Incidence and risk factors of colorectal delayed post‐polypectomy bleeding in patients taking antithrombotics

Objective In this study we aimed to investigate the incidence and risk factors for delayed post‐polypectomy bleeding (DPPB) in Chinese patients taking antithrombotics including antiplatelet agents and anticoagulants. Methods A retrospective study was conducted in patients who underwent colorectal polypectomy from January 2017 to May 2020. Their demographic characteristics, features of the polyps including number, size, morphology, and location, and use of antiplatelet agents and anticoagulants were collected. The incidence and risk factors for DPPB were compared between the patients with and without antithrombotic use. Results A total of 5152 polyps from 2267 patients were resected under endoscopy. Of these patients, 35 (1.54%) experienced DPPB. Compared with the control group who did not take antithrombotics (1.18%), the incidence of DPPB was significantly higher in patients treated with heparin bridge (HB) therapy (17.39%; P < 0.001) and clopidogrel (4.88%; P = 0.022), but did not differ in patients taking aspirin (1.28%), dual antiplatelet therapy (3.70%), warfarin alone (0%), or direct oral anticoagulants (3.85%). Using the multivariate analysis, HB therapy (odds ratio [OR] 16.735, 95% confidence interval [CI] 4.320‐64.834, P < 0.001), male sex (OR 3.825, 95% CI 1.298‐11.265, P = 0.015), polyps >1 cm (OR 4.584, 95% CI 1.782‐11.794, P = 0.002) and rectal polyps (OR 8.820, 95% CI 3.968‐19.602, P < 0.001) were independently associated with a high risk of DPPB. Conclusions HB and clopidogrel therapies significantly increase the incidence of DPPB. HB therapy, male sex, polyp size and polyps located in the rectum are significant risk factors for DPPB.

their discontinuation and resumption of antithrombotic medications.
The number, size, morphology and location of the polyps were recorded by reviewing the colonoscopy reports. Patients gave their written informed consent to undergo the endoscopic procedure and to their potential inclusion in the retrospective analysis afterwards.

| Antithrombotics
Aspirin and clopidogrel (or ticagrelor) were suspended at 7 and 5 days, respectively, prior to the endoscopic procedure, until 48 hours after the procedure. Similarly, DOAC were discontinued at least 48 hours before endoscopy and resumed within 48 hours after the procedure.
Warfarin was stopped at 5 days before endoscopy. Patients on warfarin were divided into a warfarin alone and a warfarin with HB therapy group according to their thromboembolic risk, as defined based on the European guidelines. 1 In those at a low risk of thromboembolism, a polypectomy was performed when their international normalized ratio (INR) decreased to <1.5, and warfarin was resumed at the night of endoscopy. While in patients at a high risk of thromboembolism, lowmolecular-weight heparin (LMWH) was started 2 days after warfarin was discontinued and stopped at least 24 hours before the endoscopic procedure. Warfarin was resumed at the night of endoscopic procedure and LMWH was continued until the INR returned to the therapeutic range (1.8-2.5).

| Endoscopic resection of the polyps
Colonoscopic polypectomy was performed using an electric endoscope (CF-HQ290I or CF-H260AI; Olympus, Beijing, China). Generally, polypectomy was performed by endoscopic mucosal resection or a hot snare polypectomy using an electrosurgical device (VI0200S; ERBE China, Shanghai, China). Prophylactic clipping was routinely performed after polyps larger than 5 mm were resected by using the hot polypectomy, and hemostatic clipping or thrombin spraying was considered when delayed bleeding occurred. All procedures were performed by five experienced endoscopists, each of whom had performed more than 10 000 colonoscopies.

| End-points
The primary end-point of the study was DPPB, which was defined as overt melena or hematochezia occurring from 24 hours to 30 days after the polypectomy. Follow-up visit was performed via telephone every other week after patient's discharge to ascertain whether melena or hematochezia occurred, which guaranteed their immediate referral to a gastrointestinal consultation. For those without DPPB, a postoperative evaluation was routinely arranged at 1 month after discharge. The secondary end-points were interventional radiology or surgery, severe thromboembolic events, and death. Thromboembolic events were defined as stroke, transient ischemic attack, acute coronary syndrome, systemic embolism, pulmonary embolism or deep vein thrombosis. 6

| Statistical analysis
Statistical analyses were performed using the SPSS Statistic version 22.0 (IBM, Armonk, NY). Continuous variables were expressed as mean ± standard deviation and the differences between the two groups were analyzed by using an independent-sample t-test, whereas categorical variables were expressed as numbers and percentages and were compared by using the χ 2 test or Fisher's exact test. A logistic regression analysis was performed to determine the risk factors associated with DPPB. Variables with a P value lower than 0.1 in the univariate analyses were included in the multivariate model. Odds ratio (OR) and 95% confidence interval (CI) were calculated to identify independent prognostic factors for DPPB. A two-tailed P value of <0.05 was considered to indicate statistical significance.

| Risk factors for DPPB
The characteristics of the patients who experienced DPPB and those who did not are summarized in Table 2. Compared with patients in the non-bleeding group, those in the bleeding group there were significantly higher proportions of male subjects, those with polyps >1 cm and F I G U R E 1 Flowchart of patient enrollment. Abbreviations: DAPT, dual antiplatelet therapy; DOAC, direct oral anticoagulants; ESD, endoscopic submucosal dissection; HB, heparin bridge F I G U R E 2 Time point of delayed post-polypectomy bleeding (DPPB) pedunculated and rectal polyps. There were no significant differences in age and comorbidities between the two groups (all P > 0.05).
In the univariate analysis (Table S1)

| Emergency endoscopy for DPPB
All of the 35 patients with DPPB underwent an emergency endoscopy, including 32 patients treated with hemostasis with clipping, one with thrombin spraying on the bleeding site, and two having no active bleeding after fasting treatment and thus were not treated endoscopically. None of them required a blood transfusion, interventional radiology or surgery, and no thromboembolic events or death occurred.

| DISCUSSION
The current study showed that the overall incidence of DPPB was 1.54% after resection of colorectal polyps. The patients on antithrombotics had a significantly increased risk of bleeding than the control group who did not taking antithrombotics (2.90% vs 1.18%, P = 0.006). In previous studies, the incidence of DPPB varies significantly because of different definitions of DPPB and management with antithrombotics. Our results are consistent with those reported previously, with the incidence of DPPB being 1% in patients not on antithrombotics, 2 and 1.8%-7.0% in those who temporarily discontinue and rapidly resume antithrombotic drugs when compared with 11% in patients who have not discontinued antithrombotics. 7 Most delayed bleeding events occur within 1-2 weeks after endoscopic resection. 8 And in our patients DPPB occurred at 1-7 days after endoscopy (median 3 d).
The HB therapy subgroup had a significantly higher rate of DPPB than the control group (17.39% vs 1.18%, P < 0.001), and HB therapy was an independent risk factor for DPPB (OR 16.735). The guidelines have suggested that warfarin should be discontinued and replaced by HB therapy in patients at a high risk of thrombosis. 1-3 However, the HB therapy has been reported to be an independent risk factor for DPPB and is associated with a DPPB incidence of 14.9%-22.2% (vs 0%-1.9% in patients who did not receive HB therapy), [9][10][11] which is consistent with our results. Heparin increases the risk of DPPB in warfarin users because warfarin and heparin may be used simultaneously for several days after the procedure to achieve sufficient anticoagulation.
However, in the BRIDGE trial, HB therapy increased the risk of bleeding during invasive surgery but did not significantly prevent thromboembolism. 6 Therefore, endoscopists should strictly select patients with an indication for HB therapy in order to reduce the DPPB rate.
Use of clopidogrel was significantly associated with a higher incidence of DPPB than the control group but was not an independent risk factor for DPPB in the multivariate analysis. Two studies 12,13 indicated that clopidogrel increased the risk of DPPB, and the latter also found that clopidogrel was not an independent risk factor for DPPB.
In a recent randomized controlled trial of uninterrupted clopidogrel use, 14 the incidence of DPPB was 3.8% in the clopidogrel group and 3.6% in the placebo group. The DPPB rate in the placebo group was higher than expected, possibly due to the resumption of clopidogrel.
Although clopidogrel was discontinued during the procedure in our study, the bleeding rate remains high, which might have been related to the resumption of clopidogrel at 48 hours after polypectomy.
The aspirin group had a similar DPPB rate as that of the control group. Aspirin suspension before polypectomy is not recommended in the current guidelines, 1-3 but cessation of aspirin was documented in all patients 7 days prior to the procedure in our retrospective study.
Anecdotally, Asian practitioners seem to be more prudent about antiplatelet decisions, whereas many American physicians tend to keep aspirin administration during endoscopy, 15 although 32.5% of gastrointestinal centers instruct aspirin to be discontinued before colonoscopy in the United States. 16 Consistent with our results, most studies indicated that aspirin did not increase the risk of DPPB, [17][18][19] except in cases of large colonic lesions. 20 DAPT has been reported to increase the risk of DPPB. 13 In this study, the risk of DPPB was higher in the DAPT group than in the control group; however, this is not statistically significant, possibly because of the small number of patients receiving DAPT. Therefore, the decision to discontinue antiplatelet agents should be made based on the risks of thromboembolism and DPPB. women and thus are more likely to undergo polyp removal during a colonoscopy. Studies have also reported that polyps in the right hemicolon leads to a higher DPPB rate than the lesions located in the left hemi-colon. 9,19,24 In Chinese patients, left colon and rectal neoplasms are more common 28 than they are in Western populations. The incidence of DPPB was higher for rectal polyps, possibly because of the abundant blood vessels in the rectal mucosa, and similar results have been reported in a Japanese study. 29 Almost half of DPPB cases do not require therapeutic intervention. 30 These factors can be used to further stratify the risk of DPPB and identify patients likely to benefit from colonoscopy.
Although none of the patients developed thromboembolism in this study, thrombosis caused by the withdrawal of antithrombotics may more likely be fatal in high-risk patients. Therefore, the collaborative evaluation of the risk of bleeding and thromboembolism is needed, and the appropriate treatment timing and plan must be determined.
One strength of our study is that we enrolled a relatively large number of patients on anticoagulant or antiplatelet therapy whose bleeding data were acquired by investigators do not have access to the endoscopic interventions the patients receive. There were some limitations to the study, one of which is the retrospective singlecenter study design; eg, the distribution of patients was unequal among the groups, and the sample size of the DAPT and anticoagulant users was relatively small, which might have affected the liability in the statistical analyses (see the wide 95% CI in the logistic model) and undermined the value of extrapolating these results to clinical practice. However, similar results have also been noticed in studies 11,29 with sample sizes of HB therapy ranging from 20 to 45. Although these patients are relatively uncommon in clinical setting, our findings, along with those of other studies, imply that use of HB therapy warrants more caution than it has hitherto received. In addition, due to the limited number of patients undergoing cold snare polypectomy in our endoscopy center, these cases were not included in this study and studies on these patients will be conducted in the future.

| CONCLUSIONS
In conclusion, patients on antithrombotics had a higher risk of DPPB than those who did not receive antithrombotics. HB and clopidogrel therapies significantly increased the incidence of DPPB. HB therapy, male sex, polyp size >1 cm, and polyps located in the rectum were independent risk factors for DPPB.