Human Cytomegalovirus Infection in Lung Transplant Recipients Triggers a CXCL-10 Response

Authors


Corresponding author: Elisabeth Puchhammer-Stöckl, elisabeth.puchhammer@meduniwien.ac.at

Abstract

Human cytomegalovirus (HCMV) causes significant morbidity in lung transplant recipients (LTRs). The clinical effects of HCMV replication are determined partly by a type 1 T-helper cell (Th1) response. Because the chemokine interferon-inducible protein of 10 kilodaltons (IP-10, CXCL-10) induces a Th1 response, we investigated whether HCMV triggers IP-10 in LTRs. The IP-10 concentration and HCMV DNA load were determined in 107 plasma and 46 bronchoalveolar lavage fluid (BALF) samples from 36 LTRs. Initial HCMV detection posttransplantation was significantly associated with increased plasma IP-10, regardless of whether the patients showed HCMV DNAemia (p = 0.001) or HCMV replication only in the allograft (p < 0.0001). In subsequent episodes of HCMV detection, plasma IP-10 increased regardless of whether HCMV was detected in blood (p = 0.0078) or only in BALF (p < 0.0001) and decreased after successful antiviral therapy (p = 0.0005). Furthermore, levels of HCMV DNA and IP-10 correlated statistically (p = 0.0033). Increased IP-10 levels in HCMV-positive BALF samples were significantly associated with severe airflow obstruction, as indicated by a decrease in forced expiratory volume in one second (FEV1). Our data indicate that HCMV replication in LTRs evokes a plasma IP-10 response and that, when an IP-10 response is observed in BALF, it is associated with inflammatory airway obstruction in the allograft.

Abbreviations: 
HCMV

human cytomegalovirus

LTRs

lung transplant recipients

Th1

CD4+ T-helper cells type 1

IP-10

interferon-inducible protein of 10 kilodaltons

BALF

bronchoalveolar lavage fluid

FEV1

forced expiratory volume in one second

LuTX

lung transplantation

BOS

bronchiolitis obliterans syndrome

CXCL-10

CXC ligand 10

HRV

human rhinovirus

HRSV

human respiratory syncytial virus

COPD

chronic obstructive pulmonary disease

PCR

polymerase chain reaction

ELISA

enzyme-linked immunosorbent assay

HIV

human immunodeficiency virus

Introduction

Human cytomegalovirus (HCMV) causes significant mortality and morbidity in lung transplant recipients (LTRs) (1–3). In addition to direct clinical effects of HCMV replication after lung transplantation (LuTX), such as HCMV syndrome and tissue-invasive HCMV disease, there is evidence that HCMV also mediates inflammatory processes that have been associated with long-term sequelae of LuTX (4–8). These include acute cellular rejection and bronchiolitis obliterans syndrome (BOS), which manifest clinically as progressive graft dysfunction (9,10).

To date, the mechanisms of HCMV pathogenesis after LuTX are not entirely understood. Previous studies have indicated that the HCMV serostatus of the donor and the recipient, individual viral kinetics, and the treatment regimen have a major impact on the clinical outcome of the infection (2,11,12). Furthermore, the immune response of the host has been identified as decisive factor for development of HCMV disease (13). While the prominent role of HCMV-specific CD4+ T-helper cells with a type 1 cytokine profile (Th1 cells) has been verified in several studies, the role of chemokines associated with the Th1 response in HCMV infection is less clear (14–16).

Chemokines are small chemotactic cytokines that modulate inflammatory processes and regulate migration of leucocytes and cell–cell signaling in the immune system. Four subfamilies, based on the number and arrangement of conserved cysteines have been described (CXC, which is subdivided into non-ELR and ELR [Glu-Leu-Arg motif], CC, C and CX3C) (17,18). A restricted subset of non-ELR CXC inflammatory chemokines act as high-affinity ligands for the CXCR3 receptor found predominantly on activated Th1 cells and chemoattract effector T cells to the site of the infection, thereby inducing and shaping a Th1-polarized adaptive immune response (19–22). Interferon-inducible protein of 10 kilodaltons (IP-10), or CXC ligand 10 (CXCL-10), is one of these CXCR3-binding chemokines that exerts a stimulating effect on the directional migration of activated and memory Th1 cells and promotes the production of Th1 cytokines (23).

An association between specific viral airway infections and IP-10 production has been shown previously. In vitro studies have indicated that infections with influenza virus, human rhinovirus (HRV) and human respiratory syncytial virus (HRSV) are associated with IP-10 expression, mainly in bronchoalveolar epithelial cells and macrophages (24–30). Recently, it was demonstrated that the plasma IP-10 level can also be used as biomarker for predicting inflammatory airway obstruction after HRV infection, which results in exacerbations of asthma bronchiale and chronic obstructive pulmonary disease (COPD, 29,31).

In transplantation, and in particular in LuTX, IP-10 has been shown to be involved in acute cellular rejection and in the development of obliterative bronchiolitis, where it triggers an influx of alloreactive CXCR3-positive T cells into the airways (32). Compared to asymptomatic LTRs, IP-10 levels are apparently increased in bronchoalveolar lavage fluid (BALF) or serum from LTRs who suffer from primary graft dysfunction, acute allograft rejection or BOS (33,34).

For HCMV, which infects cells of the lung allograft and plays a significant role in LuTX, there are still no data that reveal an association between HCMV replication in LTRs and IP-10 production. However, in vitro experiments show that monocytes massively upregulate IP-10 mRNA upon stimulation with HCMV (35).

We therefore investigated in the present study whether HCMV replication, as indicated by presence of HCMV DNA in blood or the lung allograft of LTRs during the posttransplant follow-up, was associated with an increase in IP-10 levels, which might have further implications for the transplant. We show that, in LTRs, HCMV triggers a plasma IP-10 response, irrespective of the compartment where the virus replicates, and that an increase in the IP-10 level in BALF due to HCMV is associated with inflammatory airway obstruction.

Materials and methods

Patients

In this retrospective study, 36 patients were included who received a lung transplant at the Medical University of Vienna between May 2002 and August 2008. 26 LTRs were selected as study patients if their virological records revealed at least one episode of HCMV replication with viral DNA levels exceeding 1000 copies/mL in BALF and/or plasma during the posttransplant period. Ten patients without any episode of HCMV replication during the follow-up served as control group. Clinical characteristics of study patients are presented in Table 1. All patients received immunosuppression with prednisolone, mycophenolate and cyclosporine or tacrolimus as well as HCMV prophylaxis. Prophylaxis consisted of hyperimmune globulin (10 mg/kg per dose every 12 hours) at days 1, 7, 14 and 21 after LuTX, ganciclovir administered intravenously for 3 weeks, and valganciclovir administered orally (450–900 mg, twice daily, depending on the patient's weight and renal function) (Table 1). This study was conducted in accordance with the guidelines of the local ethics committee.

Table 1.  Clinical characteristics of study patients
Patient # IDSex (male/ female)Age (years)Primary diseaseType of LuTX (single/ double)Date of LuTX (month/ year)HCMV serostatus (donor/ recipient)Time point(s) (of) (day post LuTX)Compartment of first HCMV DNA detection (BALF only/plasma ± BALF)Quantitative PCR at first HCMV DNA detection (copies/mL)Histological grading of biopsy sample obtained at first HCMV DNA detection1Time point(s) of further episodes of HCMV detection within follow-up (day post LuTX)Onset of BOS (day post LuTX)Follow-up (days post LuTX)
End of HCMV prophylaxisBaseline sample was obtainedFEV1 baseline was determinedFirst HCMV DNA detectionPlasmaBALFAcute rejectionAirway inflammationPlasma ± BALFBALF only
  1. LuTX = lung transplantation; HCMV = human cytomegalovirus; COPD = chronic obstructive pulmonary disease; CF = cystic fibrosis; A1AD = alpha 1-antitrypsin deficiency; D = donor; R = recipient; BALF = bronchoalveolar lavage fluid; PCR = polymerase chain reaction; BOS = bronchiolitis obliterans syndrome.

  2. 1Evaluated according to the 1996 International Society for Heart and Lung Transplantation working formulation: A0 = no acute rejection; A1 = minimal acute rejection; A2 = mild acute rejection; B0 = no airway inflammation; B1 = minimal airway inflammation; B2 = mild airway inflammation.

 #1m58COPDDouble01/2006D+R+10591113, 191220BALF only<10002400A0B1None409440
 #2f60COPDSingle03/2006D+R+110110350, 425443BALF only<10001100A1B1None569, 684498684
 #3f56COPDDouble07/2006D−R+100100124, 158200BALF only<10001662A2B1–2235None1198
 #4m53COPDDouble07/2006D+R+663366, 94122BALF only<1000140 000A0B0–13761921231
 #5f27CFDouble06/2005D+R−135501849, 877973BALF only<10002030A0B2NoneNone11581225
 #6m65COPDDouble06/2005D+R+7737115, 140168BALF only<10003170A0B0NoneNone7111617
 #7f19CFDouble05/2002D?R+98105No FEV1 data Available189BALF only<10005830A0–1B0–1None3992459
 #8f64COPDSingle06/2007D+R+115101101, 115161BALF only<10002320A2B0–1NoneNone892
 #9f45COPDDouble06/2006D+R+8383112, 143181BALF only<10006990A0B0–1259None1147
#10f63COPDDouble09/2003D?R+12736190, 212225BALF only<100061 500A0B0NoneNone273
#11f57COPDDouble07/2007D−R+63112133, 146175BALF only<100028 400A0B0–1None369406
#12m30CFDouble07/2007D?R+848484, 119189BALF only<100020 200A1B1None371421
#13m46COPDDouble04/2008D+R+10011256, 112177BALF only<10001840A0B1–2NoneNone370
#14f60COPDDouble05/2008D+R+963796, 156178BALF only<100020 500A0B0–1None369369
#15f57COPDDouble03/2008D−R+9235107, 139171BALF only<100065 700A0B1–2None275580
#16m53FibrosisSingle08/2006D+R+9557141Plasma ± BALF16 400<1000A0B0–1NoneNone1130
#17m48COPDDouble05/2007D+R+9292165Plasma ± BALF1680No BALFNo bronchoscopyNoneNone933
#18m58A1ADDouble08/2007D−R+7526133Plasma ± BALF8050No BALFNo bronchoscopyNone364819
#19f48COPDSingle02/2006D+R+113251310Plasma ± BALF3820No BALFNo bronchoscopy576, 764None4211396
#20f50COPDDouble12/2007D−R+894141, 89147Plasma ± BALF4800119 000A0B0–1NoneNone339670
#21m58COPDDouble06/2007D+R+9090188Plasma ± BALF15 800No BALFNo bronchoscopyNone213569912
#22m61COPDDouble07/2007D+R−36284159Plasma ± BALF92 600No BALFNo bronchoscopyNone457889
#23m54COPDDouble09/2005D+R−364319394Plasma ± BALF6380No BALFNo bronchoscopy880503, 649, 758581880
#24m41CFDouble08/2008D+R−Permanent therapy30157Plasma ± BALF7690No BALFNo bronchoscopyPersistent viremiaPersistent detection245
#25f59COPDDouble08/2007D?R+417777, 99137Plasma ± BALF29301410A0B1–2189, 264None830
#26m65FibrosisDouble05/2004D+R+10835No FEV1 data Available1863Plasma ± BALF63401840A0B1NoneNone1875

Samples

All samples investigated in this study were obtained in the course of the patients’ clinical routine follow-up posttransplantation, which included weekly collection of plasma samples for two months after LuTX, and thereafter, monthly, as well as bronchoscopies, with sampling of BALF at weeks 1, 2, 4, 8, 12, 24 and 52 after LuTX, with additional procedures performed if infection or rejection was suspected clinically. When plasma HCMV loads exceeded 1000 copies/mL during follow-up, preemptive valganciclovir therapy was initiated. For the study, the following specific samples from the follow-up were selected:

  • (a) Baseline samples: From each patient, a pair of plasma and BALF samples, obtained on the same day in the posttransplant follow-up, was selected and defined as baseline based on the following criteria: (1) Clinical information obtained by retrospective chart review indicated no clinical signs of rejection, inflammation or infection on the day the samples were obtained, and the patients had received no antimicrobial therapy except valganciclovir and amphotericin B prophylaxis. (2) C-reactive protein, leucocyte and lymphocyte numbers were in normal range. (3) Histological investigation of a lung biopsy specimen obtained on the same day from all patients showed no evidence of rejection, infection, inflammation, T cell infiltration or viral inclusion bodies in the tissue, and immunohistochemistry and in situ hybridization gave negative results for HCMV and Epstein–Barr virus infection. (4) In the analysis of the baseline BALF sample, the bacterial count of the resident ororpharyngeal flora was lower than 107/mL, Candida albicans was detected at a concentration of no more than 100 colony forming units/mL, and acid-resistant rods and mycobacteria were undetectable by polymerase chain reaction (PCR) and culture. Furthermore, the BALF sample was negative for Toxoplasma gondii, Pneumocystis carinii (both by PCR) and Aspergillus fumigatus (culture), and the absence of respiratory viruses (adenovirus, HRSV, influenza A and B viruses, parainfluenza 1, 2 and 3 viruses) was verified by antigen enzyme-linked immunosorbent assay (ELISA). (5) HCMV was undetectable in plasma by PCR and in BALF by PCR and shell vial culture. The time point at which the baseline sample was obtained from each patient is shown in Table 1.
  • (b) Samples at initial HCMV detection: From each of the 26 study patients, a plasma sample was tested that was taken on the day during the posttransplant follow-up when the HCMV DNA level in plasma and/or BALF first exceeded 1000 copies/mL. In 11 LTRs, this first emergence of HCMV DNA was detected in plasma, while in 15 LTRs, HCMV was first detected at levels higher than 1000 copies/mL only in BALF (Table 1). From these 15 patients, and from 3 LTRs in whom HCMV was initially detected in plasma and BALF, HCMV-positive BALF samples were investigated.
  • (c) Samples from later episodes of HCMV detection and posttreatment samples: In addition, 23 plasma samples were analyzed that were obtained during further episodes of HCMV detection with a viral load exceeding 1000 copies/mL in plasma and/or BALF, including 8 episodes of HCMV detection in plasma and 15 episodes of HCMV replication only in BALF (Table 1). Twelve additional posttreatment plasma samples were included, which were the first HCMV-DNA-negative samples taken after oral valganciclovir treatment of either the initial episode or a later episode of HCMV replication.

Detection of HCMV

Quantitative assessment of the HCMV DNA load in plasma and BALF samples was performed by PCR using a Cobas Amplicor HCMV Monitor-Test Kit on a COBAS Amplicor Analyzer (Roche Molecular Systems, Branchburg). In addition, HCMV was detected in BALF by shell vial method and by immunofluorescence staining of immediate-early antigen in infected cells. Lung biopsy specimens were analyzed for presence of HCMV by immunohistochemistry. Gancyclovir resistance testing was performed as described previously (36).

Measurement of IP-10 levels

IP-10 levels in plasma and BALF were determined using a commercially available ELISA (BD OPTEIA Human IP-10 ELISA Set, Becton Dickinson Biosciences, San Diego, CA, USA). The lower limit of detection was 7.5 picograms (pg) per mL. All samples were stored at −20°C. Each plasma sample was tested undiluted, as well as in 10- and 20-fold dilutions. BALF samples were tested in serial dilutions of twofold steps from 1:2 to 1:64. As shown previously, the relationship between IP-10 concentration measurements in diluted and undiluted samples was not linear, but piecewise linear, with the degree of divergence increasing with increasing concentration (37). Therefore, the cutoff points for the pieces of linearity in plasma and BALF samples were determined, and the concentration in undiluted samples with higher IP-10 levels was estimated from the reverse function.

Evaluation of airway obstruction

The forced expiratory volume in one second (FEV1) was determined in 16 LTRs in whom HCMV DNA was detected in the BALF during the initial episode of viral replication posttransplant (Table 1). The FEV1 value at initial HCMV detection was compared to the arithmetically averaged mean of two FEV1 values determined at immediately preceding regular visits to the outpatient clinic when HCMV was undetectable and signs of infection, acute rejection or BOS were absent (FEV1 baseline) (Table 1).

Statistical methods

Within-patient comparison of IP-10 levels was performed using paired nonparametric t-tests (Wilcoxon matched pairs test). Groups of patients were compared by Mann–Whitney U test. Spearman's rank test was used to estimate the association between HCMV DNA and IP-10 levels. For all statistical tests, a p-value of <0.05 was considered statistically significant. GraphPad Prism version 5.0 software was used for statistical analysis.

Results

Plasma IP-10 levels and initial HCMV detection posttransplantation

First, we investigated whether initial detection of HCMV replication after LuTX was associated with a rise in plasma IP-10 levels. Therefore, we measured the IP-10 level in baseline plasma samples that had been obtained from each of the 26 LTRs included in the study at a time point when neither HCMV nor any other infection, inflammation or rejection was detectable (see Materials and Methods) and compared it to the IP-10 level in a plasma sample obtained later during the posttransplant follow-up, at the point when HCMV DNA was first detected. The samples were divided into two groups for analysis, according to the compartment where HCMV was first detected. In 11 of the LTRs, HCMV was first detected in the blood (Table 1). Analysis of samples from these 11 individuals, shown in Figure 1(A), revealed a significantly higher IP-10 concentration in HCMV-DNA-positive plasma samples than in the patients’ baseline plasma samples (Figure 1A, p = 0.001). In 15 LTRs, HCMV could initially be detected only in BALF. For each patient, the plasma IP-10 level on the day when HCMV was first detected in the BALF specimen was compared to the baseline level, as presented in Figure 1(B). Although the virus was not detected in the blood, the plasma IP-10 levels were also significantly increased compared to baseline when HCMV emerged in the BALF only (Figure 1B, p < 0.0001). Finally, we investigated whether the extent of the plasma IP-10 increase was associated with the compartment where HCMV replication was detected. The levels of plasma IP-10 in patients with HCMV DNA in their plasma were compared to those in patients with HCMV DNA in their BALF only, as shown in Figure 1(C). The HCMV-associated plasma IP-10 level was significantly lower in patients in whom HCMV was detected only in BALF than in those in whom the virus was detected in plasma (Figure 1C, p = 0.0059).

Figure 1.

IP-10 plasma levels at the time of initial detection of HCMV. Baseline IP-10 levels were measured in plasma samples that were obtained from 36 LTRs during the posttransplant follow-up at a time when neither HCMV nor any other infection, inflammation or rejection was detected. For each of the 26 study patients, the baseline IP-10 level was compared to the level in the plasma sample taken during the first episode of HCMV replication that occurred during the posttransplant follow-up (Wilcoxon matched pairs t-test). (A) In 11 of the LTRs, HCMV was first detected in the blood. A significantly higher IP-10 concentration was detected in these HCMV-DNA-positive plasma samples than in the patients’ baseline plasma samples (p = 0.001). (B) In 15 LTRs, HCMV could initially be detected only in BALF, but not in blood. The IP-10 levels in plasma samples obtained on the same day that HCMV was initially detected in these BALF samples were significantly increased relative to the corresponding baseline samples (p < 0.0001). (C) The plasma IP-10 level at the time of initial HCMV detection during the posttransplant follow-up was significantly higher in LTRs in whom HCMV DNAemia was detectable than in those in whom HCMV DNA was detected in BALF only (Mann–Whitney U test, p = 0.0059). Baseline IP-10 levels in plasma samples obtained at cessation of antiviral prophylaxis from 10 LTRs who did not show any HCMV replication episodes during the entire posttransplant follow-up served as controls.

IP-10 plasma levels during further episodes of HCMV detection

Next, we addressed the question whether further episodes of HCMV detection that occurred after the initial detection of the virus during follow-up were also associated with an increase in plasma IP-10 levels. For this purpose, we tested 23 additional plasma samples obtained from 16 LTRs during further episodes of detectable HCMV DNA replication. In 8 episodes, HCMV DNA was detected in blood, and in 15, only in BALF (Table 1). During these episodes of HCMV detection, an increase in plasma IP-10 levels compared to the baseline was detected, regardless of whether viral DNA was found in plasma (Figure 2A, p = 0.0078) or in BALF only (Figure 2B, p < 0.0001). However, as shown in Figure 2(C), significantly higher levels of plasma IP-10 were found in patients with HCMV DNAemia (p = 0.0103) than in those with HCMV in BALF only.

Figure 2.

IP-10 plasma levels during further episodes of HCMV detection. IP-10 levels were measured in 23 plasma samples obtained from 16 study patients during further episodes of detectable HCMV DNA replication during the posttransplant follow-up and compared to the levels in the corresponding baseline plasma samples from these patients, which were obtained when neither HCMV nor any other infection, inflammation or rejection was detectable (Wilcoxon matched pairs t-test). (A) In eight episodes of viral replication, HCMV DNA was detected in the blood. The IP-10 concentration was significantly higher in these HCMV-DNA-positive plasma samples than in the corresponding baseline plasma sample (p = 0.0078). (B) In 15 episodes of HCMV replication, HCMV DNA was detected in BALF only. The IP-10 concentration in plasma samples obtained on the same day that HCMV DNA was first detected in the BALF specimen was significantly higher than the corresponding baseline sample (p < 0.0001). (C) Plasma IP-10 levels during further HCMV replication episodes during the posttransplant follow-up were significantly higher in LTRs in whom HCMV DNAemia was detectable than in those in whom HCMV DNA was detected in BALF only (Mann–Whitney U test, p = 0.0103). IP-10 levels in plasma samples that met baseline criteria and were obtained at least 6 months after LuTX from 10 control patients who did not show any HCMV replication episodes during the entire posttransplant follow-up formed the control group.

Correlation and kinetics of plasma IP-10 level and HCMV DNA load

We further analyzed whether the plasma HCMV DNA load and the plasma IP-10 level correlated statistically in HCMV-positive plasma samples. As shown in Figure 3(A), a significant correlation was found between plasma HCMV-DNA load and plasma IP-10 concentration in the 19 HCMV-positive plasma samples investigated in this study (p = 0.0033, Spearman R = 0.6386). To compare the kinetics of viral load and IP-10 levels, HCMV DNA and IP-10 concentrations were measured in multiple plasma samples from one LTR (patient #24, Table 1) during the course of primary HCMV infection. As shown in Figure 3(B), in this patient, kinetics of the plasma IP-10 level paralleled those of the plasma HCMV DNA load during the entire follow-up, even when the virus level suddenly changed due to development of valgancylovir resistance and rejection treatment.

Figure 3.

Correlation and kinetics of IP-10 and HCMV DNA levels in plasma. (A) The plasma HCMV DNA load (copies/mL) correlated significantly with the plasma IP-10 level (pg/mL) in 19 HCMV-positive plasma samples investigated in this study, including samples obtained at the time of initial detection of the virus and ones obtained during episodes of HCMV replication that occurred during the posttransplant follow-up (p = 0.0033, Spearman R = 0.6386). (B) In one LTR (patient #24, Table 1) with a primary HCMV infection, the kinetics of plasma IP-10 concentration paralleled the kinetics of the HCMV DNA load during the complete posttransplant follow-up, even when the virus levels suddenly changed due to the development of valgancylovir resistance and rejection treatment.

Plasma IP-10 levels in antiviral therapy

Next, we evaluated whether the decrease in HCMV load due to successful antiviral treatment was also associated with a decrease in plasma IP-10 levels. IP-10 levels of 12 HCMV-positive plasma samples obtained from 11 LTRs before the beginning of oral valganciclovir treatment were determined and compared to those in consecutive HCMV-negative follow-up samples obtained after cessation of successful antiviral therapy. As shown in Figure 4, the decline of the HCMV load to negativity after treatment was associated with a decrease in the plasma IP-10 level in all patients (Figure 4, p = 0.0005). We further compared posttreatment plasma IP-10 levels with the baseline plasma IP-10 concentration and found that the plasma IP-10 concentration was still higher in the posttreatment samples than in baseline, although HCMV was already undetectable (Figure 4, p = 0.001).

Figure 4.

Plasma IP-10 levels during antiviral therapy. IP-10 levels in 12 HCMV-positive plasma samples obtained from 11 LTRs before oral valganciclovir treatment was administered were compared to levels in consecutive HCMV-negative posttreatment samples taken after cessation of successful antiviral therapy. The decrease in the HCMV DNA load to negativity due to antiviral treatment was associated with a significant decrease in the plasma IP-10 level (p = 0.0005). IP-10 levels in the HCMV-negative posttreatment samples were slightly, but significantly, higher than the corresponding baseline plasma IP-10 levels from these patients (p = 0.001).

IP-10 level in BALF and loss of FEV1

We then investigated whether an IP-10 response to HCMV could also be detected in BALF samples. For this purpose, we analyzed the IP-10 level in BALF samples from the 18 patients in whom HCMV was detected in BALF during the initial episode of replication posttransplant (Table 1). The BALF IP-10 levels were then compared to the patients’ baseline BALF IP-10 concentration, and in general, a higher IP-10 concentration was found during HCMV replication (Figure 5A, p = 0.0092). However, a substantial increase in the BALF IP-10 level, which was defined as more than five times the baseline value and additionally exceeding a level of 500 pg/mL, was observed in 8 LTRs but did not exceed this cutoff in the other 10 patients. When we assessed whether these differences in BALF IP-10 levels were due to different HCMV loads in BALF, we found no statistical correlation between HCMV load and IP-10 (p = 0.3662, Spearman R =−0.2265; data not shown).

Figure 5.

IP-10 level in BALF and loss of FEV1 at the time of initial detection of HCMV. (A) IP-10 levels were determined in HCMV-positive BALF samples from 18 LTRs in whom HCMV was detected in BALF during the initial episode of viral replication. These were compared to the patients’ baseline BALF samples, which were obtained when neither HCMV nor any other infection, rejection or inflammation was detectable. Overall higher IP-10 levels were found in BALF during HCMV replication infection (p = 0.0092). A substantial IP-10 response in BALF to HCMV (defined as BALF IP-10 levels more than five times the baseline value and exceeding 500 pg/mL) was found in 8 of 18 LTRs. IP-10 levels in BALF samples obtained at cessation of antiviral prophylaxis from 10 control patients who did not show HCMV replication episodes during the entire posttransplant period served as controls. (B,C) The 8 LTRs who showed a substantial IP-10 response in BALF to HCMV infection showed a significantly higher loss of respiratory function, as indicated by a higher absolute (B, p = 0.007) and relative (C, p = 0.0104) reduction in forced expiratory volume in one second (FEV1) than patients whose BALF IP-10 levels in HCMV-positive BALF samples did not exceed the cutoff of five times the baseline (BL) value and 500 pg/mL.

Finally, we investigated whether the BALF IP-10 levels in the initial episode of HCMV replication in the lung, as indicated by HCMV detection in the BALF, were associated with the patients’ loss of respiratory function. To measure acute loss of respiratory function, we compared the FEV1 measured in 16 of the 18 LTRs at the time when HCMV was initially detected in BALF with the respective FEV1 baseline. The association of the IP-10 response due to HCMV with the FEV1 change is shown in Figures 5(B) and (C). Patients with a BALF IP-10 response to HCMV that was more than five times the baseline value and exceeded a level of 500 pg/mL showed a higher absolute reduction of FEV1 (Figure 5B, p = 0.007) as well as a higher reduction compared to FEV1 baseline (Figure 5C, p = 0.0104). There was no histologic evidence of acute rejection in 7 out of 8 LTRs with a FEV1 drop and a BALF IP-10 increase due to HCMV replication, while in one patient a mild rejection episode was detected. In none of the patients HCMV pneumonitis could be verified by immunohistological staining at this time point.

Discussion

The present study provides first evidence that HCMV evokes a systemic IP-10 response in transplant recipients. Since IP-10 is a key signal of the innate immune system to recruit activated Th1 cells to the site of the infection and is a prerequisite for the establishment of a local Th1 response, this finding reveals a possible link between HCMV replication and induction of a HCMV-specific Th1 response in the organ where HCMV replicates.

It has been shown previously for acute infections with hepatitis C virus and human immunodeficiency virus (HIV) that viral replication can trigger an IP-10 response, which ultimately leads to an elevation of the plasma IP-10 level. With these viruses, a massive peak in plasma IP-10 immediately followed the early peak in viremia in acute infection (38). Consistent with these observations, data from our study reveal that episodes of HCMV DNAemia are also associated with simultaneous elevation of plasma IP-10. These findings, together with the fact that the HCMV DNA load and the IP-10 level in plasma significantly correlate, suggest that HCMV replication and IP-10 production are functionally linked, which is further supported by the finding that HCMV load and IP-10 concentration simultaneously decrease after antiviral treatment and even showed parallel kinetics in a primarily infected patient. It is possible that HCMV replication and IP-10 response are functionally linked by a direct induction of IP-10 expression in infected cells by protein components of HCMV, as it has been shown for other viruses (24).

In LTRs, HCMV replication is not always associated with viremia and is sometimes restricted to the local tissue of the transplanted lung (39,40). One main finding of the present study was that the plasma IP-10 level increases even when HCMV DNA replication is detected only in BALF, and not in blood. The significantly increased IP-10 levels in the blood in cases where HCMV replicated only in the allograft most likely originated in the parenchyma of the transplanted lung, where this chemokine was apparently produced by infected or activated cells and then efficiently secreted into the bloodstream. This indicates that, even in episodes when HCMV replication is confined to the allograft, there is a systemic stimulation of migration of Th1 cells due to an increase in the plasma IP-10 level. However, our data suggest that this stimulus is weaker when there is no hematogenous dissemination.

An interesting aspect of the present study is that, for the first time, a significant increase in plasma IP-10 levels in response to HCMV was found in LTRs, a patient population that is under immunosuppressive therapy with corticosteroids and calcineurin inhibitors. Although it has been shown that these immunomodulatory substances impair essential T cell functions and thus prevent allograft rejection, it is less clear whether they affect chemokine production and, in particular, IP-10 expression. While calcineurin inhibitors cyclosporine and tacrolimus have been shown to negatively affect IP-10 mRNA levels in leucocytes under certain stimulatory conditions in vitro, experiments with corticosteroids revealed no significant reduction of Interferon-γ-mediated or HRV-induced IP-10 expression when administered at very high concentrations (26,41,42). Data from the present study now indicate that, in vivo, a significant rise in IP-10 levels due to HCMV replication occurs despite immunosuppressive therapy and that corticoids and calcineurin inhibitors do not completely abrogate the HCMV-associated IP-10 response in LTRs, at least when the clinically applied dosage is used.

The present study also showed that IP-10 levels in BALF may increase due to local HCMV replication in the lung, but this increase was not seen consistently in all LTRs with viral replication in the allograft. From the clinical data evaluated, however, increased IP-10 levels in the BALF apparently have a clinical correlate, as we found that strong IP-10 responses to HCMV replication in the lung were significantly associated with severe inflammatory airway obstruction, as indicated by a reduction of the patients’ FEV1. This is the first time that such an association was found in the course of HCMV infection in LTRs, and this finding is consistent with previous data that show a correlation between IP-10 and FEV1 reduction in acute asthma and COPD exacerbations due to HRV infection (29,31). Based on these findings, and analogous to what has been observed with HIV-infected patients with T cell alveolitis, it seems likely that in HCMV infection of the transplanted lung allograft, IP-10 could signal CXCR3-positive T cells to infiltrate the respiratory tract and express antiviral cytokines, thereby mediating airway inflammation, with airflow obstruction as its clinical consequence (19). In the mouse model, it has already been verified that, upon inoculation with murine cytomegalovirus and a subsequent systemic IP-10 response, virus-specific effector T cells infiltrate the infected organ and mediate inflammation at the site of viral replication (20).

HCMV infection is believed to mediate so-called ‘indirect effects’, i.e. immunological processes that may lead to acute allograft rejection and development of BOS. An important aspect of the present study with regard to these ‘indirect effects’ is that IP-10 also plays a key role in attracting T cells and polarizing Th1 responses in acute and chronic allograft rejection (32–34,43). A rise in IP-10 levels due to HCMV replication could result in recruitment of not only HCMV-specific but also of alloimmune T cells to the allograft. It has been shown that trafficking of CXCR3-positive allograft-specific T cells into the lung is regulated by production of IP-10 by graft-infiltrating macrophages, and from this, it seems likely that IP-10 poses a functional link to these ‘indirect effects’ (44). Because we have shown that plasma IP-10 levels still exceeded the baseline level when HCMV replication had been reduced to undetectable levels by antiviral treatment, a long-term and cumulative effect of multiple HCMV replication episodes on the plasma IP-10 level seems possible that could be involved in immunopathological mechanisms responsible for ‘indirect effects’ of HCMV replication.

In conclusion, the results obtained in our study demonstrate that, in LTRs, HCMV triggers a systemic IP-10 response that occurs independently of whether HCMV replication is restricted locally to the lung allograft or is associated with hematogenous dissemination. The extent to which elevated IP-10 plasma levels due to HCMV replication are involved in immunological processes of acute cellular rejection and BOS needs to be analyzed in future studies. Finally, we show that IP-10, which is produced in the lung compartment when local HCMV replication is occurring, apparently mediates severe inflammatory airflow obstruction and could possibly be used as clinical marker.

Acknowledgments

The authors would like to thank Bettina Jaradat for her support in acquiring the clinical data.

Disclosure

The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation.

Ancillary