Genetic factors predicting response to interferon treatment for viral hepatitis

[Guest guest]

Genetic factors predicting response to interferon treatment for viral

hepatitis C

Commentaries

Gut April 2008

Stärkel

Department of Gastroenterology, St. Luc University Hospital, Av.

Hippocrate 10, 1200 Brussels, Belgium

Since completion of the human genome and HapMap projects it is likely

that an increasing number of studies dealing with genetic

associations in specific diseases will be published in medical

journals. Hepatitis C and especially its response to antiviral

treatment does not constitute an exception to this rule. It has been

known for quite a long time that viral factors such as genotype and

viral load have a major influence on the outcome of antiviral

treatment.1 Nevertheless, researchers have become increasingly aware

that host genetic factors including ethnicity, human leucocyte

antigens, cytokine production and differences in T cell immune

responses can modulate the response to antiviral treatment and viral

clearance.2 3 In this issue of Gut (see page

10.1136/gut.2007.129478), Persico et al report on single nucleotide

polymorphisms (SNPs) of suppressor of cytokine signalling 3 (SOCS3)

being positively and negatively associated with response to antiviral

therapy in hepatitis C virus (HCV) genotype-1-infected patients.4

Interestingly, these data were generated from peripheral mononuclear

blood cells (PMBCs), a material that would be easily available for

large-scale studies in the future. The concept of SOCS3 being

involved in modulating antiviral response mechanisms is appealing to

the scientific community. SOCS3 acts as a negative regulator of the

cytokine-induced JAK-STAT (Janus kinase-signal transducer and

activator of transcription) pathway. As a consequence of a classical

negative feedback circuit, SOCS3 inhibits the cytokine receptor-

associated JAK tyrosine kinase through either direct interaction with

JAKs or indirect inhibition by binding to specific sites on the

receptor.5 6 Inhibition of JAKs leads in turn to deficient STAT

phosphorylation and dimerisation required for nuclear translocation

of STATs and ultimately to blockage of transcriptional activity of

STAT-responsive genes involved in regulation of immune responses and

maintaining immunological homeostasis.6 Interferons (IFNs) are

naturally occurring proteins that induce an antiviral state in cells

by activating the JAK-STAT pathway.7 Type I IFN, mainly IFN, and the

type II IFN act synergistically and induce the expression of a large

number of interferon-stimulated genes, setting up an antiviral,

antiproliferative and immunoregulatory state in the host cells. Both

IFNs have been shown to inhibit HCV replication.8-11 However, HCV

seems to have developed strategies that interfere with the IFN

effector pathways attenuating their antiviral efficacy.12 13 In

addition, it has been shown that the HCV core protein is able to

induce SOCS3 and to suppress JAK-STAT signalling in cell culture.14

15 From this point of view, the results of Persico et al pointing to

SOCS3 and its gene polymorphisms as a modulator of antiviral

responses, in particular by downregulating IFN-dependent responses,

in vivo in humans is important since they open up new perspectives

for future research. A second paper in this issue of the journal (see

page 516) also supports the view that HCV evades the immune system by

interfering with IFN-stimulated genes. Asselah et al report that the

expression of three genes (IFI-6-16, IFI27 and ISG15) coding for IFN-

inducible proteins is upregulated in non-responders to antiviral

therapy.16 The authors further show that a two-gene signature

including one of these three genes (IFI27) predicts treatment outcome

reasonably well.

Although the association of SNPs of SOCS3 with antiviral responses

seems to be robust in the study of Persico et al,4 one has to keep in

mind that such an analysis is demanding from the statistical point of

view and bears a high risk of false-positive results principally due

to constraints in terms of number of samples.17 In particular, the

reader should interpret the findings of Asselah et al with caution as

the number of patients included in each subgroup analysis was rather

low, generating p values at the lower end of the spectrum of

significance.16 In addition, the sustained viral responder and non-

responder subgroups display substantial heterogeneity concerning

viral genotypes and fibrosis scores, both of which are known to

influence response to antiviral treatment. The paper also included a

responder-relapser subgroup. However, this group of patients did not

separate out as an independent subgroup as their gene expression

profiles were distinguishable neither from those of sustained

responders nor from those of non-responders. Moreover, the two-gene

signature analysis predicted 73% of them to be sustained responders

despite the fact that the final result is failure of antiviral

treatment. In principle, relapsers should be assimilated into the

group of non-responders if one considers treatment failure defined as

HCV-RNA positivity 6 months after cessation of antiviral treatment as

the primary end point. As a consequence, it is likely that the

overall predictive accuracy of treatment response lies below the 79%

claimed by Asselah et al if both subgroups are pooled together into

one treatment failure group. Considering all these different aspects,

it is therefore mandatory that the consistency and strength of

associations are verified in large-scale replication studies in

different sets of patients by independent research groups.

Although these studies tell us that a particular gene might be

important in the pathogenesis of a given disease, they do not tell us

anything about the links between these associations and disease

biology. Gene products are subjected to several levels of regulation

starting with its mRNA and extending to elaboration of the final

protein that might suppress, attenuate or amplify the functional

consequences of a given polymorphism. It is encumbent on the

researchers to explain how variation in the function of genes leads

to clinical disease and to unravel the mechanism that is responsible

for the clinical picture of a disease. In an attempt to address this

issue, Persico et al showed higher levels of basal SOCS3 mRNA and

protein expression in PMBCs in non-responders to IFN-based antiviral

therapy compared with responders. These results have to be treated

with caution because the analysis has been performed in a relatively

low number of patients (as few as six patients per group in some

experiments). In addition, it is not clear from the study whether the

results obtained in PMBCs can be extrapolated to the target organ-

that is, the liver, and what the basal expression of SOCS3 would be

in a " normal " liver or in liver disease that is not related to HCV.

High SOCS3 levels might be related to liver injury and repair

mechanisms independent of viral infection. Showing that SOCS3

expression is low in normal livers and subsequently increases in HCV-

infected non-responders as well as the absence of an association of

SOCS3 with liver diseases other than HCV would substantially

reinforce the link between SOCS3 and non-response to treatment.

Asselah et al did not address the functional consequences of their

findings, although it is striking that expression of IFN-stimulated

genes is induced in non-responders.16 These observations imply that,

in their study, the functional integrity of IFN signalling pathways

is preserved in non-responders, which is in contradiction to the data

reported in the study of Persico et al.4 A more thorough analysis of

how the three upregulated genes coding for the IFN-inducible proteins

2, 3 and 27 hamper IFN-based antiviral therapy would not only

considerably strengthen the data presented by Asselah et al, but

would also shed light on the multiple facets by which HCV finally

gets around antiviral defence mechanisms.

Some support for the findings reported in the study by Persico et al

comes, however, from two recent studies reporting SOCS3

overexpression in the livers of HCV-infected, obese patients not

responding to antiviral therapy and in livers of chimpanzees infected

with HCV that are resistant to type I and type II IFNs.18 19 Given

the complexity of gene regulation and the number of genes involved in

fine-tuning the IFN response to viral infection, it is likely that

other SNPs, such as, for example, those described in the IFN promoter

or SNPs that modulate T lymphocyte responses, will be found to be

associated with treatment response in HCV patients.3 20

In addition, interfering with the JAK-STAT pathway might not only

impact on antiviral defence mechanisms that principally involve STAT1

but may also disturb liver damage and repair mechanisms. STAT3, a

member of the JAK-STAT family, is also closely regulated by SOCS3,

and disruption of STAT3-mediated intracellular signal transduction

pathways could lead to disturbed liver regeneration and repair.21-24

Data in the literature suggest that HCV modulates STAT3 signalling,25

26 and in vivo studies in humans add further evidence that deficient

STAT3 signalling does contribute to liver fibrosis progression in HCV-

infected patients.27 28 It is conceivable that SNPs like those

described by Persico et al in the SOCS3 gene lead to a wider range of

functional consequences integrating blunted antiviral defence and

inadequate liver repair in response to injury.

Both studies fall short of suggesting answers to two well-known

phenomena. First, it is not clear how SNPs in the SOCS3 gene or

upregulation of IFN-inducible genes could explain the different

response rates to IFN-based therapies in patients with advanced

fibrosis or cirrhosis compared with their non-fibrotic counterparts

regardless of the genotype these patients are infected with. Even

though the model proposed by Asselah et al seems to predict

reasonably well the treatment outcome in patients with advanced

fibrosis, the suggested gene signature profile does not explain the

differences in response rates at baseline.

It is likely that both patient groups carry a similar distribution of

SNPs or gene expression profiles before starting antiviral therapy,

but they do not show the same response rates, these being lower in

patients with advanced fibrosis and/or cirrhosis. Genes or factors

other than those described in the two papers should account for the

different response rates to IFN-based therapy encountered in these

groups. Secondly, similar considerations could be applied to patients

infected with genotype 2 and 3 viruses. They respond extremely well

to IFN therapy although the proportion of patients carrying SNPs or

gene expression profiles supposed to influence antiviral responses

negatively is likely to be similar at baseline compared with genotype

1-infected patients. Detailed studies comparing responders with non-

responders in both groups, the difficult to treat fibrotic patients

and the easy to treat genotypes 2 and 3, are warranted as they might

help to identify those genes that are associated with non-response

independently from the genotype and the fibrosis status of a given

patient.

Genome-wide association studies are likely to define new directions

of research in the future that will help to better target antiviral

therapy to individual patients, with a maximum benefit for those who

are currently considered as belonging to the difficult to treat

patient groups. The challenge for researchers will be to determine

what SNPs in a given gene or what combination of SNPs or expression

profiles in a panel of genes finally leads to functional consequences

in vivo in humans and ultimately best predicts response and/or non-

response to antiviral therapy. Regarding the question of whether the

time has come to apply these findings to daily clinical practice, the

answer should be-not yet. Much basic science and clinical research is

still needed before genetics and genetic associations will enter

clinical decision making. However, studies such as that of Persico et

al constitute a step forward towards, for example, the elaboration of

SNP microarrays which might help to better stratify patients into

groups having a high, intermediate or low likelihood for responding

favourably to IFN-based antiviral therapy.

FOOTNOTES

Competing interests: None