: NCT03206801 [43^▪]), in which investigators use an enrichment strategy to randomize 250 patients with elevated urine CXCL10 1 : 1 into an intervention and a control arm, are expected.

Another approach of noninvasive monitoring of acute rejection compared to the use of urine CXCL10 as an individual biomarker is to integrate chemokine biomarkers into a multiparameter assessment tool. Tinel et al.[32^▪▪] developed a noninvasive multiparametric model for the diagnosis of acute rejection in kidney allografts, within a discovery and two independent validation cohorts. Optimizing the model to eight parameters including the urine chemokines CXCL9 and CXCL10 as well as clinical/laboratory parameters diagnosed acute rejection with an area under the curve (AUC) of 0.85, 95% confidence interval (CI) 0.80–0.89 in biopsies performed at a median time of 8 months posttransplant (88.2% were clinically indicated) [32^▪▪]. In addition, Van Loon et al.[33^▪▪] evaluated the diagnostic performance of urine CXCL9 and CXCL10 for detection of acute rejection when integrating these biomarkers into a multivariable model with clinical parameters including eGFR, donor-specific antibodies and BKPyV viremia, and compared the model to the histology diagnosis of acute rejection according to Banff 2019 criteria [38]. In total, 1559 biopsy-paired urine samples were included from 622 transplantations, of which 1219 belonged to surveillance biopsies (i.e., performed 3 months, 1 and 2 years posttransplant) and 340 to clinically indicated biopsies (i.e., performed up to 8 years after transplantation) [33^▪▪]. The chemokines integrated into the 5-parameter model demonstrated an AUC of 81.3, 95% CI 77.6–85.0 for the detection of acute rejection [33^▪▪]. However, borderline changes were not considered as a rejection phenotype as they were not treated in their center [33^▪▪].

To sum up, serial noninvasive surveillance of kidney transplant recipients by urine CXCL10 might become a promising tool for the detection of allograft rejection and individualized adaption of immunosuppression either used as an individual biomarker taking its confounders into consideration or integrated into a multiparameter assessment tool. An overview of the diagnostic performance (i.e., sensitivity and specificity) of urine CXCL10 for acute rejection in different studies (summarized within this review) is depicted in Fig. 1.

FIGURE 1.

Overview of the diagnostic performance (i.e., sensitivity and specificity) of urine CXCL10 for acute rejection in different studies. Green dots indicate the diagnostic performance of urine CXCL10 used as an individual biomarker. The blue triangle indicates the diagnostic characteristics of urine CXCL10 to diagnose acute rejection within the first prospective study using a biomarker-driven monitoring strategy for the performance of allograft biopsies [37^▪]. The pink square indicates the diagnostic performance when urine CXCL9 and CXCL10 chemokines are integrated within a multiparameter model [33^▪▪]. Green dots outlined in black indicate the diagnostic value of urine CXCL10 for the detection of isolated subclinical rejection.

URINE CXCL10 IN BK POLYOMAVIRUS INFECTION

BKPyV persists in tubular epithelial cells after primary infection, and it can start to replicate in significant numbers again under immunosuppressive therapy. First, the viral load in urine will increase (i.e., DNAuria), which can be accompanied by shedding of decoy cells in case of high level DNAuria defined as >7 log₁₀ copies/ml. If BKPyV replication increases further, BKPyV DNA can also be detected in the blood (i.e., DNAemia). High BKPyV DNAemia >4 log₁₀ copies/ml is associated with positive SV40-staining in renal allograft biopsies (i.e., BKPyV associated nephropathy; BKPyVAN) [44]. Many previous studies have shown that BKPyV replication leads to an inflammatory response in the allograft with elevated urine CXCL10 levels [21,23,32^▪▪]. Three groups recently reported more granular data on the kinetics of BKPyV replication and urine CXCL10 levels.

Weseslindtner et al.[45^▪▪] investigated 56 patients with a total of 148 paired urine and blood samples obtained at various stages of stages of BKPyV replication. They found a stepwise increase of urine and blood CXCL10 levels from DNAuria to DNAemia, DNAemia with decoy cells, and DNAemia with BKPyVAN, highlighting a close correlation of urine CXCL10 with the extent of BKPyV replication. Tinel et al.[46^▪▪] investigated in two large cohorts the relationship between urine CXCL10 and BKPyV replication: a cross-sectional cohort of 474 paired urine/blood/biopsy samples, and a longitudinal cohort of 1184 urine samples of 60 patients with DNAemia. Their key observation was that urine CXCL10 not only followed the DNAemia, but high urine CXCL10 at first detection of DNAemia was a prognostic marker of subsequent functional decline. This suggests that the intensity – and maybe also the duration – of the inflammatory response is critical for the outcome in patients with DNAemia. Haller et al.[47^▪] investigated 763 urine samples obtained from 235 patients participating in an interventional, randomized trial. They largely confirmed the observations made by Weseslindtner et al. and Tinel et al. In addition, they reported that low urine CXCL10 levels (i.e., <3 ng/mmol creatinine) have a negative predictive value of 97% for exclusion of significant decoy cell shedding, and a negative predictive value of 99% for exclusion of DNAemia. Furthermore, this group observed that a sudden and extreme increase of urine CXCL10 levels with stable allograft function raises the suspicion of high level DNAuria or DNAemia.

In summary, these recent data demonstrate that urine CXCL10 closely follows the extent of BKPyV replication as well as the kinetics of DNAemia. Low urine CXCL10 levels can virtually exclude relevant BKPyV replication, and high CXCL10 are a prognostic marker for poor outcome in patients with BKPyV DNAemia.

METHODOLOGICAL AND TECHNICAL ISSUES

Currently, several CXCL10 immunoassays exist for different platforms (summarized in Table 1 of reference [48^▪]; and [33^▪▪]). Most of these assays have a lower limit of detection in the 1-2 pg/ml range, which is sufficient for the purpose of renal allograft monitoring. Ho et al.[48^▪] performed a thorough multicenter validation of a urine CXCL10 assay on the Luminex platform. They demonstrate that this assay is robust, reproducible, and accurate. Further advantages of the Luminex platform are its wide availability in HLA laboratories, and the possibility to measure only a few samples in a cost-efficient way compared to ELISA plate-based assays, where often at least half a plate must be used.

Urine is a complex sample source due to its variable cell and protease composition, broad pH range, and different analyte dilution. Therefore, it is of major importance to investigate, which factors influence the measurement of urine CXCL10 and to what extent. Handschin et al.[49^▪] analyzed processed urine samples (i.e., stored at 4°C until centrifuged at 3000 rpm and stored at −70°C). They showed that CXCL10 values are very stable in processed urine at 4°C and 25°C, and not influenced by hematuria, proteinuria, urine pH, and squamous cell count. By contrast, BKPyV replication and leucocyturia are major confounders, which has been shown by other investigators as well [32^▪▪]. The remaining question is, how stable urine CXCL10 is in unprocessed urine samples. This was investigated in five urine samples, which were split immediately after collection and stored at 4°C or at room temperature for up to 3 days, while performing CXCL10 measurements 1 h, 8 h, 24 h, 2 days and 3 days after the splitting procedure. In four of the five patients’ urine CXCL10 remained stable up to 3 days after collection both at 4°C and at room temperature. All these four patients had no leucocyturia and no decoy cell shedding. Interestingly, in the remaining patient with significant BKPyV DNAemia and decoy cell shedding we observed first a doubling of urine CXCL10 peaking at 24 h and then at 3 days a drop back to levels obtained at 1 h after collection. This observation was made in both samples kept at 4°C or at room temperature but was more pronounced in the later one (see Fig. S3A/S3B in reference [48^▪]). Although based on only a few samples, we believe that CXCL10 is stable in unprocessed urine even at room temperature, if no significant amounts of leucocytes or decoy cells are present. We hypothesize that these cells might first secrete additional CXCL10, which is later degraded by proteases released after cell death. We recommend that urine is processed within 8 h after collection or kept at 4°C if this is not possible.

SUMMARY

Urine CXCL10 has emerged as a promising biomarker for posttransplant monitoring. Based on the existing data, urine CXCL10 can either be integrated into a multiparameter assessment tool or used as an individual biomarker taking its confounders into consideration. The following diagnostic approach can be envisioned (Fig. 2). If urine CXCL10 is low, the risk of rejection as well as BKPyV replication is very low. In such cases surveillance biopsies could be omitted, reducing performance of surveillance biopsies by about 60%, while missing only few subclinical rejection episodes [33^▪▪,37^▪,50]. On the other hand, if urine CXCL10 is highly elevated, significant BKPyV replication or a rejection process is very likely, after exclusion of a UTI.

FIGURE 2.

Risk of rejection and BKPyV replication dependent on urine CXCL10 levels [ng/mmol creatinine] measured on the MesoScale platform.

An important aspect to bring urine CXCL10 measurements into clinics is the availability of assays, which are approved by the regulatory authority, can be reimbursed, are inexpensive, have a fast turn-around-time, and which can be easily implemented in the laboratory for regular monitoring.

Acknowledgements

The authors would like to thank the teams of the renal transplant outpatient clinic and the HLA laboratory.

Financial support and sponsorship

This work was supported by the Swiss National Science Foundation (Grant No. 32003B_169310/1).

Conflicts of interest

There are no conflicts of interest.

REFERENCES AND RECOMMENDED READING

Papers of particular interest, published within the annual period of review, have been highlighted as:

• ▪ of special interest

• ▪▪ of outstanding interest