Urinary CXCL9 and CXCL10 Levels Correlate with the Extent of Subclinical Tubulitis

Authors


* Corresponding author: Stefan Schaub, schaubs@uhbs.ch

Abstract

Subclinical tubulitis has been associated with the later development of interstitial fibrosis and tubular atrophy (IF/TA), leading to diminished allograft survival. The aim of this study was to investigate how concentrations of urinary CXC-receptor 3 (CXCR3) chemokines (i.e. CXCL4/9/10/11) and CCL2 relate to the extent of subclinical tubulitis. Using ELISA, urinary CXCR3 chemokines, CCL2 and tubular injury markers (i.e. urinary NGAL and α1-microglobulin [α1 m]) were measured in patients with stable estimated GFR ≥40 mL/min exhibiting normal tubular histology (n = 24), subclinical borderline tubulitis (n = 18) or subclinical tubulitis Ia/Ib (n = 22), as well as in patients with clinical tubulitis Ia/Ib (n = 17) or IF/TA (n = 10). CXCL9 and CXCL10 were significantly higher in subclinical tubulitis Ia/Ib than in subclinical borderline tubulitis (p ≤ 0.03) and normal tubular histology (p ≤ 0.0002). By contrast, NGAL, α1-m, CXCL4, CXCL11 and CCL2 were not or only marginally distinctive across these patient groups. All urinary chemokines and tubular injury markers were higher in clinical tubulitis Ia/Ib than in normal tubular histology (p ≤ 0.002), but only tubular injury markers were elevated in IF/TA. These results demonstrate a correlation of urinary CXCL9 and CXCL10 levels with the extent of subclinical tubulitis suggesting potential as noninvasive screening biomarkers.

Introduction

Tubulitis is a characteristic pathological feature of most acute rejection episodes with the exceptions of pure antibody-mediated rejection and isolated or predominant arteritis (1). While significant tubulitis often leads to decreased allograft function reflected by an elevation of serum creatinine, 10–30% of patients on calcineurin-inhibitor-based triple immunosuppression can have subclinical tubulitis detected by protocol biopsies (2–4). Several studies have demonstrated that ongoing subclinical tubulitis and interstitial inflammation is associated with the development of tubular atrophy (TA) and interstitial fibrosis (IF), leading to diminished renal allograft survival (4–8).

CXC-receptor 3 (CXCR3) chemokines (i.e. CXCL9 (formerly MIG), CXCL10 (IP-10), CXCL11 (ITAC)) are important for recruitment of alloantigen-primed T cells to the site of inflammation and for enhancing proinflammatory cytokine production (9–12). They are mainly secreted by leukocytes present in the allograft, as well as tubular epithelial cells (11,13). Thus, CXCR3 chemokines might already be detectable in the urine at an early stage of the alloimmune response. Indeed, three studies have shown that high urinary CXCR3 chemokine levels are predictive for acute clinical allograft rejection (14–16). Importantly, they were already elevated a few days prior to the increase of serum creatinine suggesting their potential as early predictive biomarkers. However, their diagnostic value for detection of subclinical tubulitis has not been assessed to date. The aim of this study was to investigate whether urinary CXCR3 chemokine concentrations correlate with the extent of subclinical tubulitis. If true this would provide the rationale for future studies to evaluate urinary CXCR3 chemokines as a noninvasive screening procedure.

Materials and Methods

Patient population

All urine samples analyzed in this study were obtained with informed consent and ethics approval by the local institutional review boards. From 2003 to 2005 midstream urine samples from 214 patients were collected at the University Hospital Basel immediately before a protocol renal allograft biopsy (n = 249; scheduled at 3 and 6 months posttransplant) or clinically indicated biopsy (n = 207) was performed. These 214 patients represent 90% of all patients who had an allograft biopsy within this time frame. All biopsy specimens were evaluated by light microscopy, immunofluorescence (C4d), and immunohistochemistry (SV40-antigen). Findings were graded according to the Banff 2007 classification (1). Seventy-five of 214 patients (35%) were included because they met predefined criteria as described in detail below. Thirteen patients with subclinical tubulitis Ia/Ib were added from the Winnipeg transplant cohort to increase this diagnostic category. In total, 88 patients were analysed in this study and every patient contributed only one urine sample. Based on the allograft biopsy results, the allograft function, and the clinical course five groups with tubulointerstitial pathologies were included:

  • (i)  Normal tubular histology group: Consists of 24 urines from patients, who had stable transplant function with a MDRD-GFR ≥40 mL/min, never had clinical or biopsy-proven rejection before and had a protocol biopsy with no tubular, glomerular or vascular infiltrates (i.e. Banff i0–1 t0 g0 v0) as well as negative C4d-staining in peritubular capillaries.
  • (ii)  Subclinical borderline tubulitis group: Consists of 15 urines from patients, who had stable transplant function with a MDRD-GFR ≥40 mL/min, never had clinical or biopsy-proven rejection before and a protocol biopsy demonstrating borderline tubulitis (i.e. Banff i1–2 t1).
  • (iii)  Subclinical tubulitis Ia/Ib group: Consists of 22 urines from patients, who had stable transplant function with a MDRD-GFR ≥40 mL/min, never had clinical rejection before and a protocol biopsy demonstrating tubulitis Ia/Ib (i.e. Banff i2–3 t2–3).
  • (iv)  Clinical tubulitis Ia/Ib group: Consists of 17 urines from patients, who had a clinically indicated biopsy (serum creatinine >20% above baseline) demonstrating tubulitis Ia/Ib (i.e. Banff i2–3 t2–3).
  • (v)  IF/TA group: Consists of 10 urines from patients, who had a clinically indicated biopsy demonstrating at least moderate IF/TA (i.e. Banff ci ≥ 2 ct ≥ 2) without tubulitis (i.e. Banff i0–1 t0).

Urine analyses

Measurement of total protein (benzethonium chloride method) and creatinine (enzymatic method) were performed on a Modula clinical chemistry analyzer (Roche Diagnostics, Roche, Switzerland). Urinary chemokine measurements were performed with pH adjusted urine samples following previously described protocols (17). Besides CXCL9, CXCL10 and CXCL11, two other chemokines were selected as controls. CXCL4 (formerly platelet factor 4) is mainly secreted by platelets and binds to the chemokine receptor CXCR3B, an alternative splicing variant of CXCR3 expressed on T cells, neutrophils and monocytes (18,19). CCL2 (formerly monocyte chemoattractant protein 1) is mainly secreted by vascular endothelial and parenchymal cells and binds to the CCR2 receptor expressed on monocytes, natural killer cells and T cells (9,11). Repeated measurements of urinary chemokines by ELISA demonstrated good reproducibility with a coefficient of variation <7%. Furthermore, up to eighth freeze–thaw cycles did not affect the measured chemokine concentrations. In addition to chemokines, two established urinary tubular injury biomarkers (i.e. α1-microglobulin (α1 m) and neutrophil gelatinase-associated lipocalin (NGAL)) were measured by ELISA (Beckman-Coulter nephelometry system, Brea, CA; Antibodyshop, Gentofte, Denmark) (20,21). In order to correct for different urine dilution, excretion of urine proteins are given in relation to urine creatinine (i.e. mg, μg or ng protein/mmol creatinine).

Statistical analysis

We used JMP software version 7.0 (SAS Institute Inc., Cary, NC) for statistical analysis. Data are given as median (range). For categorical data, Fisher's exact test or Pearson's chi-square test was used. As all continuous data were not normally distributed Wilcoxon or Kruskal–Wallis rank-sum tests were used for analysis. Significant results in the Kruskal–Wallis rank-sum test were further analysed with pair-wise nonparametric tests. A p-value < 0.05 (2-tailed) was considered to indicate statistical significance.

Results

Patient characteristics

The characteristics of the patient groups are detailed in Table 1. The normal tubular histology, subclinical borderline tubulitis and subclinical tubulitis Ia/Ib groups were by design different regarding the extent of tubulointerstitial inflammation, but equal with respect to serum creatinine (p ≥ 0.56), MDRD-GFR (p ≥ 0.56), total urine protein (p ≥ 0.05) and urinary albumin (p ≥ 0.41). Patients in the clinical tubulitis Ia/Ib group and the IF/TA group had significantly higher serum creatinine (p < 0.0001), lower MDRD-GFR (p < 0.0001), higher total urine protein (p ≤ 0.04) and higher urinary albumin (p ≤ 0.005) than all other groups.

Table 1.  Baseline characteristics
VariableNormal tubular histology (n = 24)Subclinical borderline tubulitis (n = 15)Subclinical tubulitis Ia/Ib (n = 22)Clinical tubulitis Ia/Ib (n = 17)IF/TA (n = 10)p-Value
  1. 1GFR calculated with the MDRD study equation: 186*serum creatinine−1.154*age−0.203*0.742 (if subject is female).

  2. 2Most patients were either on mycophenolate-mofetil or azathioprine.

  3. 3p < 0.0001 versus all other groups. No difference between the normal tubular histology, subclinical borderline tubulitis, subclinical tubulitis Ia/Ib, and clinical tubulitis Ia/Ib groups (p ≥ 0.19).

  4. 4p = 0.59 versus each other. p < 0.0001 versus all other groups. No difference between the normal tubular histology, subclinical borderline tubulitis, and subclinical tubulitis Ia/Ib groups (p ≥ 0.56).

  5. 5p = 0.60 versus each other. p < 0.0001 versus all other groups. No difference between the normal tubular histology, subclinical borderline tubulitis and subclinical tubulitis Ia/Ib groups (p ≥ 0.56).

  6. 6p = 0.62 versus all each other. p ≤ 0.04 versus all other groups. No difference between the normal tubular histology, subclinical borderline tubulitis and subclinical tubulitis Ia/Ib groups (p ≥ 0.05).

  7. 7p = 0.43 versus each other. p ≤ 0.005 versus all other groups. No difference between the normal tubular histology, subclinical borderline tubulitis and subclinical tubulitis Ia/Ib groups (p ≥ 0.41).

Recipient sex–-no. fem (%)11 (46)7 (47)13 (59)8 (47)3 (30)0.65
Recipient age50 (18–70)55 (30–64)49 (19–61)45 (18–68)38 (20–54)0.13
Donor
 -Living–-no. (%)16 (67)6 (40)12 (55)5 (29)5 (50)0.18
 -Age47 (1–65)46 (1–62)43 (16–83)39 (1–78)48 (16–69)0.71
Time posttransplant (days)94 (80–203)110 (77–197)86 (32–184)162 (12–1315)1979 (319–4829)3<0.0001
Creatinine at biopsy1.3 (0.8–1.8)1.1 (0.8–1.8)1.2 (0.8–2.1)2.2 (1.4–7.0)42.8 (2.0–5.0)4<0.0001
GFR 154 (40–89)59 (42–86)55 (40–93)31 (8–51)523 (10–41)5<0.0001
Immunosuppression 2
 -CNI–-no. (%)21 (88)11 (73)18 (82)12 (71)9 (90)0.57
 -Sirolimus–-no. (%)8 (33)5 (33)4 (18)6 (35)1 (10)0.44
 -Steroids–-no. (%)15 (63)12 (80)17 (77)11 (65)5 (50)0.44
Allograft biopsy results
 -Glomeruli15 (7–50)15 (7–38)12 (7–48)11 (7–30)15 (7–50) 
 -Banff acute score
   -Interstitial1 (0–1)1 (1–2)2 (2–3)2 (2–3)1 (0–1) 
   -Tubular012 (2–3)2 (2–3)0 
   -Glomerular000 (0–1)0 (0–2)0 
   -Vascular000 (0–1)0 (0–1)0 
 -Banff chronic score
   -Interstitial1 (0–1)0 (0–2)0 (0–2)0 (0–2)2 (2–3) 
   -Tubular00 (0–2)0 (0–1)0 (0–1)2 (2–3) 
   -Glomerular0000 (0–1)0 (0–2) 
   -Vascular0 (0–1)0 (0–2)0 (0–1)0 (0–1)1 (0–3) 
Urine sediment
 -Erythrocytes (no/HPF)0 (0–5)0 (0–17)0 (0–9)0 (0–88)1 (0–14)0.07
 -Leukocytes (no/HPF)0 (0–22)0 (0–6)0 (0–23)2 (0–20)3 (0–14)0.04
Proteinuria
 -Protein/creat (mg/mmol)11 (5–50)18 (5–74)16 (8–74)27 (8–402)669 (7–694)60.0004
 -Albumin/creat (mg/mmol)2.6 (0.4–31)4.9 (0.6–10)3.5 (0.5–53)15 (0.5–327)725 (1.6–596)70.0002

Correlation of the extent of tubulitis with urinary chemokine levels

Urinary CXCL9 and CXCL10 levels were significantly higher in the subclinical tubulitis Ia/Ib group than in the subclinical borderline tubulitis (p ≤ 0.03) and the normal tubular histology group (p ≤ 0.0002), indicating a close correlations between chemokine levels and the extent of subclinical tubulitis (Figure 1A, B). Such a correlation with the extent of subclinical tubulitis was not observed for urinary CXCL11 (Figure 1C). Furthermore, both control chemokines concentrations (i.e. CXCL4 and CCL2) in the subclinical tubulitis Ia/Ib group were not different from the subclinical borderline tubulitis group (p = 0.64 and p = 0.62), and not or slightly higher than in the normal tubular histology group with a substantial overlap (p = 0.52 and p = 0.003) (Figure 1D, E). In patients with clinical tubulitis Ia/Ib, all urinary chemokine levels were higher than in the normal tubular histology group (p ≤ 0.002) consistent with a broad and intensive immune activation in clinical tubulitis Ia/Ib (Figure 1A–E). By contrast, urinary chemokine levels were not or only marginally higher in patients with IF/TA than in patients with normal tubular histology (p ≥ 0.03) (Figure 1A–E).

Figure 1.

Urinary chemokine concentrations of the four different groups. Differences across all groups were statistically significant (CXCL9: p < 0.0001; CXCL10: p < 0.0001; CXCL11: p = 0.0003; CXCL4: p = 0.03; CCL2: p = 0.0003; Kruskal–Wallis rank-sum tests). Individual comparisons noted in the figures were done by Wilcoxon rank-sum tests.

Correlation of the extent of tubulitis with urinary tubular injury biomarkers

Urinary α1 m and NGAL concentrations in the subclinical tubulitis Ia/Ib group were not different from the subclinical borderline tubulitis group (p = 0.19 and p = 0.15), but slightly higher than in the normal tubular histology group with a substantial overlap (p = 0.03 and p = 0.02). In patients with clinical tubulitis Ia/Ib, α1 m and NGAL levels were significantly higher than in the normal tubular histology group (p ≤ 0.0003) consistent with substantial tubular epithelial cell dysfunction in clinical tubulitis Ia/Ib (Figures 2A, B). In contrast to urinary chemokines, α1 m and NGAL levels were significantly higher in patients with IF/TA than in patients with normal tubular histology (both p = 0.003), but equal to patients with clinical tubulitis Ia/Ib (p ≥ 0.67) (Figure 2A, B).

Figure 2.

Urinary α1 m and NGAL concentrations of the four different groups. Differences across all groups were statistically significant (p ≤ 0.0002; Kruskal–Wallis rank-sum tests). Individual comparisons noted in the figures were done by Wilcoxon rank-sum tests.

Preliminary diagnostic characteristics of urinary proteins for detection of subclinical tubulitis Ia/Ib

Two studies indicated that subclinical tubulitis Ia/Ib is associated with subsequent development of IF/TA and deterioration of allograft function, while the clinical relevance of subclinical borderline tubulitis is less clear (6,22). Therefore, receiver operating characteristic (ROC) analysis was performed to obtain preliminary diagnostic characteristics of urine proteins to distinguish subclinical tubulitis Ia/Ib (n = 22) from normal tubular histology and subclinical borderline tubulitis (total n = 39). CXCL9 and CXCL10 had AUC of 0.78 and 0.79 (p = 0.006 and p = 0.003), respectively (Table 2). Combining CXCL9 and CXCL10 values (CXCL9 + CXCL10 or CXCL9 * CXCL10) did not further improve the AUC. In marked contrast, total proteinuria, α1 m, NGAL, CXCL4, CXCL11 and CCL2 had no value to separate these two groups (AUC 0.55–0.68; p ≥ 0.06) (Table 2). Applying the defined urinary CXCL10 cut-off of 5.3 ng/mmol, 0/24 patients with normal tubular histology (0%), 2/10 patients with IF/TA (20%), 4/15 patients with subclinical borderline tubulitis (27%), 15/22 patients with subclinical tubulitis Ia/Ib (68%) and 13/17 patients with clinical tubulitis Ia/Ib (76%) had CXCL10 levels above this threshold (p < 0.0001). In addition, quartiles of urinary CXCL10 correlated with increasing severity of tubulitis (p < 0.0001) (Figure 3).

Table 2.  Receiver operating characteristic (ROC) analysis
Urine proteinAUCCut-offSensitivity at cut-offSpecificity at cut-offp-Value
  1. ROC analysis was performed to obtain preliminary diagnostic characteristics of urine proteins to distinguish subclinical tubulitis Ia/Ib (n = 22) from normal tubular histology or subclinical borderline tubulitis (n = 39).

CXCL10/creatinine (ng/mmol)0.79 5.368%90%0.003
CXCL9/creatinine (ng/mmol)0.78 7.586%64%0.006
CCL2/creatinine (ng/mmol)0.6733.782%54%0.06
α1 m/creatinine (mg/mmol)0.66 5.059%74%0.08
CXCL11/creatinine (ng/mmol)0.64 2.845%87%0.08
Protein/creatinine (mg/mmol)0.6213.170%54%0.27
CXCL4/creatinine (ng/mmol)0.5572.136%82%0.30
NGAL/creatinine (μg/mmol)0.68 2.082%62%0.31
Figure 3.

Distribution of patient groups within quartiles of urinary CXCL10.

Correlation of urinary chemokines with total proteinuria and the urine sediment

No patient had a urinary tract infection at the time of the urine collection. None of the five measured urinary chemokines correlated with the number of erythrocytes (p ≥ 0.32) or leukocytes in the urine sediment. Indeed, only CCL2 showed a trend toward a correlation with leukocytes (R2= 0.04; p = 0.07), while this was not observed for the other chemokines (p ≥ 0.45). Similarly, only CCL2 showed a weak, but significant correlation with the protein/creatinine ratio (R2= 0.05; p = 0.04), while this was not observed for the other chemokines (p ≥ 0.08).

Discussion

The main observation in this study was that urinary CXCL9 and CXCL10 concentrations demonstrated a close correlation with the extent of subclinical tubulitis, while no such distinction was seen for urinary CXCL4, CXCL11, CCL2 and tubular injury markers. This supports an important role of CXCL9 and CXCL10 in the early rejection process and suggests that urinary CXCL9 and CXCL10 may have the potential to become noninvasive biomarkers to screen for subclinical tubulitis.

Tubulointerstital renal allograft rejection can be regarded as a continuous process starting with few infiltrating cells but still stable allograft function, and ending with severe tubulitis accompanied by rising serum creatinine (1,5). The invasion of alloantigen-primed T cells into the tubulointerstitial compartment is orchestrated by cytokines/chemokines produced by locally present leukocytes and parenchymal cells (e.g. tubular epithelial cells) (11). These infiltrating cells further amplify and expand the immune response leading to tubular epithelial cell damage and allograft dysfunction (23). Consistent with such a concept, we found that CXCL9 and CXCL10–-as main chemoattractants for alloantigen-primed T cells–correlated with the extent of subclinical tubulitis, while the other investigated chemokines and tubular injury markers were not distinctive at this early stage of tubulointerstitial allograft rejection. However, in acute clinical tubulitis Ia/Ib all measured urinary chemokines and tubular injury markers were significantly elevated indicating a broad immune activation at this later stage of rejection with manifest allograft dysfunction.

In contrast to tubular injury markers, urinary CXCL9 and CXCL10 were only slightly elevated in IF/TA compared to normal tubular histology. This further supports that urinary CXCL9 and CXCL10 levels correlate with active tubulointerstitial inflammation but not with chronic irreversible injury. Furthermore, urinary CXCL9 and CXCL10 levels did not correlate with proteinuria (p = 0.08 and p = 0.62) and leukocyturia in the absence of a urinary tract infection (p = 0.45 and p = 0.57). For CXCL10 this suggests that these are likely not clinically relevant confounding factors, while for CXCL9 a confounding effect of proteinuria can not be entirely excluded given the small study size (i.e. type II error), but this is likely a minor effect given that r2= 0.04. However, tubulointerstitial inflammation due to polyomavirus-BK or bacterial infections may lead to elevated urinary CXCL9 and CXCL10 levels (14,16), but these two diseases can be easily excluded by the absence of significant leukocyturia or Decoy cells, and–-if necessary–a subsequent urine culture as well as a measurement of polyomavirus-BK viremia (24). In addition, ischemia-reperfusion injury can increase urinary CXCL9 and CXCL10 levels (14), but this is an unlikely confounder beyond the first 2 months posttransplant. Therefore, after excluding polyomavirus-BK nephropathy and urinary tract infections, elevated urinary CXCL9 and CXCL10 concentrations measured beyond the first 2 months posttransplant most likely indicate active tubulointerstitial inflammation due to rejection.

The potential of urinary CXCL9 and CXCL10 as biomarkers for subclinical tubulitis Ia/Ib has to be regarded as preliminary because they are derived from selected patients with clearly defined clinicopathological phenotypes not reflecting the heterogeneity observed in the whole population. Indeed, this study population included only about a third of patients who had a protocol allograft biopsy within 3 years. Therefore, these results need to be validated in a larger, unselected and consecutive patient population, which would then allow for an accurate determination of the diagnostic characteristics of urinary CXCL9 and CXCL10 for the detection of subclinical tubulitis Ia/Ib in ‘real life’. Furthermore, sequential urine and allograft histology analyses are required to assess whether urinary CXCL9 and CXCL10 levels correlate with resolution or persistence of subclinical tubulitis in follow-up biopsies after therapeutic interventions. Of note, Hu et al. and Hauser et al. had demonstrated that urinary CXCL9 and/or CXCL10 dropped prior or simultaneously with serum creatinine after treatment of clinical allograft rejection (14,15). Finally, the results of our study are based on only one chemokine measurement per patient assuming stable chemokine excretion. Clearly, day-to-day variances of chemokine excretion need to be established for reliable interpretation of a single measurement.

In conclusion, urinary CXCL9 and CXCL10 correlated with the extent of subclinical tubulitis in this selected patient population. Further analyses are required to define a potential role of urinary CXCL9 and CXCL10 as noninvasive biomarkers to screen for subclinical tubulitis.

Acknowledgments

The authors thank the staff of the renal transplant unit and the histocompatibility laboratory for collection and processing of urine samples. SS and CH are supported by the Swiss National Foundation (grants 3200B0-109302 and PP00B3-114850). PN, DR and KH are supported by grants from the Canadian Institutes of Health Research.

Ancillary