Distinct microRNA profiles are associated with the severity of hepatitis C virus recurrence and acute cellular rejection after liver transplantation

Authors


  • See Editorial on Page 355

  • This project was supported by the National Health Service Charity of King's College Hospital.

Address reprint requests to Varuna Aluvihare, Ph.D., M.R.C.P, Institute of Liver Studies, King's College Hospital, London, United Kingdom SE5 9RS. Telephone: +44 20 3299 1766; FAX: +44 20 3299 3167; E-mail: varuna.aluvihare@kcl.ac.uk

Abstract

Recurrent hepatitis C virus (HCV) infection is associated with accelerated fibrosis rates after liver transplantation (LT) and is the leading cause of graft failure. Furthermore, distinguishing recurrent HCV from acute cellular rejection (ACR) can be problematic, and this can lead to inappropriate treatments and adverse outcomes. We hypothesized that intragraft microRNA (miRNA) expression profiles could distinguish the severity of recurrent HCV and differentiate recurrent HCV from ACR. We established meticulously matched post-LT patient cohorts in order to derive robust global miRNA expression profiles and minimize the impact of variables known to influence HCV recurrence. These cohorts consisted of patients with slow HCV fibrosis progression (Ishak stage < F2), fast HCV fibrosis progression (Ishak stage ≥ F2), ACR, and nonviral etiologies. We found increased intragraft expression of miRNA-146a, miRNA-19a, miRNA-20a, and miRNA-let7e in slow progressors versus fast progressors, and we validated these findings with quantitative PCR. This miRNA network regulates the expression of cardinal genes implicated in promoting antifibrogenic, antiangiogenic, and anti-inflammatory pathways. miRNA-19a and miRNA-20a were also specifically detected in the serum of slow progressors. Furthermore, intragraft miRNA expression distinguished fast HCV progression from ACR. Here, changes in the expression of key miRNAs regulating fibrogenic and angiogenic pathways were associated with fast HCV progression. We demonstrate specific miRNA expression signatures that discriminate the rates of fibrosis progression in patients with recurrent HCV, and we distinguish recurrent HCV from ACR after LT. A pathway analysis indicates that specific miRNAs may play a regulatory role in these processes. Selected miRNAs may serve as intragraft and serum biomarkers for recurrent HCV after LT and help to distinguish between ACR and recurrent HCV. Liver Transpl 19:383–394, 2013. © 2013 AASLD.

Abbreviations
ACR

acute cellular rejection

CCL8

chemokine ligand 8

CD40L

CD40 ligand

C-Myb

C-myeloblastosis proto-oncogene protein

EGFR

epidermal growth factor receptor

H&E

hematoxylin and eosin

HCA

hierarchical cluster analysis

HCV

hepatitis C virus

HSC

hepatic stellate cell

IGF1

insulin-like growth factor 1

IL

interleukin

IRS2

insulin receptor substrate 2

Let-7e

miRNA-let7e

LT

liver transplantation

MELD

Model for End-Stage Liver Disease

miR

microRNA

miRNA

microRNA

N/A

not available

NI

necroinflammatory

N-Myc

N-myelocytomatosis viral related oncogene

PCA

principal component analysis

SMAD4

Mothers against decapentaplegic homolog 4

STAT1

signal transducer and activator of transcription 1

TGFβ

transforming growth factor β

TGFβR2

transforming growth factor β receptor 2

TLR4

toll-like receptor 4

VEGF

vascular endothelial growth factor.

Infection with chronic hepatitis C virus (HCV) remains the leading indication for liver transplantation (LT).[1] The recurrence of HCV is universal after transplantation, but fibrosis is accelerated in transplant recipients versus nontransplant patients: 20% of patients will develop bridging fibrosis, and 20% will develop cirrhosis within 5 years of transplantation.[2] This accelerated fibrosis rate negatively affects posttransplant survival, with some studies indicating a survival rate of less than 10% at 3 years.[3, 4]

Fibrosis progression associated with HCV recurrence after transplantation is influenced by a combination of donor, viral, transplant, and recipient factors.[5] Although certain donor and transplant factors such as donor age and cold ischemia time may be optimized, recipient factors such as sex and interleukin-28B (IL-28B) genotype are clearly less modifiable.[6, 7] The molecular mechanisms implicated in HCV recurrence and consequent hepatic fibrosis include angiogenesis, inflammation, apoptosis, and the activation of profibrotic cytokines and chemokines.[8] The complex interaction of recipient adaptive and innate immune responses in conjunction with the interplay of donor, viral, and recipient factors at the time of transplantation makes the prediction of HCV recurrence after LT difficult. Furthermore, in a significant cohort of patients, it is not easy to distinguish posttransplant HCV recurrence from acute cellular rejection (ACR).[9] This has significant clinical implications because the treatment of ACR adversely affects HCV recurrence.

Patients who demonstrate a fast fibrosis progression rate are defined as those with a modified Ishak fibrosis stage F ≥ 2, and they are best identified with liver biopsy 12 months after transplantation.[10, 11] At our institution, protocol liver biopsy is performed 12 months after transplantation to stage the degree of fibrosis related to HCV recurrence. A fibrosis stage F ≥ 2 12 months after transplantation not only stratifies a patient to a fast progressor phenotype but also is an indicator for the consideration of antiviral therapy.

MicroRNAs (miRNAs) are small (approximately 22-nucleotide-long) noncoding RNAs that regulate gene expression at the posttranscriptional level by specifically binding to target messenger RNA and thus leading to its degradation or translational repression.[12] By simultaneously regulating the expression of large numbers of genes, miRNAs have been implicated in controlling diverse biological processes, including apoptosis, proliferation, inflammation, and differentiation. They have highly conserved cross-species sequence homology.[13] miRNA-122 is specifically and abundantly expressed in hepatocytes and accounts for approximately 70% of the total miRNA within the liver.[14] miRNA-122 has been shown to facilitate the replication of HCV RNA by interacting with the 5′ noncoding region of the viral genome. Serum levels have been shown to correlate with serum alanine aminotransferase activity and necroinflammatory (NI) activity in patients with HCV.[14, 15]

We hypothesized that miRNA expression profiles from liver grafts could distinguish the severity of HCV recurrence and differentiate this from ACR. We, therefore, carried out miRNA expression profiles for liver tissue obtained from well-defined cohorts of patients with HCV recurrence after LT. We demonstrate that the severity of HCV recurrence is associated with distinct miRNA profiles and that these profiles differ from those associated with ACR. Specific miRNAs regulate biological processes implicated in mediating HCV recurrence, and this indicates that they may directly influence HCV recurrence and also promote the deleterious pathological outcomes of recurrence. Our findings also indicate that miRNA-based biomarkers for delineating HCV recurrence and ACR are feasible.

PATIENTS AND METHODS

Patients

All adult patients (>18 years) who underwent transplantation for HCV-related cirrhosis and received a graft from a heart-beating donor between January 2000 and January 2010 at King's College Hospital (London, United Kingdom) were identified (n = 266). Protocol liver biopsy was performed at our institution at 12 months or earlier in the event of graft dysfunction. Specimens were assessed and scored according to the modified Ishak classification [from (0) no fibrosis to (6) cirrhosis]. A fibrosis score ≥ 2 (F ≥ 2) at 12 months classified the patient as a fast progressor with respect to HCV recurrence.[10]

Patients with recurrent HCV after transplantation were divided into 2 groups according to their fibrosis score: (A) slow progressors (n = 11) with a score < F2 at 12 months and (B) fast progressors (n = 9) with a score ≥ F2 at 12 months. Patients in groups A and B were matched according to the following variables known to heavily influence the rate of HCV recurrence: recipient age, donor age, Model for End-Stage Liver Disease (MELD) score at transplant, donor risk index, cold ischemia time, presence of diabetes, evidence of cytomegalovirus viremia, and immunosuppression regimen. All patients received dual therapy with tacrolimus (target trough level = 8-10 mg/mL) and a reducing dose of a corticosteroid after transplantation. No patient received OKT3 or an IL-2 blocker. The data capture included the postoperative clinical course, immunosuppression, number of methylprednisolone boluses, occurrence of biopsy-proven ACR, and serum liver graft function tests. Group C (n = 5) consisted of biopsy samples taken from patients with HCV and ACR (according to the Banff criteria). Group D (n = 4) consisted of biopsy samples taken from patients who underwent transplantation for alcohol-related liver disease and served as the control group. Biopsies performed in groups A and B were protocol biopsies, whereas biopsies performed in groups C and D were driven by clinical necessity. Supporting Fig. 1 demonstrates patient selection for this study. The exclusion criteria included hepatocellular carcinoma at the time of transplantation, coinfection with hepatitis B virus or human immunodeficiency virus type 1 or 2, dual liver-kidney transplantation, fibrosing HCV recurrence, survival of less than 12 months, and liver disease of an origin other than HCV infection (except for patients in group D). This study was reviewed and approved by the ethics and research and development committee of King's College Hospital (London, United Kingdom) and was performed in accordance with the Declaration of Helsinki.

HCV Virology Testing and IL-28B rs12979860 Genotyping

The quantification of serum HCV RNA was performed by the Viral Hepatitis Service of the Institute of Liver Studies with a PCR assay (Cobas Amplicor HCV Monitor test, version 2.0, Roche Diagnostics, Branchburg, NJ). To assess virus negativity, a qualitative HCV RNA PCR assay with a detection limit of 15 IU/mL (Cobas Amplicor HCV test, Roche Diagnostics) was used. DNA from patients was genotyped for the IL-28B rs12979860 polymorphisms with TaqMan single-nucleotide polymorphism genotyping assays (Applied Biosystems, Inc., Foster City, CA).

RNA Isolation From Liver Tissue

RNA was isolated from formalin-fixed, paraffin-embedded blocks. Xylene (800 μL) was added to a 10-μm section of tissue, and RNA was isolated with the High Pure FFPE RNA micro kit (Roche Diagnostics) according to the manufacturer's protocol.

GeneChip miRNA 2.0 Arrays

Total RNA samples were enriched with YM-100 columns (Millipore). Total RNA (500 ng) was diluted to 100 μL with 10 mM trishydroxymethylaminomethane (pH 8), heated at 80°C for 3 minutes, and then immediately cooled on ice for 3 minutes. Microcon columns were calibrated with 50 μL of 10 mM trishydroxymethylaminomethane (pH 8) and centrifuged for 3 minutes. Diluted RNA was added to each column and centrifuged for 7 minutes at 13,000g. RNA was labeled with the FlashTag Biotin RNA labeling kit (Genisphere) according to the manufacturer's protocol. GeneChip miRNA 2.0 arrays (Affymetrix) were processed with the GeneChip hybridization, wash, and stain kit according to the manufacturer's recommendations with fluidics script FS450_0003. Scanning was performed with Affymetrix Command Console software according to the company's guidelines. The microarray data is MIAME (minimum information about a microarray experiment) compliant and has been deposited in the GEO repository (Accession no. GSE40113).

Analysis of Scanned miRNA Arrays

The array images (CEL files) were processed with the Affymetrix miRNA QC tool with the default workflow, which included background correction summarization of multiple probes into overall probe set intensity and normalization based on quantile normalization. This procedure generated log 2 intensities for the 15,644 probe sets on the chips for each sample. Statistical analyses (analyses of variance) were carried out between different groups with Qlucore Omics Explorer version 2.1.

Qlucore Omics Explorer and MetaCore Analysis

The data table exported from the miRNA QC tool was formatted as a gedata tab-delimited text file and was imported into Qlucore Omics Explorer version 2.1 for analysis. A simple 1-way analysis of variance was employed to filter genes that were differentially regulated between different time points. P values were set as stated in the figure legends. Gene lists containing all regulated genes were displayed as heat maps to show gene expression patterns. Gene lists were also submitted to MetaCore analysis (GeneGo, Inc.).

Validation of miRNA Arrays

Selected miRNAs (miRNA-19a, miRNA-20a, miRNA-146a, miRNA-150, and let-7e) were analyzed by real-time quantitative PCR using both liver tissue and serum with the miRCURY LNA universal real-time miRNA PCR system (Exiqon) according to the manufacturer's protocol (see the Supporting Methods). All experiments were performed in triplicate. For the statistical analysis, data were expressed as means and standard deviations. The Student t test was used to determine statistical differences between groups A and B. A P value < 0.05 was considered significant.

Statistical Analysis

Continuous variables were expressed as medians and ranges (minimum to maximum). The chi-square test, the Mann-Whitney U test, and Fisher's exact test were used to establish differences between groups A and B and groups B and C. A 2-tailed P value < 0.05 was considered to indicate statistical significance. The analysis was performed with SPSS for Windows 17 (SPSS, Chicago, IL).

RESULTS

Total Patient Cohort

After rigorous patient selection, 29 patients were included in the analysis (Supporting Fig. 1). The patient demographics and clinical data are listed in Table 1. As expected, genotype 1 disease was the most prevalent and represented the HCV population seen at our center. The recipient IL-28B rs12979860 TT haplotype was equally represented across groups A, B, and C. The median HCV viral loads were also comparable. Representative examples of the histology samples used in this study to generate the miRNA signatures are shown in Fig. 1. Reticulin staining demonstrated mild periportal fibrosis for group A (Fig. 1A), with the corresponding hematoxylin and eosin (H&E) preparation showing mild hepatitis consistent with a recurrent HCV infection (Fig. 1C). In comparison, reticulin staining for group B demonstrated advanced bridging fibrosis and changes suggestive of parenchymal nodular transformation (Fig. 1B), with corresponding H&E staining suggestive of moderate portoseptal hepatitis and a recurrent HCV infection (Fig. 1D). H&E staining for group C showed a mixed portal inflammatory cell infiltrate with bile duct damage and confirmed ACR (Fig. 1E). In contrast, group D (the control group) showed essentially normal post-LT histology (Fig. 1F).

Table 1. Patient Demographics for Slow Fibrosis Progressors (Group A), Fast Fibrosis Progressors (Group B), HCV and ACR Patients (Group C), and Controls (Group D)
 Patient Groupsa
A (n = 11)B (n = 9)C (n = 5)D (n = 4)
  1. a

    There was no statistical difference between groups A and B.

  2. b

    The data are expressed as medians and ranges.

  3. c

    The data are expressed in days.

  4. d

    Fibrosis score 0-6.

Recipient age at transplant (years)b52 (36-62)46 (35-57)45 (26-56)50 (26-62)
Donor age (years)b45 (44-58)46 (29-73)55 (29-62)59 (55-69)
MELD score at transplant b15 (11-19)14 (11-18)14 (11-21)9 (8-14)
HCV genotype 1a (%)555640N/A
IL-28B rs12979860 TT genotype (%)182220N/A
Pretransplant HCV viral loadb4.77E5 (2.145-9.98E5)1.84E5 (8.58E4-5.11E5)2.1E5 (8.558E4-9.98E5)N/A
Methylprednisone boluses (n)4353
Cold ischemia time (hours)b9.0 (7.6-15.3)10.8 (7.8-18.9)8.8 (7.9-19.2)14.5 (12.8-15.6)
Biopsy-proven ACR (%)181110025
Donor risk indexb1.6 (1.2-2.3)1.7 (1.5-2.1)2 (1.8-2.4)1.8 (1.4-2.1)
Diabetes (%)45444050
Cytomegalovirus viremia (%)1811025
Variables at biopsy    
Time from transplant (months)b12 (8-15)15 (8-17)8 (5-90)c20 (10-56)
Tacrolimus monotherapy (%)6477100100
Aspartate aminotransferase (IU/mL)b55 (17-362)62 (21-263)227 (105-325)77 (38-257)
Bilirubin (mg/dL)b12 (8-23)17 (7-44)73 (48-171)16 (9-32)
Fibrosis score (0-6)bd1 (0-1)3 (3-4)0 (0-1)N/A
HCV viral load (IU/mL)b1.25E6 (1.26E4-9.89E6)5.11E5 (2.6E3-1.21E7)2.31E5 (4.41E4-3.45E6)N/A
Figure 1.

Histological analysis of slow fibrosis progressors (group A), fast fibrosis progressors (group B), ACR patients (group C), and controls (group D). (A,B) Representative reticulin staining for (A) group A and (B) group B. (C-F) Representative H&E staining for (C) group A, (D) group B, (E) group C, and (F) group D. The Ishak fibrosis and NI scores were (A) F1 and NI3 and (B) F3 and NI5.

miRNA Expression Distinguishes Between Slow and Fast HCV Fibrosis Progressors After Transplantation as Well as ACR

Principal component analysis (PCA) resolves a multidimensional data set by identifying key variables that explain the observed differences. Intragraft miRNA in all 29 patients included in our study was resolved into 3 principal components represented in a 3-dimensional scatter plot. This demonstrated a clear segregation of all 4 groups according to miRNA expression profiles at a P value of 0.01 (Fig. 2B). The supervised hierarchical cluster analysis (HCA) for miRNA expression demonstrated that a greater similarity of expression existed in each group (Fig. 2A). The graphical tree (cladogram) above the heat map also confirms this observation because it reveals stronger clustering within each group than between different groups. It also demonstrates a greater degree of similarity between groups A and B than the other 2 groups by miRNA expression. The unequivocal separation of patient groups by both PCA and HCA clearly demonstrates that these groups are functionally distinct with respect to miRNA expression. The list of miRNAs identified in this analysis and their P values are shown in Supporting Table 1.

Figure 2.

miRNA expression distinguishes slow fibrosis progressors (group A), fast fibrosis progressors (group B), ACR patients (group C), and controls (group D). (A) HCA for groups A to D (P < 0.01). The cladogram above the heat map demonstrates the degree of similarity between the samples in terms of miRNA expression. (B) PCA for the same data set. An miRNA list and associated P values are shown in Supporting Table 1. The percentage indicates how much of the variation in the data set that is explained by the corresponding PCA component.

miRNA Expression Distinguishes Between Slow and Fast HCV Progressors After Transplantation

In order to determine whether there was a difference in the miRNA expression profiles of patients with slow fibrosis and patients with fast fibrosis, we refined our analysis by directly comparing groups A and B. Patients in both groups were rigorously matched for factors known to influence the rate of HCV recurrence. Therefore, as expected, there were no demonstrable statistical differences between the 2 groups (see Table 1). At the time of liver biopsy, the patients were clinically well with satisfactory liver graft function. The median HCV viral loads were comparable both before and after LT. No patient had been exposed to antiviral therapy in the posttransplant period at the time of liver biopsy. PCA and HCA performed to a P value of 0.01 demonstrated a clear segregation of the 2 groups with respect to miRNA expression (Supporting Fig. 2). One hundred fifty-two miRNAs of statistical significance that were differentially regulated between groups A and B were identified.

miRNA Expression Is Associated With Slow HCV Progression After LT

Seventy-six of the 152 miRNAs differentially expressed between groups A and B were down-regulated, and 76 were up-regulated in group A versus group B. The data sets of both down-regulated and up-regulated miRNAs were individually analyzed with MetaCore to examine the genetic pathways known to be regulated by these miRNAs. This algorithm recognizes coregulated components of pathways or biological processes that are affected by increased or decreased expression of specific miRNAs. Up-regulation of a given miRNA results in increased inhibition (down-regulation) of its target genes, whereas reduced miRNA expression reverses this inhibition and leads to target gene up-regulation.

A MetaCore analysis of up-regulated and down-regulated miRNAs in slow progressors versus fast progressors identified 7 networks (6 up-regulated networks and 1 down-regulated network) of high statistical significance (Fig. 3A). The identification of these genetic pathways is based on known target gene regulation by miRNAs that are differentially expressed between slow and fast progressors. These networks regulate many pathways that have been implicated in mediating HCV progression, and they are outlined next.

Figure 3.

miRNA expression distinguishes slow HCV progressors (group A) from fast HCV progressors (group B) after transplantation. (A) A MetaCore analysis of group A versus group B (P < 0.01) identified statistically significant networks of miRNAs and known target genes implicated in (i) fibrogenic, (ii) angiogenic, (iii) inflammatory, (iv) apoptotic, and (v) mixed pathways. HCA and PCA for the same data set are shown in Supporting Fig. 1; an miRNA list and associated P values are shown in Supporting Table 2. (B) A quantitative PCR analysis of a select group of miRNAs that were identified by a MetaCore analysis confirmed the up-regulation of (i) miRNA-19a, (ii) miRNA-20a, (iii) let-7e, and (iv) miRNA-146a expression in group A versus group B. Small nucleolar RNA SNORD66 was used as a reference RNA for normalization. Values are expressed as means and standard deviations and are representative of 3 different experiments. P values for each miRNA are indicated on the graphs.

Antifibrotic Pathways (Fig. 3Ai)

Three statistically significant networks of miRNAs associated with antifibrotic pathways were identified: miRNA-200a and miRNA-141 (P = 2.39 × 10−18), miRNA-203 and miRNA-146a (P = 2.33 × 10−5), and miRNA-146a and miRNA-19a (P = 2.67 × 10−9). Increased expression of both miRNA-203 and miRNA-146a is known to lead to the down-regulation of SMAD4. This acts as a tumor suppressor that functions in the regulation of the transforming growth factor β (TGFβ) signal transduction pathway, which is an important profibrotic pathway. Increased miRNA200a and miRNA-141 expression is known to mediate direct down-regulation of transforming growth factor β receptor 2 (TGFβR2). Increased expression of miRNA-146a and miRNA-19a leads to the down-regulation of epidermal growth factor receptor (EGFR).

Antiangiogenic Pathways (Fig. 3Aii)

One miRNA network that is known to inhibit angiogenesis was identified: increased expression of miRNA-20a, miRNA-20b, and miRNA-205 (P = 2.02 × 10−11) leads to the down-regulation of vascular endothelium factor A (VEGFA) activity.

Anti-Inflammatory Pathways (Fig. 3Aiii)

One network was associated with regulating the expression of genes promoting anti-inflammatory pathways: miRNA-146a, miRNA-33a, and let-7e (P = 1.96 × 10−15). Increased miRNA-146a expression is known to inhibit the expression of proinflammatory mediators, including IL-8, IL-6, chemokine ligand 8 (CCL8), and CD40 ligand (CD40L). In addition, increased miRNA-33a expression leads to decreased insulin receptor substrate 2 (IRS2) activity, and increased let-7e expression leads to decreased toll-like receptor 4 (TLR4) expression.

Antiapoptotic Pathways (Fig. 3Aiv)

A single statistically significant network of down-regulated miRNAs that regulates antiapoptotic pathways was identified: miRNA 19a and miRNA 150 (P = 1.84 × 10−18). Decreased miRNA-150 expression causes up-regulation of c-Myb, and decreased miRNA-19a expression leads to the up-regulation of N-Myc expression.

Other Pathways (Fig. 3Av)

An additional network of differentially expressed miRNAs that we identified (miRNA-205, miRNA-204, miRNA-20a, miRNA-146a, and miRNA-328; P = 1.25 × 10−13) inhibits the expression of genes known to promote HCV disease progression. Increased miRNA-146a leads to decreased signal transducer and activator of transcription 1 (STAT1) expression, and increased miRNA-20a leads to decreased expression of TGFβR2.

Validation and Quantification of miRNA Expression by PCR

In order to independently validate the changes in miRNA expression that we observed in groups A and B and to quantify miRNA expression, we selected a subgroup of miRNAs from the GeneGo analysis outlined previously (Fig. 3A). A quantitative PCR analysis confirmed the up-regulation of miRNA-19a, miRNA-20a, let-7e, and miRNA-146a in group A versus group B (Fig. 3B). In addition, we validated miRNA expression changes detected in the microarray analysis by carrying out quantitative PCR for 30 randomly selected miRNAs that showed no differential expression and concordance with the array data (data not shown). In order to investigate whether any of the miRNAs that we identified could represent tractable, noninvasive, and prognostic biomarkers for HCV recurrence, a quantitative PCR analysis was also performed on serum samples taken at a median of 6 months after transplantation (range = 4-10 months) from fast and slow progressors. Our results demonstrated specific and statistically significant up-regulation of miRNA-19a and miRNA-20a (but not miRNA-150 or miRNA-146a) in group A versus group B (Fig. 4).

Figure 4.

Serum miRNA expression distinguishes slow HCV progressors (group A) from fast HCV progressors (group B) after transplantation. Serum samples were analyzed by quantitative PCR to determine the expression of (A) miRNA-19a, (B) miRNA-20a, and (C) miRNA-150 in groups A and B. SNORD66 was used as a reference RNA for normalization. Values are expressed as means and standard deviations and are representative of 3 different experiments. P values for each miRNA are indicated on the graphs. (+) represents RNA extracted from cells that are transfected with miRNA-146a.

miRNA Expression Distinguishes Posttransplant HCV-Related Fast Progression From ACR

Because of the known difficulties in distinguishing posttransplant HCV recurrence from ACR, we investigated miRNA expression in groups B and C. Groups B and C were chosen because of their proximity on the PCA plot (Fig. 2). We reasoned that differential miRNA expression between these groups might not only elucidate mechanistic insights but also identify potential biomarkers that could discriminate between these clinical entities.

HCA demonstrated distinct miRNA expression profiles for groups B and C (P < 0.01; Fig. 5A), with the cladogram above the heat map (Fig. 5A) as well as the associated PCA (Fig. 5B) confirming stronger similarities within each group than between groups. One hundred ninety miRNAs of statistical significance were differentially regulated between the 2 groups (an miRNA list and P values are shown in Supporting Table 3), with 100 miRNAs down-regulated and 90 up-regulated in group B versus group C.

Figure 5.

miRNA expression distinguishes fast HCV progressors (group B) after transplantation from patients with ACR (group C). (A) HCA and cladogram for groups B and C (P < 0.01). (B) PCA for the same data set. An miRNA list and associated P values are shown in Supporting Table 3. (C) A MetaCore analysis identified 2 statistically significant networks of down-regulated miRNAs and known target genes for individual miRNAs.

A MetaCore analysis revealed 2 highly statistically significant networks of down-regulated miRNAs in group B versus group C (Fig. 5C). The first network consists of miRNA-1336 and miRNA-223 (P = 2.64 × 10−14; Fig. 5Ci). The down-regulation of both miRNAs is associated with increased expression of insulin-like growth factor 1 (IGF1) receptor, which has been implicated in promoting liver fibrogenesis.[16] The second network consists of the down-regulation of miRNA-210 and miRNA-503 (P = 1.6 × 10−4; Fig. 5Cii) and is associated with proangiogenic pathways through the increased expression of VEGFA.

DISCUSSION

In this study, we demonstrated distinct intragraft miRNA gene expression between patients with slow fibrosis and patients with fast fibrosis after transplantation for HCV. Samples taken from groups A and B were carefully selected and rigorously matched for known factors that strongly influence HCV recurrence after transplantation (ie, donor age, cold ischemia time, donor risk index, IL-28B recipient genotype, and immunosuppression; see Table 1). Although this approach limited the sample size, it allowed us to identify changes in the expression of miRNAs of high statistical significance that are exclusively associated with HCV recurrence and fibrosis progression.[6, 17]

The clinical significance of identifying patients with a fast fibrosis phenotype cannot be overstated. The recurrence of HCV after transplantation is the leading cause of graft loss and the commonest indication for consideration of retransplantation.[3] Although the main aim of this study was to identify potential mechanisms related to HCV recurrence after transplantation, using quantitative PCR from serum, we were also able to demonstrate that specific miRNAs (miRNA-19a and miRNA-20a) could represent potential serum biomarkers for fibrosis progression. We also found that intragraft miRNA expression in patients with evidence of HCV recurrence after LT differs from that in patients with ACR and patients with normal liver biopsy findings after LT.

The miRNA networks that we have demonstrated to correlate with HCV disease progression identify molecular pathways of relevance to the pathogenesis of HCV recurrence. Although these pathways are identified through the regulation of known target gene expression, a MetaCore analysis has revealed that these pathways have high statistical significance and, therefore, relevance. According to miRNA expression, profibrotic pathways would be down-regulated in group A versus group B. Key fibrogenic mediators, including SMAD4 and TGFβR2, are known to be down-regulated in response to the up-regulation of miRNA-203 and miRNA-146a and miRNA-200a and miRNA-141, respectively. TGFβ is the main fibrogenic chemokine and is activated in response to liver injury, whereas SMAD proteins are intracellular mediators of signal transduction pathways of the TGFβ superfamily members.[18] Increased expression of miRNA-20a is known to lead to decreased expression of TGFβR2, which has been shown to attenuate the TGFβ signaling pathway and hepatic stellate cell (HSC) activation.[19] Another important pathway in the pathogenesis of hepatic fibrosis is the deposition of extracellular matrix, which originates from myofibroblastic cells derived from HSC and portal fibroblasts.[20] The activation of EGFR on extracellular matrix–producing cells has been shown to contribute to the profibrogenic phenotypic state.[21, 22] EGFR gene expression is inhibited by increased expression of miRNA-146a and miRNA-19a, as exhibited by the slow progressor group.

Angiogenesis and inflammation are also known to contribute to hepatic fibrosis. New blood vessel formation, sinusoidal modeling, and stellate cell expansion are integral and are mediated by mediators such as VEGF and platelet-derived growth factor.[23] Up-regulation of the network consisting of miRNA-146a, miRNA-205, and miRNA-20a in the slow progressor phenotype would be expected to inhibit VEGFA expression. Furthermore, increased expression of miRNA-146a is known to negatively regulate the expression of IL-8: high levels of IL-8 [chemokine (C-X-C motif) ligand 8] are associated with disease progression and a poor response to interferon-based treatment.[24] The expression of proinflammatory mediators IL-6, CCL8, and CD40L is also known to be inhibited by these changes in miRNA expression. TLR4 is the receptor for bacterial lipopolysaccharide on Kupffer cells and is also expressed on HSCs. It is essential to the inflammatory response and plays an important role in the fibrogenic response. Specific single-nucleotide polymorphisms of TLR4 have been shown to contribute to the rate of fibrosis in HCV infection.[25] Our data also showed increased expression of let-7e, a significant inhibitor of TLR4 expression.

c-Myb, the prototype oncogene, plays an essential role in the regulation of cell development and differentiation.[26] An elevation of miRNA-150 is known to result in decreased expression of c-Myb and thus cell apoptosis.[27, 28] Our finding of decreased miRNA-150 expression indicated increased c-Myb expression and reduced apoptosis and, therefore, a reduced rate of fibrosis in group A.

Our results identified an important role for miRNA-146a in the pathogenesis of HCV recurrence after LT. miRNA-146a is up-regulated in response to TLR4 stimulation in monocytes and controls endotoxin tolerance so that the innate immune response favors cell survival.[29] Murine models also suggest that miRNA-146a is critical for regulatory T cell function and acts as a molecular brake for inflammation, myeloid cell proliferation, and oncogenic transformation.[30, 31] Although intragraft miRNA-146a expression was increased in the slow fibrosis progression group, there was no comparable increase in miRNA-146a levels in serum. This phenomenon of increased expression of miRNAs in tissue but not in serum or plasma has been well described and arises because only specific miRNAs are released by the secretion of microsomes, including exosomes, and therefore only specific miRNAs are selectively packaged into microvesicles.[32, 33] We, therefore, hypothesize that miRNA-146a plays an integral role in determining the rate of HCV recurrence after transplantation but is not a suitable serum biomarker candidate.

The treatment of HCV after transplantation can be instigated in a preemptive manner under the assumption that HCV recurrence is universal or after histological evidence of recurrence is obtained (F ≥ 2). Results from 2 randomized trials that adopted a prophylactic treatment regimen were disappointing; therefore, histological evidence of recurrence has been adopted by most transplant institutions as an indicator for consideration of antiviral therapy. Currently, reported sustained virological response rates in posttransplant patients range from 30% to 35%, and these rates are considerably poorer than those for patients before transplantation.[34] Our results have identified a panel of miRNAs that not only determine whether a patient has a fast or slow fibrosis phenotype but also could potentially identify patients who require treatment for their HCV earlier. The assessment of these miRNAs once steady liver graft function is achieved could lead to early consideration of antiviral therapy.

Histology remains the gold standard for differentiating ACR from recurrent HCV infection, although it remains nonspecific and subject to interobserver and intraobserver bias.[9] HCV recurrence typically manifests as a lobular hepatitis with hepatocyte changes, including ballooning, acidophilic body formation, and steatosis. ACR is typically characterized by predominant portal changes consisting of a mixed portal inflammatory cell infiltrate, bile duct damage, and portal vein endotheliitis.[35] The 2 conditions may coexist or even trigger each other, and their histological manifestations may also overlap. It is a common clinical scenario that, if diagnosed incorrectly, can have potentially deleterious consequences, such as the inappropriate administration of corticosteroids for presumed ACR and the resultant potentiation of HCV viremia. A recent study demonstrated an increase in miRNA-122 and miRNA-148a in parallel with serum aspartate aminotransferase and alanine aminotransferase levels in patients with ACR.[36] Comparing groups B and C, we were able to identify profibrogenic miRNAs through the regulation of IGF1 receptor– and VEGFA-related pathways that could help to differentiate between recurrent HCV and ACR. The relatively small numbers in group C, however, make for statistically valid but cautious conclusions and require validation in larger prospective cohorts.

From a review of the current literature, it is clear that miRNAs are viewed as attractive, potential biomarkers because of their stable, cell-free forms in blood. Critically, what remains unclear is how miRNAs make their way into the bloodstream: is this the result of cell death, or is it secondary to active secretion from tissue cells? There is increasing evidence that serum/plasma concentrations of miRNAs are altered in multiple disease pathologies, including hepatocellular carcinoma.[37-39] Studies specific to HCV have demonstrated that miRNA-122 expression correlates with serum transaminases and NI activity on liver biopsy but not with the fibrosis stage or parameters of liver function in patients with chronic HCV.[15] miRNA-122 levels have also been shown to be decreased in patients who do not respond to antiviral therapy versus responders in a nontransplant setting.[40] Our analysis did not find miRNA-122 to be expressed differentially between groups A and B, and this is not surprising because these groups were selected exclusively by differential fibrosis rates. Quantitative PCR performed with serum found miRNA-19a and miRNA-20a to be increased in group A versus group B. Because of their potential mechanistic role in the pathogenesis of HCV recurrence, their detection in serum indicates that they may represent candidate biomarkers that can delineate fibrosis rates associated with HCV recurrence after transplantation.

In addition to providing diagnostic information, miRNAs provide the possibility of therapeutic intervention. Novel targeted anti-miRNA therapies, so-called antagomirs, have been postulated as future therapeutic targets. In 2 separate murine models, the introduction of an antisense oligonucleotide/antagomir resulted in a reduction of hepatic steatosis and tumorigenesis repression, respectively.[41] The development of antagomirs, however, remains in its infancy. To advance this field further, extensive miRNA profiling of both diseased and healthy tissue is required. Although we cannot definitively prove that the loss of miRNA-146a is a cause or consequence of HCV, a targeted antagomir or related therapeutic agent against HCV-related fibrosis and recurrence after LT would represent an exciting advance.

In conclusion, we have demonstrated that the fibrosis rate associated with HCV recurrence after LT is associated with changes in networks of intragraft miRNA expression that can regulate proinflammatory, proangiogenic, and profibrogenic pathways. Meticulous patient selection has enabled us to identify key miRNAs that provide a unique insight into potential mechanisms involved in the recurrence of HCV after transplantation. Our data suggest a pivotal role for miRNA-146a in the pathogenesis of HCV recurrence after transplantation. Using liver tissue and, more relevantly, serum, we have identified a panel of miRNAs associated with the severity of HCV recurrence that could be used as a potential biomarker. Larger translational studies are required to investigate the role of the miRNAs that we have identified in the pathogenesis of HCV recurrence and to validate tractable biomarkers that are predictive of aggressive recurrence after LT. Although the advent of the new direct-acting antiviral agents will undoubtedly transform the therapeutic landscape, the early identification of patients with rapid fibrosis rates and the ability to distinguish viral recurrence from ACR remain pivotal clinical endpoints for selecting suitable patients in the challenging post-LT setting.

Ancillary