nature publishing group
Previous studies have shown that individuals with single-nucle-
otide polymorphisms (SNPs) in genes encoding nonsteroidal
anti-inflammatory drug (NSAID)-metabolizing enzymes have a
higher risk of developing peptic ulcer disease (PUD) and/or upper
gastrointestinal bleeding (UGIB), but results have been conflict-
ing.1–6 A recent systematic review7 concluded that it is currently
not possible to assess whether there is an interaction between
exposure to NSAIDs and possession of coding variants in the
main NSAID-metabolizing enzymes, cytochrome P450 2C9, and
2C8 (CYP2C9 and CYP2C8), or whether such variants increase
the risk of developing gastrointestinal disorders independently.
The main limitations of studies undertaken to date have included
selection bias, small sample sizes, inadequate phenotyping, and
lack of assessment of the whole CYP2C gene cluster. A major limi-
tation of all the studies included in the systematic review was the
failure to include a control group unexposed to NSAIDs, which
made it impossible to determine whether any increased risk seen
was a direct effect of the variants on ulcer pathogenesis, or a result
of an interaction of the variants with NSAIDs.7
The CYP2C subfamily of enzymes comprises 20% of hepatic
CYP450 content and metabolizes 25–30% of clinically used
drugs (such as NSAIDs, proton pump inhibitors (PPIs), antide-
pressants, benzodiazepines, and clopidogrel).8 Importantly, the
CYP2C enzymes also metabolize endogenous substances such
as arachidonic acid and estrogens.9 There are four members—
CYP2C8, CYP2C9, CYP2C18, and CYP2C19—all of which have
numerous SNPs with varying frequencies in different ethnic
populations (see Supplementary Data online for those with
minor allele frequency >1%).10 However, CYP2C18 is consid-
ered clinically unimportant in humans11 (Figure 1). To date,
the Human Cytochrome P450 Allele Nomenclature Committee
has listed 14 CYP2C8, 35 CYP2C9, and 28 CYP2C19 coding
SNPs (http://www.cypalleles.ki.se/). Several of these are clini-
cally important (see Supplementary Data online), and some
have been shown to be in extensive linkage disequilibrium (LD).
For instance, 90% of Caucasian subjects with CYP2C8*3 also
carry CYP2C9*2, and ~85% of those with CYP2C9*2 also carry
CYP2C8*3.12 Similarly, CYP2C19*17 has been observed to be
in LD with wild-type CYP2C8 and CYP2C9 alleles in a Nordic
population13 and with CYP2C8*2 in populations of African
descent.14 Furthermore, two recent studies have shown that
CYP2C19*17 is in complete LD with two loss-of-function alleles,
CYP2C19*2 and CYP2C19*4 (a 1A→G mutation that abolishes
ATG initiation codon; haplotype-designated CYP2C19*4B).15,16
Whereas CYP2C19*4B leads to failure of CYP2C19 protein
expression at the level of translation,16 CYP2C19*2/*17 confers
intermediate metabolizer status.15,17 CYP2C19*4B is rare in
Caucasians (frequency 0.4).10
Single-nucleotide polymorphisms (SNPs) in the CYP2C gene cluster have been extensively investigated as
predisposing factors for nonsteroidal anti-inflammatory drug (NSAID)-induced peptic ulcer disease (PUD) or upper
gastrointestinal bleeding (UGIB). However, results have been inconclusive owing to different study designs, limited
genotyping strategies, and small sample sizes. We investigated whether eight functional SNPs in the CYP2C family
of genes—CYP2C8*3 (rs11572080 and rs10509681), CYP2C8*4, CYP2C9*2, CYP2C9*3, CYP2C19*2, CYP2C19*3, and
CYP2C19*17—are associated with PUD in 1,239 Caucasian patients. Logistic regression analysis showed that only
CYP2C19*17 was associated with PUD (odds ratio additive model: 1.47 (95% confidence interval (CI) 1.12 to 1.92);
P = 0.005; R2 16%), but not UGIB, independent of NSAID use or Helicobacter pylori infection. PUD distribution varied
(P = 0.024) according to CYP2C19*17 genotype: *1/*1, 490 (64.3%); *1/*17, 304 (71.7%); and *17/*17, 31 (73.8%).
CYP2C19*17, a gain-of-function polymorphism, is associated with PUD irrespective of etiology.
Received 6 August 2012; accepted 15 October 2012; advance online publication 26 December 2012. doi:10.1038/clpt.2012.215
CYP2C19*17 Gain-of-Function Polymorphism Is
Associated With Peptic Ulcer Disease
CO Musumba1,2,3, A Jorgensen4, L Sutton4, D Van Eker1,3, E Zhang1, N O’Hara1, DF Carr1,
DM Pritchard2,3 and M Pirmohamed1,3
1Department of Molecular and Clinical Pharmacology, University of Liverpool, Liverpool, UK; 2Department of Gastroenterology, University of Liverpool, Liverpool,
UK; 3The Royal Liverpool and Broadgreen University Hospitals NHS Trust, Liverpool, UK; 4Department of Biostatistics, University of Liverpool, Liverpool, UK.
Correspondence: M Pirmohamed (firstname.lastname@example.org)
CLINICAL PHARmACoLoGy & THeRAPeUTICS
To determine the role of CYP2C genetic polymorphisms on
peptic ulcer predisposition, we have, in the largest study under-
taken to date, evaluated the association between eight clinically
important SNPs—CYP2C8*3 (rs11572080 and rs10509681),
CYP2C8*4 (rs1058930), CYP2C9*2 (rs1799853), CYP2C9*3
(rs1057910), CYP2C19*2 (rs4244285), CYP2C19*3 (rs4986893),
and CYP2C19*17 (rs12248560)—and the risk of PUD and/
or UGIB. The eight SNPs were selected on the basis of their
known functionality from the literature14,18–21 and online
genetic databases (http://www.pharmgkb.org, http://www.
cypalleles.ki.se/) and their common occurrence in Caucasian
populations. Our rationale for investigating the entire CYP2C
cluster was as follows: studies to date have been inconclusive
and have concentrated mainly on CYP2C9/2C8; all studies have
investigated loss-of-function polymorphisms but not the most
recently reported gain-of-function SNP, CYP2C19*17; SNPs
in the CYP2C family may predispose to PUD via effects inde-
pendent of NSAID use, such as through altered arachidonic acid
metabolism; and SNPs in CYP2C19 may predispose to PUD
indirectly through altering PPI metabolism, as has been shown
for some CYP2C19 variants, in patients on gastroprotective
Patients, demographic data, and genetic analysis
Overall, 1,301 of the targeted 1,500 patients were recruited and
genotyped during the study period. The shortfall resulted from
unanticipated delays in enrollment. Of the patients, 62 (4.8%)
were excluded: 31 due to non-Caucasian ethnicity, 25 with
equivocal NSAID use, and 6 with missing PUD data (Figure 2).
Of the remaining 1,239, 67.4% had endoscopically confirmed
PUD (comprising 58.1% NSAID users and 41.9% nonusers)
and 32.6% had no endoscopic evidence of PUD (comprising
30.4% NSAID users and 69.6% nonusers). Among patients with
PUD, UGIB was present in 109 (22.5%) NSAID users and in 49
Figure 1 Schematic illustration of the CYP2C subfamily locus on
chromosome 10q24, spanning an ~500-kilobase region. It comprises four
highly polymorphic and highly homologous (>80%) genes, CYP2C8, CYP2C9,
CYP2C19, and CYP2C18, arranged in tandem. CYP2C18 is considered clinically
unimportant in humans. Each of the other genes contains several clinically
important single-nucleotide polymorphisms, some of which are in strong
linkage disequilibrium. On the basis of CYP2C19 enzyme activity, individuals
can be categorized as intermediate metabolizers (IMs; those with one wild-
type allele and one variant allele that encodes reduced or absent enzyme
function, e.g., *1/*2, *1/*3), poor metabolizers (PMs; those with two loss-of-
function alleles, e.g., *2/*2, *2/*3, *3/*3), resulting in markedly reduced or
absent activity, extensive metabolizers (EMs; those with two wild-type alleles,
i.e., *1/*1); and ultrarapid metabolizers (UMs; those who carry one or two
*17 gain-of-function alleles, e.g., *1/*17, *17/*17) (ref. 50). The thick arrows
indicate the direction of transcription of the DNA sequence into messenger
RNA. Modified from ref. 13.
Figure 2 Flowchart of patients recruited into the study. *UGIB status was unknown for 2 patients. NSAIDs, nonsteroidal anti-inflammatory drugs; PUD, peptic
ulcer disease; UGIB, upper gastrointestinal bleeding.
n = 1,301
patients (n = 1,239)
Excluded, n = 62 (4.8%)
Equivocal NSAID use 25
Non-white ethnicity 31
Ulcer status unknown 6
PUD n = 404 (32.6%)
Patients with PUD
n = 835 (67.4%)
n = 350 (41.9%)
n = 485 (58.1%)
n = 281 (69.6%)
n = 123 (30.4%)
UGIB, n = 109
n = 376 (77.5%)
UGIB, n = 49
n = 299 (85.9%)
(14.1%) nonusers. Table 1 lists the clinical and demographic
characteristics of patients with and without PUD. Seven SNPs
were included in the final analysis after exclusion of one SNP,
CYP2C19*3, which was found to be monomorphic; 100% con-
cordance was observed with all duplicate genotypes.
Associations between CYP2C polymorphisms and PUD/UGIB
All SNPs were in Hardy–Weinberg equilibrium (P > 0.05) and
in agreement with frequencies reported in other Caucasian
populations (see Supplementary Data online). The frequency
of CYP2C genotypes at each SNP is shown in Table 2. In the
logistic regression analyses using an additive model, only one
SNP, CYP2C19*17, was significantly associated with the pres-
ence of PUD (odds ratio 1.47 (95% confidence interval (CI)
1.12 to 1.92); P = 0.005, Table 3). This association with PUD
persisted when using a genotype model for analysis (overall P
value for the model = 0.014; odds ratio 1.57 (95% CI 1.15 to
2.15); for *1/*17 vs. *1/*1), although the P value did not with-
stand Bonferroni correction (see Supplementary Data online).
However, the effect did not differ depending on NSAID use (P
value for interaction term = 0.158). No significant association
was seen between CYP2C19*17/*17 and PUD (odds ratio in
additive model: 1.64 (95% CI 0.70 to 3.89)), probably owing
to the small numbers in this group (n = 42). A logistic regres-
sion model for PUD including covariates representing NSAID
type, H. pylori status (positive or negative), and CYP2C19*17
gave a total pseudo R2 of 16%. A greater proportion of patients
with PUD was seen among CYP2C19*17 variant allele carriers
as compared with wild-type homozygotes (1*/1*, 64.3%; 1*/17,
71.7%; 17*/17*, 73.8%; P = 0.024 using χ2 test, Figure 3). No
significant association was found between any of the SNPs and
binary UGIB status or bleeding PUD status (see Supplementary
Data online). Sensitivity analysis including only the cases with-
out evidence of H. pylori infection showed no significant associ-
ation with CYP2C19*17 when comparing those with or without
PUD (P = 0.062) or those with or without UGIB (P = 0.63).
CYP219*2 and CYP2C19*17 genotypes were available for
1,210 patients and were distributed as follows: CYP2C19*1/*1,
525 (43.4%); CYP2C19 *1/*2, 205 (16.9%); CYP2C19*2/*2,
18 (1.5%); CYP2C19*1/*17, 343 (28.3%); CYP2C19*2/*17, 78
(6.4%); and CYP2C19*17/*17, 41 (3.4%). To investigate the effect
of the *2/*17 combined genotype, a logistic regression using a
Table 1 Clinical and demographic characteristics of study
Patients with PUD
(n = 835)
PUD (n = 404)
Age, mean (SD)65.2 (13.6)57.6 (14.7)
Male sex, n (%)484 (58.0%)174 (43.1%)
BMI, mean (SD)
Endoscopy indication, n (%)a
26.9 (7.7) 27.3 (7.8)
Hematemesis/melena317 (38.6%)24 (6.0%)
Dyspepsia/abdominal pain228 (27.7%)189 (46.9%)
Anemia109 (13.2%)68 (16.9%)
Reflux symptoms20 (2.4%)8 (2.0%)
Other146 (17.8%)114 (28.3%)
Low-dose aspirin318 (38.1%)94 (23.3%)
High-dose aspirin 18 (2.2%)1 (0.2%)
Diclofenac104 (12.5%)18 (4.5%)
Ibuprofen77 (9.2%)15 (3.7%)
Ketoprofen4 (0.5%)1 (0.2%)
Naproxen30 (3.6%)6 (1.5%)
Oxicams 15 (1.8%)0 (0%)
Coxibs13 (1.6%) 2 (0.5%)
Ketoprofen4 (0.5%)1 (0.2%)
Indomethacin3 (0.4%)0 (0%)
Mefenamic acid0 (0%)0 (0%)
Concomitant medications, n (%)
Proton pump inhibitors299 (35.8%) 228 (56.4%)
Antisecretory medications32 (3.8%)13 (3.2%)
Antiplatelets44 (5.3%)19 (4.7%)
Anticoagulants51 (6.1%) 20 (5.0%)
Steroids 32 (3.8%)14 (3.5%)
Smoking, n (%)b
54 (6.5%)52 (12.9%)
Never 276 (33.1%)185 (45.8%)
Current247 (29.6%)106 (26.2%)
Former311 (37.3%)113 (28.0%)
Alcohol intake, n (%)
Audit score <8672 (80.5%)327 (80.9%)
Audit score 8–15126 (15.1%)57 (14.1%)
Audit score >1537 (4.4%)20 (5.0%)
Medical history, n (%)
Previous DU156 (18.7%)1 (0.2%)
Previous GU100 (12.0%)0 (0%)
Previous UGIB42 (5.0%)0 (0%)
Cardiovascular disease478 (57.2%)179 (44.3%)
Cerebrovascular disease 30 (3.6%)5 (1.2%)
Respiratory disease196 (23.5%)102 (25.2%)
Musculoskeletal conditions 435 (52.1%)151 (37.4%)
Table 1 Continued
Patients with PUD
(n = 835)
PUD (n = 404)
Helicobacter pylori status, n (%)c
134 (16.0%)47 (11.6%)
Positive445 (53.3%)56 (21.1%)
Negative376 (45.0%)208 (78.5%)
Equivocal/unknown12 (1.4%) 1 (0.4%)
BMI, body mass index; DU, duodenal ulcer; GU, gastric ulcer; NSAID, nonsteroidal
anti-inflammatory drug; PUD, peptic ulcer disease; SSRI, selective serotonin reuptake
inhibitor; UGIB, upper gastrointestinal bleeding.
a14 Unobtainable/missing (13 PUD, 1 non-PUD). b1 Unknown (PUD). c139
CLINICAL PHARmACoLoGy & THeRAPeUTICS
Table 2 Frequencies of CYP2C genotypes
(n = 1,239)
Patients with PUDPatients with no PUD
(n = 485)
(n = 350)
(n = 123)
(n = 281)
CYP2C8*3 (Arg139Lys), n = 1,232 rs11572080
Homozygous wild type (TT)948 (76.5%)370 (76.3%) 265 (75.7%)95 (77.2%) 218 (77.6%)
Heterozygous (TC) 273 (22.0%)106 (21.9%)81 (23.1%) 26 (21.1%) 60 (21.4%)
Homozygous variant (CC)11 (0.9%) 4 (0.8%)3 (0.9%)2 (1.6%) 2 (0.7%)
Missing 7 (0.6%)5 (1.0%) 1 (0.3%)0 (0.0%) 1 (0.4%)
C allele 295 (11.9%)114 (11.8%)87 (12.4%)30 (12.2%)64 (11.4%)
CYP2C8*3 (Lys399Arg), n = 1,231rs10509681
Homozygous wild type (GG)948 (76.5%)371 (76.5%) 266 (76.0%)94 (76.4%) 217 (77.2%)
Heterozygous (GA)272 (22.0%) 107 (22.1%)80 (22.9%)26 (21.1%) 59 (21.0%)
Homozygous variant (AA)11 (0.9%)4 (0.8%)3 (0.9%) 2 (1.6%) 2 (0.7%)
Missing8 (0.6%)3 (0.6%)1 (003%)1 (0.8%)3 (1.1%)
A allele294 (11.9%)115 (11.9%)86 (12.3%) 30 (12.2%) 63 (11.2%)
CYP2C8*4 (Ile264Met), n = 1,230 rs1058930
Homozygous wild type (CC)1,132 (91.4%)439 (90.5%)324 (92.6%)107 (87.0%)262 (93.2%)
Heterozygous (CG)96 (7.7%)39 (8.0%)24 (6.9%)15 (12.2%) 18 (6.4%)
Homozygous variant(GG) 2 (0.2%)2 (0.4%)0 (0.0%)0 (0.0%)0 (0.0%)
Missing9 (0.7%) 5 (1.0%)2 (0.6%) 1 (0.8%)1 (0.4%)
G allele100 (4.0%)43 (4.4%)24 (3.4%) 15 (6.1%)18 (3.2%)
CYP2C9*2 (Arg144Cys), n = 1,209 rs1799853
Homozygous wild type (CC)930 (75.1%)367 (75.7%)261 (74.6%)91 (74.0%)211 (75.1%)
Heterozygous (CT)266 (21.5%) 102 (21.0%)82 (23.4%) 22 (17.9%)60 (21.4%)
Homozygous variant (TT)13 (1.0%) 6 (1.2%)2 (0.6%)2 (1.6%)3 (1.1%)
Missing30 (2.4%)10 (2.1%)5 (1.4%) 8 (6.5%)7 (2.5%)
T allele292 (11.8%)114 (11.8%)86 (12.3%)26 (10.6%)66 (11.7%)
CYP2C9*3 (Ile359Leu), n =1,230 rs1057910
Homozygous wild type (AA)1,057 (85.3%) 420 (86.6%)296 (84.6%)108 (87.8%)233 (82.9%)
Heterozygous (AC)165 (13.3%)58 (12.0%)47 (13.4%)13 (10.6%) 47 (16.7%)
Homozygous variant (CC)8 (0.6%)4 (0.8%)2 (0.6%)2 (1.6%)0 (0.0%)
Missing9 (0.7%)3 (0.6%)5 (1.4%)0 (0.0%)1 (0.4%)
C allele181 (7.3%)66 (6.8%)51 (7.3%)17 (6.9%)47 (8.4%)
CYP2C19*2 (–G681A), n = 1,222rs4244285
Homozygous wild type (GG)918 (74.1%)340 (70.1%)272 (77.7%)94 (76.4%)212 (75.4%)
Heterozygous (GA)286 (23.1%)123 (25.4%)69 (19.7%)28 (22.8%)66 (23.5%)
Homozygous variant (AA) 18 (1.5%)11 (2.3%)4 (1.1%)0 (0.0%)3 (1.1%)
Missing17 (1.4%)11 (2.3%)5 (1.4%) 1 (0.8%)0 (0.0%)
CYP2C19*17 (–806C>T), n = 1,227
322 (13.0%)145 (14.9%) 77 (11.0%)28 (11.4%) 72 (12.8%)
Homozygous wild type (CC)761 (61.4%)282 (58.1%)208 (59.4%)80 (65.0%) 191 (68.0%)
Heterozygous (CT)424 (34.2%)183 (37.7%)121 (34.6%)38 (30.9%)82 (29.2%)
Homozygous variant (TT)42 (3.4%)16 (3.3%)15 (4.3%)4 (3.3%)7 (2.5%)
Missing12 (1.0%)4 (0.8%)6 (1.7%)1 (0.8%)1 (0.4%)
T allele 508 (20.5%)215 (22.2%)151 (21.6%) 46 (18.7%)96 (17.1%)
NSAIDs, nonsteroidal anti-inflammatory drugs; PUD, peptic ulcer disease; rs, reference single-nucleotide polymorphism.
genotype model (using *1/*1 as baseline and a covariate to rep-
resent each of these CYP2C19 genotypes) was fitted. Odds ratios
from this regression model are given in Figure 4. Of note, there
was no evidence of an association between PUD and carriage of
the CYP2C19*2/*17 genotype (odds ratio 1.14 (95% CI 0.61 to
2.12)), although numbers in this group were small (n = 78), and
so this result should be interpreted with caution.
Influence of PPIs and gender
To further explore the association of PUD status with
CYP2C19*17, we undertook an analysis to investigate whether
the association occurred as a result of interaction with PPI expo-
sure or gender. Overall, a total of 546 PPIs were used by 527
patients (some patients used more than one PPI), distributed
as follows: omeprazole (242), lansoprazole (211), esomeprazole
(53), pantoprazole (32), and rabeprazole (8); 250 (47.4%) patients
used PPIs and NSAIDs concomitantly. PPI status or gender was
introduced as a covariate into the previously fitted genetic model,
and a likelihood ratio test was used to compare the new model
with the genetic model. A total of 299 (35.8%) of the patients with
PUD and 228 (56.4%) of those without PUD were taking PPIs.
The use of PPIs was significantly associated with the absence of
PUD in all patients both when those with evidence of H. pylori
infection were included and when they were excluded (P <
0.0001). However, there was no statistically significant interac-
tion between PPI status and CYP2C19*17 in either group (P =
0.52 and 0.73, respectively). Of the whole cohort, 658 (53.1%)
were male: 484 (58.0%) with PUD and 174 (43.1%) without PUD.
Although male gender was significantly associated with the pres-
ence of PUD (P = 0.004), there was no significant interaction
between gender and CYP2C19*17 (P = 0.45). Sensitivity analysis
excluding those with non-NSAID ulcers did not show any sig-
nificant association with PUD status (P = 0.97).
Influence of ulcer location
Ulcer distribution was as follows: gastric ulcers 432 (51.9%),
duodenal ulcers 322 (38.7%), both gastric and duodenal ulcers
62 (7.5%), and pyloric ulcers 16 (1.9%). The analyses of asso-
ciation with genotype described above were repeated, but first
including only gastric ulcers as cases and later only duodenal
ulcers as cases (those without any type of ulcer were treated as
controls (n = 404) in both analyses). Cases with gastroduodenal
ulcers were excluded from both analyses. For gastric ulcers, a low
P value was achieved with only CYP2C19*17 (P = 0.022), and for
duodenal ulcers, a low value was seen in association with both
CYP2C19*17 and CYP2C8*4 (P = 0.037 for each); however, none
was significant after Bonferroni correction (see Supplementary
Data online). There was no significant interaction on sensitivity
analysis including only cases without evidence of H. pylori infec-
tion (see Supplementary Data online).
Analysis using only NSAIDs predominantly or partially metab-
olized by CYP2C
Finally, we repeated the above analysis with patients classified as
being on NSAIDs only if using NSAIDs that are predominantly
Table 3 Logistic regression analysis using additive model
showing association between all SNPs and binary ulcer status
Analysis using PUD statusa
P valueoR95% CI
rs10589301,230 0.14 0.650.36–1.15
CI, confidence interval; n, number of patients contributing to analysis; OR, odds ratio;
PUD, peptic ulcer disease; rs, reference SNP; SNP, single-nucleotide polymorphism.
aPUD = cases, n = 835; no PUD = controls, n = 404. nb = number of patients
contributing to analysis.
Figure 4 Odds ratios for peptic ulcer disease according to CYP2C19
composite genotype from logistic regression model. CI, confidence interval.
Odds ratio (95% CI)
CYP2C19 composite genotype
Figure 3 Distribution of peptic ulcer disease (PUD) status among the
different CYPC19*17 genotype groups (10 patients with PUD and two patients
without PUD had no genotyping data). MT, mutant allele; WT, wild-type allele.
WT homozygous(CC)Heterozygous(CT)MT homozygous
For all: P = 0.024
CLINICAL PHARmACoLoGy & THeRAPeUTICS
(celecoxib, ibuprofen, lornoxicam, and piroxicam) or partially
(diclofenac, flurbrufen, indomethacin, and meloxicam) metabo-
lized by CYP2C enzymes, with all other patients classified as
not being on NSAIDs (see Supplementary Data online). Again,
only CYP2C19*17 was significantly associated with having PUD
(P = 0.004). In a sensitivity analysis in which only those patients
without any evidence of H. pylori infection (n = 376) were ana-
lyzed as cases, there was no significant interaction seen with
CYP2C19*17 (P = 0.068) (see Supplementary Data online).
The interaction between CYP2C19*17 and NSAID status was
similarly nonsignificant (P = 0.44). Furthermore, there was no
significant association when the analysis was repeated between
each SNP and UGIB, even when only those patients without H.
pylori infection were included (n = 376).
In contrast to previous studies, we have adopted a much broader
approach and have investigated the role of seven functional SNPs
in the CYP2C family for association with PUD and/or UGIB.
Our data did not show any association with CYP2C9/2C8 SNPs
and NSAID-induced UGIB, as has been previously reported.1,2,5
However, we observed a novel association between the
CYP2C19*17 allele and the presence of PUD at both the allelic
and genotype levels, irrespective of NSAID exposure or presence
of H. pylori infection. Further analysis focusing on all CYP2C19
genotypes and PUD status (using *1/*1 as baseline) showed that
only *1/*17 was significantly associated with PUD. The discrep-
ancies between our results and those published to date can be
explained by various factors: (i) previous studies of the associa-
tion between CYP2C polymorphisms and ulceration have been
relatively small (16–40 cases)3,4,25 and have not included patients
with ulcers not due to NSAIDs; (ii) there is a great degree of LD
across the CYP2C gene cluster on chromosome 10, and other
studies have examined only a limited number of SNPs; and (iii)
CYP2C19*17 has not been examined previously in PUD.
Why is CYP2C19*17 associated with PUD? CYP2C19*17 is
a common gain-of-function polymorphism, with carriers hav-
ing increased rates of metabolism of substrate drugs (such as
PPIs, escitalopram, sertraline, tamoxifen, and clopidogrel), with
a resultant decrease in the plasma concentrations of some drugs
demonstrated. However, a review of the functional and clinical
consequences of the CYP2C19*17 allele concluded that it has
only a modest effect at best, which is unlikely to be clinically sig-
nificant except in CYP2C19*17 homozygotes, and only for sub-
strate drugs with a narrow therapeutic window.26 On the other
hand, one recent study15 and a recent meta-analysis27 concluded
that carriage of the CYP2C19*17 allele in users of clopidogrel
is associated with a lower magnitude of on-treatment platelet
reactivity, a lower risk of cardiovascular outcomes and stent
thrombosis, and a higher risk of major bleeding, due to enhanced
activation. Hence, one possible explanation could be that carriers
of this SNP have enhanced clearance of gastroprotective PPIs
and thereby reduced mucosal gastroprotection against aggres-
sors (such as NSAIDs and H. pylori), thus predisposing them to
developing PUD. However, analysis of our data failed to support
any interaction between CYP2C19*17 status and PPI use.
The CYP2C enzymes are involved not only in the metabolism
of xenobiotics but also in that of endogenous substances. For
instance, arachidonic acid is metabolized through three main
enzymatic pathways: the cyclooxygenases, the lipooxygenases,
and the CYP450 monooxygenases (CYP2C9, 2C8, 2C19, and
CYP2CJ).9,28,29 CYP2C19 efficiently metabolizes arachidonic
acid to four kinds of epoxyeicosatrienoic acids (EETs; 5,6-EET;
8,9-EET; 11,12-EET; and 14,15-EET) that possess diverse physi-
ological roles (including control of vascular tone, angiogenesis,
cellular migration, and proliferation and inflammation)29–31 that
are regioisomer-selective as well as species- and organ-specific.
The role of EETs in the human gastrointestinal tract is unknown.
Both 5,6-EET and 8,9-EET can be converted by cyclooxygenases
to vasoconstrictor metabolites,32 and 5,6 EET has been shown to
cause vasoconstriction in the renal microcirculation of rats33,34
and pulmonary arteries of rabbits.35 Furthermore, Fang et al.
showed that 8, 9-EET and 14,15-EET inhibited prostaglandin
E2 production in vascular smooth muscle cells.36 CYP2C cataly-
sis has also been shown to generate reactive oxygen species in
coronary artery endothelial cells, negating the beneficial effects
of EETs produced, with resultant ischemic injury.37,38 Therefore,
one possible explanation for the association we have identified is
that CYP2C19*17 alters arachidonic acid metabolism, resulting
in an impairment of gastrointestinal mucosal defenses through a
combination of reduced gastroprotective prostaglandin E2 pro-
duction, enhanced vasoconstriction in the mucosal microcir-
culation, and promotion of the production of injurious reactive
oxygen species (see Supplementary Data online). Clearly, this
requires further study.
PUD (particularly duodenal ulcers) and UGIB have been
shown to occur more frequently in males.39,40 This is thought
to be due to the higher levels of estrogens in females, leading
to greater stimulation of gastroprotective duodenal mucosal
bicarbonate secretion as well as to enhanced immune responses
(as compared with testosterone, which is immunosuppressive),
which in turn lead to greater resistance to H. pylori infection.41
Given that CYP2C19 has been shown to metabolize estro-
gens,42,43 and CYP2C19*17 is postulated to enhance estrogen
catabolism,44 the association seen with PUD in our study could
have resulted from impaired gastroduodenal mucosal protection
due to reduced estrogen levels. However, post hoc analysis of our
data did not support this hypothesis, as no significant interac-
tion was seen between CYP2C19*17 and gender or a specific
This study has several strengths. First, it is the largest study
to date that has specifically examined the association between
genetic polymorphisms in the CYP2C enzymes and PUD
among both NSAID users and nonusers, including more than
1,200 Caucasian patients. Before this, the largest similar study,
by Blanco et al., included 311 Caucasian patients (134 cases,
defined as individuals with UGIB, and 177 controls, defined
as individuals consuming similar doses of NSAIDs with no
reported adverse effects).1 A recent genome-wide association
study of 7,035 Japanese patients with duodenal ulcers showed
two susceptibility loci—in the PSCA gene at 8q24 (r2294008)
and in the ABO blood group locus at 9q34 (rs505922)—but
there was no categorization according to NSAID use.45 Patients
in our cohort were also exposed to a wide variety of NSAIDs,
including both aspirin and nonaspirin NSAIDs, which enabled
us to evaluate differences in any genetic effects seen between the
various NSAIDs. Furthermore, our study is the first to use two
groups that could be compared with the group with NSAID-
induced ulceration. The inclusion of patients with PUD but
without prior exposure to NSAIDs enabled investigation of
whether any significant associations were truly due to a gene-
by-treatment interaction rather than being a prognostic marker
for PUD as compared with studies including only those with
a history of NSAID usage. In addition, all our patients in the
three groups underwent endoscopy to ascertain the presence
or absence of PUD, UGIB, and H. pylori.
The study also has some limitations. First, we had incomplete
data for the duration and doses of PPIs used in our patient
cohort. This might have led to an underestimation of the asso-
ciation we found between PPI use and CYP2C19*17 genotype,
especially in patients on concomitant NSAIDs. Second, given
that no previous studies have reported a direct association
between CYP2C19*17 and ulcer pathogenesis, it is possible that
the positive association we found was a result of another variant
due to the LD in the CYP2C region. Third, because we did not
determine the presence of the CYP2C19*4B haplotype in our
patient cohort, we might have overestimated the frequency of
the ultrarapid metabolizers (*1/*17 or *17/*17). Logistic regres-
sion analysis of all the CYP2C19 genotypes showed that the
more frequent CYP2C19*2, a common loss-of-function allele
also in LD with CYP219*17, reduced the effect of CYP2C19*17
(odds ratio *2/*17 vs. *1/*1: 1.14, 95% CI: 0.61, 2.12; not sig-
nificant). However, the number of patients with the *2/*17
genotype was small (n = 78), and thus this result, although
biologically plausible, should be interpreted with caution.
Although CYP2C19*4B may have the same effect, it is impor-
tant to note that it is a rare variant, and we would therefore not
have the power to be able to show its effect despite the large
number of patients studied.
In conclusion, possession of the CYP2C19*17 allele was asso-
ciated with the presence of endoscopically confirmed PUD in
this large cohort of Caucasian subjects. There are plausible
biological explanations for this association, for example, the
effect of CYP2C enzymes on the metabolism of arachidonic
acid, which is well known to be involved in the pathogenesis of
PUD.46 Further well-powered studies in a different population
are now needed to validate these findings and to evaluate the
functional consequences of CYP2C19*17 on PUD pathogenesis.
Patients and controls. This was a multicenter case–control study involving
15 hospitals in the United Kingdom (see Supplementary Data online).
Patients who had undergone endoscopy for suspected PUD between
July 2005 and June 2011 were identified from endoscopy databases at the
participating hospitals and invited via telephone or letter to take part in
the study, or recruited prospectively in hospital as inpatients or day cases
during their attendance at endoscopy beginning January 2008. The study
was approved by the Liverpool (Adults) Research Ethics Committee.
Informed consent was obtained from all eligible patients and their use
of NSAIDs prior to endoscopy was determined as described below.
Following recruitment, patients were categorized into three groups:
1. Patients with endoscopically confirmed PUD within 2 weeks of
using NSAIDs (including aspirin)
2. Patients with endoscopically confirmed PUD without a history of
using NSAIDs (including aspirin) within 3 months of diagnosis
3. Patients without endoscopic evidence of PUD, whose use or nonuse
of NSAIDs prior to endoscopy was determined
Patients in groups 1 and 2 were recruited both retrospectively and
prospectively from all sites, whereas those in group 3 were recruited
prospectively from randomly selected outpatients undergoing upper
gastrointestinal endoscopy at the Royal Liverpool University Hospital.
To be eligible, all patients had to be older than 18 years at time of recruit-
ment, with capacity to give consent. Exclusion criteria were inability to
contact patients, inability/refusal to give consent, malignant peptic ulcers,
previous gastrointestinal surgery, Zollinger–Ellison syndrome, and blood-
borne transmissible disease. Detailed epidemiological data for the patients
recruited from the Royal Liverpool Hospital are available elsewhere.47
Demographic characteristics and drug use. Participants were inter-
viewed using a structured questionnaire regarding their demographic
characteristics, comorbidities, medical history, drug use (both pre-
scription and over the counter), alcohol intake, smoking habits, and
previous H. pylori eradication. Pre-endoscopy interview forms and
endoscopy reports were scrutinized for details of patients’ clinical pres-
entation, ulcer characteristics, and use of ulcerogenic drugs, includ-
ing NSAIDs. Case notes were interrogated or general practitioners
contacted for additional information on NSAID use where necessary.
Patients were defined as NSAID users if they had taken a drug, as part
of a continuous period of treatment lasting for 1 week or more, during
the 2 weeks prior to endoscopy. Nonusers were defined as those who
had not used any NSAID in the 3-month period before their endos-
copy. PPI use was defined as use of any PPI within the 3 months pre-
ceding and including the time of endoscopy.
Endoscopy and H. pylori infection. PUD was defined either from the
endoscopy reports as a mucosal break ≥3 mm diameter, a widely
used criterion for defining endoscopic ulcers in studies,48 or from
the description of the endoscopist. UGIB was defined as hemate-
mesis, melena, or anemia, and/or endoscopic stigmata of recent
hemorrhage according to the modified Forrest criteria.49 During
endoscopy, most patients had gastric biopsies done for a rapid ure-
ase test (Pronto Dry, Medical Instruments Corporation, Solothurn,
Switzerland, or a Campylobacter-like organism (CLO) test) and/or
histology to assess for H. pylori infection per standard practice in
the endoscopy units involved. H. pylori infection was deemed to be
present when either of the biopsy-based tests was positive; where
these were negative or not done, a positive serology test was taken
to indicate current infection and a negative serology test was taken
to indicate absence of infection. Following enrollment, two venous
blood samples were obtained and sent to a centralized laboratory,
where plasma was frozen at −80 °C.
Genetic analysis. Genomic deoxyribonucleic acid was extracted from
ethylenediaminetetraacetic acid whole blood using the Chemagen
5 ml whole-blood deoxyribonucleic acid extraction kit on the
Chemagic Magnetic Separation Module I according to the manufac-
turer’s protocol (PerkinElmer chemagen Technologie, Baesweiler,
Germany). All DNA extracted had an A260/280 ratio between 1.8
and 2.0. Genotyping was performed by KBioscience, using its propri-
etary fluorescence-based competitive allele-specific polymerase chain
reaction assay (http://www.kbioscience.co.uk/reagents/KASP.html).
KBioscience laboratory staff were blinded as to which patients (i.e.,
cases or controls) the samples were obtained from. As part of quality
CLINICAL PHARmACoLoGy & THeRAPeUTICS
control, 10% duplicates and one negative control were included in each
96-well plate sent to KBioscience.
Statistical analysis. A sample-size calculation was performed in
advance of performing the study. We aimed to recruit 500 patients in
each group, basing our power calculations on two possible scenarios:
(i) minor allele frequency = 5%; odds ratio = 3; power = 99%; and (ii)
minor allele frequency = 20%; odds ratio = 2; power = 99% (both when
comparing group 1 vs. group 2). Each CYP2C SNP was investigated
separately, and genotype frequencies were tested for Hardy–Weinberg
equilibrium with P value <0.01 assumed to indicate deviation from it.
SNPs with a call rate <90% were excluded. Patient self-reported eth-
nicity was assessed, and any non-Caucasians were excluded from the
analysis. To test for association with each SNP in turn, two logistic
regression models were fitted. The first (the “baseline model”) included
covariates to represent study site and NSAID exposure; the second (the
“genetic model”) was the same model but with the addition of a covari-
ate to represent the SNP. An additive mode of inheritance was assumed,
with wild-type homozygotes coded as 0; heterozygotes coded as 1, and
mutant-type homozygotes coded as 2. The two models were compared
using the likelihood-ratio test, which tested for association between
the SNP and risk of developing PUD, independent of NSAID use.
Given that seven SNPs were tested for association with each outcome,
a Bonferroni correction was used to account for multiplicity of testing;
a P value <0.007 (0.05/7) was therefore assumed to represent statistical
significance. If the P value was <0.05, a further model was fitted (the
“interaction model”), which was the same as the genetic model but also
included an NSAID exposure × genotype interaction term. Again, the
likelihood-ratio test was used to compare the interaction model with
the genotype model, this time to test whether the effect of the SNP
differed depending on NSAID usage (i.e., a test for significance of the
interaction term). Sensitivity analyses were also undertaken, in which
each analysis was repeated, but cases without evidence of H. pylori
infection were excluded. The analyses were also repeated for the binary
outcomes of UGIB status (cases and controls defined as those with or
without endoscopic evidence of UGIB, respectively) and bleeding PUD
(cases and controls defined as those with PUD with or without endo-
scopic evidence of UGIB, respectively). We repeated the above analyses,
this time using a genotype model for CYP2C19*17 and PUD, making
no assumptions as to the mode of inheritance. In addition, because both
the CYP2C19*17 and CYP2C19*2 alleles are common, we assessed the
relationship between all CYP2C19 genotypes and PUD status.
Principal investigator(s) and their participating hospital site are as follows:
Keith Bodger, University Hospital Aintree; Beverly Oates, Wirral University
Teaching Hospital NHS Trust; Peter Isaacs, Blackpool Victoria Hospital;
Colin Rees, South Tyneside NHS Foundation Trust; Simon Panter, South
Tyneside NHS Foundation Trust; Andrew Higham, Royal Lancaster Infirmary;
Vishal Kaushik, East Lancashire Hospitals NHS Trust; Chris MacDonald,
North Cumbria University Hospitals NHS Trust; Mike Stroud, Southampton
General Hospital; Fraser Cummings, Southampton General Hospital; Faiyaz
Mohammed, Lancashire Teaching Hospital NHS Foundation Trust; Ashref
Tawil, Northern Devon Healthcare NHS Trust; Sulleman Moreea, Bradford
Hospitals NHS Trust; Matt Rutter, University Hospitals North Tees; Carol
Francis, Countess of Chester Hospital; Babur Javaid, Whitehaven (North
Cumbria University Hospitals NHS Trust).
SUPPLEMENTARY MATERIAL is linked to the online version of the paper at
We thank the Department of Health (UK) for funding the NHS Chair of
Pharmacogenetics program, through which this study was done. M.P. is a
National Institute for Health Research Senior Investigator. We also thank
all research nurses involved in patient recruitment in the participating
C.O.M. wrote the manuscript. M.P. and D.M.P. designed the research. C.O.M.,
D.V.E., E.Z., N.O., and D.F.C. performed the research. M.P., C.O.M., A.J., L.S., and
D.M.P. analyzed the data. E.Z. contributed new reagents/analytical tools.
CONFLICT OF INTEREST
The authors declared no conflict of interest.
© 2012 American Society for Clinical Pharmacology and Therapeutics
1. Blanco, G. et al. Interaction of CYP2C8 and CYP2C9 genotypes modifies the
risk for nonsteroidal anti-inflammatory drugs-related acute gastrointestinal
bleeding. Pharmacogenet. Genomics 18, 37–43 (2008).
Carbonell, N. et al. CYP2C9*3 Loss-of-function allele is associated with acute
upper gastrointestinal bleeding related to the use of NSAIDs other than
aspirin. Clin. Pharmacol. Ther. 87, 693–698 (2010).
Ma, J., Yang, X.Y., Qiao, L., Liang, L.Q. & Chen, M.H. CYP2C9 polymorphism in
non-steroidal anti-inflammatory drugs-induced gastropathy. J. Dig. Dis. 9,
Martin, J.H., Begg, E.J., Kennedy, M.A., Roberts, R. & Barclay, M.L. Is cytochrome
P450 2C9 genotype associated with NSAID gastric ulceration? Br. J. Clin.
Pharmacol. 51, 627–630 (2001).
Pilotto, A. et al. Genetic susceptibility to nonsteroidal anti-inflammatory
drug-related gastroduodenal bleeding: role of cytochrome P450 2C9
polymorphisms. Gastroenterology 133, 465–471 (2007).
Vonkeman, H.E., van de Laar, M.A., van der Palen, J., Brouwers, J.R. & Vermes, I.
Allele variants of the cytochrome P450 2C9 genotype in white subjects from
the Netherlands with serious gastroduodenal ulcers attributable to the use of
NSAIDs. Clin. Ther. 28, 1670–1676 (2006).
Estany-Gestal, A., Salgado-Barreira, A., Sánchez-Diz, P. & Figueiras, A.
Influence of CYP2C9 genetic variants on gastrointestinal bleeding associated
with nonsteroidal anti-inflammatory drugs: a systematic critical review.
Pharmacogenet. Genomics 21, 357–364 (2011).
Ingelman-Sundberg, M., Sim, S.C., Gomez, A. & Rodriguez-Antona, C. Influence
of cytochrome P450 polymorphisms on drug therapies: pharmacogenetic,
pharmacoepigenetic and clinical aspects. Pharmacol. Ther. 116, 496–526 (2007).
WHAT IS THE CURRENT KNOWLEDGE ON THE TOPIC?
3 Genetic factors may predispose certain individuals to
developing PUD and/or UGIB. CYP2C9 and/or CYP2C8
have been found to be potential susceptibility genes for
WHAT QUESTION DID THIS STUDY ADDRESS?
3 This study addressed the question of whether eight clini-
cally relevant single-nucleotide polymorphisms in the
CYP2C gene cluster predispose individuals to NSAID-
WHAT THIS STUDY ADDS TO OUR KNOWLEDGE
3 The CYP2C19*17 gain-of-function polymorphism was
associated with PUD, irrespective of etiology, in this UK
Caucasian cohort. There are plausible biological mecha-
nisms related to CYP-mediated metabolism of endobiot-
ics for this association that need further study.
HOW THIS MIGHT CHANGE CLINICAL PHARMACOLOGY
3 The identification of CYP2C19*17 as a predisposing
allele for PUD may provide an avenue for risk strati-
fication and a novel therapeutic target, particularly in
individuals with resistant disease.
Articles Download full-text
9. Niwa, T., Murayama, N. & Yamazaki, H. Oxidation of endobiotics mediated by
xenobiotic-metabolizing forms of human cytochrome. Curr. Drug Metab. 10,
10. Martis, S., Peter, I., Hulot, J.S., Kornreich, R., Desnick, R.J. & Scott, S.A. Multi-
ethnic distribution of clinically relevant CYP2C genotypes and haplotypes.
Pharmacogenomics J. (2012); e-pub ahead of print 10 April 2012.
11. Goldstein, J.A. Clinical relevance of genetic polymorphisms in the human
CYP2C subfamily. Br. J. Clin. Pharmacol. 52, 349–355 (2001).
12. Yasar, U. et al. Linkage between the CYP2C8 and CYP2C9 genetic
polymorphisms. Biochem. Biophys. Res. Commun. 299, 25–28 (2002).
13. Pedersen, R.S. et al. Linkage disequilibrium between the CYP2C19*17 allele
and wildtype CYP2C8 and CYP2C9 alleles: identification of CYP2C haplotypes
in healthy Nordic populations. Eur. J. Clin. Pharmacol. 66, 1199–1205 (2010).
14. Suarez-Kurtz, G. Linkage disequilibrium between the CYP2C19*17 and
CYP2C8*2 alleles in populations of African descent. Eur. J. Clin. Pharmacol. 67,
15. Harmsze, A.M. et al. The influence of CYP2C19*2 and *17 on on-treatment
platelet reactivity and bleeding events in patients undergoing elective
coronary stenting. Pharmacogenet. Genomics 22, 169–175 (2012).
16. Scott, S.A., Martis, S., Peter, I., Kasai, Y., Kornreich, R. & Desnick, R.J.
Identification of CYP2C19*4B: pharmacogenetic implications for drug
metabolism including clopidogrel responsiveness. Pharmacogenomics J. 12,
17. Sibbing, D. et al. Isolated and interactive impact of common CYP2C19 genetic
variants on the antiplatelet effect of chronic clopidogrel therapy. J. Thromb.
Haemost. 8, 1685–1693 (2010).
18. Daily, E.B. & Aquilante, C.L. Cytochrome P450 2C8 pharmacogenetics: a review
of clinical studies. Pharmacogenomics 10, 1489–1510 (2009).
19. Rosemary, J. & Adithan, C. The pharmacogenetics of CYP2C9 and CYP2C19:
ethnic variation and clinical significance. Curr. Clin. Pharmacol. 2, 93–109 (2007).
20. Scott, S.A. et al. PharmGKB summary: very important pharmacogene
information for cytochrome P450, family 2, subfamily C, polypeptide 19.
Pharmacogenet. Genomics 22, 159–165 (2012).
21. Van Booven, D. et al. Cytochrome P450 2C9-CYP2C9. Pharmacogenet.
Genomics 20, 277–281 (2010).
22. Baldwin, R.M. et al. Increased omeprazole metabolism in carriers of the
CYP2C19*17 allele; a pharmacokinetic study in healthy volunteers. Br. J. Clin.
Pharmacol. 65, 767–774 (2008).
23. Hunfeld, N.G. et al. Effect of CYP2C19*2 and *17 mutations on
pharmacodynamics and kinetics of proton pump inhibitors in Caucasians. Br.
J. Clin. Pharmacol. 65, 752–760 (2008).
24. Sim, S.C. et al. A common novel CYP2C19 gene variant causes ultrarapid drug
metabolism relevant for the drug response to proton pump inhibitors and
antidepressants. Clin. Pharmacol. Ther. 79, 103–113 (2006).
25. Shiotani, A. et al. The preventive factors for aspirin-induced peptic ulcer:
aspirin ulcer and corpus atrophy. J. Gastroenterol. 44, 717–725 (2009).
26. Li-Wan-Po, A., Girard, T., Farndon, P., Cooley, C. & Lithgow, J. Pharmacogenetics
of CYP2C19: functional and clinical implications of a new variant CYP2C19*17.
Br. J. Clin. Pharmacol. 69, 222–230 (2010).
27. Zabalza, M. et al. Meta-analyses of the association between cytochrome
CYP2C19 loss- and gain-of-function polymorphisms and cardiovascular
outcomes in patients with coronary artery disease treated with clopidogrel.
Heart 98, 100–108 (2012).
28. Capdevila, J.H., Falck, J.R. & Estabrook, R.W. Cytochrome P450 and the
arachidonate cascade. FASEB J. 6, 731–736 (1992).
29. Kaspera, R. & Totah, R.A. Epoxyeicosatrienoic acids: formation, metabolism
and potential role in tissue physiology and pathophysiology. Expert Opin.
Drug Metab. Toxicol. 5, 757–771 (2009).
30. Spector, A.A., Fang, X., Snyder, G.D. & Weintraub, N.L. Epoxyeicosatrienoic
acids (EETs): metabolism and biochemical function. Prog. Lipid Res. 43, 55–90
31. Wray, J. & Bishop-Bailey, D. Epoxygenases and peroxisome proliferator-
activated receptors in mammalian vascular biology. Exp. Physiol. 93, 148–154
32. Roman, R.J. P-450 metabolites of arachidonic acid in the control of
cardiovascular function. Physiol. Rev. 82, 131–185 (2002).
33. Carroll, M.A., Garcia, M.P., Falck, J.R. & McGiff, J.C. 5,6-epoxyeicosatrienoic acid,
a novel arachidonate metabolite. Mechanism of vasoactivity in the rat. Circ.
Res. 67, 1082–1088 (1990).
34. Imig, J.D., Navar, L.G., Roman, R.J., Reddy, K.K. & Falck, J.R. Actions of
epoxygenase metabolites on the preglomerular vasculature. J. Am. Soc.
Nephrol. 7, 2364–2370 (1996).
35. Zhu, D. et al. Epoxyeicosatrienoic acids constrict isolated pressurized rabbit
pulmonary arteries. Am. J. Physiol. Lung Cell Mol. Physiol. 278, L335–L343
36. Fang, X. et al. 14,15-Epoxyeicosatrienoic acid inhibits prostaglandin E2
production in vascular smooth muscle cells. Am. J. Physiol. 275, H2113–H2121
37. Chehal, M.K. & Granville, D.J. Cytochrome p450 2C (CYP2C) in ischemic heart
injury and vascular dysfunction. Can. J. Physiol. Pharmacol. 84, 15–20 (2006).
38. Fleming, I. et al. Endothelium-derived hyperpolarizing factor synthase
(Cytochrome P450 2C9) is a functionally significant source of reactive oxygen
species in coronary arteries. Circ. Res. 88, 44–51 (2001).
39. Kurata, J.H., Honda, G.D. & Frankl, H. The incidence of duodenal and gastric
ulcers in a large health maintenance organization. Am. J. Public Health 75,
40. Wu, H.C., Tuo, B.G., Wu, W.M., Gao, Y., Xu, Q.Q. & Zhao, K. Prevalence of peptic
ulcer in dyspeptic patients and the influence of age, sex, and Helicobacter
pylori infection. Dig. Dis. Sci. 53, 2650–2656 (2008).
41. de Martel, C. & Parsonnet, J. Helicobacter pylori infection and gender: a meta-
analysis of population-based prevalence surveys. Dig. Dis. Sci. 51, 2292–2301
42. Cribb, A.E. et al. Role of polymorphic human cytochrome P450 enzymes in
estrone oxidation. Cancer Epidemiol. Biomarkers Prev. 15, 551–558 (2006).
43. Sangrajrang, S. et al. Genetic polymorphisms of estrogen metabolizing
enzyme and breast cancer risk in Thai women. Int. J. Cancer 125, 837–843
44. Justenhoven, C. et al. CYP2C19*17 is associated with decreased breast cancer
risk. Breast Cancer Res. Treat. 115, 391–396 (2009).
45. Tanikawa, C. et al. A genome-wide association study identifies two
susceptibility loci for duodenal ulcer in the Japanese population. Nat. Genet.
44, 430–434, S1 (2012).
46. Musumba, C., Pritchard, D.M. & Pirmohamed, M. Review article: cellular and
molecular mechanisms of NSAID-induced peptic ulcers. Aliment. Pharmacol.
Ther. 30, 517–531 (2009).
47. Musumba C. et al. The relative contribution of NSAIDs and Helicobacter
pylori to the aetiology of endoscopically-diagnosed peptic ulcer disease:
observations from a tertiary referral hospital in the UK between 2005 and
2010. Aliment Pharmacol Ther. 36, 48–56 (2012).
48. Laine, L. Approaches to nonsteroidal anti-inflammatory drug use in the high-
risk patient. Gastroenterology 120, 594–606 (2001).
49. Forrest, J.A., Finlayson, N.D. & Shearman, D.J. Endoscopy in gastrointestinal
bleeding. Lancet 2, 394–397 (1974).
50. Scott, S.A. et al. Clinical Pharmacogenetics Implementation Consortium
guidelines for cytochrome P450-2C19 (CYP2C19) genotype and clopidogrel
therapy. Clin. Pharmacol. Ther. 90, 328–332 (2011).
CLINICAL PHARmACoLoGy & THeRAPeUTICS