Transcript Use of GWAS and exome sequencing to identify genes relevant to

Use of GWAS and exome sequencing to
identify genes relevant to drug-induced
liver injury
Ann K. Daly
Institute of Cellular Medicine
Newcastle University
Idiosyncratic adverse drug reations
 Serious health problem which is expensive for society


Common cause of death
Loss of new drugs late in development or soon after licensing
 Typical incidence 1 in 10,000 to 100,000 patients exposed
 Local drug concentration may contribute but concentrationindependent effects important

Contribution from immune system in many but not all cases
 Examples




Drug-induced liver injury
Hypersensitivity reactions and skin rash
Cardiotoxicity
Muscle toxicity
Drug-induced toxicities associated with 28
drugs withdrawn from the US market 1976
to 2005
Range of drugs give rise to idiosyncratic DILI but due to
value of drugs and rarity of problem are still used widely
Wilke et al. Nature Reviews Drug Discovery 2007; 6, 904-916
Drug-induced liver injury (DILI)
 Rare but serious idiosyncratic toxicity
 Several phenotypes
 Cholestatic
 Effects on bile secretion
 Hepatocellular
 Inflammatory disease
 Aims
 Develop strategies to prevent DILI reactions by genotyping
patients prior to treatment
 Develop screening approaches for use during drug
development
UK-wide study on drug-induced liver injury
 Aim:
 Find genes predisposing to DILI
 Drugs
 Anti-TB medication or Flucloxacillin or Amoxicillin-clavulanate
 Multicentre collaboration involving
Newcastle, Liverpool, Nottingham and
London
 UK-wide recruitment of cases
 Retrospective and prospective
study
Worldwide study on DILI genetics-Ann Daly and Guru Aithal-co-chairs
Genome-wide association study
(GWAS) on flucloxacillin DILI
51 UK cases (DILIGEN study) and population
control group (n=282)
Illumina IM chip
No previous genetic studies on this form of
DILI
GWAS design allowed for an open study
Manhattan plot for flucloxacillin
GWAS
!
MHC region
Strongest signal with SNP in
HCP5 gene which tags
Daly et al., Nature Genetics 2009; 41:816-822.
HLA-B*57:01
Flucloxacillin DILI and B*57:01
 Association with HLA B*57:01 similar to that for
abacavir hypersensitivity
 83% of 150 cases now studied are B*57:01positive
 No indication that other forms of DILI show this
particular association
 Sensitivity of B*57:01 genotyping lower as
predictor of flucloxacillin DILI than abacavir
hypersensitivity
 Other genetic and non-genetic risk factors may
contribute
HLA genes-common thread in
serious adverse drug reactions
 HLA class I genes expressed on
most cells
 Proteins present antigens
(particularly peptides) to CD8positive cytotoxic T cells
 A, B and C genes
 HLA class II genes expressed on
antigen presenting cells
 Proteins present antigens to T cellsmainly CD4-positive T helper cells
 DR, DQ, DP genes
Amoxicillin-clavulanate-induced liver
injury
 Important cause of DILI worldwide
 Several small independent candidate gene
studies showed role for HLA class II
DRB1*15:01 allele in susceptibility
 GWAS involving international collaboration
performed to identify additional risk factors
Amoxicillin-clavulanate DILI
GWAS
MHC region
201 cases- UK DILIGEN (n=77), US DILIN (n=56), Spanish DILI (n=49),
EUDRAGENE (n=19); 532 matched controls
rs9274407
rs3135388 (DRB1*15:01)
rs2523822 (A*02:01)
MHC zoom-in
(Lucena et al., Gastroenterology 2011)
MHC zoom-in, conditioned on the
top SNP from class II (rs9274407)
HLA typing and amoxicillinclavulanate DILI
 Detailed HLA typing performed on 177 cases and 219
controls
 DRB1*15:01 and A*02:01 alleles associated
independently with risk of DILI (OR approx. 2.5 for
each)
 Additional risk also seen from individual A*02:01 and
DRB1*15:01 alleles (not on same haplotype)
 If positive for both
 OR= 9.47 (3.67-24.5), p=2.56 x 10-7
 Evidence for genetic interaction (p=0.0015)
HLA associations with DILI
Compound
No of cases
HLA allele
Odds ratio (95% p value
CI)
Flucloxacillin
51
B*57:01
80.6(22.8-284.9)
9x10-19
Amoxicillinclavulanate
201
A*02:01
2.3(1.8-2.9)
1.8x10-10
DRB1*15:01DQB1*06:02
2.8(2.1-3.8)
3.5x10-11
Lumiracoxib
41
DRB1*15:01DQB1*06:02
5.0(3.6-7.0)
6.8x10-25
Lapatinib
35
DRB1*07:01DQA1*02:01
2.9(1.3-6.6)
0.007
Ximelagatran
74
DRB1*07:01DQA1*02:01
4.4(2.2-8.9)
6x10-6
Daly, Annual Review of Pharmacology & Toxicology 2012; 52:21-35
Chemical structures and HLA
toxicity associations
DRB1*07:01-DQA1*02:01
Lapatanib
Ximelagatran
DRB1*15:01
Clavulanic acid
Lumiracoxib
B*57:01
Flucloxacillin
Abacavir
HLA and related gene associations
with DILI-underlying mechanism
 Specific HLA protein interacts with drug or
metabolite inappropriately
 Triggers T-cell response
 Local cellular damage
Abacavir-B*57:01 interaction
From Illing et al., Nature 2012
Effect of different drugs on peptide
binding in B*57:01-positive cell line
• No apparent direct effect by flucloxacillin unlike abacavir
• Different mechanism for antigen presentation to abacavir?
Norcross et al., 2012; AIDS 26: F21-F29
CD8-positive T cell clones from B*57:01positive flucloxacillin DILI cases: response
to various penicillins
 Need to incubate with flucloxacillin for 24 to 48 hours to
see cell proliferation
o Covalent adduct formed
o Hapten mechanism likely
Monshi et al., Hepatology 2013
 Very recent findings suggest possible direct effect of
flucloxacillin in some B*57:01-positive healthy volunteers
o No covalent adduct
Wuillemin et al., J Immunol, 2013
Is metabolism of flucloxacillin
relevant to DILI reactions?
 Flucloxacillin is a PXR ligand so may induce its own
metabolism?
 Induction will also affect bile acid synthesis and transporter
levels
 Our attempts to demonstrate that CYP3A4 or other
CYPs can generate 5'-hydroxymethyl metabolite
unsuccessful
•
•
Human liver microsomes with 1 mM flucloxacillin
Overlay of 5'-hydroxymethylflucloxacillin standard
Studies on pooled DILI cases
 DILI due to any prescribed drug
 Cases from UK (Diligen), Spain (Spanish
DILI) and US (DILIN)
 Total number of cases =783
All drug DILI GWAS results
All
Without
Fluclox/amoxicilli
n-clavulanate
All with
HLA taken out
• 783 cases, 3,001 population controls
• After correcting for known HLA risk alleles for flucloxacillinand amoxicillin-clavulanate-DILI, no deviation from expected
p-value distribution
• No SNPs in HLA region approaching genome-wide
significance for "other drugs" Urban et al, 2013 Pharmacogenetics and Genomics
Diclofenac-induced liver injury

Rare toxicity associated with diclofenac
use


Serious hepatotoxicity 11 per 100,000
patients
Believed to be due to metabolic
idiosyncracy

?Specific reactive metabolite which binds
covalently to proteins
DILI and related NSAIDs
Lumiracoxib
HLA-DRB1*15:01
Diclofenac
No significant HLA class II
association
• Both drugs metabolized by broadly similar pathways
Candidate gene studies on genetic
susceptibility to diclofenac-related DILI

UGT2B7*2 and ABCC2 C-24T previously
identified as significant risk factors in small
(n=24) candidate gene study


Daly et al. Gastroenterology 2007; 132:272-281.
Majority of these cases plus additional
cases from UK/US included in "all drug"
GWAS
GWAS on diclofenac DILI cases
(n=30)
GWAS QQ
Urban et al, 2013 Pharmacogenetics and Genomics
ADME gene QQ
GWAS hits on diclofenac DILI cases and
data on ADME genes as candidates
SNP
Set
SN P
P
GWAS
rs170 36170
ADME
Fu nction
Ne arest Gene
1.04 E-08 3: 1233 0411
5 PRIME_UTR
PPARG
rs6565 872
1.14 E-07 18 :739 63655
I NTERGENIC
ZNF51 6
rs4891 153
3.10 E-07 18 :739 56235
I NTERGENIC
ZNF51 6
rs7904 33
4.49 E-07 12 :879 78503
I NTERGENIC
AC0 79598 .1
rs2918 107
5.05 E-07 10 :133 07495 4 I NTRONIC
TCERG1 L
rs9034 46
0 .0006 4: 7033 6501
I NTERGENIC
UG T2B4
rs458 7017
0 .0008 4: 6994 7398
UPSTREAM
UGT2B7
rs4772 131
0 .0012 13 :993 97536
I NTRONIC
SLC15A1 (PEPT1 )
rs152 3130
0.002 3: 1194 99507
5 PRIME_UTR
NR1 I2 (PXR)
I NTRONIC
UGT2B7
rs737 5178
•
•
Coord inates
0 .0024 4: 6996 9679
PPAR-gamma association did not replicate in 26 additional cases
ADME associations in agreement with our earlier UGT2B7 finding but
PXR an interesting additional candidate
Exome sequencing and exome chip
studies on DILI
 Exome sequencing performed on 125
amoxicillin-clavulanate DILI cases
 66 from UK and 59 from Spain
 All from previous GWAS
 Exome chip studies
 112 further amoxicillin-clavulanate DILI cases
from iDILIC study and 93 from US DILIN
network for replication of exome sequencing
 73 UK flucloxacillin DILI cases
Controls for exome
sequencing/exome chip studies
 Exome sequencing
 353 NIMH (white US) plus 102 from Spain
(EGA)
 Exome chip
 3900 controls from several cohorts
 Combined studies
 4300 controls from European sequencing
project (EVS)
Amoxicillin-clavulanate: best results for
sequencing plus exome chip with
combined data also shown
SEQ_MAF_ SEQ_MAF_
SEQ_O EC_MAF_
A
U
SEQ_P
R
A
EC_MAF_U EC_P
EC_O
R
GENE
FUNCT
PPHEN
1.4E-02
0.2
ZNF304
missense_variant
0.030
3.7E-06 0.1
0.000
1.4E-02
19.1
SGSM3
splice_region_variant
benign
possibly_damagin
g
0.000
1.3E-05 27.0
0.107
0.062
6.6E-04
1.8
LAMB1
missense_variant
probably_damaging
0.068
4.0E-05 1.8
1.8
0.135
0.084
5.5E-04
1.7
MTHFS
missense_variant
benign
0.089
5.1E-05 1.7
7.3E-02
2.5
0.024
0.011
1.6E-02
2.3
C14orf101
missense_variant
benign
0.006
1.2E-04 3.1
0.287
3.2E-02
1.4
0.366
0.290
3.0E-03
1.4
PCDH15
missense_variant
NA
0.295
1.2E-04 1.4
0.411
0.493
2.8E-02
0.7
0.424
0.498
9.2E-03
1.3
NA
NA
NA
0.491
1.7E-04 0.7
0.008
0
4.2E-02
NA
0.002
0.000
5.0E-02
NA
WDR75
missense_variant
possibly_damaging
0.000
1.8E-04 80.8
77329417
0.008
0
4.2E-02
NA
0.005
0.000
7.2E-03
38.3
PSTPIP1
missense_variant
possibly_damaging
0.000
1.9E-04 21.3
0.415
0.356
9.6E-02
1.3
0.432
0.349
1.7E-04
1.5
ERAP1
missense_variant
benign
0.357
1.9E-04 1.4
6
96124330
12154443
2
0.004
0
2.1E-01
NA
0.005
0.000
2.5E-03
NA
C6orf170
missense_variant
benign
0.000
1.9E-04 78.7
5
96129512
0.267
0.226
2.0E-01
1.2
0.293
0.217
8.6E-05
1.6
ERAP1
synonymous_variant
NA
0.220
2.0E-04 1.4
15
42054450
0.008
0
4.2E-02
NA
0.005
0.000
1.4E-02
19.1
MGA
missense_variant
benign
0.000
2.0E-04 21.1
8
8748032
0.013
0
8.6E-03
NA
0.012
0.002
2.5E-03
6.4
MFHAS1
missense_variant
possibly_damaging
0.003
2.8E-04 5.4
5
82491674
0.072
0.034
1.6E-02
2.2
0.073
0.043
1.3E-03
1.9
XRCC4
missense_variant
possibly_damaging
0.041
3.0E-04 1.8
2
95947085
0.174
0.223
1.1E-01
0.7
0.176
0.226
6.2E-03
0.7
PROM2
missense_variant
benign
0.245
3.1E-04 0.7
2
73786188
0.008
0
4.2E-02
NA
0.005
0.001
3.3E-02
9.6
ALMS1
missense_variant
possibly_damaging
0.000
3.2E-04 17.6
17
34893326
0.021
0.068
4.6E-03
0.3
0.029
0.061
1.8E-02
0.5
MYO19
upstream_gene_variant
NA
0.053
3.3E-04 0.4
17
72768191
0.008
0.002
1.9E-01
3.9
0.005
0.000
2.5E-03
NA
SLC9A3R1 downstream_gene_variant
NA
0.001
4.5E-04 15.4
16
77465249
0.004
0
2.1E-01
NA
0.010
0.001
2.2E-03
9.6
ADAMTS18
benign
0.001
4.5E-04 9.6
CHR
BP
19
57868483
0
0.021
2.0E-02
0
0.005
0.027
22
0.013
0.002
6.3E-02
5.8
0.005
7
40803845
10759980
6
0.114
0.075
6.3E-02
1.6
15
80137560
0.136
0.081
1.6E-02
14
57092112
0.030
0.012
10
55566406
0.360
15
2
27890491
19033118
0
15
5
missense_variant
SGSM3: Small G Protein Signaling Modulator 3
EVS_MA
F
P
OR
Amoxicillin-clavulanate-exome chip
only
SNP
CHR
exm-IND5-149355072 5
exm813093
10
exm1621613
22
exm2267521
13
exm748945
9
exm1074032
13
exm198643
2
exm177347
2
exm1026884
12
exm2270519
6
exm941422
11
exm2269450
3
exm-IND19-62843150 19
exm1564814
exm848765
exm681525
exm1389227
BP
149374880
18439900
50687883
51133655
35618065
86368449
68622914
25141529
94965488
128709493
74883577
53282188
58151339
MAF_A
0.505
0.015
0.010
0.320
0.015
0.029
0.193
0.010
0.012
0.129
0.156
0.317
0.498
21 31587859 0.090
10 101667814 0.068
8 10469846 0.007
18 56400644 0.007
MAF_U
0.398
0.001
0.000
0.445
0.001
0.006
0.283
0.000
0.001
0.077
0.097
0.417
0.398
P
2.3E-06
5.4E-06
6.1E-06
6.8E-06
1.1E-05
2.2E-05
2.9E-05
2.9E-05
3.2E-05
4.1E-05
5.2E-05
6.4E-05
6.4E-05
0.049
0.029
0.000
0.000
9.6E-05
1.2E-04
1.2E-04
1.2E-04
OR
GENE
FUNCTION
POLYPHEN
1.6
NA
NA
NA
23.3 CACNB2
missense_variant
benign
NA MAPK12 downstream_gene_variant
NA
0.6
NA
NA
NA
19.3
CD72
missense_variant
benign
5.1 SLITRK6
missense_variant
possibly_damaging
0.6
PLEK
missense_variant
benign
76.9 ADCY3
missense_variant
possibly_damaging
24.1 TMCC3
missense_variant
probably_damaging
1.9
PTPRK
intron_variant
NA
1.8 SLCO2B1
missense_variant
benign
0.6
TKT
intron_variant
NA
1.5
NA
NA
NA
D21S2091
2.0
E
upstream_gene_variant
NA
2.3
DNMBP
missense_variant
benign
NA
RP1L1
missense_variant
probably_damaging
NA
MALT1
missense_variant
probably_damaging
CACNB2 not well covered in exome sequencing
Flucloxacillin – Exome chip
SNP
exm518448
exm1161045
exm-rs1497546
exm2255202
exm-rs4984390
exm-rs6582630
exm305107
exm781146
exm1329395
exm1223806
exm1223776
exm359484
exm359277
exm1588866
exm1588898
exm1588914
exm952418
CHR
6
15
3
2
15
12
3
9
17
16
16
3
3
22
22
22
11
BP
17602870
50785016
98034526
241516094
94939508
38743508
43602803
127618776
42927723
20966362
20959918
151166058
151161112
21989142
21990823
21991120
104879658
P
7.26E-26
1.46E-23
5.36E-06
8.85E-06
1.77E-05
6.09E-05
6.71E-05
8.72E-05
9.23E-05
9.65E-05
0.0001076
0.0001385
0.0001416
0.0001429
0.0001429
0.0001429
0.000163
OR
20.77
11.49
4.326
4.441
0.3766
2.091
0.4259
19.75
3.943
2.657
3.084
3.397
3.39
4.222
4.222
4.222
3.219
F-A
0.2571
0.5071
0.1071
0.09286
0.1786
0.6286
0.2214
0.02143
0.08571
0.1786
0.1214
0.09286
0.09286
0.07143
0.07143
0.07143
0.1071
F-U
A1
0.02458 A
0.1445
G
0.02752 A
0.02424 C
0.3613
G
0.4595
G
0.4029
C
0.003151 A
0.025
A
0.07941 C
0.05063 A
0.03592 A
0.03613 G
0.02248 A
0.02248 A
0.02248 A
0.03765 A
GENE
FUNCTION (PPH)
NA
NA (NA)
USP8
missense (probably-damaging)
NA
NA (NA)
RNPEPL1 coding-synonymous (unknown)
NA
NA (NA)
NA
NA (NA)
ANO10
missense (benign)
WDR38
missense (benign)
EFTUD2,HIGD missense (possibly-damaging)
DNAH3
missense (probably-damaging)
DNAH3
missense (probably-damaging)
IGSF10
missense (benign)
IGSF10
missense (benign)
CCDC116 missense (probably-damaging)
CCDC116
missense (benign)
CCDC116
missense (benign)
CASP5
missense (possibly-damaging)
• USP8: ubiquitin specific protease 8
•
• Chromosome 6 includes processed pseudogene with strong homology so
finding may due to LD with B*57:01
CASP5-caspase 5
• Replication studies on this looking positive
Sequencing and exome chip studies on
DILI: progress so far and where next?
 Studies show little evidence that rare coding
variants are consistently contributing to risk of
DILI due to amoxicillin-clavulanate and
flucloxacillin
 Findings fairly similar to those for other complex
diseases
 Moving towards whole genome sequencing is
challenging but may be valuable
 Further GWAS involving new sample collection
with improved power may identify novel
associations
Successfully adjudicated new cases
in iDILIC to date
Co-amoxiclav
Flucloxacillin
Statins
Diclofenac
Anti-TB drugs
Other NSAIDs
Thiopurines
Nitrofurantoin
Other Penicillins
Quinolones
Macrolides
Anti-TNF agents
Thiourylenes
Estrogen containing drugs
Terbinafine
Co-trimoxazole
Cephalosporins
Carbamazepine
Flupirtin
Proton pump inhibitors
Disulphiram
Sartans
Venlafaxine
Valproate
Sertraline
Others
0
50
100
150
Number of cases
200
250
 Cases mainly from Europe
 GWAS-exome chip in progress on 747 of these cases
 Agreed collaboration with DILIN may improve power
Summary
 Genomics has increased understanding of individual risk for
idiosyncratic drug-induced liver injury
 HLA associations

Strong association for flucloxacillin toxicity with HLA-B*57:01



Underlying mechanism could involve inappropriate T cell response
Probably does not involve metabolism
HLA genotype not predictor of DILI with all hepatotoxic drugs
 Increasing evidence that genetically-determined metabolism to reactive
intermediate is an important step in some idiosyncratic DILI reactions

Diclofenac
 Attempts to identify additional risk factors by exome sequencing
largely unsuccessful to date
 Larger sample collections for individual drugs and genome
sequencing may provide more complete understanding and
better application to drug development process
Acknowledgements
Newcastle
Chris Day
Pete Donaldson
Julian Leathart
Julia Patch
Ruth Wake
Pallav Bhatnagar
Heather Cordell
Tom Chamberlain
Sally Coulthard
Guru Aithal (Nottingham)
Munir Pirmohamed (Liverpool)
Kevin Park (Liverpool)
Dean Naisbitt (Liverpool)
Yufeng Shen (Columbia)
Aris Floratos (Columbia)
Mark Daly (Harvard)
Jackie Goldstein (Harvard)
Matt Nelson (GSK)
Paul Watkins (North Carolina)
Tom Urban (Duke)
Maribel Lucena (Malaga)
Raul Andrade (Malaga)
Mariam Molokhia (London)
Supported by the UK Department of Health
Pharmacogenetics Initiative and International Serious
Adverse Events Consortium
Cumulative incidence of ALT elevations in
lapatinib-plus-letrozole treatment group for HLADQA1*02:01 carrier and noncarrier subsets,
compared with letrozole-only treatment group
Spraggs C F et al. JCO 2011;29:667-673
Human diclofenac toxicity: possible
relationship to metabolism
Liver injury
Covalent adducts
Other oxidative metabolites
Diclofenac
acylglucuronide
MRP2
UGT2B7
Various
CYPs
Diclofenac
CYP2C9
4'-OH-diclofenac
Biliary excretion
Abacavir hypersensitivity-success
story for pharmacogenomic screening
 Abacavir is a reverse-transcriptase inhibitor
 AIDS treatment
 Approx. 5% of patients develop hypersensitivity
reactions which resolve on discontinuation
 Rechallenge results in more severe reaction
 Up to 100% of proven hypersensitivity cases
have one or two HLA B*57:01 alleles
 Not all patients with this genotype will show
detectable reaction