The role of CYP2C9*2, CYP2C9*3 and VKORC1-1639 variants on the susceptibility of upper gastrointestinal bleeding: A full case-control study

Purpose: To investigate whether interindividual variability in the CYP2C9 (*2 and *3 alleles) and VKORC1 (rs9923231) genes is associated with increased risk of upper gastrointestinal bleeding (UGIB) in users of non-steroidal anti-inflammatory drugs (NSAIDs) or low-dose aspirin (LDA).

Methods: A full case-control study including 200 cases of patients diagnosed with UGIB and 706 controls was conducted in a Brazilian hospital complex. To perform an analysis of NSAIDs dose-effect, the defined daily dose (DDD) for NSAIDs was calculated in the 7-day etiologic window preceding the data index. Three categories of DDD, considering the genotypes of the genetic variants, were established: non-users of NSAIDs (DDD = 0), DDD ≤0.5, and DDD >0.5. Genetic variants and LDA or NSAIDs use synergism was estimated through Synergism Index (SI) and Relative Excess Risk Due To Interaction (RERI).

Results: For DDDs of NSAIDs upward of 0.50, a risk of UGIB was identified in carriers of the *3 allele (OR: 15,650, 95% CI: 1.41–174.10) and in carriers of the variant homozygous genotype (TT) of rs9923231 (OR: 38,850, 95% CI: 2.70–556.00). In LDA users, the risk of UGIB was observed to be similar between carriers of the wild type homozygous genotype and carriers of the variant alleles for the CYP2C9 and VKORC1 genes. No synergism was identified.

Conclusion: Our findings suggest an increased risk of UGIB in carriers of the variant allele of rs9923231 and in carriers of the *3 allele associated with doses of NSAIDs greater than 0.5. Hence, the assessment of these variants might reduce the incidence of NSAIDs-related UGIB and contribute to the safety of the NSAIDs user.

Highlights

- This is the first full case-control study to assess the role of rs9923231 (VKORC1 gene) in the risk of UGIB according to the genotype and the use of LDA and the DDD of NSAIDs consumed.

- This study is groundbreaking for evaluating CYP2C9 alleles while considering not only the use of NSAIDs use but the use of LDA.

- Our findings suggest an increased risk of UGIB in carriers of the heterozygous (CT) and variant homozygous (TT) genotypes of rs9923231 (VKORC1 gene) and in carriers of the *3 allele (rs10587910, CYP2C9 gene) associated with daily doses of NSAIDs greater than 0.5.

- CYP2C9*3 allele might be used as a predictive marker of UGIB risk for CYP2C9-metabolized NSAIDs.

- Considering that NSAIDs are one of the most commonly prescribed classes in the world, these findings are of a high clinical interest because understanding individual genetic factors associated with UGIB risk might contribute to promoting patient safety and reducing harm associated with medication.

Introduction

While several non-steroidal anti-inflammatory drugs (NSAIDs) are considered safe for over-the-counter use, especially when used short-term, they have the potential for serious adverse drug events, such as gastrointestinal toxicity and peptic ulcer (1). However, although the use of NSAIDs, including low-dose aspirin (LDA), is considered a risk factor for the development of peptic ulcer, only a few people develop peptic ulcer disease and its complications (e.g., perforated ulcer and upper gastrointestinal bleeding (UGIB)), which suggests idiosyncratic responses associated with possible individual risk factors or genetic susceptibility (2).

One hypothesis for idiosyncratic responses to the use of NSAIDs and LDA is the presence of variants in genes involved in the drug metabolism of these drugs or in physiological functions in the gastrointestinal tract system and platelet aggregation cascade (3). In a previous study we evaluated six variants in PTGS1 and NOS3 genes and an increased magnitude of UGIB risk was observed in NSAIDs and LDA users (4). In order to continue this investigation, we identified other genetic variants potentially involved in the risk of gastric injury and UGIB and the *2 (Arg144Cys, rs1799853) and *3 alleles (Ile359Leu, rs10587910) of CYP2C9 gene stand out (3).

The gene encoding the cytochrome P-450 2C9 isoenzyme (CYP2C9) is highly polymorphic and it is responsible for metabolizing most NSAIDs, including aspirin, and drugs with a narrow therapeutic range (5). The presence of CYP2C9 alleles may be associated with reduced isoenzyme activity (6) and several studies have investigated the relationship between the presence of these slow-metabolism variants and the risk of UGIB in warfarin users. Regarding the role of these variants in modifying the risk of UGIB in NSAIDs users, the data are inconsistent (7) and little is known about this risk in LDA users (3).

In warfarin users, the assessment of the *2 and *3 alleles of CYP2C9 is often associated with the rs9923231 (-1639), located in a promoter region of the vitamin K epoxide reductase complex subunit 1 (VKORC1 gene). Only one study assessed the influence of this variant in NSAIDs and LDA users and an increase of up to sevenfold in the risk of UGIB was identified in Romanians bearing the variant homozygous allele (8). However, the association data were reported grouping the use of LDA and NSAIDs; the dose of NSAIDs consumed was not considered, which prevents the gene-dose effect analysis; and it is not clear how drug exposure was measured. Besides, the authors did not assess whether the increase in UGIB risk is due either to the interaction between the genetic variant and NSAIDs and LDA use (synergism) or directly to the presence of the variant.

Along these lines, there is a lack of knowledge regarding the evidence linking genetic variation in CYP2C9 and VKORC1 genes to an increased rate of UGIB as an adverse drug event of NSAIDs and LDA use (1) and little is known about pharmacogenetic markers in the Latin American population (3). Hence, we intended to investigate whether interindividual variability in the CYP2C9 and VKORC1 genes is associated with increased risk of UGIB considering the use and gene-dose effect of LDA and NSAIDs, in addition to assessing the interaction or modification of effect between the genetic variants and LDA or NSAIDs intake on the risk of UGIB.

Methods

Study design, setting and ethical aspects

A full case-control study was conducted in the hospital complex of Clinical Hospital of the Ribeirão Preto Medical School of the University of São Paulo, Brazil, which comprises four hospitals that serve the population of the northwestern São Paulo sanitary region (49 municipalities and about 2,461,143 patients).

The Research Ethics Committee of the São Paulo State University (UNESP) (CAAE 53753115.4.3001.5426; protocol number 1.657.615) and Clinical Hospital of the Ribeirão Preto Medical School of the University of São Paulo (USP-RP) (CAAE 53753115.4.0000.5440; protocol number 1.536.886) approved the protocol of this study. Signed informed consent was obtained from all participants prior to their enrollment in this study.

The report of this study was based on Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) (9).

Definition of cases and control participants

Patients over 18 years old admitted to the hospital complex with signs and symptoms of UGIB and diagnosed by upper digestive endoscopy (UDE) or surgical intervention (laparoscopy) were determined as our study group.

UGIB was defined as: i) presence of endoscopically proven ulcers, perforation, or hemorrhagic erosions and ii) presence of dark or “coffee grounds” vomiting, melena, hematemesis, hematochezia, epidegastric pain, sudden loss, heavy sweating, and/or pallor.

Patient exclusion criteria were (i) bleeding from gastric or esophageal varices or neoplasm; (ii) presence of cirrhosis, Mallory-Weiss tears, and/or Dieulafoy lesions; (iii) serious health condition; (iv) UDE performed after 48 h of hospital admission; (v) hospitalization within 15 days prior to the current hospital admission; and (vi) in-hospital UGIB.

For each recruited case participant, controls were matched by sex, age (±5 years), and recruitment data (3 months). Control participants were admitted to preoperative units of the same hospital complex for mild surgery unrelated to gastrointestinal disorders (i.e., inguinal/umbilical hernia correction; plastic surgery; phacectomy (eye cataract); and prostatectomy).

Participants were recruited regardless of the use of NSAIDs and LDA in order to verify whether the proposed genetic variants are associated with the risk of UGIB or whether there is synergism between the variants and the use of these drugs in the risk of UGIB. Hence, it is essential that both case and control groups include NSAIDs or LDA-exposed and NSAIDs or LDA-unexposed individuals to assess the likely direct effect of functional variants on risk of suffering UGIB (7).

In order to reduce possible bias, only biologically unrelated participants were included. Participants with history of neoplasia, immunodeficiency syndrome, coagulopathies, nasogastric or percutaneous tube holders; patients who use narcotics; and non-residents of the study region for at least 3 months were excluded.

The inclusion and exclusion criteria for the participants are described in detail in our three previous studies (4, 10, 11).

Data collection

Data collection was conducted face-to-face by four previously trained researchers (MF, GU, TRN and PCM), during the period from July 2016 to March 2020.

The interviews were carried out using a questionnaire previously designed for this study and validated to be applied in Brazil (12). The questionnaire comprises sociodemographic variables (sex; age; self-reported race (white, black, mixed, and East Asian); schooling); clinical (body mass index; previous personal history of gastrointestinal disorders (ulcer; dyspepsia; and bleeding); Helicobacter pylori infection; comorbidities; and drug therapy in use); and lifestyle habits (smoking, alcohol, and coffee intake).

The interviews were conducted with the patient and/or family member and the data were also consulted in secondary sources (electronic medical records, medical prescriptions, and laboratory tests). At the end of each interview, the researchers assigned a score from zero to 10 according to the quality of the information recalled (consistency of the interview).

After the interview, the venous blood (5 mL) was collected in ethylenediamine tetraacetic acid (EDTA) from all participants for posterior analysis of the proposed genetic variants and serology for Helicobacter pylori infection.

Definition of drug exposure and lifestyle

An index date was considered, which is deemed for each case as the day of appearance of dark vomiting or in coffee grounds, melena, and/or hematemesis (signs and symptoms of UGIB) and for controls the index date was defined as the date of the interview (13). To assess an association between the use of NSAIDs and LDA with the risk of UGIB, a 7-day etiologic window dated from the index date was considered, in line with other published studies (12, 14).

For CYP2C9 analysis, NSAIDs use was grouped into NSAIDs metabolized at least 50% by CYP2C9 (piroxicam, celecoxib, naproxen, aceclofenac, indomethacin, diclofenac, and ibuprofen) and into NSAIDs metabolized less than 50% by CYP2C9 (other NSAIDs) (12). It is known that NSAIDs, as they are substrates of CYP2C9, are not metabolized at the same proportion of this enzyme and to perform an analysis considering the proportion of each NSAIDs metabolized by CYP2C9 increases the sample size power (7). LDA use was deemed the continuous use of aspirin at doses below 300 mg per day in the indication of prevention of primary and secondary cardiovascular events (15).

To perform an analysis of dose-effect, two researchers calculated the defined daily dose (DDD) by the World Health Organization (WHO) for NSAIDs for all participants. DDD was defined as the average maintenance dose per day of a drug used for its main indication in adults in the 7-day etiologic window preceding the data index. The dose-response effect was assessed using three categories: NSAIDs non-users (NSAIDs DDD = 0), NSAIDs users of 0.50 DDD or less (>0 NSAIDs DDD ≤0.50), and NSAIDs users of over 0.50 DDD (NSAIDs DDD >0.50) based on the proposal by Figueiras et al. (12). This approach considered each type of NSAIDs and the recommended DDD, enabling different NSAIDs to be compared with one another (7).

Regarding lifestyle, the mean daily consumption of tobacco, alcohol, and coffee over the 2 months preceding the index was calculated. Smoking habit was stratified according to the number of cigarettes consumed per day: non-smokers and ex-smokers (zero cigarette); 1 to 15 cigarettes/day; and >15 cigarettes/day. Alcohol intake was stratified in abstainers (0 g), >0 to ≤30 g of alcohol/day, and >30 g of alcohol. Coffee intake was stratified into none consumed (0 mL), >0 mL ≤ 100, >100 > mL ≤ 300, and >300 mL.

Helicobacter pylori serology

For all the participants, the presence of IgG antibodies for Helicobacter pylori infection (anti-H. pylori IgG antibodies) was determined using the chemiluminescence technique. This parameter was stratified into reagent (≥1.10 U IgG/mL), non-reagent (≤0.90 U IgG/mL), and inconclusive (0.91–1.09 U IgG/mL). In addition, all participants were queried about previous Helicobacter pylori infection.

Selection of genetic variants and genotyping

The genetic variants proposed in this study were selected through a systematic review conducted by our research group (3). The single nucleotide polymorphisms CYP2C9 Arg144Cys (rs1799853, A > C; Assay C_25625805_10); CYP2C9 Ile359Leu (rs10587910, C > T, Assay C_27104892_10); and rs9923231 (−1,639, C > T, Assay C_30403261_20) were evaluated.

Genomic DNA was extracted from venous blood samples using the Maxwell® 16 Blood DNA Purification Kit (Promega, Madison, WI, United States). DNA concentration and purity were determined by fluorescence technique using the Invitrogen Qubit 4 Fluorometer and the Qubit ™ dsDNA HS Assay kit (Applied Biosystems, Foster City, United States).

Genetic variants genotyping was performed on genomic DNA by real-time polymerase chain reaction technique using the TaqPath ProAmp SNP Genotyping Assay system (Applied Biosystems, Foster City, United States) and StepOne Plus equipment (Applied Biosystems, Foster City, United States).

The cycling conditions were validated as 60°C for 1 min; 95°C for 10 min; 40 cycles of 95°C for 15 s and 60°C for 1 min; and final extension of 60°C for 1 min.

For internal quality control, for each of the variants, 10% of the samples were randomly selected for repeat genotyping (90 samples per variant). The agreement between the two sets of genotyping was 100%.

Genotyping plots were analyzed using the Data Connect cloud from Applied Biosystems (Foster City, United States).

Statistical analysis

Intergroup comparison for qualitative variables was calculated using the χ2 test or Fisher’s exact test, if appropriate, and Student's t-test for quantitative variables. The difference between the allele frequency of the genetic variants between the case and control groups was tested with χ2 test.

Hardy-Weinberg Equilibrium

The genotype distributions of the genetic variants proposed were verified in the control group conforming to Hardy–Weinberg Equilibrium using χ2 test (BioEstat 5.0 program).

Unconditional regression models

The association between the CYP2C9 and VKORC1 variant alleles and the risk to develop UGIB was estimated by odds ratio (OR) with 95% confidence interval (95% CI) using unconditional logistic regression models (SPSS v.26 software (IBM Company, Chicago, IL, United States)).

Firstly, a bivariate analysis was carried out to test the effect of the potentially confounding variables. Variables with p-value ≤0.10 were selected to compose the unconditional logistic regression models. The variables sex and age were not included in the models, since they were controlled in the study design by matching the participants of the case and control groups.

The unconditional logistic regression model was built with all confounding variables selected in the bivariate analysis (insertion method of the SPSS Program).

To evaluate the CYP2C9 alleles, participants were grouped in carriers of CYP2C9*1, CYP2C9*2 and CYP2C9*3 alleles and in carriers of genotypes without *3 allele (*1/*1; *1/*2, or *2/*2) and carriers with *3 allele (*1/*3; *3/*3, or *2/*3). For the VKORC1 alleles, the participants were grouped in carriers of wild type homozygous, heterozygous, and variant homozygous genotypes. Both groups considered the dose of NSAIDs consumed or LDA use.

Multiplicative and additive scale interaction

For interaction analysis, the participants were divided into four groups according to genotype and NSAIDs/LDA use status, being:

00: absence of genetic variation and no use of NSAIDs and/or LDA

01: absence of genetic variation and use of NSAIDs and/or LDA

10: genetic variation and no use of NSAIDs and/or LDA

11: genetic variation and use of NSAIDs and/or LDA

Multiplicative interactions between the presence of the CYP2C9 and VKORC1 alleles and the use of NSAIDs or LDA was calculated by fitting the unconditional regression models.

The additive interaction between the presence of the CYP2C9 and VKORC1 alleles and the use of NSAIDs or LDA was estimated by employing the Synergy Index (SI) and Relative Excess Risk Due to Interaction (RERI) parameters (16) as proposed by Anderson and collaborators (17). 95%CI was constructed according to the method proposed by Figueiras et al. (18).

$\mathrm{S}\mathrm{I}=\left({\mathrm{O}\mathrm{R}}_{11}–1\right)/\left(\left({\mathrm{O}\mathrm{R}}_{01}–1\right)+\left({\mathrm{O}\mathrm{R}}_{10}–1\right)\right)$

$\mathrm{R}\mathrm{E}\mathrm{R}\mathrm{I}={\mathrm{O}\mathrm{R}}_{11}–{\mathrm{O}\mathrm{R}}_{01}–{\mathrm{O}\mathrm{R}}_{10}+1$

SI > 1 or RERI >0 indicated a significant additive interaction, which represents that the combined effects of the presence of the variant alleles and exposure to NSAIDs or LDA on UGIB risk are greater than the effects of drug exposure or presence of variant alleles alone.

Regression models outlined for multiplicative and additive interactions analysis were fitted for the selected confounding variables in the bivariate analysis (p-value ≤0.10).

Model fit measures

To avoid distortions of the model effect measures, the potential problem of multicollinearity between independent variables was evaluated based on correlation matrix of regression coefficients.

Four methods were used to measure the goodness and quality of the statistical models’ fit: Akaike information criterion (AIC), Bayesian information criteria (BIC), Hosmer and Lemeshow, and likelihood ratio test (2LL) (2LL block zero – 2LL in block one). A low AIC and BIC scores indicate a better model fit for predicting outcome.

Results

Baseline characteristics of case and Control Participants

Most participants were male (case group: 72.5% and control group: 72.5%), self-declared white (case group: 81.3% and control group: 84.2%), with zero to 4 years of schooling (case group: 52.5% and control group: 52.4%), and with a mean age of 60.0 (±15.9) for case group and 59.9 (±15.9) years for control group. The demographic and clinical characteristics of the participants are described in Table 1.

TABLE 1

TABLE 1. Demographic and clinical characteristics of the participants of case (n = 200) and control (n = 706) groups.

As indicated in Table 1, participants in the case group received more NSAIDs and LDA (21.5% and 25.5%) when compared with the participants of the control group (10.0% and 12.7%). Concomitant use of NSAIDs and LDA was reported by 4.0% of cases and 1.1% of controls. 14.5% of the participants in the case group were using NSAIDs metabolized by CYP2C9 and 7.0% of the participants in the control group, and the most commonly used NSAIDs were diclofenac and nimesulide. In relation to DDD, 2.5% of cases and 0.8% of controls used NSAIDs in DDD of 1.0 upward.

The three variants were in accordance with the Hardy-Weinberg Equilibrium: CYP2C9*2 (p-value 0.6773 for the case group and 0.7144 for the control group), CYP2C9*3 (p-value 0.9492 for the case group and 0.9570 for the control group), and VKORC1 (p-value 0.5407 for the case group and 0.5076 for the control group). No differences were observed between the frequency of CYP2C9 and VKORC1 genotypes in the case and control groups (Table 2).

TABLE 2

TABLE 2. Frequency of CYP2C9 and VKORC1 genotypes among participants in the case (n = 200) and control (n = 706) groups.

Association of genetic variants and the risk of UGIB

Risk of UGIB associated with each allele was analyzed taking the NSAIDs dose consumed into account (Table 3).

TABLE 3

TABLE 3. Risk for upper gastrointestinal bleeding associated with CYP2C9 and VKORC1 genotypes and dose of non-steroidal anti-inflammatory drugs (NSAIDs) consumed.

For DDDs of NSAIDs up to 0.50, the risk of UGIB was similar for wild type homozygous (OR: 4.871, 95% CI: 1.98–11.96) and carriers of the *2 allele (OR: 6.244, 95% CI 1.56–24.80). It was not possible to evaluate UGIB risk among carriers of the *3 allele in use of DDDs of NSAIDs up to 0.50, as there was no case in this category. For DDDs of NSAIDs upward 0.50, the risk of UGIB was much lower for the wild type homozygous (OR: 6.676, 95% CI: 1.79–24.90) when compared with carriers of the *3 allele (OR: 15.650, 95% CI: 1.41–174.10). From DDDs of NSAIDs upward 0.50, a greater risk of UGIB was observed among carriers of CYP2C9*3 genotypes (15.086, 95% CI 1.34–169.66) when compared to carriers of genotypes with *2 allele and/or wild type homozygous (OR: 7.368, 95% CI: 2.19–24.74).

For the VKORC1 gene, a higher risk of UGIB was observed in carriers of the variant homozygous genotype (TT) in use of DDDs of NSAIDs upward of 0.50 (OR: 38.850, 95% CI: 2.70–556.00) when compared to carriers of the heterozygous genotype (CT) (OR: 6.025, 95% CI: 2.40–15.12).

In LDA users, the risk of UGIB was observed to be similar between carriers of the wild type homozygous genotype and carriers of the variant alleles for the CYP2C9 and VKORC1 genes (Table 4).

TABLE 4

TABLE 4. Risk for upper gastrointestinal bleeding associated with CYP2C9 and VKORC1 genotypes and use of low-dose aspirin (LDA).

All these models were adjusted for the following confounding variables: body mass index kg/m²; personal history of ulcer and dyspepsia; Helicobacter pylori infection; cardiovascular disease; dyslipidemia; use of oral anticoagulants, other antiplatelet agents, NSAIDs drugs; tobacco consumption; alcohol intake; coffee intake; and interview consistency.

No additive interaction was identified between the presence of variant alleles of the CYP2C9 and VKORC1 genes and the consumption of NSAIDs or LDA. Although not statistically significant, the absence of the CYP2C9*3 genotype (RERI: −2.17, CI 95%: −5.16; 0.82) and rs9923231 variant (VKORC1) (RERI: −1.23, CI 95%: −4.49; 2.03) appear to be “negative” modifiers for the risk of UGIB in LDA users (Table 5).

TABLE 5

TABLE 5. Multiplicative and additive interaction between the presence of the CYP2C9 e VKORC1 alleles and the use of low-dose aspirin (LDA) and non-steroidal anti-inflammatory drugs (NSAIDs) in the risk of upper gastrointestinal bleeding.

All these models were adjusted for the following confounding variables: body mass index kg/m²; personal history of ulcer and dyspepsia; Helicobacter pylori infection; cardiovascular disease; dyslipidemia; use of oral anticoagulants, LDA, other antiplatelet agents, and NSAIDs; tobacco consumption; alcohol intake; coffee intake; and interview consistency.

Discussion

To the best of our knowledge, this is the first full case-control study to assess the role of rs9923231 (VKORC1 gene) in the risk of UGIB according to the genotype and the use of LDA and DDD of NSAIDs consumed. To consider the dose of NSAIDs is essential, since there is evidence of the presence of gene-dose effect and dose-dependency in NSAIDs-induced gastrointestinal damage (7, 12). Additionally, this study is groundbreaking for evaluating CYP2C9 alleles considering not only the use of NSAIDs use but the use of LDA and to explore additive synergism interaction between the VKORC1 and CYP2C9 variants and NSAIDs/LDA on the UGIB risk (RERI and SI).

Our findings suggest an increased risk of UGIB in carriers of the heterozygous (CT) and variant homozygous (TT) genotypes of rs9923231 (VKORC1 gene) and in carriers of the *3 allele (rs10587910, CYP2C9 gene) associated with increased dose of NSAIDs consumed. Considering that NSAIDs are one of the most commonly prescribed classes in the world, the limited understanding of the molecular mechanisms involved in idiosyncratic responses (1), and the public health and economic burden of UGIB (2), these findings are of a high clinical interest because knowing individual genetic factors associated with UGIB risk may contribute to promoting patient safety and reducing the medication harm.

The VKORC1 gene encodes the vitamin K epoxide reductase protein, a key enzyme in the vitamin K cycle (19), and the rs9923231 is one of the most studied variants. It is suggested that this variant is associated with a low-dose phenotype and changes in the transcription factor binding site of the gene (20), and luciferase assays show that the activity of the wild type allele of rs9923231 is increased by 44% over the activity of the homozygous variant allele (20). In this respect, one hypothesis for the increased risk of UGIB observed in carriers of the variant allele of rs9923231 would be this variant, located in a promoter region, which might lead to reduced promoter activity and, consequently, reduced levels of functional clotting factors and an increased likelihood of bleeding (8).

A previous study investigated whether the rs9923231 would be involved in the increased risk of UGIB in LDA and NSAID users (8), since this variant is widely evaluated in vitamin K antagonist users, as it is associated with increased sensitivity to warfarin and a low dose requirement to prevent bleeding episodes (21). The authors included 341 participants and identified about seven-fold increased risk in carriers of the variant homozygous genotype using NSAIDs or LDA (OR: 3.12–18.74) (8), whereas in our study we included 906 participants and only the users of NSAIDs in DDDs upward 0.50 bearing the variant homozygous genotype had an increased risk of UGIB (OR: 38.850, 95% CI: 2.70–556.00). It is well known that the risk of NSAID-induced gastrointestinal complications is dose-dependent and remains linear over time (22). Therefore, it is tough to compare the authors’ data with our findings, as the use of LDA and NSAIDs was grouped and the dose of NSAIDs consumed was not considered. In addition, in our study users of LDA carrying the variant allele did not present a higher risk of UGIB.

Regarding CYP2C9, variants in this gene may result in modified expression or functionality, correlating with altered drug metabolism and clearance, which may affect drug bioavailability (23) and it was suggested that the “decreased-function” and “no-function” CYP2C9 alleles are substrate-dependent (22). Indeed, a reduction in CYP2C9 activity of about 20–30% was observed in carriers of the *2 allele and up to 70% in carriers of the *3 allele (6), in addition to variations in NSAIDs metabolism as well as NSAID-induced adverse drug reactions (22, 24).

We are aware of only one previous full case-control study that evaluated the influence of CYP2C9 alleles and dose of NSAIDs consumed on UGIB risk (12), since previous studies were based on NSAIDs users (partial case control) (25, 26), which prevents assessing whether the risk of UGIB is directly related to the presence of a functional variant or interaction between slow-metabolizing variants and use of NSAIDs (7).

Figueiras et al. (2016) identified that carriers of the allele *3 users of NSAIDs in DDDs greater than 0.5 have a higher risk of UGIB (OR: 16.920, 95% CI: 4.96–57.59) (12), corroborating with our findings (OR: 15.650, 95% CI: 1.41–174.10). This finding may suggest the presence of a gene-dose effect and the risk and severity of NSAIDs toxicity are likely to be increased in individuals with a poor metabolizer phenotype of CYP2C9 because they are expected to exhibit a pronounced prolongation of drug half-life and increase in plasmatic concentration (27, 28). In line with previous studies, the *2 allele was not associated with an increased risk of UGIB (12, 25) and this allele was reported as a poor risk predictor of gastrointestinal adverse events (24).

From a practical point of view, thinking about the safety of the NSAIDs user, therapeutic recommendations involve dose reduction or alternative therapies, including NSAIDs not primarily metabolized by CYP2C9 or with pharmacokinetic not impacted by CYP2C9 genetic variants (1, 23), coupled with careful monitoring of adverse drug events (1). The care of NSAIDs users should be extended even further in the older person, considering that multimorbidity and polypharmacy increase the likelihood of drug and gene-based interactions (29). Besides, the older person is likely to be prescribed NSAIDs (29, 30) and this pharmacological class is considered inappropriate for older people (31).

Regarding the use of LDA, it was not identified modification of the magnitude of risk of UGIB in individuals bearing the *2 and *3 alleles (CYP2C9 gene) or the variant allele of rs9923231 (VKORC1 gene) and the pharmacokinetics of LDA does not seem to be significantly affected by these genetic variants (1). In contrast to NSAIDs, CYP metabolism may only play a minor role in LDA bioavailability and no strong genetic associations with aspirin-related gastrointestinal events have been described so far (32, 33).

Notwithstanding, our data corroborates evidence on the pharmacogenomics of LDA, which is less explored in Brazilian studies (21), despite UGIB being the most commom adverse drug event observerd in 72% of aspirin-related hospitalizations (34). In addition, gastrointestinal symptoms in chronic LDA users are challenging and are a result of non/poor compliance and discontinuation of treatment (35), which may be associated with an increased risk of new cardiovascular events (36).

Ultimately, although it is known that the adoption of Pharmacogenetics-guided prescribing in Brazil has been sporadic due to a combination of factors, such as the structure of the Brazilian population and barriers to its implementation (37), genetic data is a useful tool to improve pharmacotherapy, reduce the occurrence of adverse drug reactions, and change the paradigm of dosing regimens being extrapolated to entire populations (38, 39). Thus, our findings may contribute to personalized therapy and to improve the detection of the NSAIDs-related UGIB signal, since improving risk communication and signal detection is one of the demands of Brazilian Pharmacovigilance (40).

Our study has strength and limitations. This is the first time this type of study was performed to explore pharmacogenomics associations in individuals with admixed genetic ancestry in addition to collecting data through face-to-face interviews and adjusting the analyses for several UGIB confounders. The main limitation of this study is the sample size, because the stratification of participants according to genotype and DDD of NSAIDs consumed meant that some categories had a limited number of observations, which reduces statistical power and is a frequent limitation in genetic studies (41). Furthermore, additional studies including analyzes of ancestry-informative markers would be relevant, as there are differences in the distribution of variants in the CYP2C9 and VKORC1 genes, especially as the proportion of African ancestry increases (35).

In summary, in this exploratory study, our findings suggest an increased risk of UGIB in carriers of the heterozygous (CT) and variant homozygous (TT) genotypes of rs9923231 (VKORC1 gene) and in carriers of the *3 allele (CYP2C9 gene) associated with increased dose of NSAIDs consumed. Besides, no additive synergism interaction was identified and the joint biological effect of these variants and NSAIDs/LDA use on the risk of UGIB is similar to the sum of their individual effects.

Hence, our findings may contribute to personalized pharmacotherapy and to improve the detection of the NSAIDs-related adverse drug reactions.

Data availability statement

The raw data supporting the conclusion of this article will be made available by the authors, without undue reservation, to any qualified researcher. Besides, the full-methodology protocol, interview form is available at Open Science Framework (OSF), an open access repository (doi: 10.17605/OSF.IO/4SG93).

Ethics statement

The studies involving human participants were reviewed and approved by Research Ethics Committee of the São Paulo State University (UNESP) (number 1.657.615) and Clinical Hospital of the Ribeirão Preto Medical School of the University of São Paulo (USP-RP) (number 1.536.886). This study was also registered in the Registro Brasileiro de Ensaios Clínicos (REBEC-number: RBR-3hstqm). The patients/participants provided their written informed consent to participate in this study.

Author contributions

MF participated in methodology, investigation, patients’ recruitment, genetic analysis, statistical analysis, data discussion, writing-original draft, and writing-review and editing. GU and TRN participated in patients’ recruitment and writing-review. SSB and ATF participated in genetic analysis methodology, data discussion, and writing-review. PCM participated in project administration, funding acquisition, methodology, patients’ recruitment, and writing-review. All authors read and approved the final manuscript version.

Funding

The Authors thank Conselho Nacional de Desenvolvimento Científico e Tecnológico—CNPq (funding number: 401060/2014-4), Fundação de Amparo à Pesquisa do Estado de São Paulo—FAPESP (funding numbers 2017/24193-3 and 2018/07501-9), Programa de Apoio ao Desenvolvimento Científico—FCF-UNESP (PADC), Programa de Pós-graduação em Ciências Farmacêuticas e Pró-reitoria de Pesquisa UNESP (Edital PROPG-PROPe 04/2022), and Pró-reitoria de Extensão Universitária e Cultura. This study was financed in part by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior-Brasil (CAPES)—Código de Financiamento 001.

Conflict of interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Acknowledgments

The Authors thank Professor Dr. Maria Teresa Herdeiro (University of Aveiro), Professor Dr. Adolfo Figueiras Guzmán (University of Santiago de Compostela), and Professor Dr. Maruxa Zapata-Cachafeiro (University of Santiago de Compostela) for methodological contributions in this study. Ana Julia Machry and Centro de Medicina Genômica of Clinical Hospital of the Ribeirão Preto Medical School of the University of São Paulo (CMG-HC-FMRP-USP) for genetic analysis methodology contributions. Professor Dr. Romeu Magnani for the statistical support. Community Service Center (NAC) of the School of Pharmaceutical Sciences of the São Paulo State University (UNESP) and Álvaro Support Laboratory for Helicobacter pylori serology.

References

1. Theken, KN, Lee, CR, Gong, L, Caudle, KE, Formea, CM, Gaedigk, A, et al. Clinical pharmacogenetics implementation consortium guideline (CPIC) for CYP2C9 and nonsteroidal anti-inflammatory drugs. Clin Pharmacol Ther (2020) 108(2):191–200. doi:10.1002/cpt.1830

PubMed Abstract | CrossRef Full Text | Google Scholar

2. Lanas, A, and Chan, FKL. Peptic ulcer disease. The Lancet (2017) 390(10094):613–24. doi:10.1016/S0140-6736(16)32404-7

CrossRef Full Text | Google Scholar

3. Forgerini, M, Lucchetta, RC, Urbano, G, de Nadai, TR, and de Carvalho Mastroianni, P. Genetic polymorphisms associated with upper gastrointestinal bleeding: A systematic review. Pharmacogenomics J (2021) 21(1):20–36. doi:10.1038/s41397-020-00185-6

PubMed Abstract | CrossRef Full Text | Google Scholar

4. Forgerini, M, Urbano, G, de Nadai, TR, Batah, SS, Fabro, AT, and Mastroianni, Pd C. Genetic variants in PTGS1 and NOS3 genes increase the risk of upper gastrointestinal bleeding: A case–control study. Front Pharmacol (2021) 12:671835. doi:10.3389/fphar.2021.671835

PubMed Abstract | CrossRef Full Text | Google Scholar

5. Hirota, T, Eguchi, S, and Ieiri, I. Impact of genetic polymorphisms in CYP2C9 and CYP2C19 on the pharmacokinetics of clinically used drugs. Drug Metab Pharmacokinet (2013) 28(1):28–37. doi:10.2133/dmpk.dmpk-12-rv-085

PubMed Abstract | CrossRef Full Text | Google Scholar

6. Zhou, S-F, Ming Di, Y, Chan, E, Du, Y-M, Chow, V, Xue, C, et al. Clinical pharmacogenetics and potential application in personalized medicine. Curr Drug Metab (2008) 9(8):738–84. doi:10.2174/138920008786049302

PubMed Abstract | CrossRef Full Text | Google Scholar

7. Estany-Gestal, A, Salgado-Barreira, A, Sánchez-Diz, P, and Figueiras, A. Influence of CYP2C9 genetic variants on gastrointestinal bleeding associated with nonsteroidal anti-inflammatory drugs. Pharmacogenetics and Genomics (2011) 21(7):357–64. doi:10.1097/FPC.0b013e328346d2bb

PubMed Abstract | CrossRef Full Text | Google Scholar

8. Groza, I, Matei, D, Tanţău, M, Trifa, AP, Crişan, S, Vesa, ŞC, et al. VKORC1-1639 G>A polymorphism and the risk of non-variceal upper gastrointestinal bleeding. J Gastrointest Liver Dis (2017) 26(1):13–8. doi:10.15403/jgld.2014.1121.261.vko

CrossRef Full Text | Google Scholar

9. von Elm, E, Altman, DG, Egger, M, Pocock, SJ, Gøtzsche, PC, and Vandenbroucke, JP. The strengthening the reporting of observational studies in Epidemiology (STROBE) statement: Guidelines for reporting observational studies. J Clin Epidemiol (2008) 61(4):344–9. doi:10.1016/j.jclinepi.2007.11.008

PubMed Abstract | CrossRef Full Text | Google Scholar

10. Forgerini, M, Urbano, G, Nadai, Td, Zapata-Cachafeiro, M, Kemp, R, and Mastroianni, PDC. Epidemiological profile of patients with non-variceal upper gastrointestinal bleeding secondary to peptic disease in a tertiary referral Brazilian hospital. Arq Gastroenterol (2021) 58:202–9. doi:10.1590/S0004-2803.202100000-36

PubMed Abstract | CrossRef Full Text | Google Scholar

11. Forgerini, M, Urbano, G, De Nadai, TR, Zanelli, CF, Valentini, SR, and Mastroianni, PDC. VNTR polymorphism in intron 4 of the eNOS gene and the risk of gastrointestinal bleeding: A case-control study. J Gastrointest Liver Dis (2022) 31:176–83. doi:10.15403/jgld-4226

CrossRef Full Text | Google Scholar

12. Figueiras, A, Estany-Gestal, A, Aguirre, C, Ruiz, B, Vidal, X, Carvajal, A, et al. CYP2C9 variants as a risk modifier of NSAID-related gastrointestinal bleeding. Pharmacogenetics and Genomics (2016) 26(2):66–73. doi:10.1097/FPC.0000000000000186

PubMed Abstract | CrossRef Full Text | Google Scholar

13. Lanas, A, Bajador, E, Serrano, P, Fuentes, J, Carreño, S, Guardia, J, et al. Nitrovasodilators, low-dose aspirin, other nonsteroidal antiinflammatory drugs, and the risk of upper gastrointestinal bleeding. N Engl J Med (2000) 343(12):834–9. doi:10.1056/NEJM200009213431202

PubMed Abstract | CrossRef Full Text | Google Scholar

14. Mallah, N, Zapata-Cachafeiro, M, Aguirre, C, Ibarra-García, E, Palacios–Zabalza, I, Macías-García, F, et al. Influence of polymorphisms involved in platelet activation and inflammatory response on aspirin-related upper gastrointestinal bleeding: A case-control study. Front Pharmacol (2020) 11:860–10. doi:10.3389/fphar.2020.00860

PubMed Abstract | CrossRef Full Text | Google Scholar

15. Lanas, Á, Carrera-Lasfuentes, P, Arguedas, Y, García, S, Bujanda, L, Calvet, X, et al. Risk of upper and lower gastrointestinal bleeding in patients taking nonsteroidal anti-inflammatory drugs, antiplatelet agents, or anticoagulants. Clin Gastroenterol Hepatol (2015) 13(5):906–12.e2. e2. doi:10.1016/j.cgh.2014.11.007

PubMed Abstract | CrossRef Full Text | Google Scholar

16. Hosmer, DW, and Lemeshow, S. Confidence interval estimation of interaction. Epidemiology (1992) 3(5):452–6. doi:10.1097/00001648-199209000-00012

PubMed Abstract | CrossRef Full Text | Google Scholar

17. Andersson, T, Alfredsson, L, Källberg, H, Zdravkovic, S, and Ahlbom, A. Calculating measures of biological interaction. Eur J Epidemiol (2005) 20(7):575–9. doi:10.1007/s10654-005-7835-x

PubMed Abstract | CrossRef Full Text | Google Scholar

18. Figueiras, A, Domenech-Massons, JM, and Cadarso, C. Regression models: Calculating the confidence interval of effects in the presence of interactions. Stat Med (1998) 17(18):2099–105. doi:10.1002/(sici)1097-0258(19980930)17:18<2099::aid-sim905>3.0.co;2-6

PubMed Abstract | CrossRef Full Text | Google Scholar

19. Owen, RP, Gong, L, Sagreiya, H, Klein, TE, and Altman, RB. VKORC1 pharmacogenomics summary. Pharmacogenetics and Genomics (2010) 20(10):642–4. doi:10.1097/FPC.0b013e32833433b6

PubMed Abstract | CrossRef Full Text | Google Scholar

20. Yuan, H-Y, Chen, J-J, Lee, MM, Wung, J-C, Chen, Y-F, Charng, M-J, et al. A novel functional VKORC1 promoter polymorphism is associated with inter-individual and inter-ethnic differences in warfarin sensitivity. Hum Mol Genet (2005) 14(13):1745–51. doi:10.1093/hmg/ddi180

PubMed Abstract | CrossRef Full Text | Google Scholar

21. Hirata, TDC, Dagli-Hernandez, C, Genvigir, FDV, Lauschke, VM, Zhou, Y, Hirata, MH, et al. Cardiovascular pharmacogenomics: An update on clinical studies of antithrombotic drugs in Brazilian patients. Mol Diagn Ther (2021) 25(6):735–55. doi:10.1007/s40291-021-00549-z

PubMed Abstract | CrossRef Full Text | Google Scholar

22. McEvoy, L, Carr, DF, and Pirmohamed, M. Pharmacogenomics of NSAID-induced upper gastrointestinal toxicity. Front Pharmacol (2021) 12:684162. doi:10.3389/fphar.2021.684162

PubMed Abstract | CrossRef Full Text | Google Scholar

23. Guengerich, FP. A history of the roles of cytochrome P450 enzymes in the toxicity of drugs. Toxicol Res (2021) 37(1):1–23. doi:10.1007/s43188-020-00056-z

PubMed Abstract | CrossRef Full Text | Google Scholar

24. Macías, Y, Gómez Tabales, J, García-Martín, E, and Agúndez, JAG. An update on the pharmacogenomics of NSAID metabolism and the risk of gastrointestinal bleeding. Expert Opin Drug Metab Toxicol (2020) 16(4):319–32. doi:10.1080/17425255.2020.1744563

PubMed Abstract | CrossRef Full Text | Google Scholar

25. Pilotto, A, Seripa, D, Franceschi, M, Scarcelli, C, Colaizzo, D, Grandone, E, et al. Genetic susceptibility to nonsteroidal anti-inflammatory drug–related gastroduodenal bleeding: Role of cytochrome P450 2C9 polymorphisms. Gastroenterology (2007) 133(2):465–71. doi:10.1053/j.gastro.2007.05.025

PubMed Abstract | CrossRef Full Text | Google Scholar

26. Musumba, CO, Jorgensen, A, Sutton, L, Van Eker, D, Zhang, E, O’Hara, N, et al. CYP2C19*17 gain-of-function polymorphism is associated with peptic ulcer disease. Clin Pharmacol Ther (2013) 93(2):195–203. doi:10.1038/clpt.2012.215

PubMed Abstract | CrossRef Full Text | Google Scholar

27.Coxib and traditional NSAID Trialists' CNT Collaboration, Arber, N, Bhala, N, Emberson, J, Merhi, A, Abramson, S, Baron, JA, et al. Vascular and upper gastrointestinal effects of non-steroidal anti-inflammatory drugs: meta-analyses of individual participant data from randomised trials. Lancet (2013) 382(9894):769–79. doi:10.1016/S0140-6736(13)60900-9

PubMed Abstract | CrossRef Full Text | Google Scholar

28. Castellsague, J, Riera-Guardia, N, Calingaert, B, Varas-Lorenzo, C, Fourrier-Reglat, A, Nicotra, F, et al. Individual NSAIDs and upper gastrointestinal complications. Drug Saf (2012) 35(12):1127–46. doi:10.1007/BF03261999

PubMed Abstract | CrossRef Full Text | Google Scholar

29. Turner, RM, de Koning, EM, Fontana, V, Thompson, A, and Pirmohamed, M. Multimorbidity, polypharmacy, and drug-drug-gene interactions following a non-ST elevation acute coronary syndrome: Analysis of a multicentre observational study. BMC Med (2020) 18(1):367. doi:10.1186/s12916-020-01827-z

PubMed Abstract | CrossRef Full Text | Google Scholar

30. Lanas, A, Dumonceau, J-M, Hunt, RH, Fujishiro, M, Scheiman, JM, Gralnek, IM, et al. Non-variceal upper gastrointestinal bleeding. Nat Rev Dis Primers (2018) 4(1):18020. doi:10.1038/nrdp.2018.20

PubMed Abstract | CrossRef Full Text | Google Scholar

31. Schiavo, G, Forgerini, M, Lucchetta, RC, and Mastroianni, Pd C. A comprehensive look at explicit screening tools for potentially inappropriate medication: A systematic scoping review. Australas J Ageing (2022) 41(3):357–82. doi:10.1111/ajag.13046

PubMed Abstract | CrossRef Full Text | Google Scholar

32. Agundez, J, Martinez, C, Perez-Sala, D, Carballo, M, Torres, M, and Garcia-Martin, E. Pharmacogenomics in aspirin intolerance. Curr Drug Metab (2009) 10(9):998–1008. doi:10.2174/138920009790711814

PubMed Abstract | CrossRef Full Text | Google Scholar

33. Agúndez, JAG, Formea, C, Gaedigk, A, García-Martín, E, Gong, L, Grosser, T, et al. Editorial: NSAIDs pharmacogenomics. Front Pharmacol (2021) 12:798447. doi:10.3389/fphar.2021.798447

PubMed Abstract | CrossRef Full Text | Google Scholar

34. Pirmohamed, M, James, S, Meakin, S, Green, C, Scott, AK, Walley, TJ, et al. Adverse drug reactions as cause of admission to hospital: Prospective analysis of 18 820 patients. BMJ (2004) 329(7456):15–9. doi:10.1136/bmj.329.7456.15

PubMed Abstract | CrossRef Full Text | Google Scholar

35. Cayla, G, Collet, J-P, Silvain, J, Thiefin, G, Woimant, F, and Montalescot, G. Prevalence and clinical impact of upper gastrointestinal symptoms in subjects treated with low dose aspirin: The UGLA survey. Int J Cardiol (2012) 156(1):69–75. doi:10.1016/j.ijcard.2010.10.027

PubMed Abstract | CrossRef Full Text | Google Scholar

36. Sostres, C, and Lanas, A. Should prophylactic low-dose aspirin therapy be continued in peptic ulcer bleeding? Drugs (2011) 71(1):1–10. doi:10.2165/11585320-000000000-00000

PubMed Abstract | CrossRef Full Text | Google Scholar

37. Rodrigues-Soares, F, and Suarez-Kurtz, G. Pharmacogenomics research and clinical implementation in Brazil. Basic Clin Pharmacol Toxicol (2019) 124(5):538–49. doi:10.1111/bcpt.13196

PubMed Abstract | CrossRef Full Text | Google Scholar

38. Jaccard, E, Redin, C, Girardin, F, Waeber, G, Fellay, J, and Vollenweider, P. Pharmacogenomics: A toolbox to improve drug prescription. Revue Médicale Suisse (2020) 16(716):2259–63. doi:10.53738/revmed.2020.16.716.2259

PubMed Abstract | CrossRef Full Text | Google Scholar

39. Suarez-Kurtz, G. Pharmacogenetics in the Brazilian population. Front Pharmacol (2010) 1:118. doi:10.3389/fphar.2010.00118

PubMed Abstract | CrossRef Full Text | Google Scholar

40. Varallo, FR, Forgerini, M, Herdeiro, MT, and de Carvalho Mastroianni, P. Harmonization of pharmacovigilance regulation in Brazil: Opportunities to improve risk communication. Clin Ther (2019) 41(3):598–603. doi:10.1016/j.clinthera.2019.01.013

PubMed Abstract | CrossRef Full Text | Google Scholar

41. Dumas-Mallet, E, Button, KS, Boraud, T, Gonon, F, and Munafò, MR. Low statistical power in biomedical science: A review of three human research domains. R Soc Open Sci (2017) 4(2):160254. doi:10.1098/rsos.160254

PubMed Abstract | CrossRef Full Text | Google Scholar