2015 年 38 巻 5 号 p. 680-686
Clopidogrel is an antiplatelet agent widely used in combination with aspirin to limit the occurrence of cardiovascular (embolic/thrombotic) events. Consensus guidelines recommend proton pump inhibitors (PPIs) as a gastrointestinal (GI) prophylactic measure for all patients receiving dual antiplatelet therapy with clopidogrel and aspirin. The objective of this study was to analyze the effect of the simultaneous use of clopidogrel, aspirin, and PPIs on hemorrhagic and embolic/thrombotic events using the U.S. Food and Drug Administration (FDA) Adverse Event Reporting System (FAERS) database. Reports of hemorrhagic and embolic/thrombotic events between 2004 and 2013 were analyzed with a reporting odds ratio (ROR) algorithm and logistic regression methods. The Medical Dictionary for Regulatory Activities Preferred Terms was used to identify such events. Regarding hemorrhagic events, the adjusted RORs of the concomitant use of aspirin and clopidogrel and those of PPIs prescribed with aspirin and clopidogrel were 4.40 (95% confidence interval [CI], 4.02–4.81) and 3.40 (95% CI, 2.84–4.06), respectively. For embolic/thrombotic events, the adjusted RORs of the concomitant use of aspirin and clopidogrel and those of PPIs prescribed with aspirin and clopidogrel were 2.37 (95% CI, 2.16–2.59) and 2.38 (95% CI, 2.00–2.84), respectively. Among patients included in the FAERS database, the concurrent use of aspirin and clopidogrel with PPIs reduced the adjusted ROR of GI hemorrhagic events. PPIs had little influence on the adjusted ROR of embolic/thrombotic events. These results support the use of PPIs as a preventive measure against GI hemorrhagic events for patients receiving clopidogrel and aspirin.
Clopidogrel is an antiplatelet agent commonly used in combination with aspirin to reduce the incidence of cardiovascular events. Clopidogrel alone, aspirin alone, and their combination are associated with increased risk of gastrointestinal (GI) hemorrhagic events.1) Observational data suggest that proton pump inhibitors (PPIs) reduce the rate of recurrent GI hemorrhagic events in patients on antiplatelet therapy. Consensus guidelines recommend prescribing PPIs as a GI prophylactic measure for all patients receiving dual antiplatelet therapy with clopidogrel and aspirin.1) The effects of PPIs on GI hemorrhagic events have been evaluated in a case-control study,2) cohort studies,3,4) and a randomized trial,5) but are not well understood in clinical practice.
PPIs are believed to interact with clopidogrel by competitively inhibiting the CYP2C19 enzyme, which is needed to convert clopidogrel to its active metabolite. Observational studies have suggested an interaction between clopidogrel and PPIs that could have clinical significance, i.e., increased embolic/thrombotic events.6,7) In 2010, the U.S. Food and Drug Administration (FDA) added a requirement to include a boxed warning on the clopidogrel label to draw attention to the fact that patients with poor CYP2C19 metabolic activity are at increased risk for adverse cardiovascular outcomes due to the consequently reduced efficacy of the clopidogrel they take.8)
In contrast, a number of observational studies and commentaries showed no interaction between clopidogrel and PPIs.5,9,10) Significant controversy exists regarding cardiovascular adverse events that could arise from a drug interaction between clopidogrel and PPIs. Given the conflicting data, the optimal dosing for patients who require concomitant therapy with clopidogrel and PPIs is unknown.
The FDA Adverse Event Reporting System (FAERS) includes several million spontaneous reports of drug-associated adverse events and is routinely used to evaluate drug safety profiles.11–14) It is one of the primary tools used for post-marketing surveillance and pharmacovigilance, because it is the largest, most well-known database worldwide, and it reflects the realities of clinical practice.
In pharmacovigilance analysis, data mining algorithms have been developed to identify drug-associated adverse events as signals that are reported more frequently than expected, by estimating the expected reporting frequencies on the basis of information available for all drugs and all events in the database.15,16) The reporting odds ratio (ROR),17) the proportional reporting ratio (PRR),18) the information component (IC),19) and the empirical Bayes geometric mean (EBGM)20) are widely used: the PRR is currently used by the Medicines and Healthcare Products Regulatory Agency, United Kingdom: the ROR by the Pharmaceuticals and Medical Devices Agency in Japan and the Netherlands Pharmacovigilance Centre: the IC by the World Health Organization: and the EBGM by the FDA.
The FAERS is a passive reporting database and is therefore subject to numerous sources of selection bias, including over-reporting, under-reporting, and lack of a denominator.17,21) In general, the ROR is inapplicable to the inference of comparative strength of causality and only offers a rough indication of the signal strength (i.e., generating hypotheses to search for unknown potential adverse reactions).22) It may be acceptable to compare several RORs of a particular event derived from stratified analysis within a particular context to examine the possibility of an interaction between two drugs and a single event (drug–drug interaction).23,24) The effects of drug combinations can be studied using a statistical interaction term in a logistic model to calculate the RORs.22–25)
To our knowledge, our study was the first to evaluate the effects of the interactions between PPIs, clopidogrel, and aspirin using the FAERS database.
Adverse events recorded from January 2004 to March 2013 in the FAERS database were downloaded from the FDA website (www.fda.gov). It contains reports on adverse drug events submitted by physicians, pharmacists, other healthcare professionals, manufacturers, and consumers from the U.S. and other countries. The FAERS database structure complies with ICH E2B, the international safety reporting guidelines. The database consists of 7 data tables: patient demographics and administrative information (DEMO), drug/biologic information (DRUG), adverse events (REAC), patient outcomes (OUTC), report sources (RPSR), drug therapy start and end dates (THER), and indications for use/diagnosis (INDI). The adverse events in REAC are coded according to the terminology preferred by the Medical Dictionary for Regulatory Activities (MedDRA).26)
The drugs selected for this investigation were clopidogrel (Plavix) and PPIs (omeprazole, esomeprazole, lansoprazole, dexlansoprazole, rabeprazole, and pantoprazole). Before analyzing the data, a text-mining approach was used that stated the drugs in terms of their generic names. The FAERS database permits contributors to register drugs under any name, including a trade name and an abbreviation. DrugBank, a reliable drug database, was used as a dictionary for the batch conversion of drug names. The DrugBank database contains drug information used globally, including 1447 FDA-approved small molecule drugs.27) We built a database that integrated the FAERS database and the DrugBank data using the FileMaker Pro 12 software (FileMaker, Inc.). We followed the FDA’s recommendation to adopt the most recent CASE number in order to identify duplicate reports of the same patient that come from different reporting sources and excluded these from the analysis as is described in the downloadable file ‘Asc_nts.doc’ from the FAERS database website.
Definition of Hemorrhagic and Embolic/Thrombotic EventsThis study relied on the definitions provided by the MedDRA version 15.0.26) To evaluate the effect of PPIs in combination with dual clopidogrel and aspirin treatment on hemorrhagic events, we used the Standardized MedDRA Query (SMQ) for hemorrhagic events (SMQ code: 20000038) and the System Organ Class (SOC) for gastrointestinal disorder, and extracted only reports that met both criteria. The number of selected preferred terms for hemorrhagic events, limited by SOC (gastrointestinal disorder), was 68. Furthermore, to evaluate the effect of PPIs on the antiplatelet (embolic/thrombotic) activity of clopidogrel, we used 63 preferred terms that matched the SMQ for embolic and thrombotic events (SMQ code: 20000081) and the SOC for nervous system disorder.
AnalysisAll reported adverse events of interest (“hemorrhagic” or “embolic/thrombotic”) were defined as “cases” and all reported other adverse events as “non-cases.” For example, in the case of no PPI use, the reports were divided into 3 index groups: reports of patients who used aspirin, but not clopidogrel; reports of patients who used clopidogrel, but not aspirin; and reports of patients who used aspirin and clopidogrel concurrently. These index groups were compared with a reference group in which no aspirin and clopidogrel were used. The analysis of drug–drug interactions is based on the assumption that specific events may occur more (or less) frequently when both drugs are used concomitantly compared to when they are used separately.23) To compare one of the index groups with the reference group, we calculated the crude RORs as (a : c)/(b : d)23) (Table 1). RORs were expressed as point estimates with 95% confidence intervals (CIs). For signal detection, general qualitative judgments are viable; whether a signal is detected or not depends on whether the signal indices exceed predefined thresholds: ROR values <1 indicate no potential exposure-event associations and estimates >1 indicate potential exposure-event safety signals. Safety signals are considered significant when the ROR estimates and the lower limits of the corresponding 95% CI are ≥2.17)
| ‘Suspected event’ reported | No ‘suspected event’reported | ||
|---|---|---|---|
| Medication | Index group | a | b |
| Reference group | c | d |
Next, we refined the signal with a dedicated correction to detect possible confounders present in the database. The RORs were adjusted for age and sex and were then calculated using logistic regression analysis. To construct the logistic model, we included the terms for aspirin, clopidogrel, and PPI. Furthermore, the terms for the two-way interaction between aspirin and PPI (Asp*P), aspirin and clopidogrel (Asp*C), clopidogrel and PPI (C*P), and the term for the three-way interaction (Asp*C*P) were coded. The logistic model used to calculate the adjusted ROR was as follows:
 |
A probability (p) value of 0.05 or less was statistically significant. Data analyses were performed using JMP, version 9.0 (SAS Institute Inc., Cary, NC, U.S.A.).
After exclusion of duplicates following the FDA recommendation, 3522995 of the 4746890 reports in the FAERS database were of use. We excluded a further 1265093 reports because the age and sex of the patients were not reported and finally analyzed 2257902 reports.
For the hemorrhagic events, including GI hemorrhagic events, the characteristics (age, sex, aspirin, clopidogrel, and PPI) of cases and non-cases are summarized in Table 2. The age of patients in the cases group was significantly higher (p<0.01, t-test) than that in the non-cases group. The ROR with 95% CIs for hemorrhagic events for aspirin or clopidogrel was >2 (ROR of clopidogrel, 3.63 (95% CI, 3.50–3.77); ROR of aspirin, 3.09 (95% CI, 3.01–3.16)) (Table 2). The distribution of drug use for cases and non-cases, which was stratified for the use of PPI, clopidogrel, and aspirin, is shown in Table 3. In the case of no PPI use, 16999 reports referred to clopidogrel and not aspirin, 88113 referred to aspirin and not to clopidogrel, and 14490 included references to both drugs (clopidogrel and aspirin). The corresponding RORs with 95% CIs are shown in Table 3. If no clopidogrel was used, the crude ROR for the use of aspirin was 3.08 (95% CI, 2.99–3.18). If the patient took clopidogrel, the crude ROR of aspirin was 1.73 (95% CI, 1.59–1.88). For the combination with PPIs, 6035 reports referred to clopidogrel and not aspirin, 30396 referred to aspirin and not clopidogrel, and 6107 included references to both drugs. If no clopidogrel was used, the crude ROR for the use of aspirin was 1.51 (95% CI, 1.43–1.60). If the patient used clopidogrel, the crude ROR was 1.32 (95% CI, 1.14–1.52). The unadjusted and adjusted RORs with age and sex and those 95% CIs for the logistic analysis are summarized in Table 4. The pseudo R2 values for the adjusted model and for the unadjusted model were 0.0313 and 0.0209, respectively. The values of the interaction terms of β6, β7, and β8 were negative and statistically significant (Table 4). That of β9 was also statistically significant but was positive. The adjusted ROR for the concomitant use of aspirin and clopidogrel was 4.40 (95% CI, 4.02–4.81); that for aspirin and clopidogrel with PPIs was 3.40 (95% CI, 2.84–4.06) (Table 4).
| Cases, n (%), n=50879 | Non-cases, n (%), n=2207023 | Reporting odds ratio (95% CI) | |
|---|---|---|---|
| Mean age (years) | 60.1 | 53.0 | |
| Male | 24865 (48.9) | 862291 (39.1) | 1.49 (1.46–1.52) |
| Clopidogrel | 3227 (6.3) | 40404 (1.8) | 3.63 (3.50–3.77) |
| Aspirin | 8283 (16.3) | 130823 (5.9) | 3.09 (3.01–3.16) |
| PPIs | 8293 (16.3) | 197122 (8.9) | 1.99 (1.94–2.03) |
* Selected preferred terms for haemorrhagic events (SMQ: 20000039), limited by SOC (gastrointestinal disorder) were assigned in MedDRA v15.0, including abdominal wall haematoma, abdominal wall haemorrhage, anal haemorrhage, anal ulcer haemorrhage, anorectal varices haemorrhage, bloody peritoneal effluent, chronic gastrointestinal bleeding, colonic haematoma, diarrhoea haemorrhagic, diverticulitis intestinal haemorrhagic, diverticulum intestinal haemorrhagic, duodenal ulcer haemorrhage, duodenitis haemorrhagic, enterocolitis haemorrhagic, gastric haemorrhage, gastric ulcer haemorrhage, gastric ulcer haemorrhage obstructive, gastric ulcer perforation, gastric varices haemorrhage, gastritis alcoholic haemorrhagic, gastritis haemorrhagic, gastroduodenal haemorrhage, gastroduodenitis haemorrhagic, gastrointestinal angiodysplasia haemorrhagic, gastrointestinal haemorrhage, gastrointestinal ulcer haemorrhage, gingival bleeding, haematemesis, haematochezia, haemorrhagic ascites, haemorrhagic erosive gastritis, haemorrhoidal haemorrhage, intestinal haemorrhage, intra-abdominal haemorrhage, large intestinal haemorrhage, large intestinal ulcer haemorrhage, lip haematoma, lip haemorrhage, lower gastrointestinal haemorrhage, Mallory–Weiss syndrome, melaena, melaena neonatal, mesenteric haematoma, mesenteric haemorrhage, mouth haemorrhage, oesophageal haemorrhage, oesophageal ulcer haemorrhage, oesophageal varices haemorrhage, oesophagitis haemorrhagic, pancreatic haemorrhage, pancreatitis haemorrhagic, parotid gland haemorrhage, peptic ulcer haemorrhage, peritoneal haematoma, peritoneal haemorrhage, proctitis haemorrhagic, rectal haemorrhage, rectal ulcer haemorrhage, retroperitoneal haematoma, retroperitoneal haemorrhage, small intestinal haemorrhage, small intestinal ulcer haemorrhage, stomatitis haemorrhagic, tongue haematoma, tongue haemorrhage, tooth pulp haemorrhage, tooth socket haemorrhage, and upper gastrointestinal haemorrhage.
| Cases | Non-cases | Total | Crude ROR (95% CI) | |||
|---|---|---|---|---|---|---|
| PPIs not present | Clopidogrel not present | Aspirin not present | 35393 | 1897492 | 1932885 | 3.08 (2.99–3.18) |
| Aspirin present | 4790 | 83323 | 88113 | |||
| Total | 40183 | 1980815 | 2020998 | |||
| Clopidogrel present | Aspirin not present | 996 | 16003 | 16999 | 1.73 (1.59–1.88) | |
| Aspirin present | 1407 | 13083 | 14490 | |||
| Total | 2403 | 29086 | 31489 | |||
| PPIs present | Clopidogrel not present | Aspirin not present | 5850 | 157027 | 162877 | 1.51 (1.43–1.60) |
| Aspirin present | 1619 | 28777 | 30396 | |||
| Total | 7469 | 185804 | 193273 | |||
| Clopidogrel present | Aspirin not present | 357 | 5678 | 6035 | 1.32 (1.14–1.52) | |
| Aspirin present | 467 | 5640 | 6107 | |||
| Total | 824 | 11318 | 12142 | |||
* Selected preferred terms for haemorrhagic events (SMQ: 20000039), limited by SOC (gastrointestinal disorder) were assigned in MedDRA v15.0, including abdominal wall haematoma, abdominal wall haemorrhage, anal haemorrhage, anal ulcer haemorrhage, anorectal varices haemorrhage, bloody peritoneal effluent, chronic gastrointestinal bleeding, colonic haematoma, diarrhoea haemorrhagic, diverticulitis intestinal haemorrhagic, diverticulum intestinal haemorrhagic, duodenal ulcer haemorrhage, duodenitis haemorrhagic, enterocolitis haemorrhagic, gastric haemorrhage, gastric ulcer haemorrhage, gastric ulcer haemorrhage obstructive, gastric ulcer perforation, gastric varices haemorrhage, gastritis alcoholic haemorrhagic, gastritis haemorrhagic, gastroduodenal haemorrhage, gastroduodenitis haemorrhagic, gastrointestinal angiodysplasia haemorrhagic, gastrointestinal haemorrhage, gastrointestinal ulcer haemorrhage, gingival bleeding, haematemesis, haematochezia, haemorrhagic ascites, haemorrhagic erosive gastritis, haemorrhoidal haemorrhage, intestinal haemorrhage, intra-abdominal haemorrhage, large intestinal haemorrhage, large intestinal ulcer haemorrhage, lip haematoma, lip haemorrhage, lower gastrointestinal haemorrhage, Mallory–Weiss syndrome, melaena, melaena neonatal, mesenteric haematoma, mesenteric haemorrhage, mouth haemorrhage, oesophageal haemorrhage, oesophageal ulcer haemorrhage, oesophageal varices haemorrhage, oesophagitis haemorrhagic, pancreatic haemorrhage, pancreatitis haemorrhagic, parotid gland haemorrhage, peptic ulcer haemorrhage, peritoneal haematoma, peritoneal haemorrhage, proctitis haemorrhagic, rectal haemorrhage, rectal ulcer haemorrhage, retroperitoneal haematoma, retroperitoneal haemorrhage, small intestinal haemorrhage, small intestinal ulcer haemorrhage, stomatitis haemorrhagic, tongue haematoma, tongue haemorrhage, tooth pulp haemorrhage, tooth socket haemorrhage, and upper gastrointestinal haemorrhage.
| Covariates | Estimated beta | Likelihood ratio test | Adjusted ROR (95% CI)** | Unadjusted ROR (95% CI)*** | |
|---|---|---|---|---|---|
| Aspirin | β3 | 0.91 | <0.0001 | 2.47 (2.40–2.55) | 3.08 (2.99–3.18) |
| Clopidogrel | β4 | 0.92 | <0.0001 | 2.51 (2.35–2.68) | 3.34 (3.13–3.56) |
| PPIs | β5 | 0.61 | <0.0001 | 1.84 (1.79–1.89) | 2.00 (1.94–2.05) |
| Aspirin*PPIs | β6 | −0.64 | <0.0001 | 2.39 (2.24–2.55) | 3.02 (2.87–3.17) |
| Aspirn*Clopidogrel | β7 | −0.34 | <0.0001 | 4.40 (4.02–4.81) | 5.77 (5.45–6.10) |
| Clopidogrel*PPIs | β8 | −0.59 | <0.0001 | 2.55 (2.25–2.90) | 3.37 (3.03–3.75) |
| Aspirin*Clopidogrel*PPIs | β9 | 0.37 | <0.0001 | 3.40 (2.84–4.06) | 4.44 (4.04–4.88) |
| Age (years) | β1 | 0.02 | <0.0001 | 1.02 (1.01–1.02) | |
| Gender male | β2 | 0.33 | <0.0001 | 1.39 (1.36–1.41) | |
* Selected preferred terms for haemorrhagic events (SMQ: 20000039), limited by SOC (gastrointestinal disorder) were assigned in MedDRA v15.0, including abdominal wall haematoma, abdominal wall haemorrhage, anal haemorrhage, anal ulcer haemorrhage, anorectal varices haemorrhage, bloody peritoneal effluent, chronic gastrointestinal bleeding, colonic haematoma, diarrhoea haemorrhagic, diverticulitis intestinal haemorrhagic, diverticulum intestinal haemorrhagic, duodenal ulcer haemorrhage, duodenitis haemorrhagic, enterocolitis haemorrhagic, gastric haemorrhage, gastric ulcer haemorrhage, gastric ulcer haemorrhage obstructive, gastric ulcer perforation, gastric varices haemorrhage, gastritis alcoholic haemorrhagic, gastritis haemorrhagic, gastroduodenal haemorrhage, gastroduodenitis haemorrhagic, gastrointestinal angiodysplasia haemorrhagic, gastrointestinal haemorrhage, gastrointestinal ulcer haemorrhage, gingival bleeding, haematemesis, haematochezia, haemorrhagic ascites, haemorrhagic erosive gastritis, haemorrhoidal haemorrhage, intestinal haemorrhage, intra-abdominal haemorrhage, large intestinal haemorrhage, large intestinal ulcer haemorrhage, lip haematoma, lip haemorrhage, lower gastrointestinal haemorrhage, Mallory–Weiss syndrome, melaena, melaena neonatal, mesenteric haematoma, mesenteric haemorrhage, mouth haemorrhage, oesophageal haemorrhage, oesophageal ulcer haemorrhage, oesophageal varices haemorrhage, oesophagitis haemorrhagic, pancreatic haemorrhage, pancreatitis haemorrhagic, parotid gland haemorrhage, peptic ulcer haemorrhage, peritoneal haematoma, peritoneal haemorrhage, proctitis haemorrhagic, rectal haemorrhage, rectal ulcer haemorrhage, retroperitoneal haematoma, retroperitoneal haemorrhage, small intestinal haemorrhage, small intestinal ulcer haemorrhage, stomatitis haemorrhagic, tongue haematoma, tongue haemorrhage, tooth pulp haemorrhage, tooth socket haemorrhage, and upper gastrointestinal haemorrhage. ** The adjusted RORs were calculated with the estimates of betas. *** The unadjusted RORs were calculated without terms of age and sex, and therefore do not correspond to the estimated betas in the second column on the left.
For the embolic/thrombotic events, the data for the mean age, sex, number of reports per index, and reference group are presented in Table 5. The mean age of the cases group was significantly higher than that of the non-cases group (p<0.01, t-test). If no PPI and no clopidogrel was used, the crude ROR for the use of aspirin was 2.13 (95% CI, 2.06–2.20). If the patient used clopidogrel, the crude ROR of aspirin was 0.93 (95% CI, 0.85–1.01) (Table 6). For the use of PPIs without clopidogrel, the crude ROR for the use of aspirin was 1.72 (95% CI, 1.62–1.83). If the patient used PPIs and clopidogrel, the crude ROR of aspirin was 1.03 (95% CI, 0.89–1.18) (Table 6). The unadjusted and adjusted RORs with age and sex and those 95% CIs for the logistic analysis are summarized in Table 7. The values of the β6, β7, and β8 interaction terms were negative, and β6 and β7 were statistically significant (Table 7). The β9 value was positive and statistically significant. The adjusted RORs for the concomitant use of aspirin and clopidogrel, and the concomitant use of aspirin and clopidogrel with PPIs, were 2.37 (95% CI, 2.16–2.59) and 2.38 (95% CI, 2.00–2.84), respectively (Table 7). The pseudo R2 values for the adjusted model and for the unadjusted model were 0.0199 and 0.0085, respectively.
| Cases, n (%), n=61779 | Non-cases, n (%), n=2196123 | Reporting odds ratio (95% CI) | |
|---|---|---|---|
| Mean age (years) | 60.2 | 53.0 | |
| Male | 26234 (42.5) | 860922 (39.2) | 1.14 (1.13–1.16) |
| Clopidogrel | 3228 (5.2) | 40403 (1.8) | 2.94 (2.84–3.05) |
| Aspirin | 7454 (12.1) | 131652 (6.0) | 2.15 (2.10–2.21) |
| PPIs | 6867 (11.1) | 198548 (9.0) | 1.26 (1.23–1.29) |
* Selected preferred terms for embolic and thrombotic events (SMQ: 20000081), limited by SOC (nervous system disorder) were assigned in MedDRA v15.0, including basal ganglia infarction, basilar artery occlusion, basilar artery thrombosis, capsular warning syndrome, carotid arterial embolus, carotid artery occlusion, carotid artery thrombosis, cerebellar artery occlusion, cerebellar artery thrombosis, cerebellar embolism, cerebral artery embolism, cerebral artery occlusion, cerebral artery thrombosis, cerebral hypoperfusion, cerebrovascular insufficiency, cerebrovascular stenosis, ischaemic cerebral infarction, ischaemic stroke, lacunar infarction, precerebral artery occlusion, spinal artery embolism, spinal artery thrombosis, stroke in evolution, transient ischaemic attack, vertebral artery occlusion, vertebral artery thrombosis, cerebral venous thrombosis, intracranial venous sinus thrombosis, superior sagittal sinus thrombosis, transverse sinus thrombosis, basal ganglia stroke, brain stem infarction, brain stem stroke, brain stem thrombosis, cerebellar infarction, cerebral infarction, cerebral infarction foetal, cerebral ischaemia, cerebral thrombosis, cerebrospinal thrombotic tamponade, cerebrovascular accident, cerebrovascular disorder, diplegia, embolic cerebral infarction, embolic stroke, foetal cerebrovascular disorder, haemorrhagic cerebral infarction, haemorrhagic stroke, haemorrhagic transformation stroke, hemiparesis, hemiplegia, monoparesis, monoplegia, paraparesis, paraplegia, paresis, quadriparesis, quadriplegia, spinal cord infarction, thalamic infarction, thrombotic cerebral infarction, thrombotic stroke, and visual midline shift syndrome.
| Cases | Non-cases | Total | Crude ROR (95% CI) | |||
|---|---|---|---|---|---|---|
| PPIs not present | Clopidogrel not present | Aspirin not present | 48024 | 1884861 | 1932885 | 2.13 (2.06–2.20) |
| Aspirin present | 4534 | 83579 | 88113 | |||
| Total | 52558 | 1968440 | 2020998 | |||
| Clopidogrel present | Aspirin not present | 1312 | 15687 | 16999 | 0.93 (0.85–1.01) | |
| Aspirin present | 1042 | 13448 | 14490 | |||
| Total | 2354 | 29135 | 31489 | |||
| PPIs present | Clopidogrel not present | Aspirin not present | 4560 | 158317 | 162877 | 1.72 (1.62–1.83) |
| Aspirin present | 1433 | 28963 | 30396 | |||
| Total | 5993 | 187280 | 193273 | |||
| Clopidogrel present | Aspirin not present | 429 | 5606 | 6035 | 1.03 (0.89–1.18) | |
| Aspirin present | 445 | 5662 | 6107 | |||
| Total | 874 | 11268 | 12142 | |||
* Selected preferred terms for embolic and thrombotic events (SMQ: 20000081), limited by SOC (nervous system disorder) were assigned in MedDRA v15.0, including basal ganglia infarction, basilar artery occlusion, basilar artery thrombosis, capsular warning syndrome, carotid arterial embolus, carotid artery occlusion, carotid artery thrombosis, cerebellar artery occlusion, cerebellar artery thrombosis, cerebellar embolism, cerebral artery embolism, cerebral artery occlusion, cerebral artery thrombosis, cerebral hypoperfusion, cerebrovascular insufficiency, cerebrovascular stenosis, ischaemic cerebral infarction, ischaemic stroke, lacunar infarction, precerebral artery occlusion, spinal artery embolism, spinal artery thrombosis, stroke in evolution, transient ischaemic attack, vertebral artery occlusion, vertebral artery thrombosis, cerebral venous thrombosis, intracranial venous sinus thrombosis, superior sagittal sinus thrombosis, transverse sinus thrombosis, basal ganglia stroke, brain stem infarction, brain stem stroke, brain stem thrombosis, cerebellar infarction, cerebral infarction, cerebral infarction foetal, cerebral ischaemia, cerebral thrombosis, cerebrospinal thrombotic tamponade, cerebrovascular accident, cerebrovascular disorder, diplegia, embolic cerebral infarction, embolic stroke, foetal cerebrovascular disorder, haemorrhagic cerebral infarction, haemorrhagic stroke, haemorrhagic transformation stroke, hemiparesis, hemiplegia, monoparesis, monoplegia, paraparesis, paraplegia, paresis, quadriparesis, quadriplegia, spinal cord infarction, thalamic infarction, thrombotic cerebral infarction, thrombotic stroke, and visual midline shift syndrome.
| Covariates | Estimated beta | Likelihood ratio test | Adjusted ROR (95% CI)** | Unadjusted ROR (95% CI)*** | |
|---|---|---|---|---|---|
| Aspirin | β3 | 0.53 | <0.0001 | 1.71 (1.65–1.76) | 2.13 (2.06–2.20) |
| Clopidogrel | β4 | 0.90 | <0.0001 | 2.46 (2.32–2.61) | 3.28 (3.10–3.48) |
| PPIs | β5 | 0.02 | 0.2609 | 1.02 (0.99–1.05) | 1.13 (1.10–1.17) |
| Aspirin*PPIs | β6 | −0.13 | 0.0001 | 1.52 (1.42–1.63) | 1.94 (1.84–2.05) |
| Aspirn*Clopidogrel | β7 | −0.57 | <0.0001 | 2.37 (2.16–2.59) | 3.04 (2.85–3.24) |
| Clopidogrel*PPIs | β8 | −0.11 | 0.0674 | 2.25 (2.00–2.53) | 1.96 (1.77–2.16) |
| Aspirin*Clopidogrel*PPIs | β9 | 0.23 | 0.0091 | 2.38 (2.00–2.84) | 6.56 (5.93–7.26) |
| Age (years) | β1 | 0.02 | <0.0001 | 1.02 (1.02–1.02) | |
| Gender male | β2 | 0.08 | <0.0001 | 1.08 (1.06–1.10) | |
* Selected preferred terms for embolic and thrombotic events (SMQ: 20000081), limited by SOC (nervous system disorder) were assigned in MedDRA v15.0, including basal ganglia infarction, basilar artery occlusion, basilar artery thrombosis, capsular warning syndrome, carotid arterial embolus, carotid artery occlusion, carotid artery thrombosis, cerebellar artery occlusion, cerebellar artery thrombosis, cerebellar embolism, cerebral artery embolism, cerebral artery occlusion, cerebral artery thrombosis, cerebral hypoperfusion, cerebrovascular insufficiency, cerebrovascular stenosis, ischaemic cerebral infarction, ischaemic stroke, lacunar infarction, precerebral artery occlusion, spinal artery embolism, spinal artery thrombosis, stroke in evolution, transient ischaemic attack, vertebral artery occlusion, vertebral artery thrombosis, cerebral venous thrombosis, intracranial venous sinus thrombosis, superior sagittal sinus thrombosis, transverse sinus thrombosis, basal ganglia stroke, brain stem infarction, brain stem stroke, brain stem thrombosis, cerebellar infarction, cerebral infarction, cerebral infarction foetal, cerebral ischaemia, cerebral thrombosis, cerebrospinal thrombotic tamponade, cerebrovascular accident, cerebrovascular disorder, diplegia, embolic cerebral infarction, embolic stroke, foetal cerebrovascular disorder, haemorrhagic cerebral infarction, haemorrhagic stroke, haemorrhagic transformation stroke, hemiparesis, hemiplegia, monoparesis, monoplegia, paraparesis, paraplegia, paresis, quadriparesis, quadriplegia, spinal cord infarction, thalamic infarction, thrombotic cerebral infarction, thrombotic stroke, and visual midline shift syndrome. ** The adjusted RORs were calculated with the estimates of betas. *** The unadjusted RORs were calculated without terms of age and sex, and therefore do not correspond to the estimated betas in the second column on the left.
The present results suggest that PPIs decreased GI hemorrhagic events in the dual antiplatelet therapy with clopidogrel and aspirin and that PPIs had little influence on the embolic/thrombotic events. Using the interaction terms of drugs, the logistic regression analysis offers the possibility to control for covariates.
GI hemorrhagic events are the most common bleeding complications associated with the use of antiplatelet drugs. Clinical decisions regarding the concurrent use of a PPI and aspirin should be made based on benefit-risk assessments of cardiovascular events and GI hemorrhagic events, also because the adjusted ROR of aspirin and clopidogrel was decreased with the concomitant use of PPIs (Table 4). Although further research is necessary to determine the optimal approach for reducing the risk of adverse GI events in patients receiving antithrombotic therapy, the prophylactic use of PPIs provides a promising tool for this purpose.
Clopidogrel is a well-established antiplatelet drug that reduces the likelihood of embolic/thrombotic events.1–4) However, the ROR for embolic/thrombotic events for clopidogrel was >2 (Table 5; ROR, 2.94 (95% CI, 2.84–3.05)). Patients at high risk for the occurrence or recurrence of embolic events are treated with clopidogrel. In this disproportionality analysis, the patients treated with clopidogrel and who were at high risk for embolic events were compared with the patients treated with other medications and who were assumed to be at low risk for embolic events. Therefore, the increased ROR for clopidogrel seems to be an inverse association due to confounding by indication.
Observational studies suggest that the interaction between clopidogrel and PPIs could have significant clinical effects, such as an increase in embolic/thrombotic events.6,29) However, the potential for cardiovascular adverse events arising from the concomitant use of clopidogrel and PPIs remains highly controversial.30,31) For example, there was no apparent cardiovascular interaction between clopidogrel and omeprazole in the COGENT study.5) A number of recent commentaries from experts have argued that there is no evidence that this potential drug interaction has any clinical impact.4,5,9,10,32–34) The ACCF/ACG/AHA 2010 Expert Consensus Document is neutral with respect to a clinical interaction between PPIs and clopidogrel, citing the benefit/risk balance between the overall cardiovascular benefit and gastrointestinal complications.1) Our study suggests that the combined administration of clopidogrel and aspirin with PPIs has little influence on the reporting of embolic/thrombotic events (Table 7), which supports the Expert Consensus Document guidelines. These results of our evaluation of the FAERS database provide essential knowledge to improve our understanding of this issue. This information may be particularly beneficial to prescribers.
The ROR is clear and easy to understand and it is an easily applicable technique that allows for adjustment through logistic regression analysis.23,24) Logistic regression offers the advantageous possibility of controlling for covariates and it can be used to analyze the use of interaction terms in more detail.23–25) However, other research groups have criticized the use of logistic regression to assess drug interactions.35) Although research on the performance, accuracy, and reliability of different data mining algorithms is in progress, there is no recognized gold standard methodology.25,36)
In the Table 3 (for hemorrhagic events) and Table 6 (for embolic/thrombotic events), the crude ROR offered a rough indication of the signal strength for the influence of PPIs on combined administration of clopidogrel and aspirin. As shown in Table 3, the reporting of the hemorrhagic events decreased when PPIs were used along with concomitant clopidogrel and aspirin. Among clopidogrel users, when PPI is not used, the cases referred to aspirin use more often than the non-cases with a crude ROR of 1.73; when PPI is used the cases referred to aspirin use compared with the non-cases with a crude ROR of 1.32 (Table 3).
The covariates used in logistic regression analyses should be carefully selected. In the present study, we adjusted for the variables of age and sex in the calculation of the adjusted ROR. However, the FAERS database reporting might be influenced by the year of reporting and/or the data source. Furthermore, the presence of an anti-thrombotic agent (e.g., warfarin, ticlopidine, heparin) may act as a confounder, and influence the reporting of hemorrhagic and/or embolic/thrombotic events. Since the pseudo R2 values for the adjusted model were higher than those for the unadjusted model in both the hemorrhagic and embolic/thrombotic events, the adjusted model provided a better model fit than did the unadjusted model. The pseudo R2 values for the hemorrhagic events (0.0313) and for the embolic/thrombotic events (0.0199) were low. These low pseudo R2 values in logistic regression are common and this presents a problem when reporting these for an audience accustomed to normal R2 values.28,37) However, higher values of pseudo R2 is indicative of better model fit in model selection using the same dataset and outcome measurements.
In our logistic regression analysis, adjustment variables might have a strong influence on the reporting ratio of the adverse events of interest (GI hemorrhagic or embolic/thrombotic events). For example, the unadjusted ROR for the concomitant use of aspirin and clopidogrel and for both drugs combined with PPIs were 5.77 and 4.44, respectively (Table 4). The interpretation of the unadjusted RORs should be conducted carefully because of reporting bias. Using logistic regression analysis, these unadjusted RORs for the concomitant use of aspirin and clopidogrel and for both drugs combined with PPIs were adjusted to 4.40 and 3.40, respectively (Table 4). One possible reason is that different dosages of aspirin is used for the different treatment purposes, as an anticoagulant or as an analgesic drug; this may have influence the reporting strength of the events. The aspirin dosage used could also be associated with age and gender in the clinical practice. It might therefore be better to have taken into consideration the different aspirin dosages that were prescribed in various clinical settings in our calculations. Further studies of these variables as covariates are necessary.
Despite the limitations inherent to spontaneous reporting, the FAERS database is a rich resource and data mining indices provide a powerful means of identifying potential associations between drugs and adverse events. After considering the causality restraints of the current analysis, robust epidemiological studies are recommended to further test the hypothesis to draw conclusions that contribute to the clinical practice.
The present analysis demonstrates that PPIs decreased the reporting of GI hemorrhagic events in the patients receiving dual antiplatelet therapy with clopidogrel and aspirin and that PPIs had little influence on the reporting of the embolic/thrombotic events. Our study strengthened the prophylactic use of PPIs in patients receiving antithrombotic therapy using FAERS data that reflect the realities of clinical practice. The logistic regression analysis was deemed to be a useful tool for the correction of covariates. However, additional studies are required to confirm and determine the strength of a true drug–drug interaction.
This research was partially supported by JSPS KAKENHI Grant Number, 24390126.
The authors declare no conflict of interest.
