Supplemental Material
Contents
1. Principal components analysis to evaluate population structure (page 3).
2. Q-Q plots of ADME-restricted and autoimmune-restricted GWAS including all DILI cases (page 4).
3. Sensitivity analysis of putatively associated variants in the discovery cohort (page 5).
4. Sensitivity analysis of putatively associated variants in the replication cohort (page 6).
5. Results of GWAS restricted to long latency (> median time to onset) DILI cases not due to flucloxacillin or amoxicillin/clavulanate (pages 8-9).
6. Distribution of p-values for SNPs in the extended MHC region vs. all other autosomal regions (page 10).
7. Sensitivity analysis of putative ABCC2 association with all-cause DILI (page 11).
8. Results of GWAS restricted to short latency (≤ median time to onset) DILI cases (pages 12-13).
9. Results of GWAS restricted to long latency (> median time to onset) DILI cases (pages 14-15).
12. Results of GWAS restricted to short latency (≤ median time to onset) DILI cases not due to flucloxacillin or amoxicillin/clavulanate (pages 16-17).
13. Results of GWAS restricted to early-onset (≤ median age of onset) DILI cases (pages 18-19).
14. Results of GWAS restricted to late-onset (> median age of onset) DILI cases (pages 20-21).
15. Results of GWAS restricted to early-onset (≤ median age of onset) DILI cases not due to not due to flucloxacillin or amoxicillin/clavulanate (pages 22-23).
16. Results of GWAS restricted to late-onset (> median age of onset) DILI cases not due to flucloxacillin or amoxicillin/clavulanate (pages 24-25).
17. Results of GWAS restricted to pediatric DILI cases (age < 18yr) (pages 26-27).
18. Results of GWAS including only male DILI cases vs. male controls (pages 28-29).
19. Results of GWAS including only female DILI cases vs. female controls (pages 30-31).
20. Results of GWAS restricted to DILI cases with cholestatic injury (pages 32-33).
21. Results of GWAS restricted to DILI cases with hepatocellular injury type (pages 34-35).
22. Results of GWAS restricted to DILI cases with mixed injury type (pages 36-37).
23. Results of GWAS restricted to DILI cases with cholestatic injury not due to flucloxacillin or amoxicillin/clavulanate (pages 38-39).
24. Results of GWAS restricted to DILI cases with hepatocellular injury not due to flucloxacillin or amoxicillin/clavulanate (pages 40-41).
25. Results of GWAS restricted to DILI cases with mixed injury not due to flucloxacillin or amoxicillin/clavulanate (pages 42-43).
26. Results of GWAS restricted to DILI cases presenting with eosinophilia (pages 44-45).
27. Results of GWAS restricted to DILI cases presenting with fever (pages 46-47).
28. Results of GWAS restricted to DILI cases presenting with rash (pages 48-49).
29. Results of GWAS restricted to DILI cases presenting with any extrahepatic features (eosinophilia, fever, rash) (pages 50-51).
30. Results of GWAS restricted to DILI cases of mild or moderate severity (pages 52-53).
31. Results of GWAS restricted to moderate-hospitalized, severe or fatal DILI cases (pages 54-55).
32. Results of GWAS including only non-steroidal anti-inflammatory drug (NSAID) DILI cases (pages 56-57).
33. Results of GWAS including only DILI cases due to isoniazid-containing drug regimens (pages 58-59).
34. Results of GWAS including only beta-lactam antibiotic-induced DILI cases, excluding flucloxacillin and amoxicillin/clavulanate (pages 60-61).
35. Results of GWAS including only sulfamethoxazole/trimethoprim-induced DILI cases (pages 62-63).
36. Results of GWAS including only diclofenac-induced DILI cases (pages 64-65).
37. Results of GWAS including only fluoroquinolone antibiotic-induced DILI cases (pages 66-67).
38. Results of GWAS including only macrolide antibiotic-induced DILI cases (pages 68-69).
39. Results of GWAS including only statin-induced DILI cases (pages 70-71).
40. Results of GWAS including only tetracycline antibiotic-induced DILI cases (pages 72-73).
41. Results of GWAS including only nitrofurantoin-induced DILI cases (pages 74-75).
42. Results of GWAS including only nitrofurantoin- or minocycline-induced DILI cases (pages 76-77).
43. Results of GWAS including only valproic acid-induced DILI cases (pages 78-79).
Supplemental Figure S1. Principal components analysis (PCA) shows matched population structure of DILI cases (black circles) and population controls (red circles).
Supplemental Figure S2. Quantile-quantile (QQ) plots of ADME-restricted (A) and autoimmune SNP-restricted (B) analyses of all DILI cases (n = 783), conditioned on genotypes at the known MHC SNPs associated with flucloxacillin- or amoxicillin/clavulanate-induced DILI.
Discovery Cohort (all) / Discovery Cohort (causality at least “probable” or RUCAM ≥ 6)Phenotype / SNP Set / SNP / Nearest Gene / Na / p-value / OR (95%CI) / Nb / p-value / OR
All DILI / ADME / rs3740065 / ABCC2 / 783 / 2.6 X 10-5 / 1.45 (1.22-1.72) / 622 / 2.8 X 10-5 / 1.49 (1.24 – 1.81)
Cholestatic / AI and GWAS / rs2476601 / PTNP22 / 187 / 5.4 X 10-7 / 2.20 (1.62-3.00) / 161 / 1.1 X 10-4 / 1.95 (1.39 – 2.74)
Hepatocellular / AI / rs7574865 / STAT4 / 256 / 4.5 X 10-4 / 1.45 (1.18-1.79) / 219 / 1.2 X 10-3 / 1.45 (1.16 – 1.82)
Diclofenac / GWAS / rs17036170 / PPARG / 30 / 1.0 X 10-8 / 11.3 (4.9-25.9) / 23 / 1.8 X 10-8 / 13.9 (5.57 – 34.8)
Fluoroquinolones / ADME / rs17862876 / UGT1A / 12 / 3.4 X 10-5 / 11.6 (3.6-37.4) / 8 / 4.9 X 10-6 / 23.4 (6.05 – 90.2)
NSAIDs / GWAS / rs9376256 / IL22RA2 / 68 / 6.7 X 10-8 / 3.42 (2.19-5.34) / 39 / 6.3 X 10-4 / 2.93 (1.58 – 5.42)
Statins / AI / rs7574865 / STAT4 / 27 / 4.9 X 10-5 / 3.35 (1.87-6.00) / 20 / 4.0 X 10-4 / 3.36 (1.72 – 6.57)
Supplemental Table S1. Sensitivity analysis of putatively associated variants in the discovery cohort. Inclusion of all cases results in generally similar test statistics and effect size estimates (odds ratio, OR) as analyses restricted to only cases with causality of “probable” or greater (defined as DILIN causality score of “probable”, “very likely”, or “definite”, or a RUCAM score ≥ 6).
Replication Cohort (all) / Replication Cohort (causality at least “probable”)Phenotype / SNP Set / SNP / Nearest Gene / Nb / p-value / OR / N / p-value / OR
All DILI / ADME / rs3740065 / ABCC2 / 304 / 0.49 / 0.90 (0.68-1.20) / 170 / 0.79 / 0.95 (0.66 – 1.37)
Cholestatic / AI and GWAS / rs2476601 / PTNP22 / 105 / 0.26 / 0.74 (0.44-1.24) / 52 / 0.44 / 0.75 (0.36 – 1.55)
Hepatocellular / AI / rs7574865 / STAT4 / 168 / 0.011 / 1.39 (1.08-1.78) / 74 / 0.22 / 1.26 (0.87 – 1.84)
Diclofenac / GWAS / rs17036170 / PPARG / 26 / 0.088 / 3.59 (0.83-15.6) / 22 / 0.487 / 2.05 (0.27 – 15.5)
Fluoroquinolones / ADME / rs17862876 / UGT1A / 10 / 1.00 / N/A / 4 / 1.00 / N/A
NSAIDs / GWAS / rs9376256 / IL22RA2 / 12 / 0.42 / 0.44 (0.06-3.23) / 5 / 1.00 / N/A
Statins / AI / rs7574865 / STAT4 / 16 / 0.58 / 0.77 (0.31-1.91) / 9 / 0.78 / 1.24 (0.29-5.38)
Supplemental Table S2. Sensitivity analysis of putatively associated variants in the replication cohort. Analyses restricted to only cases with causality of “probable” or greater do not improve the associations compared with tests including all cases in the replication cohort. The majority of cases without a score of probable or greater had not yet undergone causality assessment at the time of writing.
Supplemental Figure S3. Manhattan (A) and QQ (B) plots of GWAS results for analysis restricted to all DILI cases with time to onset greater than the median of 35 days, for cases not due to flucloxacillin or amoxicillin/clavulanate (n = 213). Also shown are the QQ plots from the restricted ADME (C) and autoimmune (D) SNP subsets.
SNP Set / SNP / P / Coordinates / Function / Nearest GeneGWAS / rs166580 / 7.73E-07 / 6:128158459 / INTRONIC / THEMIS
rs4672268 / 1.98E-06 / 2:59146423 / WITHIN_NON_CODING_GENE / AC007092.1
rs35709459 / 2.17E-06 / X:91516427 / INTRONIC / PCDH11X
rs7340004 / 2.93E-06 / 1:246374889 / INTRONIC / SMYD3
rs35925943 / 3.58E-06 / X:91516089 / INTRONIC / PCDH11X
ADME / rs1437135 / 0.0004 / 16:69757828 / INTRONIC / NQO1
rs1800566 / 0.0004 / 16:69745145 / NON_SYNONYMOUS_CODING / NQO1
rs4148083 / 0.0006 / 21:43620158 / INTRONIC / ABCG1
rs4507768 / 0.001 / 8:70642018 / INTRONIC / SLCO5A1
rs4148082 / 0.0011 / 21:43619415 / UPSTREAM / ABCG1
AI / rs7574865 / 0.0025 / 2:191964633 / INTRONIC / STAT4
rs229527 / 0.0082 / 22:37581485 / NON_SYNONYMOUS_CODING / C1QTNF6
rs2476601 / 0.0207 / 1:114377568 / NON_SYNONYMOUS_CODING / PTPN22
rs6679677 / 0.0225 / 1:114303808 / UPSTREAM / RSBN1
rs9388489 / 0.0246 / 6:126698719 / INTERGENIC / CENPW
Supplemental Table S3. Top hits from GWAS and SNP subset analyses, restricted to all DILI cases with time to onset greater than the median of 35 days, for cases not due to flucloxacillin or amoxicillin/clavulanate (n = 213).
Supplemental Figure S4. Distribution of p-values for SNPs in the extended MHC region vs. all other autosomal regions. (A), observed p-value distribution in the MHC region; (B), observed p-value distribution among non-MHC autosomal SNPs; (C), distribution of the median p-value in the MHC region among 1000 permutations. The median p-value in the MHC region in the observed (unpermuted) dataset was 0.381, yielding an empirical p-value for the difference in MHC association vs. expectation under the null of p=0.015.
Supplemental Table S4. Sensitivity analysis of the association between rs3740065 (near ABCC2) and risk of DILI due to any drug. Note that the association statistics remain static as the sample is reduced to the earlier-onset cases, despite lower sample size, and the estimate of effect size increases with decreased time to onset, consistent with the expectation of a pharmacokinetic risk factor for DILI.
Conditions / N / Z / P / OR (95% CI)All Drugs / 783 / 4.21 / 2.55E-05 / 1.45 (1.22 - 1.72)
Causality >= Probable / 622 / 4.19 / 2.79E-05 / 1.50 (1.24-1.81)
Causality >= Probable & onset<180d / 535 / 4.37 / 1.24E-05 / 1.55 (1.28-1.90)
Causality >= Probable & onset<90d / 501 / 4.47 / 7.82E-06 / 1.58 (1.30-1.94)
Causality >= Probable & onset<30d / 338 / 4.71 / 2.48E-06 / 1.74 (1.38-2.19)
Causality >= Probable & onset<15d / 194 / 4.15 / 3.32E-05 / 1.86 (1.39-2.49)
Causality >= Probable & onset<7d / 85 / 4.15 / 3.32E-05 / 2.37 (1.58-3.57)
Supplemental Figure S5. Manhattan (A) and QQ (B) plots of GWAS results for analysis restricted to all DILI cases with time to onset less than or equal to the median of 24 days (n = 359). Also shown are the QQ plots from the restricted ADME (C) and autoimmune (D) SNP subsets. The contribution of flucloxacillin and amoxicillin/clavulanate-DILI cases to the analysis explains the association signal in the MHC region on chromosome 6 (c.f. Figure S5).
SNP Set / SNP / P / Coordinates / Function / Nearest GeneGWAS / rs2076533 / 8.66E-11 / 6:32363527 / INTRONIC / BTNL2
rs2076529 / 1.02E-10 / 6:32363955 / SYNONYMOUS_CODING / BTNL2
rs4424066 / 1.06E-10 / 6:32354428 / UPSTREAM / AL662796.2
rs3817973 / 1.24E-10 / 6:32361111 / INTRONIC / BTNL2
rs9268473 / 2.45E-10 / 6:32355683 / UPSTREAM / AL662796.2
ADME / rs3740065 / 5.64E-06 / 10:101605693 / INTRONIC / ABCC2
rs8187710 / 0.0006 / 10:101611294 / NON_SYNONYMOUS_CODING / ABCC2
rs8187707 / 0.0008 / 10:101610533 / SYNONYMOUS_CODING / ABCC2
rs11816708 / 0.0008 / 10:101615015 / DOWNSTREAM / ABCC2
rs7904678 / 0.0008 / 10:101596851 / SYNONYMOUS_CODING / ABCC2
AI / rs2395029 / 2.46E-08 / 6:31431780 / WITHIN_NON_CODING_GENE / HCP5
rs3135388 / 7.17E-06 / 6:32413051 / DOWNSTREAM / HLA-DRA
rs9271366 / 2.55E-05 / 6:32586854 / INTERGENIC / HLA-DRB1
rs3129934 / 0.0001 / 6:32336187 / INTRONIC / C6orf10
rs3135338 / 0.0004 / 6:32401217 / INTERGENIC / HLA-DRA
Supplemental Table S5. Top hits from GWAS and SNP subset analyses, restricted to all DILI cases with time to onset less than or equal to the median of 24 days (n = 359). The contribution of flucloxacillin and amoxicillin/clavulanate-DILI cases to the analysis explains the association signal in the MHC region on chromosome 6 (c.f. Table S5).
Supplemental Figure S6. Manhattan (A) and QQ (B) plots of GWAS results for analysis restricted to all DILI cases with time to onset greater than the median of 24 days (n = 356). Also shown are the QQ plots from the restricted ADME (C) and autoimmune (D) SNP subsets.
SNP Set / SNP / P / Coordinates / Function / Nearest GeneGWAS / rs166580 / 1.21E-07 / 6:128158459 / INTRONIC / THEMIS
rs34566 / 1.15E-06 / 5:106600215 / INTERGENIC / RP11-513C4.1
rs3814258 / 1.18E-06 / 13:113915485 / INTRONIC / CUL4A
rs4672268 / 1.19E-06 / 2:59146423 / WITHIN_NON_CODING_GENE / AC007092.1
rs34564 / 1.28E-06 / 5:106599519 / INTERGENIC / RP11-513C4.1
ADME / rs12035560 / 0.0012 / 1:97641771 / INTRONIC / RP11-526F14.1
rs16864191 / 0.0013 / 1:171177416 / INTRONIC / FMO2
rs2020865 / 0.0013 / 1:171176912 / NON_SYNONYMOUS_CODING / FMO2
rs12087971 / 0.0013 / 1:171158499 / INTRONIC / FMO2
rs28369860 / 0.0013 / 1:171165803 / FRAMESHIFT_CODING / FMO2
AI / rs2395029 / 0.0003 / 6:31431780 / WITHIN_NON_CODING_GENE / HCP5
rs2476601 / 0.0028 / 1:114377568 / NON_SYNONYMOUS_CODING / PTPN22
rs6679677 / 0.0032 / 1:114303808 / UPSTREAM / RSBN1
rs9388489 / 0.0043 / 6:126698719 / INTERGENIC / CENPW
rs12928822 / 0.0053 / 16:11403893 / UPSTREAM / MIR548H2
Supplemental Table S6. Top hits from GWAS and SNP subset analyses, restricted to all DILI cases with time to onset greater than the median of 24 days (n = 356).