High Pretransplant Serum Levels of CXCL10/IP ... - Wiley Online Library

C Blackwell Munksgaard 2004 Copyright

American Journal of Transplantation 2004; 4: 1466–1474 Blackwell Munksgaard

doi: 10.1111/j.1600-6143.2004.00525.x

High Pretransplant Serum Levels of CXCL10/IP-10 Are Related to Increased Risk of Renal Allograft Failure Mario Rotondia , Alberto Rosatic , Andrea Buonamanoa , Laura Lasagnia , Elena Lazzerib , Fabio Pradellad , Vittorio Fossombronid , Calogero Ciramic , Francesco Liottab , Giorgio La Villab , Mario Serioa , Elisabetta Bertonic , Maurizio Salvadoric and Paola Romagnania,∗ a

Department of Clinical Pathophysiology, and Department of Internal Medicine, University of Florence, Florence, Italy c Center for Nephrology, Dialysis and Transplantation, Florence, Italy d Immunogenetics Unit, Azienda Ospedaliera di Careggi, Florence, Italy ∗ Corresponding author: Paola Romagnani, [email protected] b

In experimental models, the chemokine CXCL10/IP-10 is required for graft failure owing to both acute and chronic rejection. In the present study, pretransplantation sera from 316 cadaver kidney graft recipients were tested for serum CXCL10 and CCL22/MDC levels by an ELISA assay. Kidney graft recipients with normally functioning grafts showed significantly lower serum CXCL10 levels than patients who experienced graft failure, whereas no differences for serum CCL22 levels were observed. After the assignment of all patients to four groups according to serum CXCL10 levels, the death-censored survival rates of grafts were 97.5%, 93.6%, 89.7%, 78.7% (p = 0.0006) at 5 years, while no influence was observed on patient survival. Accordingly, patients with the highest CXCL10 levels showed an increased frequency and severity of rejection episodes. Serum C-reactive protein (CRP) level was also assayed in the same samples. Increase of serum CRP levels represented a predictive parameter for death, but not for graft failure. Multivariate analysis demonstrated that among the analyzed variables, CXCL10 had the highest predictive power of graft loss (RR 2.787). Thus, measurement of pretransplant serum CXCL10 levels might represent a clinically useful parameter to identify subjects who are at high risk of severe rejection and graft failure. Key words: CXCR3, graft failure, graft outcome, IP-10, rejection Received 4 December 2003, revised and accepted for publication 9 April 2004

1466

Introduction Despite the administration of immunosuppressants, allograft rejection remains the primary cause of human renal transplant failure. Acute rejection (AR) is particularly frequent in the early post-transplant period and, although graft loss occurs in only a small percentage of subjects, represents a leading cause of morbility and hospitalization, and predisposes renal allografts to the development of chronic allograft nephropathy (CAN) (1). Nowadays, CAN is the main cause of failure of kidney transplants and raises major clinical problems because of the lack of effective treatment (1). Chemokines are a large family of cytokines implicated in the regulation of leukocyte migration, angiogenesis, tumor growth, and metastasis (2). The chemokine CXCL10/IP10 is a potent chemoattractant for activated lymphocytes and dendritic cells (3–7), mediates vascular pericyte proliferation (8–11), and acts as a powerful angiostatic agent (12–14). Growing evidence suggests that this chemokine may be important for the immune response to transplant. Indeed, in knock-out models of the CXCL10 gene or of its receptor gene, CXC chemokine receptor 3 (CXCR3), cardiac transplant is not acutely rejected and undergoes permanent engraftment (15,16). Furthermore, neutralization of CXCL10 with monoclonal antibodies (mAbs) prolongs the allograft survival in both cardiac and small bowel allograft rejection (16,17). There is also a rapidly developing literature on the expression of CXCL10 and its receptor in human allografts. Of particular importance, the intragraft expression of CXCL10 has been reported in association with clinical rejection of renal (18), lung (19) and cardiac (20,21) allografts. In the sole study on renal transplant and chemokines, induction of CXCR3 and CXCL10 mRNA was associated with acute rejection (18). Thus, CXCL10– CXCR3 interactions play an important role in the pathogenesis of graft failure owing to rejection in multiorgan models. Recent evidence indicates that CXCR3 and CXCL10 are also highly expressed in conjunction with the development of transplant vasculopathy in cardiac allografts (22) and of chronic allograft nephropathy in biopsies of renal transplanted patients (23). In the present study, we investigated whether pretransplant CXCL10 serum levels may predict the recipient’s risk of kidney graft rejection and transplant failure. As control,

Pretransplant Serum CXCL10 and Graft Failure

serum levels of MDC/CCL22 were assessed in the same samples. CCL22 was chosen because its serum levels are increased in several immune-mediated inflammatory disorders (24–27), but targeting of the CCL22 receptor, CCR4, has no effect on graft survival (28). The results showed that, consistently with the well-established involvement of CXCL10 in the immune response towards the transplant, only high pretransplant CXCL10 serum levels represent an important predictive risk factor for the development of severe rejection and transplant failure; therefore, assessment of pretransplant CXCL10 serum levels may be helpful in prospective individualization of immunosuppression therapy for renal transplantation.

Materials and Methods Patients A total number of 376 consecutive cadaveric kidney transplants were carried out at the Florence Transplant center, between January 1991 and June 2001. Sixty patients had no available pretransplant serum samples and were excluded. The 316 transplant recipients who had pretransplant sera available were tested retrospectively for serum CXCL10 and CCL22 content levels. All patients were on hemodialysis before transplant and underwent measurements of main hematochemical parameters every 6 months. An aliquot was obtained from the serum specimen which was used for final cytotoxic cross-match at the time of transplantation (within 1 month before the date of transplantation). The other aliquote was randomly selected from all available samples obtained from the same patients at least 6 months before. Specimens were frozen at −70 ◦ C until the time of assay. All the recipients were Caucasians. The demographic characteristics of the recipients were the followings (see also Table 1): age: 46.33 ± 6.4 year, gender M/F: 209/107, dialytic age: 38.3 ± 2.1 months, original disease: glomerulopathies 43.3%, ADPKD 21.5%, interstitial nephritis 12%, nephroangiosclerosis 11.5%, and others 11.7%. Ninety-three percent of the patients received cyclosporinebased immunosuppression, while the remaining 7% received a FK-based immunosuppression. Only one patient had more than one transplant. HLA

Table 1: Clinical characteristics of the kidney transplant recipients and distribution of the analyzed variables between patients with survived grafts and patients undergoing graft failure Variable

Survived

Failed

p-value

No. patients Sex (M/F) Age at transplantation (years) Time on dialysis (months) Type of dialysis PRA ≥ 5% Mismatch A (n) Mismatch B (n) Mismatch DR (n) Mismatch Total (n) CXCL10 (pg/mL) CCL22 (pg/mL) Rejection (n) Sex donor (M/F) Donor age (years) Cold ischemia time (h)

284 187/97 46.0 ± 0.67

32 22/10 49.3 ± 20.6

NS NS NS

38.7 ± 2.3

34.2 ± 3.7

NS

243/41 246/38 1.20 ± 0.04 1.06 ± 0.04 0.36 ± 0.03 2.62 ± 0.06 129.0 ± 6.8 790.0 ± 40.3 0.21 ± 0.02 160/124 44.4 ± 0.98 20.6 ± 0.38

26/6 29/3 1.12 ± 0.12 0.97 ± 0.13 0.53 ± 0.09 2.62 ± 0.17 218.0 ± 28.9 761.2 ± 88.4 0.57 ± 0.09 15/17 56.2 ± 2.19 21.5 ± 1.12

NS NS NS NS NS NS 0.0007 NS 0.00003 NS 0.0001 NS

American Journal of Transplantation 2004; 4: 1466–1474

typings and panel reactive lymphocytotoxic antibody (PRA) determinations were performed at the Tissue Typing Laboratory of Azienda Ospedaliera Careggi of Florence. Patients with 5% antibody reactivity against a randomly chosen lymphocyte test panel were categorized as sensitized, and represented 13% of all subjects. HLA-A + B + DR mismatches were ≤3 in 44% and ≥3 in 56% of patients. Donor age was 45.5 ± 0.93 years, and cold ischemia time was 20.2 ± 0.36 h. Information on graft function and patient survival was documented at 1, 3, 6 months and at 1, 2, 3, 4, 5 years. The median follow-up time after transplant, which included patients who experienced graft failure, was 39 months. The procedures used in this study were in accordance with the Regional Ethics Committee on human experimentation. Only biopsy-proved ARs were used for statistical analysis. Acute rejection episodes were classified according to Banff classification (29) and treated with steroids. Acute rejection episodes, showing vascular components at biopsy (grade II or III) and failure to reverse after a course of steroids, were treated with ATG (rabbit antihuman thymocyte globulin). Sex and agematched adult healthy subjects (n = 48) and patients with end-stage renal disease (ESRD) shortly before the onset of dialysis (n = 25) therapy were used as controls.

Serum assays Serum CXCL10 levels were assayed by a quantitative sandwich immunoassay (25) using a commercially available kit (R & D Systems, Minneapolis, MN), with a mean minimum detectable dose of 1.67 pg/mL and a maximum detectable dose of 500 pg/mL. The intra- and interassay coefficients of variation were 3% and 6.1%, respectively. Serum CCL22 were assayed by a quantitative sandwich immunoassay using a commercially available kit (R & D Systems), with a mean minimum detectable dose of 62.5 pg/mL and a maximum detectable dose of 4000 pg/mL. The intra- and interassay coefficients of variation were 2.1% and 6.3%, respectively. C-reactive protein (CRP) concentrations were measured by a commercially available quantitative enzyme-linked immunosorbent assay (ALPHA Diagnostic International, San Antonio, TX), showing a mean minimum detectable dose of 0.35 ng/mL and a maximum detectable dose of 10000 ng/mL. The intra- and interassay coefficients of variation were 2.1% and 3–7%, respectively. All serum samples that appeared greater than the maximum detectable levels of the assays were diluted and assessed again and only the final value was considered. Samples were assayed in duplicate. Quality control pools at low, normal, and high concentrations for all parameters were present in each assay, respectively.

Statistical analysis Statistical analysis were performed using SPSS software (SPSS, Inc., Evanston, IL). Owing to a nonparametric distribution, comparisons of serum CXCL10 levels among different groups were performed by Mann–Whitney U-test for unpaired data. Correlation between two variables was ascertained by linear regression analysis and Spearman’s correlation test as appropriate. To test the independent effects of different variables on rejections and kidney allograft survival, multiple regression analysis was used and partial correlation coefficients were computed. Kaplan-Meier estimates were used to generate an overall survival curve for transplanted grafts and differences among groups were assessed by log-rank test. Graft failure was defined as the need to return to dialysis. Frequencies of allograft failure were compared among groups by v 2 test. A p-value < 0.05 was considered statistically significant. Results are expressed in the text as mean ± SEM unless otherwise stated.

1467

Rotondi et al.

Figure 1: Comparison of serum CXCL10 and CCL22 levels between kidney graft recipients and healthy controls. (A) CXCL10 serum levels were significantly increased in kidney graft recipients before transplantation (ESRD; n = 316) as well as in CRF patients (25) who never experienced dialytic treatment (CRF-ND, chronic renal failure patients not on dialysis) in comparison with healthy controls (n = 48); (∗ p = 0.0006 and ∗ p = 0.000002, respectively). (B) Serum CCL22 levels were not significantly different in the three groups. (Mann–Whitney U-test for unpaired data; ). The data are expressed as median and 25th and 75th percentiles in boxes and 5th and 95th percentiles as whiskers.

Results Patients with end stage renal disease (ESRD) display stably increased serum CXCL10 levels Pretransplant serum levels of CXCL10 and CCL22 were assessed in 316 adult kidney graft recipients. Kidney graft recipients had a significantly higher median pretransplant serum CXCL10 content before transplantation than 48 adult healthy controls, (97.1 vs. 75.6 pg/mL; p = 0.0006) (Figure 1A), while serum CCL22 levels were not significantly different in the two groups 678.4 vs. 631.4 pg/mL; (NS) (Figure 1B). To determine whether increased serum CXCL10 levels were related to dialysis or to chronic renal failure (CRF), serum CXCL10 levels were also assessed in 25 patients with severe CRF, who never experienced dialytic treatment. These patients, in comparison with controls, displayed significantly increased median serum CXCL10 levels (121.6 pg/mL vs. 75.6 pg/mL, p = 0.000002; Figure 1A) and comparable median serum CCL22 (670.8 pg/mL vs. 631.4 pg/mL, NS; Figure 1B), respectively, suggesting that increased serum CXCL10 levels were mainly related to chronic inflammation driven by severe CRF. In order to assess if pretransplant serum CXCL10 levels in graft recipients were stable over time, 265/316 patients (84%) for whom an additional serum sample was available underwent a second CXCL10 serum assay. The time interval between the two serum collections was at least 6 months. Linear regression analysis demonstrated a strong relationship between the two assays (R 2 = 0.793; p = 0.000001), thus indicating that pretransplant serum CXCL10 levels were a stable parameter (Figure 2A). CXCL10 levels were also analyzed in serum samples obtained from 105 patients coming to the consulting room for post-transplant routine controls. Linear regression analysis still demonstrated a strong relationship between pretransplant and post-transplant serum levels of CXCL10 (R 2 = 0.61; p = 0.001, Figure 2B). By contrast, no significant relationship

1468

between post transplant serum CXCL10 levels and time after transplant could be demonstrated (R 2 = 0.019), further suggesting that serum CXCL10 levels represent a stable parameter over time. CXCL10, but not CCL22, pretransplant serum levels predict the recipient’s risk of graft failure Demographic and clinical characteristics of all patients in relation to subsequent graft outcome have been compared in Table 1. Patients with normally functioning grafts showed significantly lower median pretransplant serum CXCL10 levels than patients who experienced graft failure (93.0 vs. 157.4 pg/mL; p = 0.0007) (Figure 3A). Such a difference was even more significant in patients who lost the graft within 2 years after transplantation (93.0 vs. 162.2 pg/mL; p = 0.0003). No differences were observed in serum pretransplant CXCL10 levels with regards to age, gender, year of transplantation, original disease, transplant number, type or duration of dialysis (data not shown). Only eight patients were highly sensitized and they did not show significantly increased median serum CXCL10 levels (75.8 pg/mL). Furthermore, serum CCL22 levels were unrelated with graft outcome (678.4 vs. 675.0 pg/mL; NS, Figure 3B). When life time analysis was performed after the assignment of all patients to four groups according to the 0–25th (64 and 97 and 157 pg/mL, n = 80) percentiles based on serum CXCL10 levels, the death-censored survival rates of grafts were 97.5%, 93.6%, 89.7%, 78.7% (p = 0.0006 in total; p = 0.0002 and p = 0.008 between group 1 and 2 vs. 4, respectively; and p = 0.03 between group 1 vs. 3) at 5 years (Figure 4). Despite differences in graft survival rates were maintained at 5 years’ follow up, the effect of serum CXCL10 levels appeared to be mainly



sayed in the same samples in which CXCL10 was assessed (n = 265). Although a partial overlap was observed between patients with increased pretransplant CXCL10 levels and those with increased CRP levels, regression analysis did not demonstrate any significant correlation between the two parameters (R 2 = 0.095, Figure 5). Accordingly, no significant correlation between pretransplant CRP values and graft failure was found, while pretransplantation median CRP levels in our cohort of transplant recipients were significantly increased in patients who died for any cause (2885.6 vs. 7835.5; p = 0.01). Pretransplant serum CXCL10 levels predict the recipient’s risk of early, severe, acute graft rejection The frequency of rejection episodes in the first month after transplant increased in relation to pretransplant serum CXCL10 levels in the four groups (v 2 = 11.412; p = 0.009). In particular, patients with serum CXCL10 levels above the 75th centile (>157 pg/mL) showed an increased frequency of rejection in comparison with patients with serum CXCL10 levels below the 75th centile (34.8% vs. 21.4%; p = 0.01). Accordingly, the 74 patients who underwent rejection episodes within the first month after transplantation, showed higher serum CXCL10 levels when compared with nonrejectors (117.8 vs. 90.3 pg/mL; p = 0.006) (Figure 6A).

Figure 2: Serum CXCL10 levels in graft recipients are a stable parameter. (A) Strong relationship between the assays of CXCL10 in two different pretransplant serum samples obtained from the same patients (n = 265) (p = 0.000001). (B) Strong relationship between the assays of CXCL10 in serum samples obtained from patients (n = 105) coming to the consulting room for post-transplant routine controls in comparison with pretransplant serum levels (p = 0.001).

operating within the first 2 years after transplant (Figure 4). As additional confirmation, life-time analysis performed on both serum samples obtained from the 265 patients, or on the mean of the two samplings, yielded similar results (data not shown). No influence of pretransplant serum CXCL10 on patient survival could be observed and no differences for the length of surveillance period were found among groups showing different pretransplant CXCL10 serum levels. As expected, life-time analysis performed in relation to CCL22 serum levels revealed no difference in graft survival rates. Pretransplant serum CRP levels were also asAmerican Journal of Transplantation 2004; 4: 1466–1474

Moreover, among patients experiencing early rejection episodes, the 14 patients who required ATG treatment because of severe rejection, showing vascular components at biopsy (grade II or III) and failure to reverse after a course of steroids, were characterized by significantly higher pretransplant median serum CXCL10 levels in comparison with 60 patients with less severe rejection episodes (216.1 vs. 112.4 pg/mL; p = 0.04) (Figure 6B). Interestingly, among the 105 patients coming to the consulting room for posttransplant routine control patients, those who experienced successfully treated rejection episodes showed a slight reduction (192.7 vs. 163.7 pg/mL) in serum CXCL10 levels once the rejection episode was overcome, but the difference with pretransplant serum CXCL10 levels was not statistically significant. Predictive values of pretransplant serum CXCL10 levels and other factors for graft failure and early acute rejection In order to estimate the relative risks for developing both graft failure and early rejection episodes in patients with an increased serum CXCL10 level (>157 pg/mL) two separate multivariate analyses were performed. In particular, multivariate Cox regression analysis was performed with graft failure as a dependent variable. Recipient age, gender, number of HLA-A, -B and -DR mismatches, primary disease, type of immunosuppression, PRA, number of previous transplant, cold ischemia time, donor age and gender were considered as covariables. The results confirmed a significantly increased risk (risk ratio 2.787; C.I. 1.306–5.949; p = 0.008) for developing graft failure in such 1469

Rotondi et al.

Figure 3: Serum CXCL10, but not CCL22 levels are increased in patients undergoing graft failure. (A) Pretransplant CXCL10 serum levels were significantly higher in patients (n = 32) who experienced graft failure in comparison with patients with normally functioning grafts (n = 284). (B) Pretransplant CCL22 serum levels were comparable in patients (n = 32) who experienced graft failure in comparison with patients with normally functioning grafts (n = 284). (Mann–Whitney U-test for unpaired data; ∗ p = 0.0007). Data are expressed as median and 25th and 75th percentiles in boxes and 5th and 95th percentiles as whiskers.

patients (Table 2). Similarly, logistic regression analysis performed with early rejection episodes, as a dependent variable with the same covariables, demonstrated a significantly increased risk (risk ratio 2.233; C.I. 1.173–4.247; p = 0.01) of rejection within the first month after transplantation. This finding was not altered when the variables were considered as continuous rather than dichotomous variables. Multivariate analysis of graft failure revealed the importance of another variable, the donor age, that differed significantly between patients who lost the graft and those who did not (Table 2). However, donor age had a lower dis-

Figure 4: High pretransplant serum CXCL10 levels predict allograft failure. Recipients with very low serum CXCL10 levels (157 pg/mL; n = 78) and recipients with serum CXCL10 levels >97 and 65 and 157 pg/mL. In this analysis, death with a functioning graft was not counted as graft failure (KaplanMeier life-time analysis and log-rank test).

1470

Figure 5: Lack of significant correlation between pretransplant serum CXCL10 and CRP levels. Pretransplant serum CRP levels were assayed in the same samples in which CXCL10 was assessed (n = 265). Although a partial overlap can be observed between patients with increased pretransplant CXCL10 levels and those with increased CRP levels, regression analysis does not demonstrate any significant correlation between the two parameters.

criminatory power than the pretransplant serum CXCL10 level, as evidenced by the lower relative risks, and the two variables independently and significantly acted in determining graft failure, as demonstrated by the lack of any overlap between the two intervals of confidence in the multivariate analysis (Table 2). As expected, the effect of the two parameters on graft outcome was additive (Figure 6). Indeed, life-time analysis performed after the assignment of American Journal of Transplantation 2004; 4: 1466–1474


Figure 6: Association between high pretransplant serum CXCL10 levels and increased frequency and severity of rejection during the first month after transplantation. (A) Patients who experienced rejection episodes within the first month after transplantation showed higher pretransplant serum CXCL10 levels. (Mann–Whitney U-test for unpaired data; ∗ p = 0.006.). (B) Significant association of severe vascular rejection with higher CXCL10 serum levels. In the entire group of patients who underwent rejection episodes in the first month after transplantation, recipients who required ATG treatment for a severe rejection with vascular components at biopsy (grade II or III) and failure to reverse after a course of steroids showed significantly higher pretransplant serum CXCL10 levels. (Mann–Whitney U-test for unpaired data; ∗ p = 0.04.) The data are expressed as median and 25th and 75th percentiles in boxes and 5th and 95th percentiles as whiskers.

all patients to four groups in relation to the combined effect of the two parameters (CXCL10 serum levels less or greater than the 75th centile = 157 pg/mL and donor age less of greater than the 75th centile = 58 years) revealed survival rates of 94.4% in patients with serum CXCL10 < 157 pg/mL and donor age < 58 years, and rates of 91.2% in patients with serum CXCL10 < 157 pg/mL and donor age > 58 years, while patients with serum CXCL10 > 157 pg/mL and donor age < 58 years (86.5%), and even more so those with serum CXCL10 > 157 pg/mL and donor age > 58 years (67.8%), had a significantly reduced survival

rate (p = 0.00001 1 vs. 4; p = 0.003 2 vs. 4; p = 0.01 3 vs. 4) (Figure 7).

Discussion Allograft failure is one of the four most common causes of end-stage renal disease. Half of the patients who

Table 2: Multivariate analysis showing the relative risk of allograft failure according to selected risk factors

Cold ischemia time CXCL10 Recipient gender Original disease Type of dialysis Transplant no. MMA MMB MMDR Donor gender PRA Recipient age Type of immunosuppression Donor age

95.0% CI

p-value

Relative risk

Lower

Upper

0.159 0.008 0.640 0.510 0.997 0.978 0.663 0.629 0.310 0.547 0.745 0.401 0.891

1.045 2.787 0.817 1.033 1.002 0.972 0.876 0.854 1.499 1.265 0.803 0.984 1.164

0.983 1.306 0.350 0.938 0.372 0.129 0.483 0.449 0.687 0.588 0.213 0.948 0.133

1.110 5.949 1.908 1.136 2.695 7.347 1.589 1.622 3.273 2.724 3.022 1.021 10.203

0.002

1.052

1.020

1.086

CI = confidence interval. Significant variables are in bold


Figure 7: Combined effect of serum CXCL10 and donor age on kidney graft survival. Recipients with serum CXCL10 levels and donor age below 75th centile (