See Editorial on Page 5
Genetics of IL28B and Treatment Response
In 2009, three groups reported that variation in and near the IL28B gene strongly associates with response to treatment of chronic hepatitis C virus (HCV) infection using the standard-of-care treatment, pegylated interferon-alpha (Peg-IFN-α) plus ribavirin (RBV).1-3 In the first study, Ge et al.1 used patients from the Initiating Dialysis Early and Late study4 to carry out a genome-wide association study (GWAS) on sustained virological response (SVR). They reported a P value of approximately 10−24 for rs12979860, the most strongly associated single-nucleotide polymorphism (SNP) in patients of European ancestry. Strikingly, this common polymorphism upstream of the IL28B gene was associated with a 2.5-fold higher relative rate of response among non-Hispanic Caucasian subjects carrying the responsive C/C genotype, compared with the treatment-resistant T/T genotype. Ge et al. also found that the C/C genotype is associated with improved treatment responses in both Hispanics and in African Americans.
At around the same time, Tanaka et al.3 and Suppiah et al.2 also reported genome-wide significant association for variants in the IL28B region in Japanese and European ancestry populations, respectively. Interestingly, in the Tanaka et al. study, the beneficial effect of the C/C genotype was considerably greater than that reported by Ge et al. or by Suppiah et al.
Ge et al. also suggested that IL28B variation is likely to also influence natural clearance of hepatitis C based on observing a lower frequency of the C allele in the chronically infected cohort than in ethnically matched population controls. This suggestion was confirmed shortly thereafter by study of a cohort of individuals acutely infected with HCV, which showed clear allele-frequency differences between individuals that do and do not clear hepatitis C5 and, subsequently, in a GWAS study of natural clearance.6
Since these initial reports, there have been several hundred articles published describing aspects of the relationship between IL28B variation and hepatitis C treatment, many of them clearly replicating the relationships between IL28B and HCV clearance. Although the initial discoveries were found in samples of patients predominantly infected with HCV genotype 1, subsequent studies have explored the effects of IL28B on outcomes from other viral genotypes and in more specific clinical situations, as reviewed in this issue and elsewhere.7, 8 Several other features of this association are worth noting.
IL28B Genotyping Is Now in Wide Clinical Use With “Home Brew” Tests Being Offered by Both LabCorp and Quest.
Given this clinical application, one question that has emerged is which variant is most appropriate to genotype diagnostically. As far as is known, there are several variants that are all equivalently informative in populations of European or Japanese ancestry.9 In populations of African ancestry, however, rs12979860 is clearly the most informative, along with the amino-acid replacement polymorphism, rs8103142, encoding the amino-acid substitution, Lys70Arg. For now, therefore, rs12979860 is an appropriate choice of variant for diagnostic testing, especially given that rs8103142 appears more difficult to genotype given close homology in this region between IL28B and other genes encoding type III interferons. It should be noted, however, that no clear case has yet been made for any causal variant(s) responsible for this association, and if one or more is identified, it may be necessary to adjust the diagnostic tests.
IL28B Explains Much of the Population-Based Variation in Response to Treatment.
It is well documented that African Americans are less likely to respond to treatment than European Americans and that East Asians respond best of all.10-12 Strikingly, much of this variation in average population SVR rates is, in fact, the result of differences in the favorable IL28B allele frequency among different human population groups, with Africans having the lowest frequencies and East Asians the highest. Specifically, Ge et al. also estimated that over half of the difference in response rates between European Americans and African Americans is the result of this allele-frequency difference.1
There May Be Other Important Genetic Variants to Be Found in the IL28B Region (or Elsewhere).
Ge et al. noted that when a test for association is carried out conditioned on the rs12979860 variant, the most significant P value in the region is then 10−6, which strongly suggested that there are further variants in the IL28B region that contribute an effect that is independent of that associated with rs12979860. Because identified functional variants may well help to elucidate the mechanism underlying this association, it is a priority to identify these putatively independent causal variants.
IL28B Influences Responses to Treatment, Including HCV Protease Inhibitors.
The U.S. Food and Drug Administration has recently approved two new drugs that act by inhibiting the HCV NS3/4A protease. The drugs have been tested in triple therapy, which adds a protease inhibitor (PI) to the Peg-IFN-α and RBV components. Although response rates improve considerably overall, IL28B still continues to influence response in the presence of the PIs.13 The rs12979860 C/C group continues to be more likely to be cured and also may be appropriately treated with a shorter course of therapy. The presence of a PI does attenuate the difference between the C/C responders and T/T nonresponders, and indeed the T/T group, in fact, benefits the most from the addition of a PI, on average. The presumption is that IL28B continues to exert its effect because of the interferon (IFN) backbone in these treatments, and it would seem most likely that in the setting of IFN-sparing treatment, IL28B variation would have little or no predictive value. Nevertheless, there are ways that the virus could respond to the host genotype that could cause a difference even in IFN-sparing treatments (see below), so this will require further evaluation.
Biology of Interferon-Lambda
IFNs represent the first line of defense against viral pathogens and act both directly on viral replication and indirectly through activation of host immune response genes.14 The type I interferon, IFN-α, has received particular attention in the treatment of chronic HCV infection, because recombinant IFN-α is a major component of the standard treatment of HCV.15-17 The recent discovery of the type III interferon-lambda (IFN-λ) family, spurred, in large part, by the association between IL28B genotype and HCV treatment response, has opened new avenues of research into a novel mechanism of antiviral activity.18
The IFN-λs or type III IFNs bind to a unique receptor complex,19, 20 but otherwise share many functional characteristics with the type I IFNs.18 This family comprises three members, designated IL28A (IFN-λ2), IL28B (IFN- λ3), and IL29 (IFN- λ1). The nomenclature used to describe the IFN-λ family reflects their structural and functional similarity to both the interleukin family of cytokines (specifically, IL10) and the type I IFNs.20 Like type I IFN, IFN-λs have been shown to be up-regulated in the presence of viruses and double-stranded DNA and to have antiviral activity.18 The IFN-λs are distinguished from type I IFNs by their binding to a unique heterodimeric receptor complex formed by the IFN-λ-specific alpha subunit (IL28RA) and the IL-10 beta receptor subunit (IL10RB).19, 20 Binding of IFN-λ to this complex leads to activation of the Janus kinase (JAK) and protein tyrosine kinase 2, which leads subsequently to phosphorylation and activation of the signal transducer and activator of transcription (STAT) protein kinases.18 Phosphorylation of STAT proteins leads to dimerization (either as homodimers or STAT1/STAT2 heterodimers) and translocation to the nucleus, followed by downstream activation, through transcriptional activation, of a host of genes with immunomodulatory functions, called IFN-stimulated genes (ISGs).14 The precise complement of genes up-regulated by the IFN-λs is not completely known, but numbers in the hundreds.
Although the second messengers (i.e., JAK/STAT) utilized by IFN-λs are shared with the type I IFNs, the IFN-λs are known to activate other signaling pathways, including the v-akt murine thymoma viral oncogene homolog kinase and the mitogen-activated protein kinase, Jun N-terminal kinase, which are not believed to be targets of type I IFN signaling.18 It is likely that the particular antiviral properties that are specific to type III IFNs are the result of the precise second-messenger proteins that are activated by this unique family of IFNs. Additionally, the kinetics of ISG activation mediated by IFN-λs appears be distinct from IFN-α, with IFN-λ having relatively slow onset and more prolonged ISG activation in cell-culture models of HCV infection.21 Like type I IFNs, the IFN-λs have been shown to have antiviral properties both in vitro and in vivo,21-23 including activity against HCV replication, and recent work has shown that the HCV inhibitory activity of IFN-λ3 is largely dependent on signaling through the JAK/STAT pathway.24 Although it appears that IFN-λs may be less-potent inhibitors of HCV replication than IFN-α,22 the expression of IFN-λ receptors appears to be more restricted, with particularly high expression in the liver,25 which may indicate that IFN-λs may be particularly relevant to hepatotropic viruses.
Efforts to elucidate the functional mechanism behind the association between IL28B and SVR would benefit tremendously from identification of the causal variant or variants in the IL28 region, including, in particular, variants that change the structure and/or activity of the IFN-λ3 protein. To date, direct mechanistic studies of individual IL28B variants have been limited to a single study of the nonsynonymous coding variant, Lys70Arg, in a cell-culture model of HCV replication.26 Huh7.5 cells hosting a subgenomic HCV replicon were treated with recombinant IFN-λ3-70Lys and IFN-λ3-70Arg, and the two treatment conditions compared in terms of inhibition of HCV replication and induction of ISG expression. No appreciable differences in activity of the protein sequence variants of IL28B were observed. However, these results were obtained using a single experimental model over a relatively short time frame (i.e., 24 hours); therefore, the results should be interpreted narrowly and do not rule out a major role for the Lys70Arg variant in HCV treatment response. In addition to the Lys70Arg variant, several other noncoding variants exist in the IL28B region that are strongly correlated with the top-associated SNPs from GWAS. One or more of these noncoding variants may contribute mechanistically to the IL28B/SVR association, perhaps through regulation of IL28B expression.
Several groups have examined the relationship between IL28B genotype and messenger RNA (mRNA) expression of IL28B itself or expression of ISGs in different tissues. The poor-response IL28B genotype has been associated with reduced IL28 mRNA expression in whole blood in several reports2, 3; however, similar studies using peripheral blood mononuclear cells did not show such an association.1 When IL28B genotype has been associated with its mRNA expression, the magnitude of the effect has been fairly weak (approximately 30%-50% difference in mean expression level observed between good-response and poor-response genotypes). Thus, IL28B genotype may be related to IL28B mRNA expression in peripheral immune cells, though the biological relevance of this is unclear. It is possible that IL28B mRNA expression is temporally regulated and/or cell type specific, and that large differences in IL28B production by genotype may occur in particular cell populations at critical stages of infection.
In several independent studies of gene expression in liver biopsy samples from individuals chronically infected with HCV, IL28B genotype was not associated with IL-28 mRNA expression26, 27; however, the poor-response genotype was associated with generally higher expression of ISGs in the liver.26-28 High baseline ISG expression in liver tissue had previously been associated with a poorer response to treatment.29-33 It has been argued that this association between ISG expression and treatment outcome may be primarily a consequence of IL28B genotype, though this remains controversial.26-28 Nonetheless, it appears that the relationship between IL28B genotype, ISG expression, and treatment outcome is in the counterintuitive direction that the favorable host genotype and treatment outcome are associated with lower baseline ISG expression. Given that ISGs are presumed to be the final mechanism by which IFNs bring about viral clearance, this is somewhat of a paradox. One compelling explanation is that high baseline hepatic ISG expression may be a sign of a maladaptive response to infection, perhaps the result of exhaustion of the IFN pathway by suboptimal IFN-λ-mediated ISG induction. On the other hand, the relatively quiescent ISG status at baseline in treatment responders (or individuals with the good-response IL28B genotype) may render them more sensitive to the effects of pharmacologic IFN-α. Such a scenario is consistent with recent data showing that IFN-λ signaling may act as a negative regulator of IFN-α responsiveness.34 If this scenario is true, the favorable IL28B genotype might, in fact, be expected to produce lower IFN-λ activity; this, however, remains to be demonstrated.
Perhaps related to this is the similarly counterintuitive finding that the good-response IL28B genotype is associated with higher HCV viral load at baseline, whereas high viral load is generally predictive of poor treatment outcomes. It has been demonstrated through simulation studies that this relationship may be explained by a kind of selection bias, in which patients with both low baseline viral load and the good-response IL28B genotype are particularly likely to spontaneously resolve HCV infection, so that patients carrying the good-response genotype that progress to chronic infection (i.e., those ascertained in chronic HCV cohorts) are more likely to carry high viral loads, compared to poor-response IL28B genotypes.35 Whether this indeed occurs has yet to be determined and will require prospective study of HCV viral kinetics as well as immune and liver-specific responses in infected patients from the acute phase through establishment of chronic infection. Such studies will be crucial to our understanding of the effect of IL28B genotype on both spontaneous and treatment-induced clearance of HCV and may shed light on the relevance of hepatic ISG expression and peripheral IFN-λ production to HCV clearance.
Examination of the relationship between IL28B genotype and early viral kinetics may shed some light on the possible mechanisms for the genetic association; however, studies to date have shown mixed results. Several reports36-39 have suggested that the protective IL28B genotype is associated with a steeper first-phase decline (i.e., decrease in viral titer over the first several days of treatment), with a generally weaker effect on the second-phase decline (i.e., 2-28 days after treatment initiation), suggesting, per Neumann et al.,40 that the major mode of IL28B action may be on the clearance of free virus. Consistent with this, a study of Taiwanese chronic HCV patients employing a constrained version of the Neumann model suggested that the primary effect of IL28B genotype may be on the viral clearance rate41; however, others have suggested that the assumptions underlying this constrained model may be unrealistic and may complicate the interpretation.42 In contrast, a study employing a smaller sample size, but a denser sampling scheme in the initial phase, suggested that the primary IL28B effect may be on the death rate of infected hepatocytes (δ), though there was a trend toward an association between IL28B genotype and first-phase decline as well.43 A better understanding of the precise relationship between IL28B genotype and viral kinetics will require more detailed, well-powered prospective studies.
There is some evidence in favor of an interplay between IL28B and natural killer (NK) cell activity in HCV responses. A number of genetic association studies have suggested that certain killer immunoglobulin-like receptor (KIR) genes, and certain classes of human leukocyte antigen C (HLA-C) alleles, or combinations thereof, are associated with HCV outcomes, including resistance to infection,44 spontaneous resolution of infection,44-46 and treatment-induced clearance.44, 47 It has been shown that expression of the inhibitory NK receptor, NKG2A, is up-regulated on CD56dim NK cells in chronic HCV infection, and that NKG2A receptor expression is associated with greater rates of treatment response,48 and, more recently, that the expression of NKG2A is associated with IL28B genotype.49 These findings suggest that the effect of IL28B on HCV outcomes may be modified or complemented by NK cell activity. However, it should be pointed out that the associations between KIR or HLA-C and HCV outcomes have not yet been assessed using contemporary methods and standards of human genetic studies, including explicit correction for population structure and multiple testing, and thus the true significance of these findings remains to be determined.
Recently, it has been shown that the diversity of HCV nonstructural protein (NS)3/4A protease amino-acid sequence and activity in human immunodeficiency virus/HCV-coinfected individuals is associated with both treatment response and host IL28B genotype, with lower amino-acid diversity in the viral NS3/4A protease observed in individuals with the favorable IL28B genotype.50 Interestingly, when NS3/4A mutants from the most abundant quasispecies in each patient were tested for their ability to cleave the host IFN-stimulatory protein, Cardif, it was observed that patients experiencing treatment failure were more likely to carry mutants deficient in Cardif-cleaving activity. As mentioned above, the direction of this relationship is counterintuitive, because this implies that treatment responders are more likely to carry viruses that better inhibit host IFN signaling. Again, the explanation for this may be that a more quiescent endogenous IFN activation state at baseline may be preferable, allowing for a stronger response to pharmacologic IFN treatment.
One major impediment to the investigation of the mechanism for the IL28B effect on treatment response has been the absence of robust animal models of HCV infection. However, a recent study of HCV infectivity and treatment outcome in human hepatocyte chimeric mice showed a striking similarity to the results obtained in humans with HCV infection: Mice implanted with hepatocytes taken from donors carrying the favorable IL28B genotype and inoculated with HCV RNA or live genotype 1b HCV virus tended to have higher viral loads, compared to those carrying poor-response genotype51; however, mice with the favorable IL28B genotype showed a greater responsiveness to IFN-α treatment in terms of viral RNA decline and hepatic ISG expression, consistent with studies in humans.33 These results suggest that at least part of the IL28B genetic effect occurs through the liver itself, because the extrahepatic immune response in mouse recipients seems unlikely to be modified significantly by the genotype of the implanted hepatocytes, though this possibility cannot be ruled out. Studies of IL28B genotype in the setting of liver transplantation in HCV infection have shown that both donor and recipient IL28B genotypes have a significant effect on treatment outcomes.52 Although it is tempting to imagine a simple, tissue-specific mechanism for the IL28B effect on treatment response, these results collectively suggest an influence of IL28B genotype from both hepatic and extrahepatic tissues.
Among the agents in development for treatment of HCV is the pegylated formulation of human IFN-λ1, which has shown antiviral activity both in cellular models and in vivo and is currently in phase II clinical trials.21-23 Given the influence of host genetic variation in the IFN-λ pathway on treatment response, recombinant IFN-λ products represent a promising avenue in the future of HCV therapeutics and may also provide valuable information regarding the mechanism of the IL28B genetic effect.