The impact of selective serotonin receptor inhibitors on post-endoscopic sphincterotomy bleeding, alone or with concurrent aspirin or nonsteroidal anti-inflammatory drugs

Introduction

Endoscopic retrograde cholangiopancreatography (ERCP) in the modern era is mainly therapeutic, while endoscopic sphincterotomy (ES) remains the cornerstone for a successful procedure [1]. The main complications of therapeutic ERCP include post-ERCP pancreatitis, post-ES bleeding, perforation, cholangitis, and adverse cardiopulmonary events [1-4]. The incidence of post-ES bleeding depends on the definition applied, and ranges from 2-7% [1-3]. All recently published guidelines [5-7] on antiplatelet use during endoscopic procedures recommend withholding thienopyridines while continuing aspirin (ASA) and nonsteroidal anti-inflammatory drugs (NSAIDs) in patients undergoing ES.

Selective serotonin reuptake inhibitors (SSRIs) have a well-known antiplatelet activity [8]. Platelets release serotonin in response to vascular injury; this triggers vasoconstriction and platelet aggregation, resulting in hemostasis [8]. Long-term treatment with SSRIs upregulates the expression of glucose-synthase kinase 3-β (GSK3B) on platelets [9]. GSK3B acts as a negative regulator of platelets and thrombosis and might contribute to bleeding risk with SSRI use [9]. Thus, SSRIs, either alone or in combination with ASA or NSAIDs, have been associated with an increased risk of upper gastrointestinal bleeding, post-surgery or after percutaneous endoscopic gastrostomy (PEG) [10-14].

In the literature, there is only one recent small retrospective cohort study that investigated the influence of SSRIs and serotonin–norepinephrine reuptake inhibitors on immediate and delayed bleeding post-ES and found no association [15]. The aim of our study was to investigate the possible impact of SSRIs, alone or combined with ASA or NSAIDs, on the incidence of post-ES bleeding.

Patients and methods

This study was conducted after obtaining necessary approval from the Institutional Review Board of our hospital. We retrospectively analyzed the medical records of all patients who underwent therapeutic ERCP between January 2000 and December 2017. The patients were categorized according to whether they continued to take SSRIs, alone or with ASA or NSAIDS, until the day of the procedure, or had never been on SSRIs. Patients’ SSRIs and ASA/NSAIDs dosages were not included, since the aim of the study was to analyze the bleeding risk with the aforementioned therapies. Data comprised demographics, clinical history of medical comorbidities, including renal, cardiovascular and hepatic issues, coagulation disorders, bleeding disorders, coagulation profile, blood results, endoscopic findings (periampullary diverticulum, naïve papilla), procedure details (conventional ES or precut technique, mechanical lithotripsy, type of stent placed) and complete drug history, focusing on the use of SSRIs alone, or combined with ASA or NSAIDs, at the time of the procedure. All ESs were performed by an experienced pancreatobiliary endoscopist (PK). A microprocessor-controlled ERBE electrosurgical generator was used and all ES procedures were performed with endocut current (120 W cut; 60 W coagulation, ERBE USA, Atlanta, Georgia, USA). Minimal, self-limited bleeding during the procedure was not recorded as bleeding.

The type of post-ES bleeding was classified as intraprocedural, early (bleeding episode within 24 h after ES but not evident during procedure), or delayed (bleeding episode occurred up to 20 days after ES and manifested as melena, hematemesis or hematochezia associated with a decrease in hemoglobin level). Bleeding severity was defined by consensus criteria [16] graded as follows: 1) mild, clinical evidence with a decrease in hemoglobin level (>3 g/dL), with blood transfusion not required; 2) moderate, transfusion required (≤4 units) but not angiographic intervention or surgery; or 3) severe, transfusion of 5 or more units of blood or the need for surgical or angiographic intervention.

Endoscopic hemostasis was carried out with injection of 1:10000 epinephrine solution, or with coagulation current applied by the apex of the sphincterotome or the tip of a polypectome snare, and rarely by a heater probe. The primary endpoint of the study was to investigate the impact of SSRIs, alone or with concomitant ASA or NSAID use, on the incidence, type and severity of post-ES bleeding.

Results

Using the database of our departments we identified 12,204 patients who underwent therapeutic ERCP from 2000-2017; 3058 of these were included in the final analysis. At the time of the procedure, 322 patients were taking SSRIs, 37 patients SSRIs plus NSAIDs and 98 patients SSRIs plus ASA therapy (SSRIs group) (Table 1). The control group (non-SSRIs) comprised 1925 patients who had not received SSRIs, of whom 613 and 121 patients had taken ASA or NSAIDs, respectively (Table 1). The patient characteristics of the total cohort are presented in Table 1. No statistically significant difference was found between the 2 groups regarding sex distribution, indication for ERCP, or comorbid diseases; however, patients were younger in the non-SSRIs group (P<0.05) (Table 1). The technical characteristics and the results of therapeutic ERCP are shown in Table 2, indicating no statistically significant difference between the 2 groups.

Table 1 Clinical characteristics of patients

Table 2 Technical characteristics and result of therapeutic ERCP

The incidence of post-ES bleeding was 3.9% in the SSRIs group and 3% in the non-SSRIs group, a non-significant difference (P=0.754). Likewise, there were no differences regarding the type (P=0.145) or severity (P=0.623) of post-ES bleeding between the 2 groups (Table 3).

Table 3 Incidence and severity of post.ES bleeding

Table 4 shows the association between the type of antiplatelet drug and the severity of post-ES bleeding.

Table 4 Type of antiplatelet drug and severity of post.endoscopic bleeding

Univariate analysis yielded the following significant risk factors for post-ES bleeding: precut sphincterotomy (OR 2.56, 95%CI 1.23-363; P=0.001), anticoagulation use (OR 1.89, 95%CI 1.23-3.63; P=0.022), ASA use (OR 2.05, 95%CI 1.14-3.94; P=0.018), and NSAID use (OR 1.72, 95%CI 1.02-3.03; P=0.037) (Table 5). However, multivariate analysis demonstrated only precut sphincterotomy as a significant risk factor (OR 1.98, 95%CI 1.16-4.21; P=0.001) for post-ES hemorrhage.

Table 5 Risk factors for post.endoscopic sphincterotomy bleeding

Discussion

Since its introduction, ERCP continues to be the main procedure for the endoscopic treatment of pancreatobiliary diseases [17,18]. Post-ES bleeding is a well-known complication of ES, and therefore ES is stratified as a high-risk procedure. The current guidelines of all gastrointestinal societies [5-7] include recommendations for the use of antiplatelet drugs (ASA, NSAIDs, aminopyridines) during ES, but surprisingly they do not mention the antiplatelet effect of SSRIs on post-ES bleeding.

In this retrospective cohort study, we found no statistically significant difference in post-ES bleeding associated with SSRI use alone, or with concurrent ASA or NSAIDs. Indeed, there were no differences in the incidence, type or severity of post-ES bleeding between the 2 groups studied. Specifically, the incidence of post-ES bleeding was 3.9% in the SSRIs group and 3% in the non SSRIs group, consistent with other published studies [1-3]. We also found that SSRIs, alone or with concurrent ASA or NSAIDs, were not associated with an increased risk of post-ES bleeding (Table 3). Moreover, there was no difference in the type or severity of post-ES hemorrhage, and no statistical difference in the risk factors for post-ES bleeding between the 2 groups (Table 4). However, multivariate analysis of risk factors, including SSRIs and SSRIs plus ASA or NSAIDs, showed that precut sphincterotomy was the only studied risk factor for post-ES bleeding. Precut sphincterotomy per se has been reported to have high bleeding rates, reaching up to 27% in some series [19,20]. Comparable results regarding the effect of SSRIs on post-ES bleeding were also reported in a recent small retrospective cohort study [15].

Our findings contradict those of other studies that found SRRIs were associated with a greater incidence of bleeding manifestations, including upper and lower gastrointestinal tract hemorrhage [13,14,21-23]. Indeed, in a meta-analysis of 15 case-control studies and four cohort studies, SSRIs alone were found to be associated with a modest increase in the risk of upper gastrointestinal bleeding (OR 1.68, 95%CI 1.13-2.50); this risk, however, was significantly elevated when SSRI medications were used in combination with NSAIDs (OR 4.25, 95%CI 2.82-6.42) [13]. In another retrospective study, the use of SSRIs 24 h prior to PEG was associated with an increased risk of bleeding (OR 4.1, 95%CI 1.1-13.4; P=0.04) [14].

SSRIs are used as first-line pharmacotherapy for depression and are approved for the treatment of various psychogenic disorders because they have a favorable safety profile compared with older-generation antidepressants [13]; however, they are known to demonstrate antiplatelet activity [8] and to act on reduction in platelet/endothelial activation synergistically with concurrent antiplatelet drugs, such as ASA, NSAIDs and aminopyridines [13]. SSRI use has been reported to be associated with an increased risk of upper gastrointestinal bleeding [12] and post-procedure bleeding after PEG [14]. Moreover, a retrospective study of more than 530,000 adults who underwent major surgery showed that SSRI use increased the rate of late post-surgical bleeding slightly but significantly. Additionally, a recent meta-analysis showed that the concomitant use of SSRIs and NSAIDs or ASA is associated with a greater increase in the risk of upper gastrointestinal bleeding than the use of either type of drug alone [9]. One possible explanation for these contradicting results is that during ES a cutting current that also coagulates is used, thus ablating the exposed vessels; it is well known that endocut mode supplies electrosurgical current in intermittent pulses, automatically reducing its voltage after the starting phase and releasing a “soft” coagulation current, insufficient to cut tissue, between pulses of cut current [24]. This is different to the procedure for PEG and upper and lower gastrointestinal bleeding, where no thermic method is applied to the exposed vessels. Indeed our results are in accordance with a retrospective cohort study regarding the impact of SSRIs on post-ES bleeding [15].

It is important to note that clinicians must consider the risk–benefit ratio of discontinuing an SSRI before an elective operative procedure. Discontinuing SSRI medications may result in discontinuation syndrome, symptom recrudescence, or relapse of depression [23,25]. Although the morbidity might be greater in patients under SSRI treatment, the mortality is still low [11].

Our study presents important limitations. No definite guidelines can be derived from this retrospective study, while all procedures were performed by an experienced pancreatobiliary endoscopist and this may have affected the generalizability of our results. It is well-known that a low case volume [26-28] on the part of the endoscopist may reflect less precise control of the incision, increasing the “zipper” cut phenomenon and therefore the post-ES bleeding. In addition, the sample size of SSRIs users was small, exposing our analysis to the possibility of type II error. Although we had 457 SSRIs users in our study and found an incidence of post-ES bleeding of 3.9% in the SSRIs group and 3% in the non SSRIs group, the necessary number of patients enrolled in each arm to detect a statistically significant difference would be approximately 5300. We consider that a prospective, appropriately powered trial is indicated to further investigate the impact of SSRIs, alone or in combination with ASA and NSAIDs, on post-ES bleeding.

In conclusion, our study found no evidence for an effect of SSRIs, alone or in combination with ASA or NSAIDs, on post-ES bleeding, though powerful prospective multicenter studies are needed to confirm our results.