Max Moldovan Bioinformatics Division, WEHI

moldovan@wehi.edu.au

Bioinformatics Seminar September 28, 2009

HCV research: Recent findings and future

challenges

Presentation plan

 Introduce Hepatitis C Virus and give information about treatment and related problems

 Summarize the recent findings reported in the literature

 Indicate the remaining questions to answer

Hepatitis C Virus (HCV)  Known from 1970s, but officially discovered in

1989  Around 200mln people affected worldwide  Affects liver  Often asymptomatic or with mild symptoms e.g.

fatigue, poor appetite, joint pains etc.  Chronic infection can lead to fibrosis, cirrhosis

and liver cancer  The standard, only approved and currently most

effective treatment is pegylated interferon-α plus ribavirin

HCV infection transmission by source (USA)

Source: Center for Disease Control and Prevention

Source: www.hepatitisctreatmentcenter.com

HCV infection by genotype

HCV infection by genotype

Genotype 1: 77%

HCV genotype 1 progression

HCV infection

Clearance (~20%)

Chronic HCV (~80%)

No treatment response (~50%)

Treatment response (~50%)

Source: Based on NIH information

Approved treatment

 Pegylated interferon-α plus ribavirin (PEG-IFN-α/RBV) for 24 to 48 weeks

 It is assumed that interferon mobilizes body’s natural defense against viral infection

 The mechanism of action of ribavirin is not completely understood

 Treatment leads to a number of side effects

Common side effects of PEG-IFN-α/RBV treatment

 fatigue  muscle aches  headaches  nausea and vomiting  skin irritation at the

injection site  low-grade fever

 weight loss  irritability  depression  mild bone marrow

suppression  hair loss (reversible)

Uncommon side effects of PEG-IFN-α/RBV treatment (~2%)

 autoimmune disease (especially thyroid disease)  severe bacterial infections  marked thrombocytopenia  marked neutropenia  seizures  depression and suicidal ideation or attempts  retinopathy (microhemorrhages)  hearing loss and tinnitus

The problem summary

 The viral infection (HCV) potentially leading to life-threatening liver damage

 There is only one approved treatment (PEG-IFN-α/RBV) with a number of side effects

 Only about 50% of infected people respond to treatment

 It is not known who is gong to respond

Genome-wide association study

(GWAS)  Among chronic HCV genotype 1 affected

individuals, treatment non-responders are taken as cases and responders as controls

 Individuals are genotyped using a high throughput technology i.e. SNP Chips

 Genotypes are assessed with respect to association with the case-control status

Expected GWAS outcomes

 Identification of a genetic marker or a set of genetic markers truly associated with the phenotype

 This can point to markers specific to cases (controls) and further assist with the case-control status prediction

 The location of detected markers can point to specific genes and potentially reveal underlying biological mechanisms

Recent findings reported in the literature within a single month

Nature: online 16th of August, 2009

Nature Genetics online 13th of September, 2009

Comparison of the three GWA studies

Study Ancestry (sample size)

Genotyping platform

Case/Control Associated SNPs

Ge et al. Cauc/Afric/Hisp (N=1615)

Illumina 610-Quad

R/NR rs12979860 (OR=3.10)

Suppiah et al. Cauc (N=293) Illumina CNV370-Quad

R/NR rs8099917 (OR=1.98)

Tanaka et al. Jap (N=154) Affymetrix 6.0 NVR/VR rs8099917 (OR=12.10)

R/NR – sustained virological response/no sustained virological response NVR – null virological responders VR – viralogical responders: subject who respond to treatment, but do not necessary clear the virus)

Ge et al. and Suppiah et al. case-control split

Case NO-SVR

Control SVR

SVR – Sustained Virological Response: Absence of HCV RNA in blood 6 months after treatment

Tanaka et al. case-control split

Case NO-SVR

Control SVR

TVR – Transient Virological Response: Substantial reduction but not absence of HCV RNA in blood

Tanaka et al. case-control split

Case NVR

Control TVR+SVR

NVR – Null Virological Response: No reduction of HCV RNA in blood

The main common finding:

Source: Tanaka et al. Nature Genet

rs8099917

rs12980275

LD structure of IL28B genomic region

Source: Suppiah et al. Nature Genet

LD structure of IL28B genomic region

Source: Suppiah et al. Nature Genet

rs12979860

Ge et al. specific findings  Rates of treatment response vary across

populations together with frequency of specific genotypes

Suppiah et al. specific findings  Expression levels of IL28A and IL28B in

healthy individuals vary with genotype frequencies at rs8099917

Tanaka et al. specific findings

 Alternative study design: NULL virological responders (cases) vs. virological responders (controls)

 This more distinct phenotypical discordance leads to much higher odds ratios

IL28B  Location: Chromosome 19, q13.13 (44426112

to 44427451: 1339 bp)  Protein: interferon-λ3 (interleukin 28B)  One of three genes (the other two are IL28A and

IL29) known as type III of λ interferons  Type III interferons are shown to be unregulated

by viral infection and other interferons  IFN-λ has effects similar to IFN-α, but more

selective (i.e. it can produce less side effects)

The gap between genetic information and medical innovation Size of GenBank database

The number of new SNPs

The number of new drug applications submitted to FDA

The number of new drugs approved by FDA

Source: Nature

Remaining challenges  Precise identification of antiviral

mechanism (mostly biological challenge)

 Accurate prediction of PEG-IFN-α/RBV treatment response

Pharmacogenomics

 Composite discipline covering the range of applied biomedical research areas, from bioinformatics to molecular chemistry

 Opens entirely new opportunities such as treatment personalisation and genomic information assisted clinical trials

 Attractive from the clinical research point of view, as well as from the financial viability point of view (e.g. given $10bln size of HCV treatment market)

Identification of antiviral mechanism

 This is achieved mainly through a series of biological experiments

 Likely to lead to new treatments  Even if the discovery process is successful, the

way to clinical practice acceptance can take years

Prediction of PEG-IFN-α/RBV treatment response

Types of relevant predictors:

 Genomic information (SNPs, CNVs etc.)  Clinical baseline factors (age, BMI, viral load

etc.)  Unknown/unmeasurable environmental factors

e.g. stress-level

Approaches to building a clinically functional prediction model

 More sensitive statistical significance testing procedures, e.g. efficiency robust tests (to identify additional association signals missed otherwise)

 Identification of non-genetic predictive variables e.g. age, BMI etc. (account for confounding!)

 Finite sample model selection procedure e.g. penalised regressions and cross-validation

 Biologically motivated variable selection  Out-of-sample validation

Work in progress

Bioinformatics, WEHI Melanie Bahlo

AGRF Rust Turakulov

Math and Stats, UniMelb Hugh Miller

MBS, UniMelb Chris Lloyd

Acknowledgments

Millenium Institute & Westmead Children’s Hospital, Sydney

Vijay Suppiah

David Booth

Jacob George

Funding

ARC Linkage Grant