Innate immune genes synergize to predict increased risk of chronic disease in hepatitis C virus infection

Abstract

Hepatitis C is a common infection with significant morbidity and mortality, and only a minority of patients successfully clear the infection. Identification of factors that influence disease progression in HCV infection is difficult owing to the lack of well-defined patient cohorts. However, recent evidence supports a role for the innate immune system in virus clearance. In this study, we investigated innate immune genes for their contribution to disease progression in a unique cohort of well-controlled HCV-infected patients. The Irish cohort of HCV patients is uniquely homogenous; patients were infected with a single genotype of HCV from contaminated anti-D Ig. We genotyped 543 infected patients, including 247 patients who spontaneously resolved infection, for natural killer (NK) cell-associated killer cell Ig-like receptors (KIR) genes and the recently reported IL28B (IFNλ3) SNP. The NK cell gene KIR2DS3 was significantly increased in patients with chronic infection [odds ratio (OR) 1.90, 95% confidence interval (CI) 1.25–2.90, P < 0.002]. The IL28B “T” allele was also significantly increased in chronically infected patients (OR 7.38, 95% CI 4.93–11.07, P < 10⁻⁸). The presence of both markers synergized to significantly increase the risk of chronic infection over either factor alone (OR 20.11, 95% CI 9.05–44.68, P < 10⁻⁷). In functional experiments, we found that IL28A significantly inhibited IFN-γ production by NK cells. Thus, we demonstrate a functional link between NK cells and type 3 IFN. Our findings may contribute to the development of a prognostic test for HCV and identify therapeutic strategies for the clinical management of HCV-infected patients.

Hepatitis C virus (HCV) is a common infection associated with significant morbidity and mortality worldwide (1). A successful immune response to HCV infection is thought to underlie the spontaneous resolution of infection that is observed in ≈20% of patients (1–3). However, most patients develop chronic infection, often leading to cirrhosis of the liver, hepatic cellular carcinoma, and liver failure (4). Identification of patients who naturally resolve HCV infection is a significant clinical challenge because these patients may not be aware of their infection status owing to mild clinical symptoms. This lack of well-defined cohorts has hindered identification of host immune genetic factors involved in viral persistence and clearance (5), thereby limiting opportunities for development of immunotherapies or immune-based prognostic tests for HCV infection.

Although a role for the adaptive immune response has been well established (6–9), more recent evidence supports a role for the innate immune system in response to HCV infection. This includes detection of viral infection, activation of effector cells, particularly natural killer (NK) cells, and the production of cytokines, of which the type 1 IFNs are particularly important (10, 11). Indeed, the only current treatment for chronic HCV infection is type 1 IFN (usually given in combination with ribavirin). NK cells can directly induce apoptosis of HCV-infected hepatocytes and themselves produce a range of antiviral cytokines (12, 13). Evidence is also accumulating that NK cell depletion or dysfunction may contribute to HCV persistence (14). NK cell activities are regulated in part through cell-surface receptors, including the killer cell–Ig-like receptors (KIR), encoded by a family of genes located on human chromosome 19q13.4. NK cells detect virally infected cells through KIR interactions with HLA class I. Several studies have investigated the role of KIR and HLA in HCV infection, but low sample numbers and heterogeneity of patient cohorts in terms of sex, age, ethnicity, route of infection, and HCV genotype have resulted in conflicting results (15–17). One larger study identified an inhibitory KIR gene, KIR2DL3, associated with resolution of HCV infection, and this was dependent on a homozygous HLA class I ligand background (11, 18, 19). More recent evidence to support a role for the innate immune system in HCV has come from a series of reports on SNPs in the IL28B gene region that can predict responsiveness of patients with chronic HCV to type 1 IFN treatment (20–22). One of these SNPs, rs12979860 (20), was also significant in predicting spontaneous resolution of HCV infection (23, 24). Although not within a coding region of a gene, SNP rs12979860 is found adjacent to the IL28B gene that encodes for a unique type 3 IFN called IFN-λ3. Little is known about the IFNλ family, but evidence is mounting to support a role for them in the immune response to viral infection (25, 26). It is of particular interest that both the NK cell-associated KIR genes and the IL28B-associated SNP are found in the same region of the genome. In light of these reports, we investigated the hypothesis that multiple immune-related loci in chromosome region 19q13 are involved in spontaneous clearance of HCV virus.

The patients in the present study were all infected with HCV through contaminated anti-D blood products in Ireland in 1977/8. As such, this cohort is extremely homogenous: all patients are female, fertile, and were aged 16–44 (mean 27.4 ± 5.5) y when infected. They come from a homogenous genetic background (Irish Caucasians) and were all infected through the same route with a similar, low level of inoculum and, importantly from an epidemiological viewpoint, they were all infected with a single source of HCV of defined genotype (genotype 1b) (27). Thus, they represent a uniquely informative cohort for the study of HCV that excludes many of the confounding variables associated with other studies. These women have been extensively studied and have a relatively high rate of resolution (≈45%) (27). They are therefore particularly interesting in the context of defining immunogenetic factors that determine whether an individual will eliminate virus or succumb to a lifetime chronic infection. To test our hypothesis, we undertook to investigate the contribution of KIR, HLA class I, and IL28B innate immune genes to resolution of infection in these patients.

Results

KIR2DS3 Gene Frequency Is Increased in Chronic HCV Infection.

The present study investigated a total of 543 HCV-infected patients, of whom 296 were chronically infected and 247 had resolved infection. All patients were first typed for 14 KIR genes by PCR-SSP (sequence-specific primers) (Table 1). A single activatory gene, KIR2DS3, was significantly increased in chronically infected patients (0.314, n = 92) compared with those who resolved infection (0.194, n = 47, P = 0.002). The carrier frequency of the KIR2DS3 gene in healthy, noninfected, Irish women was found to be 0.233 (n = 116), which was similar to the HCV-infected group as a whole (0.256, n = 543). Although not reaching statistical significance, the inhibitory gene KIR2DL5 was also more frequent in chronically infected patients (0.526, n = 153) vs. resolvers (0.454, n = 109, P = 0.100). KIR gene haplotypes are broadly categorized as being either A-type, with a restricted gene content (characterized by KIR3DL3, KIR2DL3, KIR2DL1, KIR2DL4, KIR3DL1, KIR2DS4, and KIR3DL2), or B-type, which varies both in number and content of KIR genes (includes additional genes that encode activatory receptors) (28, 29). Because KIR2DS3 and KIR2DL5 are both genes associated with B-type KIR haplotypes, this suggested that certain B-haplotypes were increased in frequency in patients with chronic HCV infection.

Table 1.

KIR2DS3 gene is increased in chronic HCV infection

	Frequencies		Chronic vs. resolvers
KIR gene	Chronic HCV (n)	Resolvers (n)	χ2 (P)	OR (95% CI)
KIR2DS2	0.519 (152)	0.469 (115)	1.30 (0.25)	1.22 (0.86–1.74)
KIR2DL3	0.898 (265)	0.918 (223)	0.59 (0.44)	0.79 (0.42–1.49)
KIR2DL2	0.527 (155)	0.486 (118)	0.92 (0.34)	1.18 (0.83–1.68)
KIR2DL5	0.526 (153)	0.454 (109)	2.70 (0.10)	1.33 (0.93–1.91)
KIR2DS3	0.314 (92)	0.194 (47)	9.89 (0.002)*†	1.90 (1.25–2.90)*
KIR2DL1	0.976 (287)	0.959 (235)	1.31 (0.25)	1.76 (0.61–5.21)
KIR3DL1	0.966 (283)	0.979 (235)	0.12 (0.73)	0.84 (0.28–2.45)
KIR3DS1	0.405 (115)	0.389 (91)	0.14 (0.71)	1.07 (0.74–1.55)
KIR2DS5	0.284 (83)	0.321 (79)	0.86 (0.35)	0.84 (0.57–1.24)
KIR2DS1	0.381 (111)	0.395 (96)	0.10 (0.75)	0.94 (0.66–1.36)
KIR2DS4	0.966 (284)	0.963 (237)	0.03 (0.87)	1.08 (0.40–2.93)

Carrier frequency of KIR genes in 296 chronic HCV patients compared with 247 spontaneous resolvers. Differences in frequency distribution between populations were tested for significance by χ² test. A positive OR indicates an association with increased risk for chronic HCV.

*Significant value.

^†Bonferroni corrected P value for multiple comparisons, P = 0.022.

KIR2DS3 Predisposes to Chronic HCV Infection in the Presence of HLA-C2.

KIR and HLA genes are inherited on different chromosomes, and both ligand and receptor must be present for particular functional interactions to occur (30). We tested whether KIR2DL3 was associated with resolution of HCV infection on an HLA-C1 homozygous background, as has been previously reported. HLA class I genes can be stratified according to the ligands provided to KIR molecules. In brief, C1 refers to HLA-C alleles that encode ligands for the KIR2DL3 receptor (including its common allotype, KIR2DL2), and C2 refers to HLA-C alleles that encode ligands for the KIR2DL1 receptor (31, 32). Although not reaching statistical significance, we found a trend in our cohort toward a higher frequency of KIR2DL3+ on HLA-C1/C1 background in resolved patients (0.429, n = 91) vs. chronically infected patients (0.356, n = 95, P = 0.101; Table 2), a finding previously reported by Khakoo et al. (18). However, examination of the interaction on receptor and ligand homozygous backgrounds showed no significant effect on resolution of HCV infection (KIR2DL3/KIR2DL3/C1/C1: 0.191, n = 53 and 0.236, n = 54 for chronic and resolved patients, respectively, P = 0.243; Table 2). On the basis of the finding that the KIR2DS3 gene frequency is significantly increased in patients with chronic HCV infection (Table 1) and that multiple KIR genes with different HLA specificities are inherited on haplotypes, and the fact that HLA can profoundly influence the NK cell response, we tested whether the association of KIR2DS3 with chronic infection was influenced by the HLA-C genetic background. When analyzed in terms of HLA class I ligands, KIR2DS3 was only significantly increased in patients with chronic HCV infection when present on a HLA-C2⁺ genetic background (Table 2).

Table 2.

KIR2DS3 is associated with chronic HCV infection only in the presence of HLA-C2

	Frequencies		Chronic vs. resolvers
KIR–HLA	Chronic HCV (n)	Resolvers (n)	χ2 (P)	OR (95% CI)
KIR2DL3+ C1/C1	0.356 (95)	0.429 (91)	2.68 (0.10)	0.73 (0.50–1.08)
KIR2DL3/KIR2DL3/C1/C1	0.191 (53)	0.236 (54)	1.36 (0.24)	0.77 (0.49–1.20)
KIR2DL2+ C1/C1	0.179 (50)	0.182 (41)	0.01 (0.93)	0.98 (0.61–1.59)
KIR2DL2/KIR2DL2/C1/C1	0.041 (12)	0.017 (4)	2.70 (0.10)	2.53 (0.75–9.43)
KIR2DS3+ C1/C1	0.079 (22)	0.060 (14)	0.72 (0.39)	1.35 (0.64–2.86)
KIR2DS3+ C2+	0.198 (55)	0.098 (23)	9.75 (0.002)*†	2.26 (1.30–3.94)*

Frequency of KIR2DL3 and KIR2DL2 genes in combination their HLA C ligand (C1) were compared in 296 chronic HCV patients and 247 spontaneous resolvers, as well as KIR2DS3 with either HLA-C1 homozygotes or HLA-C2 carriers. KIR⁺ indicates carriers for that gene; HLA/HLA indicates a particular HLA genotype; HLA⁺ indicates carriers for that HLA allele; KIR2DL2/KIR2DL2, KIR2DL3/KIR2DL3, and C1/C1 indicate homozygosity for that KIR gene or HLA type. Calculations as for Table 1.

*Significant value.

^†Bonferroni corrected P value for multiple comparisons, P = 0.012.

Identification of a Specific KIR Haplotype Associated with Chronic HCV Infection.

Given that KIR receptors are encoded for by multiple functionally related genes at a single locus, it is likely that the inheritance of a haplotype rather than a single KIR gene is the source of the significant association found. Furthermore, because our results show that KIR2DS3 and KIR2DL5 genes are increased in patients with chronic HCV infection and that these are in high linkage disequilibrium (LD) together (D' = 0.89), this suggested KIR B-haplotype involvement (33). The KIR2DL4 gene has previously been shown to separate KIR genes in the centromeric from the telomeric end of the KIR gene cluster, with LD stronger between the genes within each end (33, 34). Haplotypes were reconstructed using KIR3DL3, KIR2DS2, KIR2DL3, KIR2DL5, KIR2DS3, and KIR2DL1 genes, which are all found centromeric of the KIR2DL4 gene. A total of 25 haplotypes (of which the 12 most frequent are shown in Table 3), accounting for 100% of haplotypes, were generated and used in subsequent analysis. Comparison of patient groups demonstrated that a single centromeric KIR haplotype was significantly increased in patients with chronic infection (0.154, n = 90.0) compared with resolved patients (0.091, n = 44.2, P = 0.002), and as predicted, it contained KIR2DL5 and KIR2DS3 genes (Table 3). Analysis of the frequency distribution of reconstructed telomeric gene haplotypes (KIR2DL4, KIR3DL1/S1, KIR2DS1, KIR2DS4, and KIR3DL2) revealed no differences between the patient groups (Table S1), further supporting the lack of involvement of the telomeric end of the KIR gene locus with chronic HCV infection.

Table 3.

Genes contributing to the association with chronic HCV infection are found in the centromeric part of the KIR haplotype

Estimated haplotype frequencies for centromeric KIR genes in chronic HCV patients were compared with those in spontaneous resolvers. Shaded boxes indicate presence of a gene (with paler shading and a “2” for the KIR2DL2 allele of a KIR2DL3), and unshaded boxes indicate absence of a gene on a haplotype. Differences in haplotype distribution between populations were tested for significance by χ² test or Fisher exact test where appropriate.

*Significant value.

IL28B Adjacent SNP, rs12979860, Is Associated with Chronic HCV Infection.

All of the HCV study cohort were typed for the rs12989760 SNP (hereafter referred to as IL28B). This SNP is defined by either a C or a T nucleotide. When the C and T allele frequencies were compared between the spontaneously resolving and the chronic HCV-infected groups, the T allele was found to be significantly increased in the chronic group [Table 4; 0.134 vs. 0.395, P < 10⁻⁸; odds ratio (OR) 4.20, 95% confidence interval (CI) 3.05–5.79]. Further analysis showed that both the CT and the TT genotypes were significantly increased in the chronically infected group compared with the spontaneous resolvers, indicating that the IL28B-T allele has a dominant effect (Table 4). This conclusion was also supported by analysis showing that IL28B genotypes were in Hardy-Weinberg equilibrium (HWE) for the spontaneous resolvers (P > 0.05) but were out of HWE, with an excess of heterozygotes, in the chronically infected group. Typing for IL28B-associated SNP in a group of healthy Irish controls (n = 173) confirmed the genotypes to be in HWE and at a similar frequency to the complete HCV cohort.

Table 4.

IL28B is associated with an increased risk of chronic HCV

	Frequencies		Chronic vs. resolvers
IL28B C/T genotype	Chronic HCV (n)	Resolvers (n)	χ2 (P)	OR (95% CI)
CC	0.296 (87)	0.756 (183)
CT	0.619 (182)	0.219 (53)
TT	0.085 (25)	0.025 (6)	112.61 (<10−8)*	n/a
C	0.605 (356)	0.866 (419)
T	0.395 (232)	0.134 (65)	89.78 (<10−8)*	4.20 (3.05–5.79)*
CC	0.296 (87)	0.756 (183)
CT+TT	0.704 (207)	0.244 (59)	112.49 (<10−8)*	7.38 (4.93–11.07)*
CC	0.296 (87)	0.756 (183)
TT	0.085 (25)	0.025 (6)	27.91 (10−7)*	8.76 (3.26–24.82)*

Genotype and allele frequency distributions of the IL28B SNP (rs12979860) were compared in 294 chronic HCV patients and 242 spontaneous resolvers. Differences in frequency distribution between populations were tested for significance by χ² test. A positive OR indicates an association with increased risk for chronic HCV. n/a, test not applicable.

*Significant value.

Combination of IL28B-T and KIR2DS3 Synergize to Increase the Risk of Developing Chronic HCV Infection.

To test for independence of the KIR2DS3 association, in the light of the much stronger association of the IL28B SNP and the fact that both loci are found in the same chromosomal region (19q13), carrier frequencies for KIR2DS3 were stratified by the presence of the IL28B-T allele and found to be significantly associated with the chronic group irrespective of the IL28B-T allele. Equally, when stratified by the presence/absence of KIR2DS3, the IL28B-T allele carriers were still significantly increased in the chronic group (Table S2). Thus, each locus on its own increases the risk of developing chronic HCV infection.

Because these two genes encode for molecules involved in the innate immune response to virus, we tested for the possibility of interaction between these two risk factors. Statistical interaction between KIR2DS3 and IL28B-T was tested for by departure from additivity (35–37). ORs were calculated by multinomial logistic regression, and the presence of both alleles (i.e., KIR2DS3 carriers and IL28B-T carriers) compared with the absence of both (KIR2DS3 negative and IL28B-CC) resulted in a strikingly increased OR in chronic HCV-infected patients (Fig. 1; OR 20.11, 95% CI 9.05–44.68, P < 10⁻⁷). This synergy was confirmed to be statistically significant, with a Synergy index (S) of 2.43 (95% CI 1.03–5.68) and with the attributable proportion due to interaction (AP) 55.9% (95% CI 20.80–91.00%) (Fig. 1). No additional risk was provided by the presence of HLA-C2 (Table S3). In summary, the presence of both “risk” alleles, KIR2DS3 and IL28B-T, synergizes to significantly increase the risk of chronic HCV infection compared with the presence of either marker alone.

Fig. 1.

Presence of both KIR2DS3 and IL28B-T synergize to increase the risk of developing chronic HCV infection. ORs for KIR2DS3 alone, IL28B-T carriers (CT or TT) alone, and KIR2DS3 with IL28B-T compared with neither risk factor present were calculated from multinomial logistic regression and used to calculate S and AP. The contribution of individual and combined risk factors to the ORs is graphically represented in the bar chart. Positive OR indicates an association with increased risk for chronic HCV. ^aOdds ratio (95% confidence interval). ^bSynergy index (95% confidence interval). ^cAttributable proportion for interaction (95% confidence interval).

IL28A Inhibits IFN-γ Production by Human NK Cells.

Our genetic data demonstrated synergy between NK cell receptors and type 3 IFNs in increasing risk of chronic HCV infection. However, no functional link between NK cells and type 3 IFNs has ever been shown. We therefore carried out experiments to explore a direct link between NK cells and some of the type 3 IFN family of cytokines. Our experiments focused on IFN-γ production by NK cells as it is a key component of the NK cell response to virus. We found that IL28A (and IL28B to a lesser extent) significantly inhibited IFN-γ production by human NK cells (Fig. 2; P < 0.05). Although the inhibition was not profound, eight of 11 donors showed decreased production of IFN-γ in response to IL12+IL15 in the presence of IL28A. Furthermore, three individual donors within the group seemed to be particularly sensitive to the effects of IL28A, and in these, IFN-γ production was completely inhibited. Changes in CD69 expression, an antigen expressed by activated lymphocytes, correlated with the cytokine results (n = 8). In donors sensitive to IL28A inhibition of IFN-γ production, IL28A also robustly inhibited CD69 expression on NK cells, whereas there was no change in CD69 expression in donors that had low or no inhibition of IFN-γ production in response to IL28A. Our data demonstrate a direct interaction between NK cells and type 3 IFN and suggest that IL28A has some inhibitory effects on NK cell functions and that these are particularly potent in certain individuals. However, there was no apparent correlation between the IL28B SNP and sensitivity to the effects of either IL28A or IL28B.

Fig. 2.

IL28A inhibits IFN-γ production by human NK cells. Peripheral blood mononuclear cells (n = 11) from healthy normal donors were stimulated with PBS or IL12 (30 ng/mL)/IL15 (100 ng/mL) in the presence or absence of IL28A (500 ng/mL). Percentage of CD56⁺ CD3⁻ NK cells expressing IFN-γ as measured by intracellular staining is shown. Horizontal lines indicate the median percentages, and vertical lines indicate the range of values. A paired Student t test was used to compare data.

Discussion

Multiple factors, many immunological, have been implicated in determining disease outcome in HCV infection (1, 2). Some individuals naturally clear infection, whereas most succumb to chronic disease, resulting in liver cirrhosis and/or hepatic carcinoma, which may require liver transplantation (4). Therefore, identification of factors involved in the persistence of viral infection is important and may lead to new therapeutic interventions, the development of specific prognostic tests, and/or improved treatment management for patients. Although emphasis has previously been placed on the importance of an effective adaptive immune response (6–9), evidence from this study and others (18, 23) highlights a role for the innate immune response in regulating disease progression in HCV infection.

The Irish cohort of HCV-infected patients is extremely important because we can directly compare patients who resolve infection or develop chronic infection. Conventional analysis of individual genes identified KIR2DS3 as an NK cell-associated KIR gene that was at significantly increased frequency in patients who resolved infection compared with patients who developed chronic HCV infection. Although trends for particular HLA associations were observed (Table S4), no definitive associations were found for HLA class I genes in terms of their provision of ligands for NK cell receptors (C1, C2, Bw4). Importantly, our data also support the previously reported finding of a role for KIR genes in resolving HCV infection in a large independent cohort (18). Although not reaching statistical significance, we found that KIR2DL3+ on an HLA-C1/C1 background in our cohort was increased in patients who resolve infection and conversely, the absence of KIR2DL3 (defined by KIR2DL2 allele homozygosity) was associated with an increase in chronic infection on an HLA-C1 genetic background (OR 2.53, 95% CI 0.75–9.43). Our study provides information for a revised model of the role of NK cells in HCV infection because it identifies a strong influence of a KIR B haplotype (and HLA-C2) on the development of chronic infection, whereas previously a beneficial effect of a KIR A haplotype (with HLA-C1) was observed (18). Differences in host genetics in terms of the combinations of NK cell receptors and ligands present clearly affect disease outcome.

Within our patient cohort, we found a strong association between the presence of the NK cell-associated KIR2DS3 and chronic HCV infection. KIR2DS3 has previously been implicated as a risk factor in different viral infections (38, 39) and as a risk factor for graft-vs.-host disease after stem cell transplant for hematological malignancies (40, 41). However, the little that is known about the biology of the receptor does not support a role for it as a functional receptor. Indeed, the KIR2DS3 gene itself has been lost from some populations, and KIR2DS3 protein has been shown to have little or no cell-surface expression in transfection experiments (42). Furthermore, no avidity for HLA-C1, HLA-C2, or any other HLA class 1 epitope tested was found, and its ligand remains unknown (43). Although it is tempting to speculate a direct role for KIR2DS3 in the immune response to HCV, it is likely that KIR2DS3 is a marker for a haplotype that is contributing to the development of chronic viral infection. The association between KIR2DS3 and chronic infection was only seen on an HLA-C2 genetic background. Because HLA-C2 is not a ligand for KIR2DS3 (43), the receptor interacting with HLA-C2 is probably encoded for by a gene in LD with KIR2DS3 (either within the NK complex or the wider leukocyte receptor complex genomic region). Indeed, our analysis led to identification of a specific KIR haplotype in the centromeric haplotype block that is associated with chronic HCV infection in patients. This haplotype also included KIR2DL2, KIR3DL3, KIR2DS2, and KIR2DL1 genes, some of which have been found to be associated with chronic HCV infection in other studies (16, 18). Indeed, because KIR2DL1 and KIR2DL2 both encode receptors that bind to HLA-C2 (32, 44), these are possible candidate genes for explaining the increased risk of chronic infection associated with KIR2DS3 and HLA-C2.

Even more striking than the association with the KIR genes was the association of the IL28B SNP as a predictor of developing chronic HCV infection. The presence of a T allele was sufficient to confer an increased risk of chronic infection. These data confirm the previous finding of Thomas et al. (23) in a second large cohort of spontaneous HCV-resolving patients. The statistical strength of the association found between IL28B-T and chronic infection is very robust; however, because the SNP is located in the intergenic region between IL28A and IL28B genes, the biological mechanism behind the genetic association remains to be elucidated. Our data suggested a synergistic interaction between the unique type 3 IFN and NK cells in an antiviral immune response. We investigated this at a functional level and found that IL28A inhibited IFN-γ production by NK cells. It has been shown by numerous studies that NK cells in patients with chronic HCV infection have normal or relatively higher cytotoxic activity (14, 45, 46), leading to the suggestion that functional or activated NK cells may contribute to the chronic persistence of infection in these patients. Indeed, a relative expansion of CD56^bright cells, known to produce IFN-γ, has previously been reported for the present cohort of patients (14), and high levels of IFN-γ expression have been reported in the livers of both humans and chimpanzees with chronic HCV infection (47, 48). Attenuation of chronically activated NK cells may therefore be beneficial in the treatment of persistent HCV infection. Our findings support this concept: we have shown that one of these cytokines, IL28A, can significantly inhibit NK cell activation. Further experiments will be required to elucidate the full extent of the functional synergy between NK cells and type 3 IFN in the immune response to virus.

The homogeneous nature of our cohort (all female, all infected with the same genotype of virus) allowed identification of factors involved in the development of chronic HCV infection that might not have been identified in a heterogeneous cohort. From a clinical perspective, we found that the combination of KIR2DS3 and the IL28B-T allele dramatically increased the OR for development of chronic HCV (OR 20.11, 95% CI 9.05–44.68) compared with either risk factor alone (OR 2.27, 95% CI 1.28–4.02 for KIR2DS3 and OR 7.59, 95% CI 4.86–11.87 for presence of IL28B-T allele; S 2.43, 95% CI 1.03–5.68). This was independent of the HLA background of the patients. Although this synergistic response needs to be validated in a heterogeneous HCV patient cohort, these data may contribute to the development of a relatively simple and specific host genotype test for high-risk patients or healthcare workers that will predict the clinical outcome in HCV infection more accurately that any current prognostic indicator. In summary, our data provide a significant advance in terms of understanding the role of the immune system in disease progression during HCV infection. This has clinical potential to lead to new therapeutic interventions and may also contribute to the development of a robust prognostic test for HCV-infected individuals.

Materials and Methods

Patient Cohort.

The study population consisted of a well-defined cohort of females who had been inoculated with HCV (genotype 1b)-contaminated anti-D immunoglobin, as described in detail elsewhere (27, 49). On a 17-y follow-up, only 55% of subjects who were antibody-positive had chronic HCV infection (27). Of the 543 patients involved in this study, 296 had developed chronic HCV infection, and 247 had spontaneously resolved infection. Informed written consent was obtained from each patient, and the study received ethical approval from the Research and Ethics Committee at St. Vincent's University Hospital.

Genotyping for KIR Genes and the IL28B-Associated SNP rs12979860.

DNA was isolated from blood, using the Qiamp DNA blood Mini Kit system (Qiagen, Hilden). The presence or absence of 14 KIR genes (KIR3DL3, KIR2DS2, KIR2DL3, KIR2DL2, KIR2DL5, KIR2DS3, KIR2DL1, KIR2DL4, KIR3DL1, KIR3DS1, KIR2DS5, KIR2DS1, KIR2DS4, and KIR3DL2) was determined using a PCR-SSP method as described by Vilches et al. (50). The PCR-SSP discriminated between full-length and deleted versions of KIR2DS4 alleles, and the presence of two bands indicated heterozygosity for these forms (50). Genotyping for the rs12979860 SNP was performed using the ABI Taqman allelic discrimination kit (23). For routine quality control purposes, ≈10% of samples were retyped anonymously, and no mismatches were found.

Cell Stimulation and IFN-γ Intracellular Staining for Flow Cytometric Analysis.

Peripheral blood mononuclear cells were isolated from venous blood of healthy normal donors by density gradient centrifugation. Cells were stimulated for 18 h at a density of 1.5 × 10⁶ cells/mL, the last 4 h in the presence of Golgi-Plug (BD Pharmingen); 100 ng/mL rhIL15, 30 ng/mL rhIL12 (Strathmann Biotec), and 500 ng/mL of either IL28A or IL28B (R&D Systems) were used. NK cells were stained and analyzed for intracellular production of IFN-γ as previously described (51).

Statistical Analysis.

Genotype, allele, and carrier frequency differences and OR trend tests between populations were tested for significance by direct counting using a χ² test, or when sample size was <5 for the contingency table, a Fisher exact test as implemented by EPI-INFO 3.5.1. HWE was estimated using GenePop 4.0 (www.genepop.curtin.edu.au). KIR haplotypes were reconstructed using PHASEv2.1.1 (52, 53). A number of constraints were placed on the haplotype reconstruction on the basis of LD and known haplotype structure within the KIR gene complex (33); these are detailed in SI Text.

Differences in haplotype distribution between populations were tested for significance using a χ² or Fisher exact test. Statistical interaction between KIR2DS3 and IL28B was evaluated by departure from additivity using the method developed by Andersson et al. (35–37) based on ORs derived from multinomial logistic regression (release 16.0; SPSS Inc.). An S >1 and an AP >0 indicate a significant synergystic interaction.

A paired Student t test performed using PRISM software (version 4.0; Graphpad Software Inc.) was used to investigate the effect of either IL28A or IL28B on IFN-γ production by human NK cells, and P < 0.05 was considered statistically significant.