Idiosyncratic DILI: It’s in the Genome

Matthew R. Nelson

Why DILI Idiosyncrasy? The Immune System and BeyondSilver Spring, MD March 14, 2012

Many factors may influence iDILI risk

Environment

GeneticsAge

Dose

Compliance

Diet

Disease

OtherMeds

Image from Ramachandran R and Kakar, (2009) J Clin Pathol 62:481-492

The age-old question

Genes Env

Questions

What do we know about the contribution of genetics to iDILI risk?

How does what we know inform what we can learn and the studies needed?

Can we estimate the genetic contribution to iDILI?

From Daly et al. (2009) Nat. Genet. 41:816-9 5

Combined DILIN/iSAEC cross-drug analysis783 cases, 3001 controls

Key findings

– No major cross-drug risk factors– STAT4 associated with hepatocellular DILI

(OR = 1.4, p = 1.5×10-5)

From Urban et al., submitted

DILI genetic risk factors

Drug  Genetic Risk Factor

Prevalence Risk Allele Freq. Rel. Risk Convincing risk factorsCross drug <0.001 STAT4 rs7574865 0.24 1.4Ximelagatran  0.08 HLA-DRB1*07:01 0.08 4Augmentin <0.001 HLA-DRB1*15:01 0.15 4

HLA-A*02:01 0.27 3PTPN22 R620W 0.12 2

Lapatinib  0.09 HLA-DRB1*07:01 0.08 9Lumiracoxib  0.013 HLA-DRB1*15:01 0.15 13Ticlopidine  <0.001 HLA-A*33:03 0.14 36Flucloxacillin  <0.001 HLA-B*57:01 0.04 81

Hyperbilirubinemia risk factorsPazopanib 0.12 UGT1A1*28 0.3 13Tranilast  0.12 UGT1A1*28 0.3 48

Unconfirmed/suspect risk factorsIsoniazid CYP2E1*1 & NAT2

GSTM1 & GSTT1 NullTroglitazone GSTM1 & GSTT1 NullDiclofenac UGT2B7*2

ABCC2 C-24TIL4 C-590AIL10 C-627A

Tacrine IL6

The genetic effects that we knowDisease-related associations in NHGRI GWAS Catalog

From http://www.genome.gov/gwastudies

The genetic effects that we knowPharmacogenetic effects

We are largely limited to discovering major iDILI risk factors

1001K10K100KN =

α = 5×10-8, N Cases = Controls, log-additive OR

HeritabilityThe genetic contribution to phenotypic variation

Narrow-sense heritability

Broad-sense heritability

P

A

VVh 2

P

IDA

VVVVH

2

Birth weightBMI, age 20-29BMI, age 30-39

Height (clinically measured)Age at menarche

Age at menopauseMale baldness

Alanine aminotransferaseAspartate aminotransferase

ApoEInsulin concentration

Total cholesterolTriglycerides

Diastolic blood pressureSystolic blood pressure

Alcohol consumptionCloninger: Harm avoidance

Cloninger: PersistenceVoting behavior

Major depressive disorder

Hay feverLiability to asthma

Osteoarthritis in womenSusceptibility to Migraine

0 10 20 30 40 50 60 70 80 90 100

Heritability (H2)**From Hill et al. 2008, PLoS Genet. 4:e10000008via Zuk et al. 2012, PNAS, 109:1193

Estimating heritability via familial resemblance

rMZ and rDZ are the phenotypic correlations between MZ and DZ twin pairs

H2 = 2( rMZ ― rDZ )

Broad sense!

Estimating heritability viafamilial resemblance

From Visscher et al., Nat Rev Genet (2008) 9:255-66http://powerofthegene.com/joomla/index.php/genetic-inheritance/recessive

h2 = b

Estimating heritability in population samples with genome-wide data

×

Use the genomic estimates of genetic relatedness to derive an estimate of heritability (h2) from the joint genotypic and phenotypic resemblance.

From Yang et al. (2010) Nat Genet 42:565-71

Examples of variance explained by genetics

Trait N h2 (%)Visscherh2 (s.e., %) GWAS (%)

Height 11,576 80-90 42 (3.0) ~10BMI 11,558 42-80 16 (2.9) ~1.5vWF 6,641 66-75 25 (5.1) ~13QTi 6,567 37-60 17 (5.2) ~7

From Yang et al. (2011) Nat Genet 43:519-27

iDILI heritability estimates (Visscher)

From Casey Overby and Yufeng Shen, in preparation

Opportunities in iSAEC phase 2

iSAEC DILI consortium (IDILIC) expanding case collection

– Range of case recruitment strategies

–Academic networks

–HMOs

–Hospital-based EHRs

– >2,000 DILI cases

– Multi-ethnic

–Includes Chinese centers

Drug-specific sample sizes to identify major risk factors

Large overall sample to dip further into more modest cross-drug risk factors

Drug/ClassPhase I Cases

Phase I & 2 Cases

Other 52 725Co-amoxiclav 176 448Flucloxacillin 75 219

TB 29 202Thiopurine 6 128

Zileuton 71 71Statin 8 81

Diclofenac 27 74Ibuprofen etc. 44 55

Nimesulide 15 47Nitrofurantoin 6 36

Total 509 2,086

Conclusions

Several drug-specific risk factors known

–Largely limited to adaptive and innate immune response–ADME-related associations tenuous

They do not explain all of the DILI risk for their respective drugsSome drugs probably don’t have major common genetic risk factorsHeritability due to common variants for Augmentin and flucloxacillin

appears modest (25-35%)

–Rare variants may be responsible for some additional fraction of the risk

Genetic risk factors are important at least for some forms of DILI

–May be necessary for true DILIHowever, in most instances, non-genetic factors are important