Improvement in Hepatic Fibrosis Biomarkers Associated With Chemokine Receptor Inactivation Through Mutation or Therapeutic Blockade

Abstract

Background

The C-C chemokine receptor Type 5 (CCR5) is a key receptor for human immunodeficiency virus type 1 (HIV-1) entry into T-cells and a variant allele, CCR5 delta-32, is associated with decreased viral replication and disease progression. Active HIV-1 replication is highly associated with accelerated rates of hepatic fibrosis. We postulated that CCR5 plays a role in the development of hepatic fibrosis and evaluated the longitudinal effect of natural or drug-induced CCR5 mutation and blockade on biomarkers of liver fibrosis in HIV-1 patients.

Methods

To accomplish this goal, we examined 2 distinct cohorts. First, we evaluated fibrosis markers in the Multicenter Hemophilia Cohort Studies (MHCS), which included subjects with HIV and hepatitis C virus (HCV) coinfection with the CCR5 delta-32 allele. We also evaluated an HIV-1 infected cohort that was treated with a dual CCR5/CCR2 antagonist, cenicriviroc. The enhanced liver fibrosis (ELF) index was validated against liver histology obtained from HCV/HIV and HCV patients and demonstrated strong correlation with fibrosis stage.

Results

In both the MHCS patients and patients treated with cenicriviroc, CCR5 mutation or blockade was associated with a significant decrease in the ELF index. Among the patients with the delta-32 allele, the ELF index rate significantly decreased in sequential samples as compared to CCR5 wild-type patients (P = .043). This was not observed in control subjects treated with efavirenz nor with a lower dose of 100 mg cenicriviroc.

Conclusion

These findings suggest that hepatic fibrosis in HIV-1 infected patients can be modulated by the mutation of CCR5 and/or use of CCR5/CCR2 blockade agents.

Clinical Trials Registration

NCT01338883.

Natural gene mutations in a gene called CCR5, or specific blockade of CCR5/CCR2 function, may slow liver fibrosis in HCV/HIV coinfection.

Chemokines play a key role in leukocyte-trafficking through their interaction with their cognate receptor. The C-C chemokine receptor Type 5 (CCR5) is a biologically important protein that is present on the surface of a variety of cell types, including T cells and other immune system cell types, hepatocytes, and probably stellate cells [1, 2]. In humans, CCR5 is a receptor for key ligands include MIP-1 alpha and MIP-1 beta, as well as RANTES, a potent chemotactic cytokine. CCR5 also serves as a coreceptor for human immunodeficiency virus (HIV) entry into CD4+ T cells [3].

In human populations, a deletion mutation of 32 basepairs creates a stop codon that results in formation of a nonfunctional CCR5 receptor. The mutation is moderately prevalent in people of western European origin, possibly due to selection pressure exerted by Yersinia pestis-associated plague/smallpox epidemics [4, 5]. In modern times, individuals with homozygotic or heterozygotic alleles containing the CCR5 delta-32 mutation are either resistant to infection with CCR5-utilizing HIV or progress more slowly toward AIDS [6–9].

Patients with HIV infection demonstrate systemic immune activation due to gut barrier breakdown with leakage of bacterial endotoxin into the portal system [10, 11]. Furthermore, these patients are at greater risk for coinfections with hepatitis C virus (HCV), hepatitis B virus (HBV), and frequently demonstrate faster fibrotic progression than those with monoinfections of hepatitis viruses [12, 13]. However, HIV alone is associated with fibrotic progression in the liver [14–17]. We postulated that individuals with either naturally occurring CCR5 mutations, or those exposed to agents that block CCR5, would have slower rates of fibrotic progression than those without such mutations. To examine this hypothesis, biomarkers of fibrotic progression were evaluated in 2 distinct cohorts.

METHODS

Study Populations

Multicenter Hemophilia Cohort Studies

Initiated in 1982, the Multicenter Hemophilia Cohort Study (MHCS) was an 18-year study of patients with hemophilia that sought to understand the natural history of HIV infection and AIDS in a population at high risk because of the use of blood products and coagulation factor concentrates. The Second MHCS (MHCS-II), 2000–2006, prospectively followed a cohort of 2566 hemophiliacs with HCV infection (including 442 from MHCS-I) from 54 treatment centers. These large cohort studies provided long-term longitudinal evaluations of patients with hemophilia from multiple sites, which collected prospective blood samples, and clinical data [18]. A subset (N = 997) of the total MHCS cohort was typed for CCR5 homo/heterozygosity, and samples for analysis were selected from this group. To be eligible for additional analysis, subjects were first selected by the following criteria: (a) HCV or HCV/HIV infected, (b) homo- or heterozygotic for CCR5 delta-32, and (c) 3 sequential samples available with at least 2 years between 3 samples. Following selection, controls were matched to CCR5 delta-32 mutation subjects by race, age at earliest draw within 5 years, sex, time between mid and late sample within 1 year, and HCV monoinfection or HCV/HIV coinfection status.

Cenicriviroc Phase 2 Trial Cohort

Cenicriviroc (CVC) is an oral, once-daily drug with dual antagonism of CCR5 and CCR2, which was studied in a randomized, controlled phase 2 trial (NCT01338883) of HIV-infected subjects to determine safety and efficacy of response. The multicenter treatment trial randomized HIV treatment-naive subjects to 1 of 3 treatment arms (2:2:1) including CVC 100 mg + tenofovir/emtricitabine, CVC 200 mg + tenofovir/emtricitabine or a control arm with efavirenz (EFV, a non-nucleoside reverse transcriptase inhibitor) + tenofovir/emtricitabine. Patients with HCV, HBV, or known cirrhosis were excluded. Serum ALT elevation was excluded only if >2.6× upper limit of normal. All patients had CCR5-tropic HIV. Patients were treated for 48 weeks [19]. Samples for fibrosis biomarker and cytokine analysis were available from baseline and 48 weeks (end of treatment). Current clinical trials in patients with NAFLD/NASH utilize an updated formulation of CVC (150 mg) that most closely approximates the 200 mg dose evaluated in this study.

Enhanced Liver Fibrosis Index and Cytokine Measurement

Serial measures of hepatic fibrosis were determined by calculation of the Enhanced Liver Fibrosis (ELF) index using the following formula: [ELF Index = −7.412 + (ln (HA) × 0.681) + (ln(PIIINP) × 0.775) + (ln(TIMP1) × 0.494) + 10], where HA is hyaluronic acid, TIMP1 is tissue inhibitor of metalloproteases-1, and PIIINP is procollagen-3. Levels were quantified using the following assays: Human PIIINP ELISA Kit (Cloud-Clone Corp), Human HA Quantikine ELISA kit (R&D Systems), and Human TIMP-1 ELISA (Abcam). The ELF index was validated using serum from patients who underwent liver biopsy, representing all Metavir fibrosis stages in patients with HCV/HIV coinfection. It has previously been validated in patients with HCV and NASH and other chronic liver diseases [20–23]. Cytokines including TGF-beta, TNF-alpha, MIP-1 (alpha and beta), interferon alpha, interferon gamma, interleukin (IL)-4, IL-6, IL-8, IL-12p70, IL-13, IL-17, IL-22, IL-33, RANTES, MIG, GCP-2, ITAC, IP-10, and MCP-1 were analyzed using Luminex multiplex assays (EMD Millipore).

Microarray

An Affymetrix Clariom S Chip was utilized to analyze transcriptome-wide gene expression from over 20000 genes. Frozen peripheral blood mononuclear cells (PBMCs) were thawed for total RNA extraction using the RNeasy mini kit (Qiagen) with on-column DNase treatment. Microarray services and bioinformatics were performed at the University of Michigan DNA Sequencing Core, Ann Arbor, Michigan.

Statistical Analysis

Cohort data were analyzed using parametric and nonparametric methods as appropriate. A Spearman rank correlation of the ELF index between HIV/HCV coinfected patients as stratified by Metavir fibrosis stage (F0-F4) was conducted. Evaluation of fibrosis change over time was performed using robust regression analysis. This methodology utilizes a form of regression analysis that overcomes limitations due to heteroscedasticity [24]. A weighted linear model designed for microarray analysis was utilized to compute the contrasts of interest [25]. Samples were weighted based on gene-by-gene algorithms, which downweigh chips deemed less reproducible [26]. Probe sets with a variance < 0.15 were filtered out and focused on probesets with a fold change of ≥2. P-values were adjusted for multiple comparisons using false discovery rate (FDR) as described by Benjamin and Hochberg [27]. The oligo and limma packages were implemented in the R statistical environment (R version 3.3.0).

RESULTS

Validation of ELF Index

Serum samples were previously collected from HCV/HIV coinfected patients enrolled in observational and treatment trials who provided permission for serum/tissue collection and use (ClinicalTrials.gov Identifier: NCT00055341, NCT00078403, NCT00545558) and a University of Cincinnati repository. The ELF index was determined in 80 HCV/HIV coinfected patients with liver biopsies that were staged for fibrosis using the Metavir scoring system. Figure 1 demonstrates the correlation between ELF index and degree of liver fibrosis. A significant correlation between ELF index and fibrosis stage was observed (r = 0.8018, P < .0001).

Figure 1.

Enhanced liver fibrosis (ELF) index validation. Highly significant analysis of variance (ANOVA) of ELF index between human immunodeficiency virus/hepatitis C virus coinfected patients (stratified by metavir fibrosis stage F0–F4) (r = 0.8018, P < .0001). Abbreviation: ELF, enhanced liver fibrosis.

Evaluation of Hepatic Fibrosis in MHCS

Change in ELF index over time in HCV/HIV coinfected subjects with the CCR5 delta-32 allele was compared to matched enrollees with 2 wild-type alleles (Figure 2). In a separate analysis, HCV monoinfected subjects with homozygotic CCR5 delta-32 alleles were compared to HCV monoinfected controls with wild-type CCR5. Demographic and viral characteristics of cohort subjects carrying delta-32 alleles and controls are shown in Table 1. Using a robust regression model and data from all 3 time points to compare CCR5 wild-type patients to matched patients with the delta-32 allele, the rate of increase in ELF index over the 4-year sample period is significantly reduced by the presence of the mutation (P = .043). However, this effect was not detected among matched pairs with HCV alone. The absolute ELF index difference for those carrying the delta-32 allele in the HCV/HIV coinfected subjects is 0.32 ELF units less than the wild-type homozygotics (95% confidence limits = −0.62 to −0.01). In contrast, patients with HCV alone progressed at a lower rate (0.11 to 0.12 ELF units), and the absolute difference between CCR5 wild-type and delta-32 variants was not statistically different.

Figure 2.

Longitudinal change in enhanced liver fibrosis index over a 4-year interval among hepatitis C virus/human immunodeficiency virus coinfected subjects enrolled in the Multicenter Hemophilia Cohort Studies (P = .043). Abbreviations: D32, delta-32; ELF, enhanced liver fibrosis; WT, wild-type.

Table 1.

Demographic and Viral Characteristics of Multicenter Hemophilia Cohort Studies Subjects

	Controls (wt/wt) (n = 51)	Cases (wt/d32 or d32/d32) (n = 51)
Sex, n (%)
Male	51 (100)	51 (100)
Race, n %
White	50 (98)	50 (98)
Black	1 (2)	1 (2)
Age first draw, years, mean (range)	29.3 (8.7–69.8)	28.8 (4.7–64.8)
Infection status
HCV-HIV-	1	1
HCV+HIV-	10	10
HCV+HIV+	40	40
CD4 count, n,	564, n = 31,	703, n = 24,
(range)	(200–1309)	(134–1714)

Abbreviations: D32, delta-32; HCV, hepatitis C virus; HIV, human immunodeficiency virus; WT, wild-type.

Evaluation of Hepatic Fibrosis in CVC Cohort

In the parent study, 41/56 subjects were enrolled in the CVC 200 mg arm and completed treatment per protocol, and 36 had paired pre-/post-treatment sample for analysis [19]. In the CVC 100 mg arm, 42/59 were randomized and completed treatment, but only 37 had paired pre-/post-treatment samples. Demographic and viral characteristics are shown in Table 2. Baseline ALT abnormalities—defined as ≥30 U/mL for men and ≥19 U/mL for women—were present in 25%–34% of subjects in each treatment arm. At baseline, the mean ELF index was 9.90 ± 0.97 in the CVC cohort receiving 200 mg/day and 9.95 ± 0.93 among those who received 100 mg CVC. Among the 28 patients randomized to EFV, 17 completed treatment and 16 had paired samples available. The mean baseline ELF score in the EFV controls was 9.13 ± 0.98. The mean change in ELF score after treatment remained essentially unchanged in patients who received CVC 100 mg. However, the mean change in ELF index after 48 weeks of 200 mg CVC was −0.87 compared to a mean increase of 0.16 ELF units in the patients receiving EFV (P = .01). This change corresponds to a mean decrease of 9% in the ELF index compared to a 2% increase in the control arm. Figure 3 shows the difference in slope over time between the 2 treatment arms, and a repeated measures analysis finds that ELF did not change among those that received EFV but that there was a significant decline in ELF index in those who received 200 mg CVC (P < .0001).

Table 2.

Demographic and Viral Characteristics of Cenicriviroc Subjects

	Controls (EFV) (n = 16)	Cases (CVC 100 mg) (n = 37)	Cases (CVC 200 mg) (n = 36)
Sex, n (%)
Male	13(81)	35 (95)	36 (100)
Female	3 (19)	2 (5)
Race, n %
White	11 (69)	23 (62)	28 (78)
Black	5 (31)	13 (35)	4 (11)
Other	0 (0)	1 (3)	4 (11)
Age first draw, years, mean (range)	33.2 (20–49)	34 (19–63)	37.8 (21–57)
Infection status
HCV-HIV+	16 (100)	37 (100)	36 (100)
CD4 Count, (range)	358 (177–605)	393 (207–774)	417 (190–1017)
HIV Viral Load, copies/mL, mean	53337	43450	66424
Baseline BMI	25.5	26.2	26.8

Abbreviations: BMI, body mass index; CVC, cenicriviroc; EFV, efavirenz; HCV, hepatitis C virus; HIV, human immunodeficiency virus.

Figure 3.

Longitudinal change in enhanced liver fibrosis (ELF) index over a 48-week interval among human immunodeficiency virus type 1–infected patients randomized to receive cenicriviroc vs. efavirenz (EFV) with a tenofovir-emtricitibine backbone. Abbreviations: CVC, cenicriviroc; ELF, enhanced liver fibrosis.

Change in Cytokines During Treatment With CVC vs. EFV

Multiple cytokines and chemokines were analyzed for change before and after 48 weeks of treatment. TGF-beta1 levels increased significantly in those exposed to CVC and were unchanged in patients taking EFV (P = .03). MIP-1beta (CCL4) was also significantly increased in patients with CVC versus EFV (P = .0001). MCP-1 (CCL2) levels were increased by a mean of 2505 pg/mL, whereas those receiving EFV were unchanged (P < .001).

Effects of CVC on the Transcriptome

Six matched pairs (CVC vs. EFV) were selected based on sample availability and quality of RNA extract. Significant changes (2-fold or greater) were observed in a large number of genes following 48 weeks of exposure to CVC when compared to baseline. However, changes were also observed in those treated with EFV. Therefore, we focused our comparison of the differential changes in up- and downregulation that were observed between EFV and CVC therapy, as shown in the heatmap in Figure 4. Table 3 shows the top 10 gene changes that were statistically significant between groups. Of note, integrin beta-3 (P = .0062) and the cadherin-binding inhibitory receptor (KLRG1) (P = .004) were significantly modulated. Integrin beta-3 was downregulated, and KLRG1 was upregulated. Both molecules are thought to have significant effects on immune functions that could modulate hepatic fibrosis as discussed below.

Figure 4.

Heatmap of gene expression changes following 48 weeks of combination antiretroviral therapy (cART) treatment in hepatitis C virus/human immunodeficiency virus coinfected patients. Gene expression changes are depicted as the ratio of gene expression at 48 weeks comparing efavirenz controls (left) to cenicriviroc treated (right) patients.

Table 3.

Top 10 Most Differentially Regulated Genes Comparing Human Immunodeficiency Virus Type 1–infected Patients Treated With Cenicriviroc Versus Efavirenz Following 48 Weeks of Antiretroviral Therapy

Symbol	Gene Name	logFC	AveExpr	T	P Value
FAM153C	family with sequence similarity 153, member C, pseudogene	1.13	1.72	4.29	.000188
GNG11	G protein subunit gamma 11	−1.43	3.1	−4.27	.000198
BMP6	bone morphogenetic protein 6	−1.07	2.55	−4.03	.000384
C11orf54	chromosome 11 open reading frame 54	−1.12	2.58	−3.86	.000605
KLRG1	killer cell lectin like receptor G1	1.04	4.59	3.11	.00428
ITGB3	integrin subunit beta 3	−1.33	3.9	−2.95	.00628
PRKAR2B	protein kinase cAMP-dependent type II regulatory subunit beta	−1.14	3.29	−2.69	.0118
MSR1	macrophage scavenger receptor 1	−1.47	1.72	−2.65	.0131
MAMLD1	mastermind like domain containing 1	−1.19	2	−2.56	.016
PPBP	pro-platelet basic protein	−1.01	6.27	−2.19	.0372

DISCUSSION

The role of CCR5 and CCR2 in hepatic fibrosis is unclear. Both chemokine receptors play a central role in pathways involving cell trafficking and immune responses. They also may modulate pathways associated with stellate cell activation, collagen deposition, and perhaps collagen reabsorption [28]. In patients with HIV, the etiology of hepatic fibrosis is multifaceted and includes chronic viral infections that may lead directly to hepatic injury, hepatotoxicity associated with antiretroviral agents or other drugs, alcohol, and steatosis [29]. HIV causes leakiness in the gut barrier, resulting in increased levels of endotoxin reaching the liver through portal circulation. There is evidence that HIV itself may bind to hepatocytes via the CCR5 receptor and trigger downstream cytokine responses that are pro-inflammatory and pro-fibrotic in nature [30]. Indeed, work in our lab and others has shown that HIV suppression is directly associated with changes in intrahepatic inflammation and markers of injury [31–33]. This is supported by studies that show rates of hepatic fibrotic progression inversely correlate with HIV viral load [13, 34].

We hypothesized that CCR5 and/or CCR2 plays a central role in the development of hepatic fibrosis among those with HIV and sought information about hepatic fibrosis markers from 2 complementary cohorts. In MHCS-1, the majority of patients were coinfected with HCV and HIV. The cohort was established prior to development of potent antiretroviral therapies. Less than 10% of cohort was heterozygotic for CCR5 delta-32 mutations, and virtually none of the HIV-infected patients were homozygotic for this mutation. The overall prevalence of CCR5 delta-32 homozygosity is 1%, and we evaluated nearly 1000 enrolled subjects to identify the small subgroup with this mutation. We validated the efficacy of the ELF index in HIV-infected patients and used this marker as a tool because liver biopsies were not available in a cohort whose underlying medical problem was a bleeding diathesis. Indeed, our data support the hypothesis that CCR5 delta-32 mutation is associated with a decreased rate of hepatic fibrosis as measured by the ELF index. Subjects were well matched in terms of factors otherwise associated with variation in fibrotic progression. Although it is possible that the changes reflect alterations in extrahepatic markers that are associated with fibrosis elsewhere in the body, Schuppan and Kim note that ELF index is reflective of endpoints such as hepatic decompensation and death that are the outcomes of progressive hepatic fibrosis [28]. ELF has also been previously validated in other cohorts with HIV compared to other biomarker assays [35].

The second study cohort included HIV-infected patients enrolled in a phase 2 treatment trial. The efficacy of HIV suppression was similar between the arm treated with CVC, a dual-active CCR5/CCR2 inhibitor, and EFV, a nonnucleoside antiretroviral agent with a completely different mechanism of action. However, significant differences were observed in the ELF index, with a meaningful decrease in putative hepatic fibrosis in those receiving CVC vs. no change (numerical increase) in those treated with EFV. Of note is that enrolled subjects did not have HCV, HBV, or other evidence of active liver injury; however, this does not mean enrolled subjects did not have liver disease. Clinical trial entry criteria permitted patients with ALT levels up to 2.6 times the upper limit of normal, a range that includes many patients with silent, but progressive liver disease. Indeed, Sterling et al. biopsied patients similar to those enrolled in this treatment trial and reported that 14% were cirrhotic, and all had an Ishak fibrosis score of 1 or more. Features of NAFLD were common [36]. Up to one third of subjects in the parent study arms had ALT levels above a level typically associated with histologic presence of inflammatory liver disease. Our finding suggests that the ELF index reduction—which is highly specific for liver fibrosis—is directly attributable to CCR5 or CCR2 blockade and not just to HIV suppression because the EFV arm did not exhibit comparable reductions in the ELF index. This finding is supported by an independent analysis of the same cohort by Thompson and colleagues, which reported that CVC-treated subjects demonstrated a significant decrease in both FIB-4 and APRI after 48 weeks of therapy, and that similar declines were not seen in the EFV arm [37]. To further investigate this phenomenon, we evaluated serum cytokine display focused on proinflammatory cytokines. Interestingly, we identified increases in TGF-β, as well as MCP1 and MIP-1. The increases in MCP1 (CCL2) and MIP-1 (CCL4) likely represent accumulation of these chemokines in serum following blockade of their binding sites (CCR2 and CCR5, respectively). TGF-β plays a major fibrogenic role through both direct and indirect mechanisms [38–41]. Several animal studies of CCR2 blockade or knockout suggest that inhibition of monocyte/macrophage chemotaxis is an important mechanism of injury prevention in studies of acetaminophen-induced injury, NASH-associated injury, and a sclerosing cholangitis model [42–44]. However, only CCR5 mutation/inhibition was common to both cohorts. It is possible that the dual activity exhibited by CVC enhances the antifibrotic effect. However, there are both laboratory and clinical data that support a relatively central role for CCR5. Stellate cells treated with maraviroc, which only blocks CCR5, led to a decrease in collagen and a down-regulation of metalloproteases (TIMP-1,2) [45]. Gonzalez and colleagues reported that use of maraviroc over a 2-year period in those with HCV/HIV coinfection improved hepatic fibrosis markers without HCV treatment [46]. Another study suggests that use of maraviroc blunted development of hepatic fibrosis markers in those with HCV/HIV coinfection [47]. Taken together, these studies suggest a role for CCR5 that does not negate additional potential benefit from CCR2 inhibition.

Our study has several limitations. The lack of liver biopsies in both cohorts prevents direct assessment of hepatic fibrosis and changes in fibrosis. It is possible and even likely that NAFLD/NASH and ASH play a role in our observations, but we are unable to directly assess it for the same reason. The time-period is short for both studies, although others have described the effect of fibrosis modulatory therapies in 2- or 3-year periods; our MHCS cohort included a 4-year observation window. We do not have samples after completion of the CVC trial, so we are unable to assess for rebound in biomarkers. The transcriptome study utilized PBMCs and not hepatic tissue. It is possible that the effect of CVC on fibrosis biomarkers is not directly mediated through blockade of CCR5 or CCR2. Transcriptome analysis suggests that there is downregulation of integrin beta-3 following exposure to CVC as compared with EFV. This integrin has been implicated in TIMP-1 binding in development of cardiac fibrosis [48] and antibodies directed against HIV gp120 that binds to CCR5 also bind to integrin beta 3 (CD61) [49], suggesting a linkage between CCR5 blockade and integrin-mediated fibrosis.

Other lines of evidence suggest a role of CCR2 and/or CCR5 blockade in modulation of hepatic fibrosis. A recent phase 2 trial of CVC reported preliminary results of a trial in patients with NASH and F1-3 hepatic fibrosis. Paired biopsies were evaluated before and after 1 year of treatment. The key outcome measure endpoint was not achieved. However, a secondary endpoint, improvement in hepatic fibrosis stage, was described [50]. This study led to the development of a larger phase 3 trial in patients with NASH with an endpoint of improvement in hepatic fibrosis rather than NASH itself. This trial is now in progress. Animal studies with CVC also support the concept that hepatic fibrosis is reduced following NASH or other injury [51, 52].

In conclusion, our data provide evidence from 2 different cohorts that CCR5 and CCR5/CCR2 play a significant role in fibrosis progression, as determined by a sensitive but indirect measure, the ELF index. Additional investigation on the role of both CCR5 and CCR2 blockade in those with HIV infection is warranted.

Notes

Author contributions. K. E. S.: study concept and design; obtained funding; analysis and interpretation of data; statistical analysis; drafting of the manuscript; critical revision of the manuscript; study supervision; E. A.: acquisition of data; analysis and interpretation of data; critical revision of the manuscript; S. D. R.: acquisition of data; analysis and interpretation of data; critical revision of the manuscript; technical and administrative support; M. T. S., J. T. B.: study concept and design; analysis and interpretation of data; critical revision of manuscript; P. S.: acquisition of data; analysis and interpretation of data; B. K., L. P.: acquisition of study samples; acquisition of data; analysis and interpretation of data; critical revision of the manuscript; P. S. H.: analysis and interpretation of data; S. K.: study concept and design; acquisition of study samples; analysis and interpretation of data; statistical analysis; drafting of the manuscript; critical revision of the manuscript.

Acknowledgments. The authors would like to thank Dr. Scott L. Friedman, Icahn School of Medicine at Mount Sinai, New York, for his thoughtful review and comments related to the manuscript, and Craig Johnson, PSM, University of Michigan DNA Sequencing Core, Ann Arbor, Michigan, for bioinformatics support of the gene microarray data. Tobira (Allergan) and the National Heart, Lung, and Blood Institute BioLINCC repository generously provided patient samples under Material Transfer Agreements with the University of Cincinnati.

Disclaimer. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Financial support. This work was supported by the National Institute of Allergy and Infectious Diseases of the National Institutes of Health under Award Number R01 AI065256-06-A1.

Potential conflicts of interest. K. E. S. has received grants and contracts paid to the University of Cincinnati College of Medicine from the following sources: AbbVie, BMS, Gilead, Innovio, Intercept, MedImmune, and Merck. He has served on advisory boards for Abbott Labs, Gilead, Shionogi, Merck and MedImmune and on Data Safety Monitoring Boards for MedPace and Watermark. He received support from Tobira (now Allergan). S. K. reports grants from Gilead Sciences, grants from Merck, and funding from American Gene Technology, during the conduct of the study. The other authors have no conflicts to report. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.