By continuing to browse this site you agree to us using cookies as described in About Cookies
Notice: Wiley Online Library will be unavailable on Saturday 7th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 08.00 EDT / 13.00 BST / 17:30 IST / 20.00 SGT and Sunday 8th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 06.00 EDT / 11.00 BST / 15:30 IST / 18.00 SGT for essential maintenance. Apologies for the inconvenience.
This paper was presented, in part, at the Annual Meeting of the German Gastroenterological Association, Nürnberg, September 2003, and the Hepatitis C Therapy Session of the Annual Meeting of the American Association for the Study of Liver Diseases, Boston, October 2003, and was published in abstract form in Z Gastroenterol 2003;41:798–799 and HEPATOLOGY 2003;38:245A.
The response to antiviral therapy for chronic hepatitis C virus (HCV) is complex and is determined by both environmental and genetic factors. Recently, interacting gene polymorphisms of the chemokine RANTES have been shown to affect HIV disease progression. Our aim was to assess if these RANTES variants are associated with response to anti-HCV therapy. Three linked RANTES single nucleotide polymorphisms (403 G/A, Int1.1 T/C, and 3′ 222 T/C) were determined in 297 Caucasian patients who were treated for chronic HCV infection and 152 control subjects. Characteristic nucleotide combinations on single chromosomes (haplotypes) were reconstructed and tested for disease association. Four common RANTES haplotypes (prevalence 73%) were identified in patients and controls. There was a strong association of RANTES haplotype distribution with outcome of antiviral combination therapy (P = .007). Specifically, RANTES haplotypes carrying Int1.1 C and 3′ 222 C alleles were more frequent in nonresponders than in patients with a sustained response to antiviral therapy (odds ratio 1.9, P = .01). The influence of these RANTES haplotypes on the outcome of therapy was more pronounced in patients infected with HCV genotypes 1 and 4 (odds ratio 2.3, P = .02). Because RANTES haplotypes carrying Int1.1 C are known to down-regulate RANTES transcriptional activity in vitro, the haplotype analysis fits the hypothesis of a diminished T helper 1 lymphocyte response in patients with a negative response to antiviral therapy. In conclusion, RANTES haplotypes might contribute to the polygenic interaction between HCV and the host immune system and could help to risk stratify patients prior to antiviral therapy. (HEPATOLOGY 2004;40:327–334.)
The CC chemokine RANTES (regulated on activation normally T cell expressed and secreted; systematic name CCL5) attracts T lymphocytes to inflamed tissues and is produced by T lymphocytes, monocytes, fibroblasts, and endothelial cells.1, 2 In hepatitis C virus (HCV)-infected liver RANTES messenger RNA has been shown to be up-regulated, and its intrahepatic expression levels correlate with serum alanine aminotransferase activities.3, 4 Furthermore, expression of full-length HCV complementary DNA induces the expression of RANTES in hepatocyte cell lines,5 possibly mediated through activation of nuclear factor κB.6 Interestingly, a 32–base pair deletion in the gene of the CC chemokine receptor 5 (CCR5 Δ32), which results in decreased receptor expression,7 has recently been linked to a negative response to interferon monotherapy in hepatitis C.8 Because the CCR5 ligand RANTES preferentially attracts T helper 1 lymphocytes,9 RANTES might also be associated with treatment response to antiviral therapy, which is thought to depend on a sufficient proinflammatory cytokine response.10, 11
The human RANTES gene spans 8.5 kb on chromosome 17q11-q12 and has the characteristic three exons/two introns organization of the CC chemokine family.12 Upon sequencing of the entire RANTES gene, seven single nucleotide polymorphisms (SNPs) were identified using a DNA panel of Caucasian and African individuals.13 Four SNPs were located in the promoter region (-403 G/A, -109 C/T, -105 C/T, and -28 C/G), two were located in the first intron (Int1.1 T/C and Int1.2 G/A), and one was located in an Alu-related repeat region of the 3′ untranslated region (3′ 222 T/C).13 Of note, both the SNP at position -403 of the RANTES promoter and the SNP within the first intron (Int1.1 T/C) have been shown to be functionally relevant with respect to RANTES transcription in vitro.13, 14 Furthermore, these SNPs have been associated with susceptibility to HIV infection and AIDS progression13, 15, 16 and atopic dermatitis,14 further supporting their pathophysiological relevance.
SNPs within mammalian genes result from single mutational events that are associated with other variant alleles on the same chromosomal segments (i.e., linkage disequilibrium). A haplotype is defined as such a distinct set of gene variants on a single chromosome.17 In complex diseases haplotype analysis can provide more information than analysis of individual SNPs,18 and under certain circumstances a genotype–phenotype association might even be missed if SNPs are investigated instead of haplotypes.19 Therefore, we determined RANTES haplotypes and correlated these haplotypes to selected outcome parameters in a cohort of Caucasian patients with chronic hepatitis C and a carefully selected control population described previously.20, 21
This study provides evidence that RANTES haplotypes carrying the gene variants Int1.1 C and 3′ 222 C are associated with a negative response to antiviral combination therapy in chronic hepatitis C. We speculate that these haplotypes might be predictive parameters that could guide more effective antiviral therapies.
HCV, hepatitis C virus; SNP, single nucleotide polymorphism.
Patients and Methods
Overall, 297 patients with chronic hepatitis C infection were included in the study. All patients were of Caucasian origin and originated from two German university hospitals (205 patients from Berlin, 92 patients from Aachen). The diagnosis of chronic hepatitis C was based on consistently detectable serum HCV RNA, using qualitative reverse-transcriptase polymerase chain reaction with a lower sensitivity of 100 copies (50 IU)/mL (Amplicor HCV test 2.0, Roche Diagnostics, Mannheim, Germany) and histological examination of liver biopsies. For quantification of HCV RNA, a branched DNA assay (Versant Quantitative 3.0 Assay, Bayer Diagnostics, Leverkusen, Germany) was employed.20, 22 All patients included in the study were anti-HCV positive but were negative for the hepatitis B surface antigen and anti-HIV 1 and 2 (all assays from Abbott, Wiesbaden, Germany). HCV genotypes were determined using reverse hybridization assay (INNO LiPA HCVC-II, Innogenetics, Gent, Belgium).
All patients included in this study received their first course of antiviral therapy and were treated for 24 to 48 weeks with pegylated (n = 120) or standard interferon α-2a or α-2b (n = 177) in combination with a body weight–adjusted dose of ribavirin according to current consensus protocols, as described previously.20 The virological response to therapy was assessed via repeated measurements of HCV RNA, as described above. According to the qualitative HCV RNA results, patients were defined as either sustained virological responders (no detectable HCV RNA after 24 or 48 weeks of treatment and 6 months afterwards) or nonresponders (including virological response with relapse within 24 weeks after the end of treatment, viral breakthrough, and virological nonresponse with continued presence of HCV RNA at the end of treatment).23 According to the respective study protocols, treatment was discontinued in virological nonresponders at week 24 of therapy based on a positive qualitative HCV RNA test.
Our control population comprised 152 Caucasian patients (86 males, 56.6%) with a median age of 45 years (range 20–75 years) who were admitted to the Department of Medicine III in Aachen for diseases other than hepatitis C. In all control subjects, chronic hepatitis C was excluded by a negative anti-HCV assay (Abbott, Wiesbaden, Germany), and no patient had clinical or biochemical signs of acute HCV infection. Controls were comparable to HCV patients with respect to age and gender. An in-hospital control population was selected because the anticipated risk for HCV infection in this cohort approximates that of our cases, compared with the markedly lower risk of HCV infection in young, healthy blood donors. Therefore, in accordance with epidemiological criteria recently proposed by Schulz and Grimes,24 our control population might better approach the “background” viral exposure and minimize selection bias, as previously discussed.21, 25
The Ethics Committees approved the study protocol in Berlin and Aachen.
Liver Histology and Chemistry Tests.
Liver biopsies were performed in all patients prior to antiviral therapy. Liver specimens were fixed, paraffin-embedded, and stained. Hepatic inflammation (grade) and fibrosis (stage) were assessed in a noncentralized manner by local pathologists unaware of the study protocol employing a semiquantitative histological score,26 as described previously.20 Routine clinical chemistry was performed with a Hitachi 717 analyzer (Roche Diagnostics).
Genotyping of RANTES Single Nucleotide Polymorphisms.
We genotyped all known RANTES SNPs with minor allele frequencies greater than 5% in Caucasian populations to gain sufficient statistical power to detect significant differences between cases and controls.27RANTES variants -403 G/A and Int1.1 T/C were determined using restriction fragment length polymorphism analysis. For SNP -403 G/A, polymerase chain reaction primers were 5′-GCCTCAATTTACAGTGTG-3′ and 5′-TGTTATTCATTACAGATGTT-3′. A restriction enzyme site for MaeIII was created by changing the sequence at position 19 (underlined) in the antisense primer. For SNP Int1.1 T/C, primers 5′-CCTGGTCTTGACCACCACA-3′ and 5′-GCTGACAGGCATGAGTCAGA-3′ were used for polymerase chain reaction, and MboII was used for restriction fragment length polymorphism analysis. All restriction enzymes were purchased from New England Biolabs (Beverly, MA). SNP 3′ 222 T/C was determined via direct sequencing of polymerase chain reaction products (primers 5′-CTGTCCCGGTACTGACAACTC-3′ and 5′-CCCGAGTAGGTGGGACTACA-3′). Fluorescence-based automated cycle sequencing was performed with an automated sequencer (ABI PRISM 310, Applera Corporation, Norwalk, CT), using the dye-terminator method according to the manufacturer's instructions (Big Dye Terminator Cycle Sequencing-Ready Reaction Kit, Applera). Detailed polymerase chain reaction protocols can be obtained from the corresponding author (F.L.).
RANTES Haplotype Analysis.
Genotyping in case-control studies provides two alleles at each SNP, which are observed as a multisite genotype, or diplotype. The unique combination of linked SNPs present at a chromosomal locus, or haplotype, can be experimentally determined by using phase information from parental genotypes, which does not seem feasible in infectious diseases. Alternatively, haplotypes can be statistically inferred from diplotypes.28 In our study, haplotype reconstruction was performed using the PHASE v2.0.2 algorithm (available from www.stat.washington.edu/stephens/phase.html), a Bayesian method that provides a complete solution and is more accurate than the common expectation maximization algorithm.29 The resulting RANTES haplotypes were named according to An et al.13 A permutation test was performed to detect differences in haplotype distribution between groups.29 The permutation test checks the null hypothesis that case and control haplotypes are a random sample from a single set of haplotype frequencies versus the alternative that cases are more similar to each other than to controls. Haplotypes of responders (cases) and non-responders (controls) were permuted 100,000 times. It has been recommended that, to estimate a P value of α, one should carry out 10/α replicates. Typically, to detect association at a significance level of .01, one would perform at least 1,000 permutations.30 Subsequently, a two-sided Fisher's exact test was performed to detect differences between groups for each haplotype separately.
Additional Statistical Analysis.
Accordance of genotype frequencies with Hardy-Weinberg equilibrium was assessed with an exact test employing software developed by Wienker and Strom (available from http://ihg.gsf.de). Continuous data are given as median and ranges. The age of patients and controls was compared with the non-parametric Mann-Whitney U test. Variables in contingency tables were compared with a two-sided Fisher's exact test and odds ratios with 95%-confidence intervals were calculated. Two-sided P-values of less than .05 were considered significant.
Common Risk Profile of the HCV Study Cohort.
Table 1 displays the demographical and clinical data of our HCV study population. All patients with chronic hepatitis C and our control population were of Caucasian origin. The main risk factors for HCV infection were blood transfusions (n = 96, 32.3%), injection drug use (n = 69, 23.2%), and unknown causes of transmission (n = 122, 41.1%). Ten individuals (3.4%) had an occupational transmission of HCV. Of note, there were no hemophiliacs in our study cohort.
Table 1. Demographic and Clinical Parameters of Patients With Chronic Hepatitis C (n = 297)
NOTE: Continuous data are given as median (ranges).
Normal range: <23 U/L for male patients, <18 U/L for female patients.
Normal range: <17 U/L for male patients, <15 U/L for female patients.
The upper panel of Fig. 1 illustrates the three RANTES SNPs (-403 G/A, Int1.1 C/T, and 3′ 222 T/C) determined in this study. Because these SNPs display minor allele frequencies of greater than 5% in Caucasian populations, they were considered to detect genotype–phenotype effects with sufficient power in our cohort (see Patients and Methods). Overall, there were no differences in the minor allele frequencies of all tested SNPs between HCV-infected patients and controls, respectively (-403 A, 0.206/0.220; Int1.1 C, 0.125/0.136; 3′ 222 C, 0.069/0.065; all P > .05). As expected, distributions of tested polymorphisms were in Hardy-Weinberg equilibrium, and there were no differences concerning the number of heterozygotes or homozygotes between patients and controls (P > .05).
Reconstruction of RANTES Haplotypes.
Because the detection of an association between marker alleles and a certain phenotype depends critically on the extent of linkage disequilibrium between disease associated polymorphisms,31 we reconstructed characteristic SNP allele combinations on single chromosomes (haplotypes) in our populations. Out of eight possible haplotypes, four were detected with prevalence rates greater than 3% in our cohort. Two additional rare haplotypes were found at frequencies of 0.4% and 0.2%, respectively, and were not considered for further analysis. Figure 1 depicts the four most common RANTES haplotypes and their prevalence rates. The overall haplotype distribution did not differ between patients with chronic hepatitis C and controls (P > .05).
RANTES Haplotype Distribution Differs Between Sustained Responders and Nonresponders to Antiviral Combination Therapy.
In our cohort of HCV patients, RANTES genotypes were not associated with quantitative traits such as HCV viral load or grading and staging in liver biopsies. However, antiviral treatment response as a qualitative trait was significantly associated with RANTES haplotypes in our study. Antiviral combination therapy was performed in all 297 HCV-infected patients with interferon α and ribavirin. Overall, 146 patients (49.2 %) achieved a sustained virological response, and 151 patients (50.8%) were nonresponders according to established criteria.23 Using a permutation test with 100,000 permutations, the overall distribution of RANTES haplotypes was significantly different between sustained responders and nonresponders (P = .007, Table 2). In the subgroup of patients treated with pegylated interferon and ribavirin, responders (n = 61) and nonresponders (n = 59) also differed significantly with respect to RANTES haplotype distribution (P = .02 by permutation testing).
Table 2. RANTES Haplotype Frequencies in HCV Patients
RANTES Haplotypes Carrying Int1.1 C and 3′ 222 C Alleles Are Associated With a Negative Response to Antiviral Combination Therapy.
After demonstration of a significant difference in the overall distribution of RANTES haplotypes, we analyzed specific haplotypes with respect to treatment outcome. Table 2 shows that the haplotype R4 (defined by the presence of Int1.1 C and 3′ 222 C) was detected more frequently in nonresponders (9.5%) compared with responders (5.1%; odds ratio [OR] 1.9, confidence interval [CI] 1.01–3.70, P = .04). At the genotype level, the Int1.1 C allele (P = .03) but not the 3′ 222 C allele (P = .09) was more prevalent in nonresponders compared with responders; therefore, we also analyzed the RANTES haplotypes R3 and R4, both of which carry Int1.1 C in combination. Both haplotypes together were significantly more prevalent in nonresponders (12.8%) compared with responders (8.2%; OR 1.9, CI 1.14–3.27, P = .01; see Table 2). In contrast, there was a trend to a higher frequency of haplotype R2 in responders compared with nonresponders (P = 0.07).
RANTES Haplotypes Also Affect Treatment Response in Patients Infected With HCV Genotypes 1 and 4.
To account for the well-known impact of HCV genotypes on treatment response to antiviral therapy,20 we analyzed RANTES haplotypes separately for HCV genotypes 2/3 and 1/4, respectively.
The subgroup of patients infected with HCV genotypes 2 and 3 had an overall sustained virological response rate of 89.9%. This implies that it is statistically difficult to identify additional predictive markers, and indeed none of the haplotypes was associated with treatment response when analysis was restricted to HCV genotypes 2 and 3.
Overall, 79 patients (36.7%) infected with HCV genotypes 1 and 4 were sustained responders, whereas 136 patients (73.3%) did not respond to therapy (see Table 2). Figure 2 illustrates that in this clinically relevant subgroup of patients, the RANTES haplotype R4 was observed more frequently among nonresponders (9.6%) compared with responders to antiviral therapy (3.8%; OR 2.66, CI 1.07–6.61, P = .04). Furthermore, the nonresponders more often carried RANTES haplotypes R3 or R4 (14.7%) compared with sustained responders (7.0%; OR 2.29, CI 1.14–4.60, P = .02; see Table 2), suggesting that the tagging SNPs of both haplotypes might be important with respect to treatment outcome. As for analysis of patients with all HCV genotypes, there was a trend (P = .08) to a higher prevalence of haplotype R2 in responders compared with nonresponders (Fig. 2).
The results presented here provide evidence that RANTES haplotypes carrying the functional variants Int1.1 C and 3′ 222 C are associated with a negative outcome of antiviral combination therapy in patients with chronic HCV infection. Because the current consensus treatment of HCV infection with interferon α and ribavirin is only effective in approximately 50% of all cases, stratification of individuals with regard to their individual chances for positive treatment outcome might be relevant for the design of new optimized treatment strategies.32
In recent years, several lines of evidence have indicated that patients who either spontaneously clear HCV RNA or respond successfully to antiviral therapy display a predominantly proinflammatory (T helper 1 lymphocyte–associated) cytokine profile.10, 11 RANTES specifically binds to the chemokine receptor CCR5 and attracts proinflammatory CD4+ T helper 1 lymphocytes.9 Interestingly, the CCR5 Δ32 mutation leads to a decreased expression of the functional CCR5 protein7 and has been linked to a negative outcome of interferon monotherapy in HCV infection,8 although these results have not yet been reproduced for combination therapies.21, 33–35 We tested the hypothesis that RANTES haplotypes carrying functional and interacting polymorphisms might be linked to chronic HCV infection, and—more specifically—whether or not they could be correlated to the outcome of antiviral therapy. The study design is based on the expectation that the frequencies of specific RANTES gene variants will differ between HCV-infected patients and healthy controls with a comparable exposure risk or between patients with a positive response to antiviral therapy and those with a negative response.
Few studies have investigated variations in other gene loci in relation to outcome of the antiviral therapy for chronic HCV infection, including cytokine genes,36–40 the mannose-binding lectin gene,41 the gene for interferon-inducible MxA protein,42 the low molecular mass polypeptide 7 gene,43 and the low-density lipoprotein receptor.44 However, all of these studies had less than 80% power to detect moderate effects of gene variants because of a sample size of less than 100 individuals per subgroup.45 Furthermore, the results of these studies conflict with one another,36–38 and a single study provided extended haplotype information of the target gene.44
Previously, two RANTES promoter polymorphisms (-28 C/G and -403 G/A) have been investigated in patients with chronic HCV infection. Promrat et al.46 and Hellier et al.35 could not detect an association of these polymorphisms with treatment response; however, analysis was restricted to promoter polymorphisms, and no haplotype data of the RANTES gene were provided. Moreover, in Caucasian populations the promoter polymorphism -28 C/G (which up-regulates RANTES transcription) has been shown to have a minor allele frequency of less than 5%,13, 15 which might not be sufficient to detect moderate phenotypic effects in medium-scale association studies.47 Thus, in anticipation of only moderate effects of single RANTES polymorphisms, we performed a combined genetic analysis of all known variants with an allele frequency of greater than 5%, including the promoter polymorphism -403 G/A, a SNP in the first intron (Int1.1 T/C), and a SNP in the 3′-untranslated region (3′ 222 T/C).
Because the use of single SNPs in many current association studies has little ability to detect genotype–phenotype correlations, it has been recommended that the overall sequence variation of candidate genes be assessed via haplotype analysis.31 This is particularly true for cohorts of patients with infectious diseases such as chronic hepatitis C, which do not include family members who provide phase information on genotypes.31, 48 Thus a major aim of our study was to employ RANTES haplotype information in an HCV study population. Using a powerful and precise statistical approach to infer these haplotypes from our population genotype data,29 we identified four common haplotypes (R1–R4) with a prevalence of more than 3% in patients with hepatitis C and our control population (see Fig. 1 and Table 2).
The haplotype R3 carries the minor allele of an intron 1 polymorphism (Int1.1 C), while R4 is defined by Int1.1 C and the minor allele of the SNP in the 3′ untranslated region (3′ 222 C). From our data it is difficult to assess whether the SNP at Int1.1 or 3′ 222 drives the association with the negative outcome of antiviral therapy, because both are altered in R4. Therefore, we retrospectively analyzed the genotype data in our HCV cohorts separately for both SNPs. Albeit the Int1.1 SNP was associated with treatment outcome at the genotype level (OR 1.9, P = .03), this was not evident for the 3′ 222 SNP (P = .09). However, in contrast to our haplotype data, such analysis does not take interactions between SNPs into account. Nevertheless, it supports the notion that both the Int1.1 C and the 3′ 222 C allele might be responsible for the observed association.
Although there are currently no published functional data for the 3′ 222 SNP, in vitro studies demonstrated that intron 1 of the RANTES gene contains a strong regulatory sequence element and that constructs harboring the Int1.1 C polymorphism lead to a strong significant down-regulation of transcription in reporter gene assays.13 Therefore, the higher frequencies of RANTES haplotypes 3 and 4 in patients with a negative outcome of antiviral therapy fit perfectly with a genetically determined down-regulation of T helper 1 lymphocyte–associated responses in these patients.10, 11 This would reduce the overall chances for successful HCV clearance, as observed in the present study. In contrast, the R2 haplotype (harboring the -403 A but not the Int1.1 C allele) was more prevalent in responders compared with nonresponders, which is consistent with in vitro transient transfection studies reporting a mean fourfold higher constitutive transcriptional activity of the RANTES promoter carrying the -403 A allele.14
We also detected a significant difference in RANTES haplotype distribution between responders and nonresponders in patients who were treated with pegylated interferon α and ribavirin. Therefore, despite the heterogeneous nature of treatment protocols in our cohort of patients, our main conclusions should hold true for individuals treated with pegylated interferon α, which is now the standard of care.
In conclusion, RANTES gene variants carrying the Int1.1 C and/or 3′ 222 C alleles are likely to contribute to the complex interaction between HCV and the polygenic host immune response during antiviral therapy. These findings might help to better predict response rates based on the individual characteristics of our patients. Furthermore, the present study indicates that future case-control and prospective studies designed to discover and validate gene polymorphisms involved in chronic hepatitis C should benefit from haplotype analysis and the ongoing elucidation49 of the haplotype structure of the human genome.
The authors thank Hildegard Keppeler (Aachen) for excellent technical assistance.