Abstract
AIM
The aim of this study was to investigate the effects of co-medication with selective serotonin re-uptake inhibitors (SSRIs) on overanticoagulation during acenocoumarol maintenance treatment.
METHODS
All subjects from The Rotterdam Study who received acenocoumarol maintenance treatment between April 1 1991 and September 9 2009 were followed for the event of an international normalized ratio (INR) ≥6, until death, end of treatment or end of the study period. With the Andersen-Gill extension of the Cox proportional hazards model, risks for repeated events of overanticoagulation in relation to concomitant SSRI use were calculated.
RESULTS
The risk for overanticoagulation during acenocoumarol maintenance treatment was increased in combination with fluvoxamine (HR 2.63, 95% CI 1.49, 4.66) and venlafaxine (HR 2.19, 95% CI 1.21, 3.99). There was no increase in risk for the other SSRIs, but numbers of exposed cases were low for all SSRIs except paroxetine.
CONCLUSION
Fluvoxamine and venlafaxine were associated with a more than double risk of INR values ≥6 in acenocoumarol treated subjects.
Keywords: acenocoumarol, fluvoxamine, selective serotonin re-uptake inhibitors, venlafaxine
WHAT IS ALREADY KNOWN ABOUT THIS SUBJECT
Coumarin anticoagulants and selective serotonin re-uptake inhibitors (SSRIs) have been reported to cause bleeding. Combination of these drug groups might enhance this risk. Case reports showed an increase of prothrombin time for the combination of warfarin with fluvoxamine and fluoxetine. This has not yet been confirmed by population based studies.
WHAT THIS STUDY ADDS
Fluvoxamine and venlafaxine increased prothrombin time in users of acenocoumarol above a critical value which is associated with an increased bleeding risk. The other SSRIs had no influence on acenocoumarol effectiveness, however numbers of drug users were low. The combination of fluvoxamine and venlafaxine with acenocoumarol should be monitored by measurements of the international normalized ratio to avoid overanticoagulation.
Introduction
Coumarin anticoagulants are first choice in the treatment and prevention of arterial or venous thrombosis [1]. Warfarin, acenocoumarol and phenprocoumon (coumarins) act as vitamin-K antagonists by inhibiting the synthesis of coagulation factor II, VII, IX and X. Due to a particularly narrow therapeutic range, patients treated with these drugs have to be closely monitored by regular assessments of the international normalized ratio (INR) to warrant anticoagulation without serious bleeding. Individual dosage differs widely between patients, mainly due to genetic variation in the vitamin K epoxide reductase complex subunit 1 (VKORC1) and cytochrome P450 2C9 (CYP2C9) enzyme, in combination with age, gender and body mass index (BMI) [2–9]. Therefore, during an initiation period coumarin dosage is carefully titrated by reference to the INR to achieve a stable individual maintenance dosage. However, also within the maintenance period effects of coumarins can vary due to interactions with co-medication or development of co-morbidity. In the Netherlands, acenocoumarol and phenprocoumon are most frequently used whereas warfarin is only given on rare occasions. Although individual management of coumarin use in the Netherlands is constantly monitored in anticoagulation clinics by regular INR measurements to obtain optimal INR levels, bleeding associated with coumarins is among the leading causes of drug-induced hospital admissions [10, 11].
Selective serotonin re-uptake inhibitors (SSRIs) have also been reported to increase the risk of bleeding [12–17], possibly secondary to serotonin release which is important for platelet aggregation. The hypothesis that inhibition of serotonin re-uptake by SSRIs might induce bleeding [13] was confirmed by a significant association between the degree of serotonin re-uptake inhibition and the risk of bleeding [17]. In users of coumarins the already increased risk of bleeding may be further enhanced by SSRIs. Besides this pharmacodynamic interaction, it was suggested that SSRIs might interact with coumarins pharmacokinetically through competitive inhibition of CYP2C9, the main metabolizing enzyme of coumarins [18]. An increase of prothrombin time was described in several case reports for fluvoxamine and fluoxetine in combination with warfarin [19–22], but a risk of haemorrhage was not reported in large scale epidemiological studies [23, 24]. Although one such study demonstrated an increased risk of bleeding in users of coumarin anticoagulants who started on SSRIs [25], it was not clear whether this was associated with an increased INR.
Therefore, we investigated in a large prospective population-based cohort study whether co-medication with SSRI antidepressants was associated with an increased risk of overanticoagulation during acenocoumarol maintenance treatment and whether such an effect was modified by CYP2C9 variant alleles.
Methods
Setting
We selected all subjects from the three cohorts of The Rotterdam Study (RS-I, RS-II and RS-III). The rationale and design of The Rotterdam Study have been described elsewhere [26–28]. In brief, The Rotterdam Study is a prospective population-based cohort study, designed to study neurological, cardiovascular, locomotor and ophthalmologic diseases in a population of people of 45 years and older. All subjects signed an informed consent when included. The Rotterdam Study has been approved by the institutional review board (Medical Ethics Committee) of the Erasmus Medical Center and by the review board of The Netherlands Ministry of Health, Welfare and Sports [28]. The RS-I cohort consisted of 7983 subjects (response rate 78%), the RS-II cohort of 3011 participants (response rate 67%) and the RS-III cohort of 3932 subjects (response rate 65%). The RS-I cohort had baseline examinations during 1990–1993 with completion of standardized questionnaires, sampling of blood and isolation of DNA. The RS-II was formed as an independent cohort in 1999 with baseline examinations between 2000 and 2001 and the RS-III cohort was enrolled and examined between 2006 and 2008.
A regional anticoagulation clinic, Star Medical Diagnostic Center, monitors all inhabitants of Ommoord, a suburb of Rotterdam, with an indication for anticoagulant therapy. This clinic covers the patients from all the RS cohorts. From this clinic, since 1984, all data on dosing, laboratory and clinical information are fully computerized. The physician who treats the patient decides about the type of anticoagulant. Prothrombin times are monitored every 1–6 weeks, dependent on the target level and stability of the INR. Coumarin doses are adjusted on the basis of computerized dose calculations. More than 99% of participants fill their drug prescriptions at seven regional pharmacies, which are fully computerized. Complete data on drug use from these pharmacies were available as of 1 January 1991. In order to assess use of SSRIs, we employed data from 1 April 1991 through 9 September 2009. The pharmacy data included the Anatomical Therapeutical Chemical (ATC)-code [29], the filling date, the total amount of drug units per prescription, the prescribed daily number of units and product name of the drugs.
Cohort and outcome definition
Our study population consisted of all patients from the three RS cohorts who started with acenocoumarol in the study period between 1 April 1991 and 9 September 2009, using it consecutively for at least 42 days. Patients who used SSRIs between January 1 and April 1 1991 were excluded, to ensure that we studied only incident users. We regarded a treatment period starting 42 days after initiation with acenocoumarol as maintenance period. Steady-state of a drug is usually achieved within 5–7 half-lives of drug elimination. The (S-) enantiomeric form of acenocoumarol has a 2–5 fold higher anticoagulant potency than the (R-) form. However due to the extremely fast clearance of the (S-) enantiomeric form, treatment effects are mainly due to (R-) acenocoumarol.
For (R-) acenocoumarol with a half-life of 8 h, a period of 6 weeks taken as initiation period was considered large enough to reach steady-state [30]. In patients from The Rotterdam Study using acenocoumarol during the maintenance period we took an event of an INR of 6 or greater after baseline study enrolment as an outcome. INR levels ≥6 are associated with an exponentially increased risk of bleeding [31]. An assessment of INR ≥6 occurring in a particular individual within 21 days of an earlier event was considered as part of one event episode. When occurring more than 21 days after an earlier event, it was considered as a (subsequent) new event.
SSRI exposure and cofactors
SRRI use was defined as current use of fluoxetine, citalopram, paroxetine, sertraline, fluvoxamine, escitalopram and venlafaxine. We expected that an effect of SSRIs on the INR would be visible within 4 weeks after starting the SSRI because in case reports effects of SSRI co-medication on coumarin effectiveness was reported within a few days [21, 22] and a trial showed an interaction between sertraline and warfarin within 26 days [32]. Furthermore, any INR increase during a longer period of co-medication would have been countered at the Thrombosis Center by decreasing the dosage of acenocoumarol. The following baseline patient characteristics were considered as potential confounders or effect modifiers: gender, age, BMI and target INR level. BMI was defined as (kg m–2) and missing values were imputed with a linear regression model consisting of INR ≥6, age, gender and target INR as variables. We further adjusted for co-medication with oral non-steroidal anti-inflammatory drugs (NSAIDs) and proton pump inhibitors as these drugs have been shown to interact with acenocoumarol with effects on INR measures [33, 34]. We did not adjust for co-medication with platelet inhibitors, acetyl salicylic acid, and corticosteroids, as these drugs do not influence the INR assessment. We verified the independency of our results from this co-medication in an additional analysis (results not given). In a sub-analysis, we studied the interaction between acenocoumarol maintenance treatment with the use of the tricyclic antidepressants, nortriptyline or mirtazapine. These are frequently prescribed antidepressants that are not metabolized via CYP2C9 and thus unlikely to cause an interaction with acenocoumarol. In an additional sub-analysis we studied effect modification of a CYP2C9 variant (T-) allele (rs4086116) on an INR ≥6 during use of acenocoumarol. Within the CYP2C cluster, this SNP was found to be most strongly associated with acenocoumarol dosage variation and explained nearly as much of dosage variation as the combined CYP2C9*2/*3 genotypes [35]. This SNP was genotyped in all three RS-cohorts. We did not expect effect-modification by VKORC1 variant alleles as this gene is not involved in the mechanism or kinetics of the SSRIs. We confirmed this assumption with results in an additional analysis (results not given).
Genotyping
From all RS participants, those with available DNA were genotyped using Illumina Infinium II HumanHap BeadChips at the Department of Internal Medicine, Erasmus Medical Center following the manufacturer's protocols. RS-I participants (n = 6449) were genotyped with 550k (V.3) single and duo chips, while RS-II participants (n = 2516) were genotyped with 550k (V.3) duo and 610k Quad chips. RS-III participants (n = 2420) were genotyped with the Human 610 Quad Arrays of Illumina. Genotype calling was performed in RS-I using BeadStudio software (version 0.3.10.14), GenomesStudio in RS-II and Bead Studio (v3.2.23) in RSIII. Participants with call rates <97.5%, excess autosomal heterozygosity, gender mismatch or outlying identity-by-state clustering estimates were excluded. After quality control, 5974 RS-I participants, 2157 RS-II and 2078 RS-III participants remained with complete data on genotyping [36]. For imputation 512 349 autosomal SNPs in RS-I and 466 389 autosomal SNPs in RS-II and RS-III were used after exclusions for call rate <98%, HWE P < 10−6, and MAF <1%, in MACH (version 1.00.15 for RS-I, 1.00.16 for RS-II and RS-III) with reference to the 2 543 886 SNPs of the HapMap CEU (release 22, build 36) [36].
Statistical analysis
Within one subject, an INR ≥6 could occur more than once and co-medication with SSRIs could change during a treatment period with acenocoumarol. In order to include all information available for the whole study period, we used the Andersen-Gill extension of the Cox proportional hazards model. This model allows the study of multiple events of an INR ≥6 within one subject and uses exposure to SSRIs as a time-varying covariable [37]. All subjects on acenocoumarol maintenance therapy were followed as of April 1 1991, from their first INR assessment until the last INR assessment because of the end of their treatment or last INR ≥6 event, the end of the study period or death, whichever came first. The date on which an INR ≥6 occurred was taken as the index date on which each case was compared for SSRI exposure to all subjects who were on acenocoumarol maintenance treatment at this same index date [38]. Thus cases could serve as controls on other index dates when still being on acenocoumarol treatment until their last measure of an INR ≥6.
We computed hazard ratios (HRs) and their 95% confidence intervals (CIs) for all events of an INR ≥6. Risk estimates were adjusted for age, gender, target INR, BMI and co-medication with NSAID or proton pump inhibitors. To study effect modification by CYP2C9 genotype in an exploratory analysis, subjects from the study population who were successfully genotyped were stratified as wild type (CYP2C9 homozygous C-alleles) and variant type (CYP2C9 heterozygous C/T-alleles or CYP2C9 homozygous T-alleles). SPSS 15.0 was used for data management and SAS 9.20 for the Andersen Gill analysis.
Results
Of the 14 926 subjects in the three Rotterdam cohorts, 2755 had used acenocoumarol during the study period for longer than 42 days continuously as maintenance therapy. In 887 subjects an INR ≥6 was measured at least once. Baseline characteristics of patients with an INR ≥6 and the total cohort are shown in Table 1. From the subjects on acenocoumarol maintenance therapy 43% were male, mean age at study entry was 69 years and mean BMI 27.2 kg m−2. BMI values had to be imputed in 280 subjects (10.9%). Increasing age and higher INR target levels were associated with a significantly increased risk to develop an INR ≥6. Higher BMI measures were associated with a decreased risk of overanticoagulation. From the 2083 subjects successfully genotyped for CYP2C9, 1346 (64.6%) had a wild type genotype and 737 (35.4%) had a variant genotype. The frequency of the CYP2C9 variant T-allele was 19.4% and alleles were in Hardy-Weinberg equilibrium (HWE, P value = 0.65). In subjects being homozygous for the CYP2C9 variant allele the risk for an INR ≥6 was increased 1.49 times (95% CI 1.20, 1.85). Subjects with at least one CYP2C9 variant allele had a 1.09 times increased risk for overanticoagulation compared with the wild type genotype which just failed to reach significance (95% CI 0.98, 1.20).
Table 1.
Characteristics of acenocoumarol users with an INR of 6.0 or greater and total cohort
| Patients with INR ≥6.0 (n = 887) | Total cohort (n = 2755) | HR* | 95% CI | |
|---|---|---|---|---|
| Gender | ||||
| Male (%) | 405 (45.7) | 1192 (43.3) | 1.00 | Reference |
| Female (%) | 482 (54.3) | 1563 (56.7) | 1.25 | (1.15, 1.36) |
| Start age (years) | ||||
| 45–54 | 10 (1.1%) | 114 (4.2%) | 1.00 | Reference |
| 55–64 | 72 (8.1%) | 383 (14.0%) | 2.89 | 1.75, 4.78 |
| 65–74 | 214 (24.1%) | 907 (33.2%) | 2.74 | 1.69, 4.45 |
| 75–84 | 389 (43.9%) | 998 (36.5%) | 4.14 | 2.56, 6.68 |
| >85 | 202 (22.8%) | 334 (12.2%) | 6.10 | 2.63, 14.1 |
| BMI (kg m−2) | ||||
| 15.0–20.0 | 17 (1.9) | 45 (1.6) | 1.00 | Reference |
| 20.1–25.0 | 228 (25.7) | 718 (26.1) | 0.70 | 0.52, 0.93 |
| 25.1–30.0 | 492 (55.5) | 1446 (52.6) | 0.68 | 0.51, 0.91 |
| >30.0 | 150 (16.9) | 539 (19.6) | 0.56 | 0.42, 0.76 |
| Target level (INR) | ||||
| 2.0–2.5 | 1 (0.1) | 118 (4.3) | 1.00 | Reference |
| 2.0–3.5 | 389 (43.9%) | 1728 (62.7%) | 10.1 | 3.23, 31.2 |
| 3.0–4.0 | 482 (54.3%) | 889 (32.3%) | 27.9 | 9.00, 86.7 |
| 3.5–4.5 | 15 (1.7%) | 20 (0.7%) | 65.4 | 17.3, 247 |
| Subjects with NSAID use | 89 (10.0) | 470 (17.0) | 1.31 | 1.15, 1.50 |
| Subjects with PPI use | 198 (22.3) | 862 (31.3) | 1.46 | 1.31, 1.64 |
| CYP2C9 genotypes† | n = 648 (73.1% of all cases) | n = 2083 (75.6% of the total group) | ||
| CC | 428 (66.0%) | 1346 (64.6%) | 1.00 | Reference |
| CT | 192 (29.6%) | 659 (31.6)% | 1.03 | 0.92, 1.15 |
| TT | 28 (4.3%) | 78 (3.7%) | 1.49 | 1.20, 1.85 |
| CYP2C9 wild type genotype vs. variant alleles | n = 648 (73.1% of all cases) | n = 2083 (75.6% of the total group) | ||
| CC | 428 (66.0%) | 1346 (64.6%) | 1.00 | Reference |
| CT or TT | 220 (34%) | 737 (35.4%) | 1.09 | 0.98, 1.20 |
Univariate analysis of HR was performed with the Andersen-Gill model. HRs cannot be calculated with the numbers in this table because controls may later become cases; significant values are printed in bold.
For the CY2C9 genotype within total cohort, allelic frequency of the C-allele was 80.9%, of the T-allele 19.1%, and HWE was 0.96. INR, international normalized ratio; HR, hazard ratio; CI, confidence interval; HWE, Hardy-Weinberg equilibrium.
In our cohort in total 225 subjects (5%) used SSRIs. Eighteen subjects used fluoxetine, 32 citalopram, 97 paroxetine, 23 sertraline, 26 fluvoxamine, two escitalopram and 14 venlafaxine (Table 2). In 38 events of an INR ≥6 a SSRI was used (4.3% of all events). Numbers of exposed cases were low for fluoxetine (n = 1), citalopram (n = 3), sertraline (n = 2) and escitalopram (no cases exposed). Numbers of exposed cases were somewhat higher for fluvoxamine (n = 8) and venlafaxine (n = 6). Paroxetine was used by 18 subjects with an INR ≥6. Co-medication with fluvoxamine and venlafaxine during acenocoumarol maintenance treatment more than doubled the risk on an INR ≥6, in the case of fluvoxamine 2.63 times (95% CI 1.49, 4.66) and for venlafaxine 2.19 times (95% CI 1.21, 3.99). The other SSRIs showed no association with overanticoagulation.
Table 2.
Association between overanticoagulation (INR ≥6) under acenocoumarol maintenance treatment and SSRIs
| Number of patients with drug use within total cohort*(n = 2755) | Cases with use of a specific drug with at least one INR ≥6.0 | Andersen-Gill analysis* | ||
|---|---|---|---|---|
| SSRI | HR†(95% CI) | HR‡(95% CI) | ||
| Fluoxetine | 18 | 1 | 0.50 (0.07, 3.55) | 0.48 (0.07, 3.40) |
| Citalopram | 32 | 3 | 0.62 (0.20, 1.94) | 0.63 (0.20, 1.95) |
| Paroxetine | 97 | 18 | 1.15 (0.81, 1.65) | 0.99 (0.68, 1.44) |
| Sertraline | 23 | 2 | 1.07 (0.34, 3.34) | 1.02 (0.33, 3.18) |
| Fluvoxamine | 36 | 8 | 2.46 (1.39, 4.34) | 2.63 (1.49, 4.66) |
| Escitalopram | 2 | 0 | P = 0.95 | P = 0.95 |
| Venlafaxine | 14 | 6 | 2.37 (1.37, 4.10) | 2.19 (1.21, 3.99) |
| Sub-analysis | ||||
| Nortriptyline | 12 | 2 | 0.60 (0.25, 1.44) | 0.55 (0.21, 1.47) |
| Mirtazapine | 45 | 8 | 1.01 (0.54, 1.88) | 1.09 (0.58, 2.04) |
Statistically significant values are printed in bold.
In this time-dependent analysis, exposure in case patient and in the rest of the cohort is assessed at the time of the outcome in each case patient (index date). As control patients can be used multiple times, the number of assessments in the reference group is much larger than the number of individuals. Hence, crude RRs cannot be calculated from the data in this table.
Adjusted for age, gender, BMI and target INR; if none of the cases was exposed, P values are given instead of HRs.
Adjusted for age, gender, BMI, target INR and use of non-steroidal anti-inflammatory drugs (ATC-group M01A) or proton pump inhibitors (A02 BC) if none of the cases was exposed, P values are given instead of HRs.
For fluvoxamine and venlafaxine we investigated whether the CYP2C9 genotype modified the interaction with acenocoumarol (Table 3). Within users of SSRIs as co-medication an INR ≥6 occurred in one of the two subjects being homozygous for the CYP2C9 variant allele and in 24 of the 128 subjects with a wild type genotype. We did not find a multiplicative effect measure modification for an interaction between fluvoxamine or venlafaxine and at least one CYP2C9 variant allele (P value = 0.49). To investigate presence of an additive effect measure modification we formed four groups for the combinations of fluvoxamine or venlafaxine use ‘yes’/’no’ with absence and presence of CYP2C9 variant alleles. No additive effect measure modification was detected either (P value = 0.17).
Table 3.
Association between overanticoagulation (INR ≥6.0), exposure to fluvoxamine or venlafaxine, stratified by CYP2C9 genotype
| Number of cases exposed to fluvoxamine or venlafaxine (n = 9) | Total cohort of genotyped acenocoumarol users (n = 2083) | HR*(95% CI) | P value for interaction | |
|---|---|---|---|---|
| P value of multiplicative effect modification between fluvoxamine or venlafaxine and CYP2C9 variant alleles (wild type genotype/variant alleles) | 0.49 | |||
| P value for an additive effect modification between fluvoxamine or venlafaxine and CYP2C9 variant alleles (wild type genotype/variant alleles) | 0.17 | |||
| No use of fluvoxamine or venlafaxine, CYP2C9 wild type genotype | 425 | 1346 | 1.00 reference | |
| No use of fluvoxamine or venlafaxine, CYP2C9 variant alleles | 219 | 702 | 1.08 (0.98, 1.19) | |
| Use of fluvoxamine or venlafaxine, CYP2C9 wild type genotype | 6 | 20 | 2.10 (1.18, 3.71) | |
| Use of fluvoxamine or venlafaxine, CYP2C9 variant alleles | 3 | 15 | 4.52 (0.63, 32.3) |
Statistically significant values are printed in bold.
Adjusted for age, gender, BMI, target INR and use of nonsteroidal anti-inflammatory drugs, ATC-group M01A and proton pump inhibitors (A02 BC) In this time-dependent analysis, exposure in case patient and in the rest of the cohort is assessed at the time of the outcome in each case patient (index date). As control patients can be used multiple times, the number of assessments in the reference group is much larger than the number of individuals. Hence, crude RRs cannot be calculated from the data in this table.
Discussion
In our study population, use of fluvoxamine and venlafaxine was associated with a more than double risk of overanticoagulation during acenocoumarol maintenance treatment. For co-medication with the other SSRIs, fluoxetine, citalopram, sertraline or escitalopram, an association with an INR ≥6 was not demonstrated but only for paroxetine was the number of exposed cases high enough to draw a conclusion. As only eight cases were exposed to fluvoxamine and six to venlafaxine, the clinical significance of our results should be interpreted cautiously. However, our findings for fluvoxamine are supported by two case reports for INR increase during warfarin treatment [19, 22]. In agreement with our results, clinical trials which focused on an interaction of fluoxetine, citalopram or sertraline with warfarin reported no or only very small increases in prothrombin time which were not regarded to be of clinical relevance [32, 39, 40]. However, this is not a guarantee for complete safety as such trials usually enrol people with a relatively low risk profile. A potential pharmacokinetic explanation for our findings might be that these SSRIs competitively inhibit CYP2C9. This would prolong the effectiveness of the coumarins and result in increased INR. SSRIs were reported to differ in their potency to inhibit CYP2C9. However, there is no consistent information for all SSRIs concerning their inhibitory potential on CYP2C9. Fluvoxamine and fluoxetine have been reported to inhibit CYP2C9 in vitro most strongly within the group of SSRIs in one study [41]. A review concluded that fluoxetine, fluvoxamine and paroxetine appeared as SSRIs with the highest potential for CYP2C9 interactions but had insufficient data to make a prediction for venlafaxine [42]. A Dutch textbook on drug information reported only CYP2C9 inhibition for fluvoxamine and sertraline but not for the other SSRIs or venlafaxine [18]. For the coumarins, CYP2C9 is the principal catalyst for the (S-) enantiomeric forms of warfarin and acenocoumarol [43]. In general, the (S-) isomers of the coumarins are more effective than the (R-) enantiomeric forms. However, for acenocoumarol, the (S-) form has an extremely short half live of 2 h and acenocoumarol effects mainly depend on the (R-) enantiomeric form of which about 60% is metabolized via CYP2C9. Thus interactions via CYP2C9 inhibition may be more relevant to warfarin than to acenocoumarol. From exploratory analysis no modification was noted of a CYP2C9 variant allele on the effect of fluvoxamine or venlafaxine on INR increase within acenocoumarol users. However, numbers of exposed cases were low and more cases are needed to obtain enough power for a genotype-stratified analysis that can confirm the effect of the CYP2C9 genotypes on a pharmacokinetic interaction between acenocoumarol and SSRIs. Moreover, the effect of CYP2C9 variant alleles on acenocoumarol effectiveness will rapidly become invisible by dosage titration of acenocoumarol to the personal maintenance dosage so that effects of additional inhibition by co-medication may no longer differ between wild type genotypes and carriers of variant alleles.
Besides a pharmacokinetic interaction that leads to increased INR levels, SSRIs deplete platelet serotonin concentrations and impair platelet aggregation with a subsequent increase of the risk of haemorrhage [12–17]. This risk might be higher for fluoxetine, paroxetine and sertraline as they inhibit the re-uptake of serotonin more than citalopram and fluvoxamine. Unfortunately, in our study population of 2755 acenocoumarol users and within these 222 SSRI users, we did not have enough bleeding events for stratified analysis for the separate SSRIs to assess this hypothesis. Other studies on warfarin reported no risk increase for the combination with SSRIs on upper gastrointestinal bleeding [23, 44, 45]. For acenocoumarol one case control study found a risk increase for non gastrointestinal bleeding in combination with SSRIs [25]. However, this study did not stratify for the different SSRIs.
Our study is the first observational cohort study on a risk increase of overanticoagulation with acenocoumarol in combination with SSRIs during acenocoumarol maintenance treatment. We consider the chance of bias and confounding due to study design negligible. First, The Rotterdam Study is a prospective cohort study and the regional anticoagulation clinic covered a complete area of more than 1 million inhabitants in the Rotterdam area. Consequently, everyone who is treated with a coumarin anticoagulant as an outpatient will be registered as such and selection bias is unlikely. Second, all medication use of all subjects was almost completely covered by the pharmacy data we retrieved and SSRIs were only available on prescription via pharmacies. Any lack of compliance would move our results into the direction of the null hypothesis, and would tend to make our results conservative. Third, we followed the cohort members during their whole period of acenocoumarol maintenance therapy and compared cases with all other cohort members available at the index dates for SSRI exposure. This analysis made use of all data available and adjusted for time varying effects. However, in observational studies there is always a risk of confounding by indication. We therefore performed the same analysis for nortriptyline and mirtazapine, two antidepressants used for the same indication but not known for inhibitory potential of CYP2C9. We did not find an association for these drugs with an increased risk on achieving INR ≥6 during acenocoumarol treatment. Hence confounding by indication is highly unlikely.
In conclusion, in this population-based cohort study among outpatients of an anticoagulation clinic using acenocoumarol for maintenance treatment, fluvoxamine and venlafaxine were associated with a more than double risk on an INR ≥6. Extra INR monitoring for the combination of fluvoxamine and venlafaxine with acenocoumarol might prevent overanticoagulation. These findings should be confirmed by studies with more exposed cases. Paroxetine had no association with an increased INR. Although also fluoxetine, citalopram, sertraline and escitalopram had no association with increased INR measures during acenocoumarol maintenance treatment the numbers of exposed cases were too low to judge an association.
Acknowledgments
The generation and management of GWAS genotype data for the Rotterdam Study is supported by the Netherlands Organisation of Scientific Research NWO Investments (nr. 175.010.2005.011, 911–03-012). This study is funded by the Research Institute for Diseases in the Elderly (014–93-015; RIDE2) and the Netherlands Genomics Initiative (NGI)/Netherlands Consortium for Health and Aging (NCHA) project nr. 050–060-810,
The Rotterdam Study is funded by Erasmus Medical Center and Erasmus University, Rotterdam, Netherlands Organization for the Health Research and Development (ZonMw), the Research Institute for Diseases in the Elderly (RIDE), the Ministry of Education, Culture and Science, the Ministry for Health, Welfare and Sports, the European Commission (DG XII) and the Municipality of Rotterdam.
We thank Pascal Arp, Mila Jhamai, Dr Michael Moorhouse, Marijn Verkerk and Sander Bervoets for their help in creating the GWAS database. The authors are grateful to the study participants, the staff from the Rotterdam Study and the participating general practitioners and pharmacists.
Competing Interests
There are no competing interests to declare.
REFERENCES
- 1.Ansell J, Hirsh J, Poller L, Bussey H, Jacobson A, Hylek E. The pharmacology and management of the vitamin K antagonists: the Seventh ACCP Conference on Antithrombotic and Thrombolytic Therapy. Chest. 2004;3(Suppl.):204S–233S. doi: 10.1378/chest.126.3_suppl.204S. [DOI] [PubMed] [Google Scholar]
- 2.Geisen C, Watzka M, Sittinger K, Steffens M, Daugela L, Seifried E, Mueller CR, Wienker TF, Oldenburg J. VKORC1 haplotypes and their impact on the inter-individual and inter-ethnical variability of oral anticoagulation. Thromb Haemost. 2005;94:773–9. doi: 10.1160/TH05-04-0290. [DOI] [PubMed] [Google Scholar]
- 3.Sconce EA, Khan TI, Wynne HA, Avery P, Monkhouse L, King BP, Wood P, Kesteren P, Daly AK, Kamali F. The impact of CYP2C9 and VKORC1 genetic polymorphism and patient characteristics upon warfarin dose requirements: proposal for a new dosing regimen. Blood. 2005;106:2329–33. doi: 10.1182/blood-2005-03-1108. [DOI] [PubMed] [Google Scholar]
- 4.D'Andrea G, D'Ambrosio RL, Di Perna P, Chetta M, Santacroce R, Brancaccio V, Grandone E, Margaglione M. A polymorphism in the VKORC1 gene is associated with an interindividual variability in the dose-anticoagulant effect of warfarin. Blood. 2005;105:645–9. doi: 10.1182/blood-2004-06-2111. [DOI] [PubMed] [Google Scholar]
- 5.Carlquist JF, Horne B, Muhlestein J, Lappé D, Whiting B, Kolek M, Clarke JL, James BC, Anderson JL. Genotypes of the cytochrome p450 isoform, CYP2C9, and the vitamin K epoxide reductase complex subunit 1 conjointly determine stable warfarin dose: a prospective study. J Thromb Thrombolysis. 2006;22:191–7. doi: 10.1007/s11239-006-9030-7. [DOI] [PubMed] [Google Scholar]
- 6.Aquilante CL, Langaee TY, Lopez LM, Yarandi HN, Tromberg JS, Mohuczy D, Gaston KL, Waddell CD, Chirico MJ, Johnson JA. Influence of coagulation factor, vitamin K epoxide reductase complex subunit 1, and cytochrome P450 2C9 gene polymorphisms on warfarin dose requirements. Clin Pharmacol Ther. 2006;79:291–302. doi: 10.1016/j.clpt.2005.11.011. [DOI] [PubMed] [Google Scholar]
- 7.Oldenburg J, Bevans C, Fergin A, Geisen C, Müller-Reible C, Watzka M. Current pharmacogenetic developments in oral anticoagulation therapy: the influence of variant VKORC1 and CYP2C9 alleles. Thromb Haemost. 2007;98:570–8. [PubMed] [Google Scholar]
- 8.Teichert M, Eijgelsheim M, Rivadeneira F, Uitterlinden A, van Schaik R, Hofman A, De Smet PA, van Gelder T, Visser LE, Stricker BH. A genome-wide association study of acenocoumarol maintenance dosage. Hum Mol Gen. 2009;18:3758–68. doi: 10.1093/hmg/ddp309. [DOI] [PubMed] [Google Scholar]
- 9.Qazim B, Stöllberger C, Krugluger W, Dossenbach-Glaninger A, Finsterer J. Dependency of phenprocoumon dosage on polymorphisms in the VKORC1 and CYP2C9 genes. J Thromb Haemost. 2009;28:211–4. doi: 10.1007/s11239-008-0252-8. [DOI] [PubMed] [Google Scholar]
- 10.van der Hooft C, Dieleman JP, Siemes C, Aarnoudse ALHJ, Verhamme KC, Stricker BHC, Sturkenboom MC. Adverse drug reaction-related hospitalisations: a population-based cohort study. Pharmacoepidemiol Drug Saf. 2008;17:365–71. doi: 10.1002/pds.1565. [DOI] [PubMed] [Google Scholar]
- 11.Leendertse A, Egberts A, Stoker L, Bemt van der P. Frequency of and risk factors for preventable medication-related hospital admissions in the Netherlands. Arch Intern Med. 2008;168:1890–6. doi: 10.1001/archinternmed.2008.3. [DOI] [PubMed] [Google Scholar]
- 12.de Abajo F, Rodríguez L, Montero D. Association between selective serotonin reuptake inhibitors and upper gastrointestinal bleeding: population based case-control study. BMJ. 1990;19:1106–9. doi: 10.1136/bmj.319.7217.1106. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Oksbjerg Dalton S, Johanson C, Meelmkjaer L, Norgard B, Sorensen H, Olsen J. Use of selective serotonin reuptake inhibitors and risk of upper astrointestinal tract bleeding. Arch Intern Med. 2003;163:59–64. doi: 10.1001/archinte.163.1.59. [DOI] [PubMed] [Google Scholar]
- 14.Walraven van C, Mamdani M, Wells P, Williams J. Inhibition of serotonin reuptake by antidepressants and upper gastrointestinal bleeding in elderly patients: retrospective cohort study. BMJ. 2001;319:655–8. doi: 10.1136/bmj.323.7314.655. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Wessinger S, Kaplan M, Choi L, Williams M, Lau C, Sharp L, Crowell MD, Keshavarzian A, Jones MP. Increased use of selective serotonin reuptake inhibitors in patients admitted with gastrointestinal haemorrhage: a multicentre retrospective analysis. Aliment Pharmacol Ther. 2006;23:937–44. doi: 10.1111/j.1365-2036.2006.02859.x. [DOI] [PubMed] [Google Scholar]
- 16.Serebruany VL. Selective serotonin reuptake inhibitors and increased bleeding risk: are we missing something? Am J Med. 2006;119:113–6. doi: 10.1016/j.amjmed.2005.03.044. [DOI] [PubMed] [Google Scholar]
- 17.Meijer WEE, Heerdink ER, Nolen WA, Herings RMC, Leufkens HGM, Egberts ACG. Association of risk of abnormal bleeding with degree of serotonin reuptake inhibition by antidepressants. Arch Intern Med. 2004;164:2367–70. doi: 10.1001/archinte.164.21.2367. [DOI] [PubMed] [Google Scholar]
- 18.Informatorium, Pharmacokinetics with Proton Pump Inhibitors. The Hague: Koninklijke Nederlandse Maatschappij ter bevordering der Pharmacie; 2008. Koninklijke Nederlandse Maatschappij ter bevordering der Pharmacie. [Google Scholar]
- 19.Yap K, Low S. Interaction of fluvoxamine with warfarin in an elderly woman. Singapore Med J. 1999;40:480–2. [PubMed] [Google Scholar]
- 20.Dent L, Orrock M. Warfarin-fluoxetine and diazepam-fluoxetine interaction. Pharmacotherapy. 1997;17:170–2. [PubMed] [Google Scholar]
- 21.Woolfrey S, Gammack N, Dewar M, Brown P. Fluoxetine-warfarin interaction. BMJ. 1993;307:241. doi: 10.1136/bmj.307.6898.241-b. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Limke K, Shelton A, Elliott E. Fluvoxamine interaction with warfarin. Ann Pharmacother. 2002;36:1890–2. doi: 10.1345/aph.1C112. [DOI] [PubMed] [Google Scholar]
- 23.Optamy L, Delaney J, Suissa S. Gastro-intestinal haemorrhage risks of selective serotonin receptor antagonist therapy: a new look. Br J Clin Pharmacol. 2008;66:76–81. doi: 10.1111/j.1365-2125.2008.03154.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Kurdyak P, Juurlink D, Kopp A, Herrmann N, Mamdani M. Antidepressants, Warfarin, and the Risk of Hemorrhage. J Clin Psychopharmacol. 2005;25:561–5. doi: 10.1097/01.jcp.0000186869.67418.bc. [DOI] [PubMed] [Google Scholar]
- 25.Schalekamp T, Klungel JH, Souverein PC, de Boer A. Increased bleeding risk with concurrent use of selective serotonin reuptake inhibitors and coumarins. Arch Intern Med. 2008;168:180–5. doi: 10.1001/archinternmed.2007.32. [DOI] [PubMed] [Google Scholar]
- 26.Hofman A, Grobbee DE, de Jong PT, van den Ouweland FA. Determinants of disease and disability in the elderly: the Rotterdam Elderly Study. Eur J Epidemiol. 1991;7:403–22. doi: 10.1007/BF00145007. [DOI] [PubMed] [Google Scholar]
- 27.Hofman A, Breteler M, van Duijn C, Krestin G, Pols H, Stricker B, Tiemeier H, Uitterlinden AG, Vingerling JR, Witteman JC. The Rotterdam Study: objectives and design update. Eur J Epidemiol. 2007;22:819–29. doi: 10.1007/s10654-007-9199-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Hofman A, Breteler M, Van Duijn C, Janssen H, Krestin G, Kuipers E, Stricker BH, Tiemeier H, Uitterlinden AG, Vingerling JR, Witteman JC. The Rotterdam Study: 2010 objectives and design update. Eur J Epidemiol. 2009;24:553–72. doi: 10.1007/s10654-009-9386-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.World Health Organization Collaborating Centre for Drug Statistics Methodology. Guidelines for ATC Classification and DDD Assignment. Oslo: World Health Organization Collaborating Centre for Drug Statistics Methodology; 2004. [Google Scholar]
- 30.Flockhart D, O'Kane D, Williams M, Watson M, Gage B, Gandolfi R, King R, Lyon E, Nussbaum R, O'Kane D, Schulman K, Veenstra D, Williams MS, Watson MS. Pharmacogenetic testing of CYP2C9 and VKORC1 alleles for warfarin. Genet Med. 2008;10:139–50. doi: 10.1097/GIM.0b013e318163c35f. [DOI] [PubMed] [Google Scholar]
- 31.Cannegieter SC, Rosendaal FR, Wintzen AR, van der Meer FJM, Vandenbroucke JP, Briet E. Optimal oral anticoagulant therapy in patients with mechanical heart valves. N Engl J Med. 1995;333:11–7. doi: 10.1056/NEJM199507063330103. [DOI] [PubMed] [Google Scholar]
- 32.Aspeloff G, Wilner K, Gerber N, Tremaine L. Effect of sertraline on protein binding of warfarin. Clin Pharmacokinet. 1997;32(Suppl. 1):37–42. doi: 10.2165/00003088-199700321-00006. [DOI] [PubMed] [Google Scholar]
- 33.Visser LE, van Schaik RHN, van Vliet M, Trienekens PH, De Smet PAGM, Vulto AG, Hofman A, van Duijn CM, Stricker BH. Allelic variants of cytochrome P450 2C9 modify the interaction between nonsteroidal anti-inflammatory drugs and coumarin anticoagulants. Clin Pharmacol Ther. 2005;77:479–85. doi: 10.1016/j.clpt.2005.02.009. [DOI] [PubMed] [Google Scholar]
- 34.Teichert M, van Noord C, Uitterlinden A, Hofman A, Buhre P, De Smet P, Straus S, Stricker BH, Visser LE. Proton pump inhibitors and the risk on hyperanticoagulation during acenocoumarol maintenance treatment. Br J Haematol. 2011;153:379–85. doi: 10.1111/j.1365-2141.2011.08633.x. [DOI] [PubMed] [Google Scholar]
- 35.Teichert M, Eijgelsheim M, Rivadeneira F, Uitterlinden AG, van Schaik RHN, Hofman A, De Smet PA, van Gelder T, Visser LE, Stricker BH. A genome-wide association study of acenocoumarol maintenance dosage. Human Molecular Genetics. 2009;18:3758–68. doi: 10.1093/hmg/ddp309. [DOI] [PubMed] [Google Scholar]
- 36.Hancock DB, Eijgelsheim M, Wilk JB, Gharib SA, Loehr LR, Marciante KD, Franceschini N, van Durme YM, Chen TH, Barr RG, Schabath MB, Couper DJ, Brusselle GG, Psaty BM, van Duijn CM, Rotter JI, Uitterlinden AG, Hofman A, Punjabi NM, Rivadeneira F, Morrison AC, Enright PL, North KE, Heckbert SR, Lumley T, Stricker BH, O'Connor GT, London SJ. Meta-analyses of genome-wide association studies identify multiple loci associated with pulmonary function. Nat Genet. 2010;42:45–52. doi: 10.1038/ng.500. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Andersen P, Gill R. Cox's regression model for counting processes: a large sample study. Ann Statist. 1982;10:110–2. [Google Scholar]
- 38.Stricker B, Steijnen T. Analysis of individual drug use as a time-varying determinant of exposure in prospective population-based cohort studies. Eur J Epidemiol. 2010;25:245–51. doi: 10.1007/s10654-010-9451-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Ford M, Anderson M, Rindone J, Jaskar D. Lack of effect of fluoxetine on the hypothrombinemic response of warfarin. J Clin Psychopharmacol. 1997;17:110–2. doi: 10.1097/00004714-199704000-00009. [DOI] [PubMed] [Google Scholar]
- 40.Priskorn M, Sidhu J, Larsen F, Davis J, Khan A, Rolan P. Investigation of multiple dose citalopram on the pharmacokinetics and pharmacodynamics of racemic warfarin. Br J Clin Pharmacol. 1997;44:199–202. doi: 10.1046/j.1365-2125.1997.00628.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Schmider J, Greenblatt D, von Moltke L, Karsov D, Shader R. Inhibition of CYP2C9 by selective serotonin reuptake inhibitors in vitro: studies of phenytoin p-hydroxylation. Br J Clin Pharmacol. 1996;44:495–8. doi: 10.1046/j.1365-2125.1997.00601.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Duncan D, Sayal K, McConnell H, Taylor D. Antidepressant interactions with warfarin. Int Clin Psychopharmacol. 1998;13:87–94. doi: 10.1097/00004850-199803000-00006. [DOI] [PubMed] [Google Scholar]
- 43.Ufer M. Comparative pharmacokinetics of vitamin K antagonists: warfarin, phenprocoumon and acenocoumarol. Clin Pharmacokinet. 2005;44:1227–12246. doi: 10.2165/00003088-200544120-00003. [DOI] [PubMed] [Google Scholar]
- 44.Kurdyak P, Juurlink D, Kopp A, Herrmann N, Mamdani N. Antidepressants, warfarin, and the risk of hemorrhage. J Clin Psychopharmacol. 2005;26:561–4. doi: 10.1097/01.jcp.0000186869.67418.bc. [DOI] [PubMed] [Google Scholar]
- 45.Optarny L, Delaney J, Suissa S. Gastro-intestinal haemorrhage risks of selective serotonin receptor antagonist therapy: a new look. Br J Clin Pharmacol. 2008;66:76–81. doi: 10.1111/j.1365-2125.2008.03154.x. [DOI] [PMC free article] [PubMed] [Google Scholar]