Franziska Geiser, MD, Klinik und Poliklink fuer Psychosomatische Medizin und Psychotherapie, Universitaet Bonn, Sigmund-Freud-Street 25, D-53105 Bonn, Germany. Email: firstname.lastname@example.org
Aims: Anxiety disorders have been shown to be correlated with an activation of coagulation and impairment of fibrinolysis. The aim of the study was to assess whether medication with a serotonergic antidepressant, which has been associated with abnormal bleeding, may modify this effect.
Methods: Thirty-one anxiety patients, mostly with comorbid depression, and 31 healthy controls were included in the study. Group differences between anxiety patients medicated with a serotonergic antidepressant, patients without serotonergic antidepressant and controls were assessed for activated partial thromboplastin time, fibrinogen, factor VII, factor VIII, von Willebrand factor, von Willebrand ristocetin cofactor activity, prothrombin fragment 1 + 2, thrombin-antithrombin complex, d-dimer, α2-antiplasmin, plasmin-α2-antiplasmin complex (PAP), tissue plasminogen activator and plasminogen activator inhibitor. Intervening variables, such as age, sex, body mass index and smoking, were accounted for.
Results: We found lower coagulation measures for fibrinogen (P = 0.03) and plasminogen activator inhibitor (P = 0.01), and higher levels of PAP (P = 0.046) in patients with serotonergic antidepressant than in patients without serotonergic antidepressant. When controlling for smoking and body mass index, differences between the two groups were significant for PAP (P = 0.02), von Willebrand ristocetin cofactor activity (P = 0.02) and activated partial thromboplastin time (P = 0.046). Coagulation scores were similar in patients with serotonergic antidepressant to those of healthy controls.
Conclusions: Serotonergic antidepressants may counteract a procoagulant effect of anxiety and/or depression in anxiety patients.
IN A PREVIOUS study,1 we found an activation of coagulation factors in the direction of a hypercoagulable state in patients with an anxiety disorder in comparison to healthy controls, particularly showing a significantly elevated level of plasminogen-activator-inhibitor antigen (PAI-1), indicating an impairment of fibrinolysis. A hypercoagulable state could be one explanation for the increased risk for cardiovascular diseases associated with anxiety symptoms.2–4 Antidepressants, more specifically selective serotonin reuptake inhibitors (SSRI), serotonin-norepinephrine reuptake inhibitors (SNRI) and noradrenergic and specific serotonergic antidepressants (NaSSA) are the first line of treatment for anxiety disorders alone or for anxiety disorders with comorbid major depression.5,6 Serotonin plays a role in primary hemostasis platelet activity, and serotonergic antidepressants have been shown to inhibit platelet activity more than tricyclic antidepressants.7–9 Also, a causal association between serotonergic antidepressants and abnormal bleeding has been stated.10 While the effect of serotonergic antidepressants on platelets has been repeatedly confirmed, the few studies focusing on the coagulation cascade in depressed patients have failed to show an effect of serotonergic antidepressants on coagulation factors.11–16 A (simplified) model of the coagulation cascade is shown in Figure 1. To our knowledge, the effect of serotonergic antidepressants on coagulation has not yet been studied in patients with an anxiety disorder as the primary diagnosis.
In our previous publication1 we reported no significant association of antidepressants with coagulation when sum scores (sums of z-scores of several parameters of coagulation or fibrinolysis) were used as dependent variables. However, the statistical effect of antidepressants on the single parameter level was not tested, and there was no differentiation between different antidepressant drugs, and more specifically between serotonergic antidepressants and tricyclic antidepressants (TCA) or anxiolytics. In the present study, we reanalyzed our data focusing on the effect of serotonergic antidepressants on markers of coagulation and fibrinolysis in patients with a severe anxiety disorder.
Thirty-one patients with an anxiety disorder as first diagnosis (panic disorder with agoraphobia [n = 15], social phobia [n = 6], panic disorder with/without agoraphobia and social phobia [n = 8], generalized anxiety disorder [n = 1], post-traumatic stress disorder with anxiety [n = 1]) admitted to the Psychosomatic University Clinic of Bonn for inpatient psychotherapy were consecutively included in this study. Thirty-one healthy volunteers, matched by sex and age, were recruited as the control group. Hemostasis measures of 29 patients have already been reported elsewhere.1 Anxiety was associated with activation of coagulation and impaired fibrinolysis, without difference between the anxiety disorders included.
The study was conducted with ethics committee approval in conformity with the provisions of the Declaration of Helsinki (revised 2004). Patients and controls gave written informed consent. Exclusion criteria were language problems, age under 18 or over 65, diabetes or other serious somatic diseases (other than mild hypertension or thyroid dysfunction with effective treatment), pregnancy, medication with oral anticoagulants or corticosteroids, and the intake of a thrombocyte inhibitor (most frequently as pain killer) within the last 2 weeks.
Diagnoses were confirmed by the structured clinical interview for DSM-IV (SCID, German version17). Sample characteristics are given in Table 1. Comorbid axis I diagnoses were mood disorders (depressive episode, mild or moderate [n = 29], severe [n = 1]), other anxiety disorders (n = 9), somatoform disorders (n = 4), eating disorders (n = 5), substance-related disorders (n = 2) and adjustment disorders (n = 1). Comorbid axis II disorders were prevalent in 21 patients (cluster B [n = 6] and cluster C [n = 15]).
Table 1. Sample characteristics (mean ± SD)
Anxiety patients without serotonergic antidepressant
Anxiety patients with serotonergic antidepressant
Significance level anova (exception: sex and comorbidity: χ2-test)
ADS, German version of the Center for Epidemiological Studies Depression Test; NS, not significant; STAI, State Trait Anxiety Inventory.
31.9 ± 10.5
31.9 ± 9.9
31.2 ± 9.7
Trait anxiety (STAI)
57.6 ± 9.3
61.1 ± 10.8
38.8 ± 11.9
P < 0.001
State anxiety (STAI)
53.8 ± 11.7
46.6 ± 12.5
35.9 ± 8.2
P < 0.001
37.3 ± 8.1
36.9 ± 11.6
13.7 ± 8.3
P < 0.001
Worry about blood drawing
0.79 ± 0.79
0.50 ± 0.52
0.29 ± 0.46
P = 0.02
Body mass index (kg/m2)
25.5 ± 4.8
22.5 ± 2.4
22.3 ± 3.0
P = 0.01
5.0 ± 7.1
16.2 ± 11.0
2.4 ± 5.7
P = 0.002
Alcohol (beer or wine)/week (cL)
1.8 (n = 14)
Comorbid axis II-diagnosis
Of the 31 patients, 16 took an antidepressant. Twelve patients were medicated with a serotonergic antidepressant, most frequently an SSRI (citalopram 20–40 mg/day [n = 8], fluoxetine 10–20 mg/day [n = 2]). One patient each took an SNRI (venlafaxine 75 mg/day) or an NaSSA (mirtazapine 30 mg/day). In this dosage, venlafaxine is considered to be mainly an SSRI and to have very little effect on norepinephrine reuptake.18 Mean duration of medication with a serotonergic drug was 8.7 weeks (range 2–15 weeks). Six patients took a tricyclic antidepressant or anxiolytic and of these, two patients were also taking a serotonergic antidepressant (escitalopram + opipramol; fluoxetine + amitriptyline) and four were medicated with a tricyclic drug only (opipramol [n = 2], doxepin [n = 1] or amitriptyline [n = 1]; without serotonergic antidepressant). Five patients were medicated with promethazine, melperone or perazine (twice in combination with a serotonergic antidepressant, three times without serotonergic antidepressant). Among the patients with serotonergic medication, one had a contraceptive, and one was taking thyroxine. Among the patients without serotonergic medication, two were taking thyroxine, and one was medicated with ramipril and bisoprolol. The control group was without any psychotropic or other medication, except one woman with contraceptives.
The patient groups with and without serotonergic antidepressant did not differ with regard to sex distribution, age, or axis II comorbidity (see Table 1).
The State Trait Anxiety Inventory (STAI) German version19 assesses trait and state anxiety with two subscales of 20 items each. State anxiety reflects a transitory emotional state that is characterized by feelings of tension and apprehension, and heightened autonomic nervous system activity. Trait anxiety refers to a relatively stable tendency to respond with anxiety to perceived threats. The Allgemeine Depressionsskala20 (ADS) is the German version of the Center for Epidemiological Studies Depression Test21 and measures depression by 19 items covering depressive affect, somatic symptoms, behavioral inhibition and negative cognitions. Because anticipation of blood drawing could be a factor interfering with the resting condition, especially in anxiety patients, we assessed worry about blood drawing by means of an additional question with a three-point answering scale (‘How do you feel about the forthcoming drawing of blood? Does not worry me, Worries me a little, Worries me a lot’).
Blood sampling and assays
Blood sampling always took place between 16.00 and 18.00 hours within the first week after admission. Patients were asked to avoid smoking, alcohol or coffee consumption and physical exertion for 2 h prior to blood collection. After 15 min of rest in a sitting position, a cubital vein was punctured using a 21-gauge butterfly system. Blood was drawn into a 10-mL syringe containing 3.8% trisodium citrate, stored in a vertical position at room temperature (22 ± 2°C), then centrifuged and frozen for storage at −70°C within 2 h. Assays were performed by commercially available test kits according to European norm accredited laboratory. Tests comprised the measurement of the activated partial thromboplastin time (aPTT), the determination of international normalized ratio (INR) based on the prothrombin time, a coagulometric determination of fibrinogen, a prothrombin time-based measurement of clotting factor VII (FVII), an activated partial thromboplastin time-based test for clotting factor VIII (FVIII) (one-stage assay), enzyme-linked immunosorbent assays (ELISA) for von Willebrand factor (vWF), prothrombin-fragment 1 and 2 (F1+2), thrombin-antithrombin-complex (TAT), d-dimer, tissue-plasminogen-activator (t-PA), PAI-1 and plasmin-α2-antiplasmin-complex (PAP), a chromogenic assay for α2-antiplasmin (α2-AP), and an aggregometric determination of ristocetin-associated von Willebrand factor activity (vWF:RCo).
For statistical analyses spss for Windows version 14.0 (spss Inc., Chicago, IL, USA) was used. The central hypothesis concerning the association of antidepressant medication with measures of coagulation was tested by univariate anova, with Bonferroni post-hoc testing of group differences. We compared three groups: anxiety patients medicated with a serotonergic antidepressant (n = 12), anxiety patients without serotonergic antidepressant (n = 19), and controls (n = 31). Group differences in possibly confounding variables (age, sex, body mass index [BMI], smoking, alcohol consumption) as well as psychological variables that could be mediating the effect of serotonergic antidepressants (trait and state anxiety, depression, worry about blood drawing) were equally tested as dependent variables by anova. When relevant, these factors were introduced into anova as covariates. Significance levels were set at P < 0.05 with two-tailed testing.
Measures of coagulation and hemostasis in anxiety patients with/without serotonergic antidepressant and controls
We found a significant main effect for differences between the three groups in the following hemostatic markers (see Table 2): fibrinogen, PAP, t-PA and PAI-1. A trend could be detected for aPTT, FVIII and vWF:RCo. In post-hoc testing, fibrinogen and PAI-1 were significantly higher in patients without serotonergic antidepressant than in patients with serotonergic antidepressant. FVIII and t-PA tended to be higher in patients without serotonergic antidepressant than in patients with serotonergic antidepressant. PAP was significantly lower in patients without serotonergic antidepressant than in patients with serotonergic antidepressant, and the same applied to aPTT as a trend. In all these measures, the control group scored in between the two patient groups. We found no significant main effect for the difference between the three groups for prothrombin time, INR, FVII, vWF-Ag, TAT, F1+2, d-dimer and α2-AP.
Table 2. Means ± standard deviations of hemostatic measures in the three groups and anova results
Group 1: anxiety patients without serotonergic antidepressant (n = 19)
Group 2: anxiety patients with serotonergic antidepressant (n = 12)
We repeated the test for difference of the mean (univariate anova) in hemostatic markers between patients with serotonergic antidepressant and patients without serotonergic antidepressant, now including as covariate those intervening factors that had shown a significant difference between the two patient groups (i.e. number of cigarettes smoked per day and BMI, see Table 1). We found a significant group effect independent of smoking and BMI on PAP (P = 0.02), vWF:RCo (P = 0.02) and aPTT (P = 0.046). The group differences in fibrinogen (P = 0.11), FVIII (P = 0.08), t-PA (P = 0.31) and PAI-1 (P = 0.27) were no longer significant. However, the influence of the covariate variables as an explanation for the loss of significance is plausible only for PAI-1, as a significant covariate effect emerged only for BMI on PAI-1 (P = 0.02), and for smoking on α2-AP (P = 0.03) and marginally on PAP (P = 0.08).
Group differences in psychological variables
Trait and state anxiety as well as depression were similar in both patient groups, but (as expected) significantly higher in patients than in the control group (see Table 1, Bonferroni adjusted post-hoc tests: no significant differences between patient groups; STAI-trait: patient groups vs controls P < 0.001; STAI-state: with serotonergic antidepressant vs controls P = 0.01; without serotonergic antidepressant vs controls P < 0.001; ADS: patient groups vs controls P < 0.001). The worry about the forthcoming drawing of blood differed significantly only between patients without serotonergic antidepressant and controls (P = 0.02), patients with serotonergic antidepressant scoring in between the two.
As worry about blood drawing had been correlated with coagulation in our previous study, we repeated the anova for mean differences in hemostatic variables, now testing the influence of worry about blood drawing as covariate. Worry about blood drawing showed a significant covariate effect only on FVIII and vWF, with reduction of a previously detected group effect (patients with/without serotonergic antidepressant) to insignificance.
In a previous publication we reported an enhanced level of hemostatic markers, with activation of coagulation and predominant inhibition of fibrinolysis, in patients with an anxiety disorder.1 Although these hemostatic variations remained within the physiological range, they might contribute to the increased statistical risk for cardiovascular events connected with anxiety, and are thus of high clinical interest.4 As in most samples of patients with severe anxiety, comorbidity with depression was very high, thus making it difficult to separate the effects of anxiety and depression. Nevertheless, the present analysis shows that medication with a serotonergic antidepressant could possibly attenuate the effect of anxiety or anxiety with depression on coagulation function: when taking a serotonergic antidepressant on a regular basis, the anxiety patients' levels of fibrinogen, FVIII, vWF:RCo, t-PA and PAI-1 were lower and the levels of aPTT and PAP were higher (significantly or as a trend) than in patients without serotonergic antidepressant. Patients with serotonergic antidepressant had levels of coagulation factors comparable to those of healthy controls. The direction of the changes points towards an activation of the coagulation system in anxiety patients, which was antagonized when taking a serotonergic antidepressant. The results for aPTT are concordant with this interpretation, as a prolongation indicates a reduction of coagulation activity. Changes were most impressive for the fibrinolytic system. Although t-PA is an activator of fibrinolysis, measured t-PA is mostly bound in t-PA/PAI-1-complexes and thus partly reflects PAI-1 (inhibiting) activity. A distinctly reduced level of PAI-1 paired with an increased level of PAP in the patient group with serotonergic antidepressant (compared to patients without serotonergic antidepressant) indicates an enhancement of fibrinolysis and therefore a reduction of coagulability. When controlling for the influence of smoking and BMI, the effect on PAP, vWF:RCo and aPTT remained stably significant.
In our previous study, of the individual hemostasis parameters, only PAI-1, α2-AP and FVII had shown a statistically significant increase associated with an anxiety disorder. However, the influence of the intake of serotonergic antidepressants was not controlled for. According to our present results, α2-AP and FVII seem to be factors less influenced by the intake of a serotonergic antidepressant. Our reanalysis suggests that an activation of other hemostasis markers associated with anxiety, in particular fibrinogen, FVIII, vWF, PAP and t-PA, might have been masked by the opposite effect of medication with a serotonergic antidepressant. As almost 40% of patients in our sample were medicated with a serotonergic antidepressant, we conclude that the previously reported association of anxiety on coagulation and fibrinolysis with a procoagulable state might be even larger than originally reported.
In previous studies, platelet hyperreactivity associated with depression was found to be normalized when medicated with an SSRI.22 This is in line with our finding that activation of coagulation and impairment of fibrinolysis associated with anxiety and depression is less pronounced when medicated with a serotonergic antidepressant. More research is called for to study a possibly beneficial effect of these drugs on cardiovascular risk in patients with an anxiety or depressive disorder.
Our study suffers from several restrictions. Due to the small sample, statistical power was restrained. A correction of the alpha error was performed in post-hoc testing concerning the three sample subgroups, but not for the repeated testing of different coagulation factors. Patients were medicated with different substances from the serotonergic antidepressant group, including in two cases a dual SNRI or NaSSA drug. Also, although the proportion of patients taking a tricyclic or other psychotropic drug was small in the two patient groups (with serotonergic antidepressant [n = 4], without [n = 7]), this medication as well as the scarce somatic medications may have interfered with our results. As serotonin levels in plasma were not assessed, conclusions about a possible pathway from serotonergic drugs to hemostatic changes remain hypothetical. Although hemostasis activity is generally higher in the morning,23 due to logistic reasons, blood was always sampled between 16.00 and 18.00 hours. This applied to all three groups (patients with and without serotonergic antidepressant and controls), therefore we do not expect time of blood sampling to bias our results.
An increased risk of abnormal bleeding has been reported to be correlated with the degree of serotonin uptake inhibition by serotonergic antidepressants.10,24,25 On the other hand, in large epidemiologic studies, such as the Enhancing Recovery in Coronary Heart Disease Patients (ENRICHD) trial and the Sertraline AntiDepressant Heart Attack Randomized Trial (SADHART), the use of SSRI has shown to be associated with a decreased risk of death or myocardial infarction as well as with decreased platelet and endothelial activation markers.26–28 To date, most research has concentrated on platelet serotonin uptake and impairment of platelet function through SSRI.2,29 Our results are in contrast to a small number of studies that found no impact of serotonergic antidepressants on coagulation factors in patients with depression.12–16 However, with the exception of the study of Lederbogen, the power of these studies was limited due to small samples (n = 5–10), most studies did not assess single coagulation factors, the severity of illness was not specified, and anxiety was not focused.
But how can serotonergic antidepressants affect coagulation factors? A simple hypothesis would be that they reduce illness severity, thus attenuating the effect of anxiety or depression on coagulation. In our study, this assumption is refuted by the fact that patients with and without serotonergic antidepressant had similar scores in depression and trait anxiety. Changes of coagulation and fibrinolysis have been explained by an activation of the sympathetic system and mediation through catecholamines.30 In our sample, patients without serotonergic antidepressant reported higher levels of acute worry about blood drawing. In the statistical analysis, the effect of serotonergic antidepressant on factor VIII and vWF decreased when controlling for the acute worry about blood drawing. Possibly, a reduction in acute situational anxiety and lower level of sympathetic activity could have been a mediator of the effect of serotonergic antidepressants on these coagulation factors.
However, higher levels of PAP, an indicator of fibrinolytic activity, and a prolonged aPTT, a clinically relevant global measure of hemostasis, in patients medicated with serotonergic antidepressants were independent of acute anxiety. There are several possible direct impact mechanisms of serotonergic antidepressants on hemostasis. Serotonin is a potent platelet activator.31 Serotonergic antidepressants are known to decrease the granular storage of serotonin in platelets and this effect has been proposed as a mechanism of reduced platelet aggregation as well as reduced activity of the coagulation cascade.11,32 Serotonin plasma levels rise at the beginning of treatment with serotonergic antidepressants, but seem to be normalized and even decreased after longer treatment,33–35 and serotonin concentrations are correlated with fibrinolytic inhibition.36 Citalopram is associated with an enhanced production of nitric oxide, which is related to endothelial function.37 Glucocorticoid receptors, which respond to treatment with antidepressants, are not only present in the brain, but also in platelets.38,39 Furthermore, recent studies have shown that treatment with serotonergic antidepressants may modify serum levels of brain-derived neurotrophic factor (BDNF), which can be found in human platelets and is released by platelet agonist stimulation. Blood levels of BDNF are reported to be associated with depression as well as with myocardial infarction.40–42 Although these mechanisms were not the subject of our study, we assume that the effect of serotonergic antidepressants on coagulation and fibrinolysis in our study cannot be explained only by reduction of acute anxiety and arousal, but also reflects the direct impact of antiserotonergic mechanisms on the hemostatic system.
This study broadens the view of reported associations between the use of serotonergic antidepressants and hemostatic complications from the focus on platelet dysfunction to the inclusion of changes in the coagulation cascade. Further exploration of the underlying physiologic processes remains an important topic for future research.