Longitudinal assessment of the CXCL10 blood and urine concentration in kidney transplant recipients with BK polyomavirus replication—a retrospective study

Summary In kidney transplant recipients (KTRs), BK polyomavirus (BKPyV) replication may progress to polyomavirus‐associated nephropathy (PVAN). In this retrospective study, we assessed the chemokine CXCL10 in urine and blood samples consecutively acquired from 85 KTRs who displayed different stages of BKPyV replication and eventually developed PVAN. In parallel to progression toward PVAN, CXCL10 gradually increased in blood and urine, from baseline (prior to virus replication) to BKPyV DNAuria (median increase in blood: 42.15 pg/ml, P = 0.0156), from mere DNAuria to low‐ and high‐level BKPyV DNAemia (median increase: 52.60 and 87.26 pg/ml, P = 0.0010 and P = 0.0002, respectively) and peaked with histologically confirmed PVAN (median increase: 145.00 pg/ml, P < 0.0001). CXCL10 blood and urine levels significantly differed among KTRs with respect to simultaneous presence of human cytomegalovirus (P < 0.001) as well as in relation to the clinical severity of respective BKPyV DNAemia episodes (P = 0.0195). CXCL‐10 concentrations were particularly lower in KTRs in whom BKPyV DNAemia remained without clinical evidence for PVAN, as compared to individuals who displayed high decoy cell levels, decreased renal function and/or biopsy‐proven PVAN (median blood concentration: 266.97 vs. 426.42 pg/ml, P = 0.0282). In conclusion, in KTRs CXCL10 rises in parallel to BKPyV replication and correlates with the gradual development of PVAN.

Although BKPyV is the main causative agent, JCPyV may also trigger PVAN [5,13]. Quantitative measurement of BKPyV with or without JCPyV DNA in urine and blood is therefore commonly undertaken during the post-transplant surveillance [2,3,14,15]. Additional correlates of clinical progression are emergence of decoy cells in the urinary sediment, an indicator for cytopathic effects due to urothelial cell infection, the detection of viral aggregates by the Haufen test, BKPyV VP1 mRNA as well as histological evidence of viral inclusions, tubular injury, and inflammatory infiltrates [2][3][4][16][17][18].
Host factors, as well as viral determinants, account for the complex pathogenesis of PVAN [3,4]. Immune responses, which on one hand, are essential to control viral replication may on the other hand mediate renal injury. In this process, like in all inflammatory processes, chemokines, structurally related, small chemotactic cytokines orchestrate cell-cell signaling and immune cell trafficking [19,20]. Since chemokines shape immune responses against specific pathogens by recruiting particular subsets of leukocytes to the infection site, they play an essential role and may be used as inflammatory biomarkers in viral diseases [21][22][23].
In this study, we longitudinally assessed levels of CXC-ligand 10 (CXCL10), a chemokine which attracts and activates immune cells with a cytokine profile of T-helper cells Type 1 (Th1), aiming to determine whether CXCL10 correlates with an increase and clinical progression of BKPyV replication [24][25][26]. Our analysis also included the chemokines CCL8, CXCL16, and CCL20, candidate markers for outcome and severity of human cytomegalovirus (HCMV) infections in solid organ recipients [22,23,27].

Patients and sample acquisition
This retrospective study included pairs of blood and urine samples from a total of 95 KTRs (35 female, 60 male) including controls, who underwent KTX between March 2008 and September 2014. All KTRs underwent virological routine testing which consisted of PCR analyses for BKPyV and JCPyV DNA in urine and blood, collected pairwise on the same day of the post-transplant follow-up, respectively, and HCMV DNA in blood. Eighty-five patients were selected for the study by retrospective review of virological test results and were included when at least one episode of BKPyV DNAemia with DNA loads ≥1000 copies/ml occurred during the post-transplant follow-up. Ten KTRs, who underwent transplantation and post-transplant surveillance during the same time period, served as controls since they lacked any evidence of BKPyV, JCPyV, or HCMV replication (see Table S1). Further information on the immunosuppressive treatment, the schedule of sample acquisition and virological routine testing as well as on the storage of samples is given in Supplemental Material and Methods. Sample collection and storage as well as the whole study protocol was approved by the local ethics committee of Medical University of Vienna, which concluded that no written informed consent from the patients was required (EK1035/2016, EK2064/2016). The study was performed in accordance with the Helsinki Declaration.

Quantitative PCR
Information on the protocols used for quantitative BKPyV, JCPyV, and HCMV PCR analyses is given in Appendix S1.

Criteria for grouping patients and selection of samples
Out of the 85 KTRs, who were included in the study because they displayed at least one episode of BKPyV DNAemia, 56 displayed BKPyV DNAemia in the absence of significant JCPyV and HCMV DNAemia. Thirteen KTRs simultaneously developed BKPyV and JCPyV DNAemia (>1000 copies/ml, respectively) and 16 simultaneously developed BKPyV and HCMV DNAemia. Among these groups, there was no significant difference with respect to gender, age, transplant type, donor characteristics, cold ischemia time, delayed graft function, HLA-mismatches, AB0 and HLA-incompatibilities, duration of follow-up and the median time point BKPyV DNAemia occurred (Table S1).
To assess individual CXCL-10 kinetics during progression of BKPyV replication, we retrieved and analyzed 148 specific blood/urine sample pairs, collected from the 56 KTRs with mere BKPyV HCMV DNAemia due to routine testing. These samples specifically fulfilled the inclusion requirement that they had been collected during different stages of BKPyV replication (mean number of sample pairs per patient: 2.64, range: 2-4). As shown in Table 1 BKPyV, JCPyV, and HCMV DNA in urine and blood, no evidence for any other infection nor rejection, normal levels of C-reactive protein and leukocytes, normal urinary cytology, no drop in renal function), (ii) BKPyV detection only in urine (BKPyV DNAuria) but not in blood, (iii) BKPyV DNAuria plus low-level BKPyV DNAemia (blood DNA levels <1000 copies/ml), (iv) BKPyV DNAuria plus BKPyV DNAemia (blood DNA levels ≥1000 copies/ml) but no significant decoy cell levels, (v) BKPyV DNAemia (>1000 copies/ml) together with detection of decoy cell levels >20%, a decrease of eGFR AE histological evidence for PVAN. The mean sampling intervals are shown in Table 1. When multiple samples were acquired from one patient during the same stage of BKPyV replication, only the initial sample that fulfilled the respective criteria was selected for the study.
In order to compare chemokine levels among individual patients, the 56 KTRs with mere BKPyV DNAemia were sub-grouped based on the clinical severity of the respective BKPyV DNAemia episode. Although their BKPyV DNA loads exceeded 1000 copies/ml, 9 of the 56 KTRs displayed no significant decoy cell levels (<5%) in urinary sediment during any BKPyV DNAemia episode, renal function remained stable and no biopsy was therefore performed during the entire follow-up. The other 47 KTRs, in contrast, displayed at least one BKPyV DNAemia episode in which significant decoy cell levels (defined as >20%) were concomitantly detected. Of those 47, 23 (48.9%) additionally showed a ≥15% decrease of estimated glomerular filtration rate (eGFR), as compared to the mean of three measurements which immediately preceded the initial occurrence of the respective high-level DNAemia episode (median interval: 41 days, range: 13-98). In 15 patients, PVAN was confirmed with a renal biopsy obtained within a week before or after blood/urine sampling. In the remaining KTRs, no biopsies were performed within this time period.
In addition to the 47 patients with mere BKPyV DNAemia, all 13 KTRs, who simultaneously displayed high-level BKPyV and JCPyV DNAemia, showed decoy cells >20%. An eGFR decrease occurred in 6 (46.2%) and PVAN was histologically confirmed in 5 of these 13 KTRs.
As described, urine/blood sample pairs from 10 KTRs who did not display any BKPyV, JCPyV, and HCMV replication were included as controls, and samples from these controls fulfilled the same criteria as baseline samples (C-reactive protein and leukocytes in normal range, normal urinary cytology, and no eGFR decrease).

Statistical methods
Within-patient comparisons of CXCL-10 in blood and urine were performed using paired nonparametric t-tests (Wilcoxon matched-pairs signed-rank test). Differences in CXCL10, CCL8, CCL20, and CXCL16 concentrations (including blood/urine ratios) between KTRs with BKPyV DNAemia with and without progression to PVAN as well as among patients with additional JCPyV, HCMV DNAemia were compared by unpaired nonparametric Kruskal-Wallis and Dunn´s multiple comparison tests. CXCL10 between baseline samples and control individuals was compared by Mann-Whitney t-test. Effect sizes are reported as d Cohen . Receiver operating characteristic (ROC) analyses were performed to analyze the association of CXCL10 with high-level BKPyV DNAemia, clinical manifestations, and/or histological evidence of PVAN. For all statistical tests, a twotailed P-value of <0.05 was considered statistically significant. GRAPHPAD PRISM version 8.0 software was used for statistical analyses.

CXCL10 in blood and urine during progression of BKPyV replication
First, we analyzed whether progression of BKPyV replication toward DNAemia and PVAN was associated with a response of the inflammatory chemokine CXCL10.
Next, we analyzed whether progression of BKPyV replication toward DNAemia and PVAN was associated with an inflammatory CXCL10 response in urine analogously to blood concentrations. Therefore, we quantified CXCL10 in urine samples obtained pairwise with blood samples and found that CXCL10 urine concentrations also increased due to progression of BKPyV replication. As shown in Fig. 1b and Table S2, CXCL10 in urine rose from baseline to BKPyV DNAuria alone (median increase: 14.15 pg/ml, 95% CI: À1.48 to 38.23, P = 0.0223) and increased from this to low-level DNAemia (median increase: 36.44 pg/ml, 95% CI: 16.59-80.91, P = 0.001). The CXCL10 urine concentration further increased from low-level BKPyV DNAemia when viral DNA loads exceeded 1000 copies/ml but no clinical manifestations of PVAN were observed yet (median increase: 87.59 pg/ml, 95% CI: 39.02-148.07, P = 0.0005). CXCL10 in urine finally peaked when, in addition to DNAemia, significant levels of decoy cells, a decrease of eGFR and/or histological evidence for PVAN were detected (median increase: 173.50 pg/ml, 95% CI: 129.27-280.41, P < 0.0001).
The individual blood/urine ratio of the CXCL10 concentration did not significantly change during different stages of BKPyV replication until an eGFR decrease occurred and/or biopsy-proven PVAN developed. As compared to the other stages, evidence for PVAN was associated with a significant decrease of the ratio due to a disproportionately strong increase of urine CXCL-10 levels (median change of the blood/urine ratio: À0.89, 95% CI: À7.38 to À0.69, P = 0.0042).

CXCL10 differences in relation to BKPyV-associated clinical manifestations
Next, we analyzed whether CXCL10 in blood and urine differed among KTRs with respect to the most severe clinical manifestation which concomitantly emerged with BKPyV DNAemia during the follow-up. Therefore, KTRs with episodes of BKPyV DNAemia were grouped as shown in Table 2. In 20 KTRs, the most severe manifestation of BKPV (AEJCPyV) replication was histological evidence of PVAN; in 12 KTRs, in whom no biopsy was performed, the most severe clinical manifestation was a decrease of eGFR. In these patients, the simultaneously detected decoy cell levels were high (>30%), while in 9 KTRs, in contrast, BKPyV DNAemia was associated with no significant levels of decoy cells (<5%) and eGFR remained stable. Although BKPyV DNA loads exceeded 1000 copies/ml in these patients, viral loads were significantly lower than in KTRs with histological and/or clinical evidence of PVAN (Table 2). Absolute eGFR values at the time point of BKPyV DNAemia were also significantly lower in the 20 KTRs with histologically verified PVAN than in the nine patients with low decoy cell levels and lower BKPyV DNA loads (median difference: À11.69 ml/min/1.73, 95% CI: À23.88 to À1.92, P = 0.0213, d Cohen = 0.96).

Association of CXCL10 with high-level BKPyV DNAemia and BKPyV-associated disease
Then, we analyzed CXCL10 levels in high-level BKPyV DNAemia (presumptive for PVAN) using 10 000 copies/ml (but not decoy cell levels or an eGFR decrease) as reference. Out of the 56 KTRs with peak BKPyV DNA levels >1000 copies/ml, 42 KTRs displayed viral loads >10 000 copies/ml, while 14 patients did not. CXCL10 serum concentrations were significantly higher in these 42 KTRs with peak DNA loads >10 000 copies/ ml than in the 14 KTRs with lower viral loads (median CXCL10 concentration: 411.22 vs. 310.15 pg/ml, 95% CI: 438.77-162.82 vs. 318.64-116.81, P = 0.0122, d Cohen = 0.70). For CXCL10 in urine, however, no statistically significant difference was observed (median concentration: 343.50 vs. 292.91 pg/ml, 95% , P = 0.7152). As shown in Fig. S1, we also performed ROC analyses to evaluate the association of the CXCL-10 concentration with manifestations of BKPyV-associated disease. Of note, the area under the curve (AUC) was slightly higher when clinical manifestations and/or histological evidence PVAN rather than high-level BKPyV DNAemia >10 000 copies/ml were used as a reference (Table 2; Fig. S1). In episodes of significant BKPyV DNAemia (DNA loads exceeding 1000 copies/ml) CXCL10 in blood was levels are significantly lower in KTRs who display lower BKPyV DNA loads, lower decoy cell levels (<5%), and no eGFR decrease (n = 9), as compared to KTRs who display a simultaneous eGFR decrease, higher viral loads, and higher decoy cell levels (>30%; n = 12), as well as in KTRs in whom PVAN is concomitantly verified by histology (n = 20). (b) Also, CXCL10 in urine is significantly higher in KTRs who display an eGFR decrease, higher viral DNA loads, and significant higher levels of decoy cells as well as in KTRs in whom PVAN is histologically verified, as compared to patients with lower BKPyV DNA loads, no significant decoy cell levels and no concurrent eGFR decrease.

562
Transplant International 2020; 33: 555-566 ª 2020 The Authors. Transplant International published by John Wiley & Sons Ltd on behalf of Steunstichting ESOT CCL8, CXCL16, and CCL20 blood levels in relation to BKPyV-associated clinical manifestations Next, we investigated whether concentrations of other T-cell associated chemokines differed in relation to the severity of BKPyV replication. Therefore, we quantified CCL8, CXCL16, and CCL20 in blood in the groups of KTR as shown in Table 2 and found that CCL8 blood concentrations were significantly higher in KTRs with BKPyV DNAemia who displayed higher viral loads, significant levels of decoy cells, and an eGFR decrease as well as in KTRs who displayed histological evidence of PVAN as compared to KTRs who did not (median concentrations: 59.40 vs. 57.79 vs. 32.34 pg/ml,, P = 0.0491 vs. eGFR decrease, P = 0.0205 vs. PVAN, d Cohen : 0.88, Fig. 4a). In contrast, CXCL16 and CCL-20 in blood did not significantly differ among KTRs with BKPyV DNAemia with respect to the severity of clinical manifestations associated with respective BKPyV DNAemia episodes (Fig. 4b,c).

Discussion
In this study, we longitudinally analyzed CXCL10 in KTRs during different stages of BKPyV replication and demonstrate that, along with progression of BKPyV replication, CXCL10 in blood and urine gradually increases, from viral DNAuria over low-and high-level DNAemia to a decrease of renal function and histological evidence PVAN. These findings suggest that CXCL10 concentrations in blood and urine might reflect the stepwise increase of inflammation and development of polyomavirus-associated disease in KTRs, a fact that should specifically be considered when CXCL10 is proposed as particular marker of allograft rejection [25,[31][32][33][34][35][36][37].
Indeed, CXCL10 attracts and activates T-cells with a Th1 cytokine profile, and since its production can again be induced upon stimulation with Th1-derived cytokines like interferon-c (IFN-c), it apparently participates in a positive feedback loop [24,25]. CXCL10 is thus essential to elicit a Th1-shaped cellular immune response against viruses, but on the other hand, contributes to tissue inflammation and injury [38]. Consequently, CXCL10 has been proposed as marker of immune activation, severity, and therapy response in diverse viral diseases but also of allograft rejection [21,25,[30][31][32][33][34][35][36].
In previous studies, elevated CXCL10 urine concentrations were equally detected when KTRs displayed BKPyV nephritis or DNAemia as well as allograft rejection [33,34,39]. Elevated blood concentrations were encountered in KTRs with BKPyV detection in urine and/or serum [8,33,40]. Our present data are in Figure 4 CCL8, CXCL16, and CCL20 in blood in relation to BKPyV-associated clinical manifestations. (a) CCL8 in blood is significantly higher in KTRs with BKPyV DNAemia (DNA loads exceeding 1000 copies/ml) who display an eGFR decrease, higher viral loads, and higher decoy cell levels (n = 12) as well as histologically confirmed PVAN (n = 20), as compared to KTRs who display lower BKPyV DNA loads, lower levels of decoy cells, and no concurrent eGFR decrease (n = 9). (b) CXCL16 and (c) CCL20 in blood do not significantly differ in KTRs with BKPyV DNAemia in relation to levels of BKPyV DNA, decoy cells, the concurrent presence of an eGFR decrease as well as to histological evidence of PVAN. accordance with and extend these former findings, indicating that a CXCL10 increase due to BKPyV can be equally detected in urine and blood, occurs stepwise and in proportion to the extent of viral replication, and correlates with a decrease of renal function and histological verification of PVAN. Of note, a similar observation has been made for Interleukin-6, which increases with rising BKPyV loads [41].
The gradual CXCL10 response we observed here, not only discloses that intrarenal BKPyV replication and CXCL10 secretion into urine and blood are functionally linked, but suggests its cumulative inflammatory effect on the occurrence of nephropathy. A correlation of urinary CXCL10 levels with tubule-interstitial inflammation and grade of tubulitis was previously reported, while upregulation of CXCL10 genes could pose the underlying functional background in patients with PVAN [33,[42][43][44].
Our particular finding that not only urine but also blood CXCL10 concentrations were elevated over baseline during mere BKPyV DNAuria, furthermore highlights that even early, clinically silent, and locally restricted BKPyV replication in the allograft is sufficient to elicit a systemic CXCL10 rise and trigger inflammation. Since increased CXCL10 levels have been associated with long-term allograft dysfunction in KTRs, an increased CXCL10 concentration due to low level but sustained BKPyV replication could also possibly adversely influence allograft survival [33,35,36,39,45,46]. Correspondingly, a constantly low CXCL10 urinary concentration was recently correlated with histologically stable allografts and freedom of rejection [47]. Thus, it is indeed possible that CXCL10 might serve as inflammatory marker to individualize immunosuppressive treatment, this, however, requires further studies.
Although we demonstrated that CXCL10 gradually increases along with BKPyV replication, the potential use of CXCL10 as biomarker for PVAN may be limited. Of note, allograft rejection may also be associated with increased CXCL10 levels, whereby multiple studies have proposed CXCL10 as a rejection marker [25,31,32,39,42,46]. In this regard, our observation that CXCL10 in blood and urine already increased for mere BKPyV DNAuria, particularly suggests that absence of BKPyV should be verified, before a CXCL10 rise can be considered an indicator of allograft rejection. Indeed, we observed that CXCL10 further increased when KTRs with BKPyV DNAemia subsequently developed allograft rejection, but considering that CXCL10 responses might vary among individuals, it seems questionable whether quantitative differences might ultimately discriminate between PVAN and rejection.
With this regard, we also demonstrated that other viral agents influence the CXCL10 concentration in KTRs, and especially HCMV DNAemia showed a strong cumulative effect on CXCL10 blood but not urine levels, an observation that is in line with recent findings that CXCL10 in blood but not in urine increased in KTRs with histologically verified HCMV infection [33]. Of note, we previously demonstrated that intrapulmonal HCMV replication in lung transplant recipients caused a systemic rise of CXCL10 in blood, while in bronchoalveolar fluid it only increased when inflammatory airway obstruction simultaneously occurred [21]. In the KTRs with HCMV DNAemia we investigated, the renal allograft might thus not have been the primary site of replication or inflammation.
Finally, it is noteworthy that among the T-cell-associated chemokines, we investigated, only CXCL10 and CCL8 significantly correlated with progression of BKPyV replication, while CCL20 and CXCL16 did not. Elevated CXCL16 concentrations were found in KTRs with hypertensive kidney injury, while increased CCL20 levels were detected in clinically manifest allograft rejection [48,49]. Interestingly, in vitro data demonstrated that renal tubular epithelial cells, when stimulated with IFN-c, strongly upregulate CXCL10 but not CCL20 [50]. Thus, due to the effect of virally induced Th1 cytokines, CXCL10 might predominate over CCL20 during BKPyV replication, indicating that different virus infections might indeed elicit different patterns of chemokine responses.
The lack of baseline samples and histological data for the whole cohort certainly pose limitations of this study. Nonetheless, in summary, the provided data demonstrate that CXCL10 in urine and blood already rises during early BKPyV replication, gradually increases in proportion with the extent of viral replication and correlates with the progression toward PVAN.
Authorship LW: designed the study, wrote the paper, performed the research and analyzed data. LH, YW and SWA: performed the research. RS and IH: participated in research design. GB and KH: participated in research design and wrote the paper.

Funding
This work was supported by the Sigrid Jus elius Foundation, the Jane and Aatos Erkko Foundation, the Medical Society of Finland (FLS), and the Helsinki University

Conflicts of interest
The authors declare no conflicts of interest.

SUPPORTING INFORMATION
Additional supporting information may be found online in the Supporting Information section at the end of the article. Figure S1. Association of CXCL10 with high level BKPyV DNAemia and BKPyV associated disease. Table S1. Clinical information of study patients. Table S2. Quantitative values of CXCL10 measurements.