Pharmacodynamics of direct oral anticoagulants associated with post-endoscopic delayed mucosal resection bleeding of colorectal neoplasms Authors: Masaki Murata, Mitsushige Sugimoto, Satoshi Ueshima, Yasuaki Nagami, Masaki Ominami, Yohei Furumoto, Tatsuya Mikami, Manabu Sawaya, Naohisa Yoshida, Ken Inoue, Yasuhiko Maruyama, Tomohiro Terai, Shigetsugu Tsuji, Takeshi Yamamura, Takanori Yamada, Yoji Takeuchi, Nobuaki Yagi, Satoshi Osawa, Yashiro Yoshizawa, Shunichiro Orihara, Shin’ichi Miyamoto, Sayana Matsuda, Daiki Hira, Tomohiro Terada, Toshiya Katsura, Mitsuhiro Fujishiro, Takashi Kawai Published: 2025 ABSTRACT AbstractAnticoagulant effects of direct oral anticoagulants (DOACs) are influenced by their plasma levels and anti-factor Xa (FXa) activity. Delayed bleeding is a major adverse event associated with endoscopic mucosal resection (EMR) of colorectal neoplasms. However, the association between post-EMR bleeding and pharmacodynamics in patients taking DOAC remains unknown. This study investigated the association between pharmacological risk factors and delayed bleeding in patients receiving DOAC. This multicenter prospective trial in Japan evaluated the incidence of post-EMR delayed bleeding in DOAC users between April 2018 and May 2022 and determined the association of post-EMR delayed bleeding with plasma levels and FXa activity. The incidence of post-EMR delayed bleeding was 10.5% (95% confidence interval [CI]: 5.3%-18.0%). There were large individual differences in plasma levels and FXa activity for each DOAC. At the trough and Tmax, plasma DOAC levels and plasma level/dose ratios were significantly correlated with FXa activity. Multivariate analysis showed that high anticoagulant ability was a risk factor for post-EMR delayed bleeding (adjusted OR, 12.211; 95% CI: 1.081-137.922, p = 0.043). DOAC users with high anticoagulant activity may be at higher risk of post-EMR delayed bleeding from colorectal neoplasms.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-025-33056-w.Keywords: Direct oral anticoagulant, Pharmacokinetics, Bleeding, Endoscopic mucosal resection, PolymorphismSubject terms: Colonoscopy, Risk factors, Predictive markers Introduction Endoscopic resection of adenomatous polyps reduces the incidence of colorectal cancer by 76%–90% and mortality by 53%1. Endoscopic mucosal resection (EMR) is a minimally invasive and versatile treatment2. However, EMR frequently causes delayed bleeding (0.6%–3.2%)3–10 due to lesion-, patient-, and treatment-related factors11–15. Although a report showed that direct oral anticoagulant (DOAC) use has no significant effect on the rate of delayed post-EMR bleeding16, the use of DOACs is generally associated with a higher rate of bleeding complications than the use of no antithrombotic drugs (14.8% vs. 3.1%)9. These findings indicate the need for substantial caution regarding resection in patients taking DOACs17. Warfarin can be monitored by the prothrombin time-international normalized ratio; however, parameters able to simply and accurately monitor the anticoagulant effect of DOACs have not been identified. While Japanese guidelines specify that all DOAC users should discontinue DOACs on the morning of the day of treatment18, the inability to monitor the anticoagulant effects indicates that the actual risk of treatment-related complications may vary among patients, especially with regard to bleeding. Clinical guidelines and consensus reports have been published in various countries and regions; however, the severity groups for the risk of post-procedure bleeding differ among guidelines, and the types of endoscopic procedures within each severity group also differ19–24. The anticoagulant effects of DOACs are pharmacologically associated with their plasma levels and anti-factor Xa (FXa) activity at the time of maximum plasma concentration (Tmax) and trough25–27. Pharmacokinetics and pharmacodynamics are associated with the risk of major bleeding events, including gastrointestinal bleeding28–31. Pharmacokinetics are influenced by the DOAC type, dose, renal and hepatic function, age, sex, body weight, drug interactions, and drug-metabolizing enzyme gene polymorphisms27,32. Therefore, it is often speculated that plasma levels and FXa activity may aid in monitoring anticoagulant efficacy in DOAC users and might allow the identification of patients with a high bleeding risk33. We previously demonstrated that when the Youden index from receiver operating curve analysis using FXa activity was used to determine the cutoff values of high anticoagulant ability with high post-endoscopic submucosal dissection (ESD) bleeding risk, the risk factor for post-ESD bleeding in DOAC users was high anticoagulant ability (odds ratio [OR]:6.056; 95% confidence interval [CI]: 1.094–33.529)33. However, no studies have investigated whether DOAC users receiving EMR with high anticoagulant activity, as analyzed by pharmacological parameters, have an increased risk of bleeding. To assess the rate of post-EMR bleeding and evaluate the association between pharmacological parameters and post-EMR bleeding, we conducted a prospective multicenter study on DOAC users with colorectal tumors. Methods Study design and patients To investigate whether the pharmacodynamic parameters of post-procedure bleeding were similar among endoscopic procedures, we conducted the study in the same way as the study design used in our previous clinical trials investigating patients who underwent ESD for early-stage gastrointestinal tumors33. This prospective multicenter trial was conducted across 14 Japanese institutions between April 2018 and May 2022. We enrolled 114 patients taking a direct factor Xa inhibitor (apixaban, edoxaban, or rivaroxaban) who were scheduled to undergo hot EMR for colorectal neoplasms. The inclusion criteria were DOAC users aged ≥ 20 years with colorectal neoplasm and receiving hot EMR using hot-snare polypectomy with an electrosurgical generator instead of cold-snare polypectomy34. The exclusion criteria were prescription of > 3 kinds of antithrombotic drugs, other significant clinical illnesses, resection by cold-snare polypectomy, endoscopically visible polyps after EMR, pathological residual polyps by EMR, and follow-up duration of < 28 days after EMR (Supplementary Fig. 1). Blood samples for determining plasma levels of DOAC and FXa activity were drawn and placed in 3.2% citrated tubes 2 h after the patient had taken a DOAC on the day before treatment as Tmax sampling, based on previous reports (apixaban and rivaroxaban, 2 h; and edoxaban, 1–2 h), and on the morning of EMR as trough sampling35,36. The study protocol conformed to the ethical guidelines of the Declaration of Helsinki and was approved by the institutional review board of each institution (the institutional review board of Tokyo Medical University: approval number, T2020-0025). Written informed consent was obtained from all patients. The registry number for this clinical trial is UMIN000042576 in the University Hospital Medical Information Network. Hot EMR for colorectal neoplasm lesions All patients underwent hot EMR using hot-snare polypectomy with an electrosurgical generator instead of cold-snare polypectomy34. Several confounding factors, such as endoscopic devices and endoscopist skill levels, may have been involved in this study. To avoid this variation, all physicians were endoscopists, as recognized by the Japan Gastroenterological Endoscopy Society. The EMR methods, type of endoscope, snare used as the cutting device, and electrosurgical generator used during endoscopic treatment were not specified as protocols, and EMR was performed as normally practiced in each hospital. Clips were used at the discretion of the endoscopists after EMR was performed. Lifestyle guidance after the EMR was provided according to the protocols of each institution. In patients taking antithrombotic drugs, EMR was performed according to the updated guidelines18. Patients receiving DOAC continued to receive DOAC orally until the day before EMR and discontinued DOAC on the morning of the procedure. On the day following the procedure, DOACs were restarted if bleeding was not clinically present. Definition of post-EMR delayed bleeding Post-procedural bleeding can be divided into intraoperative and delayed bleeding; however, this study only included delayed bleeding and not intraoperative bleeding. Post-EMR delayed bleeding was defined as bleeding with clinical signs (bloody stool or a decrease in hemoglobin > 2 g/dL since the most recent laboratory test) and confirmed via emergency endoscopy within 28 days post-EMR. The definition of post-EMR bleeding was established with reference to the guidelines for colorectal ESD and endoscopic mucosal resection37. The investigators who certified the occurrence of post-EMR bleeding were blinded to plasma FXa concentrations and activity at the time of endoscopy. Measurement of plasma concentration and FXa activity and genotyping Plasma samples were separated by centrifugation at 3000 × g for 15 min at 4 °C and stored at -80 °C until analysis. Plasma concentrations and intrinsic FXa activity were measured as previously reported, with a few modifications32,33,38,39. Using a commercially available kit (DNA Extract All Reagents Kit; Applied Biosystems, Waltham, MA, USA), DNA was extracted from leukocytes of each patient. The CYP3A5 6986 A > G (rs776746; *3), ABCG2 421 C > A (rs2231142) and ABCB1 1236 C > T (rs1128503), 2677G > T/A (rs2032582), 3435 C > T (rs1045642) polymorphisms were determined using a real-time PCR system (StepOnePlus™; Applied Biosystems)32. Sample size calculation In a retrospective study, the post-EMR bleeding rate among DOAC users was 11%40. As no studies have confirmed the relationship between post-EMR bleeding and pharmacological parameters, we set the rate of post-EMR bleeding in patients in the top 33% of plasma DOAC concentrations at 22%, and the rate of post-EMR bleeding in patients in the bottom 33% at 1.0% (rate of post-EMR bleeding in general patients)10. In addition, we assumed a sample power and α error of 80% and 0.05, respectively. This indicated the registration of 108 patients; assuming a 5% dropout rate, 114 patients were considered for enrollment. Statistical analysis Values, including age, pharmacological parameters, and renal function, are presented as mean ± standard deviation. Statistically significant differences in the mean values were determined using t-tests. Significant differences in the mean pharmacological parameters were determined using one-way analysis of variance. Categorical variables are shown as n (%), and significant differences were determined using the χ2 test. Multivariate logistic regression analyses were used to test the associations between the candidate variables and post-EMR bleeding. Furthermore, because clip suturing after EMR41 and polyp location (i.e., right-sided location)42 were considered confounding factors for post-EMR bleeding, multivariate analysis adjusted for both factors was performed and adjusted for post-EMR bleeding risk43. Although continuous and discontinuous variables were mixed, for continuous variables, the change in units indicated the OR, whereas the OR for categorical variables indicated the OR compared with the reference level. We determined the cutoff values for high anticoagulant ability with high post-EMR bleeding risk using FXa activity at both the trough and Tmax, as previously reported, with a few modifications33. All p-values were two-sided, and p < 0.05 was considered statistically significant. The calculations were performed using SPSS version 28 (IBM Inc., Armonk, NY, USA). The anticoagulant effect of DOACs is sometimes assessed based on the plasma level and FXa activity at the trough, Tmax, and the area under the curve. In this study, by creating a composite variable that combines FXa activity at trough and Tmax to comprehensively and objectively evaluate pharmacological parameters, we established a cutoff value for potential bleeding risk using a receiver operating characteristic (ROC) curve. The objective cutoff value for bleeding risk is determined as the optimal point where “sensitivity plus specificity” is maximum to establish the cutoff value and “sensitivity + specificity − 1” is the Youden’s index. In this study, patients with composite variables higher than the Youden’s index were defined as the high-risk group, whereas those with lower values were defined as the low-risk group. Results We prospectively recruited 114 patients with colorectal neoplasms resected using EMR (Supplementary Fig. 1) and analyzed the data of 105 patients after excluding 9 patients according to the exclusion criteria. Concomitant antiplatelet drugs were used by 19.0% of the patients (Table 1). Among DOACs, 28.6% used apixaban (mean dose, 8.3 ± 1.6 mg/day), 41.9% used edoxaban (38.1 ± 15.7 mg/day) and 29.5% used rivaroxaban (12.9 ± 2.5 mg/day). None of the patients required a blood transfusion or surgical treatment after EMR because of immediate intraoperative bleeding. The incidence rates of post-EMR delayed bleeding, perforation, and thromboembolism were 10.5% (11/105, 95% CI: 5.3%–18.0%), 0%, and 0%, respectively. All patients were treated endoscopically, and none required surgical treatment. The median time to delayed bleeding after EMR was 2 (1–5) days. Table 1Characteristics of DOAC users who undergo EMR for colorectal neoplasm lesions.Total patients(n = 105)Non-post-EMRbleeding(n = 94)Post-EMR bleeding(n = 11)p-valueDemographicAge (years), mean ± SD73.7 ± 7.974.3 ± 7.768.4 ± 8.80.025Male, n (%)80 (76.2%)69 (73.4%)11 (100.0%)0.061Body weight (kg), mean ± SD62.9 ± 11.262.5 ± 11.367.1 ± 9.80.216Alcohol, n (%)48 (45.7%)39 (41.5%)9 (81.8%)0.021Smoking, n (%)23 (21.9%)17 (18.1%)6 (54.5%)0.012ComorbiditiesNVAF, n (%)97 (92.4%)86 (91.5%)11 (100.0%)0.593Heart failure, n (%)18 (17.1%)17 (8.1%)1 (9.1%)0.686Hypertension66 (62.9%)59 (62.8%)7 (63.6%)0.921Diabetes, n (%)20 (19.0%)20 (21.3%)0 (0.0%)0.120Hemodialysis, n (%)0 (0.0%)0 (0.0%)0 (0.0%)-Antiplatelet drugs, n (%)20 (19.0%)19 (20.2%)1 (9.1%)0.374Aspirin, n (%)11 (10.5%)10 (10.6%)1 (9.1%)0.883Thienopyridine, n (%)2 (1.9%)2 (2.1%)0 (0.0%)0.627Others n (%)7 (6.7%)7 (7.4%)0 (0.0%)0.352Anticoagulant drug, n (%)Apixaban, n (%)30 (28.6%)28 (29.8%)2 (18.2%)0.725Edoxaban, n (%)44 (41.9%)41 (43.6%)3 (27.3%)0.299Rivaroxaban, n (%)31 (29.5%)25 (26.6%)6 (54.5%)0.051Anticoagulant drug, dosing dose a dayApixaban, mean ± SD4.50 ± 2.224.31 ± 1.9910.00 ± 0.00< 0.001Edoxaban, mean ± SD37.44 ± 16.2636.79 ± 15.6150.00 ± 17.320.166Rivaroxaban, mean ± SD13.87 ± 4.9513.80 ± 5.4514.17 ± 2.040.874LaboratoryHemoglobin, mean ± SD13.2 ± 2.113.0 ± 2.115.2 ± 2.00.002Creatinine, mean ± SD0.97 ± 0.380.98 ± 0.400.95 ± 0.130.829eGFR, mean ± SD60.0 ± 17.759.8 ± 18.462.6 ± 10.20.633PT-INR, mean ± SD1.17 ± 0.301.18 ± 0.321.12 ± 0.170.574APTT, mean ± SD35.2 ± 24.835.1 ± 26.036.8 ± 7.80.840LesionNumbers of tumors, mean ± SD2.72 ± 1.992.67 ± 1.923.20 ± 2.620.422Localization, n (Ce/A/T/D/S/R)13/68/71/21/80/3413/66/70/21/76/300/2/1/0/4/40.120Stalk type, n (%)22 (21.0%)19 (20.2%)3 (27.3%)0.694Maximum tumor size (mm), mean ± SD12.11 ± 6.4012.43 ± 6.569.36 ± 4.080.133Clipping, n (%)91 (86.7%)82 (87.2%)9 (81.8%)0.653Adenoma, n (%)91 (86.7%)81 (86.2%)10 (90.9%)0.611Colorectal cancer, n (%)13 (12.4%)12 (12.8%)1 (9.1%)0.735Invasion to SM2, n (%)2 (1.9%)2 (2.1%)0 (0.0%)0.627ProcedureProcedure time > 60 min, n (%)20 (19.0%)19 (20.2%)1 (9.1%)0.381Bleeding, n (%)11Perforation, n (%)0 (0.0%)0 (0.0%)0 (0.0%)-Thromboembolism, n (%)0 (0.0%)0 (0.0%)0 (0.0%)-p-value: statistical difference between bleeding and non-bleeding patients.Abbreviations: A, ascending colon; APTT, activated partial thromboplastin time; Ce, cecum; D, descending colon; DOAC, direct oral anticoagulant; eGFR, estimated glomerular filtration rate; EMR, endoscopic mucosal resection; NVAF, non-valvular atrial fibrillation; PT-INR, international normalized ratio of prothrombin time; R, rectum; S, sigmoid colon; SD, standard deviation; T, transverse colon. Characteristics of DOAC users who undergo EMR for colorectal neoplasm lesions. p-value: statistical difference between bleeding and non-bleeding patients. Abbreviations: A, ascending colon; APTT, activated partial thromboplastin time; Ce, cecum; D, descending colon; DOAC, direct oral anticoagulant; eGFR, estimated glomerular filtration rate; EMR, endoscopic mucosal resection; NVAF, non-valvular atrial fibrillation; PT-INR, international normalized ratio of prothrombin time; R, rectum; S, sigmoid colon; SD, standard deviation; T, transverse colon. The incidence of post-EMR delayed bleeding was significantly higher in younger patients (p = 0.025) and in those reporting alcohol consumption (p = 0.021), who were smokers (p = 0.012), and who had higher hemoglobin levels (p = 0.002) than in patients without bleeding (Table 1). The rates of delayed bleeding in patients taking apixaban, edoxaban, and rivaroxaban were 6.7%, 6.8%, and 19.4%, respectively; however, there was no significant difference in the rate of delayed bleeding (p = 0.157). Pharmacological parameters: plasma level and FXa activity Plasma level at the trough and Tmax showed significant differences among apixaban (137.3 ± 74.7 ng/mL and 255.4 ± 132.5 ng/mL, p < 0.001), edoxaban (38.5 ± 54.1 ng/mL and 153.6 ± 88.0 ng/mL, p < 0.001) and rivaroxaban (41.7 ± 55.5 ng/mL and 264.6 ± 141.9 ng/mL, respectively; p < 0.001) (Table 2). Because the plasma levels of DOACs depend on body weight and dosage, when examined using the plasma concentration/dose (C/D) ratio adjusted for both factors, the C/D ratios at the trough and Tmax differed significantly among these agents. In all patients, FXa activity was 79.8% ± 24.5% at trough and 42.5% ± 25.1% at Tmax (Table 2). Table 2Plasma concentration of DOAC users who undergo EMR.Total patients(n = 105)Non-Post-EMRbleeding(n = 94)Post-EMR bleeding(n = 11)p-valueAt troughApixaban, mean ± SD (ng/mL) (N = 30)137.3 ± 74.7144.8 ± 71.232.3 ± 42.10.037Edoxaban, mean ± SD (ng/mL) (N = 44)38.5 ± 54.140.4 ± 55.813.2 ± 11.40.408Rivaroxaban, mean ± SD (ng/mL) (N = 31)41.7 ± 55.532.0 ± 30.182.2 ± 108.30.045C/D ratio of apixaban, mean ± SD [(ng/mL)/(mg/kg/day)]1190.8 ± 738.91258.5 ± 714.8242.5 ± 333.30.059C/D ratio of edoxaban, mean ± SD [(ng/mL)/(mg/kg/day)]72.8 ± 121.776.9 ± 125.216.7 ± 12.90.415C/D ratio of rivaroxaban, mean ± SD [(ng/mL)/(mg/kg/day)]221.5 ± 284.4179.3 ± 186.5397.3 ± 521.80.092At TmaxApixaban, mean ± SD (ng/mL) (N = 21)255.4 ± 132.5255.4 ± 132.5NANAEdoxaban, mean ± SD (ng/mL) (N = 28)153.6 ± 88.0153.6 ± 88.0NANARivaroxaban, mean ± SD (ng/mL) (N = 17)264.6 ± 141.9251.7 ± 159.3295.7 ± 95.50.577C/D ratio of apixaban, mean ± SD [(ng/mL)/(mg/kg/day)]2195.4 ± 1171.92195.4 ± 1171.9NANAC/D ratio of edoxaban, mean ± SD [(ng/mL)/(mg/kg/day)]268.8 ± 147.8268.8 ± 147.8NANAC/D ratio of rivaroxaban, mean ± SD [(ng/mL)/(mg/kg/day)]1333.2 ± 771.71310.2 ± 885.01388.8 ± 472.00.856p-value: statistical difference between bleeding and non-bleeding patients.C/D, plasma concentration/dose; DOAC, direct oral anticoagulant; EMR, endoscopic mucosal resection; NA, not applicable; SD, standard deviation; Tmax, time to maximum plasma concentration. Plasma concentration of DOAC users who undergo EMR. p-value: statistical difference between bleeding and non-bleeding patients. C/D, plasma concentration/dose; DOAC, direct oral anticoagulant; EMR, endoscopic mucosal resection; NA, not applicable; SD, standard deviation; Tmax, time to maximum plasma concentration. There was no significant correlation between the plasma level at trough and Tmax (Fig. 1A); however, a significant correlation was observed for the C/D ratio (correlation coefficient: 0.628) (Fig. 1B). Both the plasma levels and C/D ratios varied widely depending on the type of DOAC used. Significant correlations between plasma levels and FXa activity for apixaban, edoxaban, and rivaroxaban were observed (Fig. 2). As shown in Fig. 3, no significant correlation occurred between FXa activity at trough and Tmax. Association of plasma DOAC level at the trough with time to maximum plasma concentration (A) and of plasma concentration/dose ratio at the trough with plasma concentration/dose ratio at maximum plasma concentration (B). Abbreviations: DOAC, direct oral anticoagulant; C/D, plasma concentration/dose; Tmax, time to maximum plasma concentration. Correlation of plasma DOAC level with FXa activity for apixaban (A), edoxaban (B) and rivaroxaban (C). Abbreviations: DOAC, direct oral anticoagulant; EMR, endoscopic mucosal resection. Association of FXa activity at the trough with FXa activity at maximum plasma concentration in patients with and without post-EMR bleeding. Abbreviations: DOAC, direct oral anticoagulant; EMR, endoscopic mucosal resection. Association with Pharmacological parameters and post-EMR bleeding At trough, plasma levels were higher in rivaroxaban-treated patients with post-EMR delayed bleeding than in patients without bleeding (32.0 ± 30.1 vs. 82.2 ± 108.3, p = 0.045). However, other plasma levels and C/D ratios at both trough and Tmax were similar between patients with and without post-EMR delayed bleeding (Table 2). The FXa activity at the Tmax in patients with post-bleeding, when including all patients, regardless of type of DOAC, was 17.2% ± 7.5%, and thus significantly lower than the 44.6% ± 25.0% in those without post-bleeding (p = 0.018). The C/D ratio of FXa activity was low (201.1 ± 199.3 vs. 82.8 ± 42.7, p = 0.002) (Table 3). The rate of post-bleeding in patients with FXa activity at both trough < 98% and Tmax < 25% was 26.7% (4/15), which was higher than in the other categories (Fig. 3). Table 3Factor Xa activity DOAC users who undergo EMR.Total patients(n = 105)Non-Post-EMRbleeding(n = 94)Post-EMR bleeding(n = 11)p-valueAt troughApixaban, mean ± SD (%) (N = 30)58.7 ± 20.556.5 ± 19.289.1 ± 16.90.028Edoxaban, mean ± SD (%) (N = 44)90.2 ± 18.889.7 ± 19.396.9 ± 5.90.530Rivaroxaban, mean ± SD (%) (N = 31)85.4 ± 23.289.2 ± 20.369.6 ± 29.70.061All patients, mean ± SD (%) (N = 105)79.8 ± 24.579.7 ± 24.680.5 ± 25.40.915C/D ratio of DOACs, mean ± SD (%)334.7 ± 224.0353.7 ± 232.3411.8 ± 110.60.547At TmaxApixaban, mean ± SD (%) (N = 21)40.6 ± 22.540.6 ± 22.5NANAEdoxaban, mean ± SD (%) (N = 28)52.6 ± 21.952.6 ± 21.9NANARivaroxaban, mean ± SD (%) (N = 17)28.2 ± 26.932.8 ± 30.917.2 ± 7.50.290All patients, mean ± SD (%) (N = 66)42.5 ± 25.144.6 ± 25.017.2 ± 7.50.018C/D ratio of DOACs, mean ± SD (%)190.9 ± 193.6201.1 ± 199.382.8 ± 42.70.002p-value: statistical difference between bleeding and non-bleeding patients.C/D, plasma concentration/dose; DOAC, direct oral anticoagulant; EMR, endoscopic mucosal resection; NA, not applicable; SD, standard deviation; Tmax, time to maximum plasma concentration. Factor Xa activity DOAC users who undergo EMR. p-value: statistical difference between bleeding and non-bleeding patients. C/D, plasma concentration/dose; DOAC, direct oral anticoagulant; EMR, endoscopic mucosal resection; NA, not applicable; SD, standard deviation; Tmax, time to maximum plasma concentration. Polymorphisms in DOAC-metabolizing enzymes in patients taking DOAC were similar between those with and without post-bleeding (Table 4). Table 4Polymorphism of DOAC-metabolizing enzymes in patients taking DOAC who undergo EMR.Total patients(n = 105)Non-Post-EMRbleeding(n = 94)Post-EMR bleeding(n = 11)p-valueABCB1 1236 C > TC/C, n (%)14 (13.3%)11 (11.7%)3 (27.3%)0.355C/T, n (%)47 (44.8%)43 (45.7%)4 (36.4%)T/T, n (%)44 (41.9%)40 (42.6%)4 (36.4%)ABCB1 2677G > T/AG/G, n (%)22 (21.0%)20 (21.3%)2 (18.2%)0.534G/T, n (%)47 (44.8%)43 (45.7%)4 (36.4%)G/A, n (%)11 (10.5%)10 (10.6%)1 (9.1%)A/A, n (%)2 (1.9%)1 (1.1%)1 (9.1%)T/A, n (%)11 (10.5%)10 (10.6%)1 (9.1%)T/T, n (%)12 (11.4%)10 (10.6%)2 (18.2%)ABCB1 3435 C > TC/C, n (%)29 (27.6%)24 (25.5%)5 (45.5%)0.505C/T, n (%)58 (55.2%)54 (57.4%)4 (36.4%)T/T, n (%)18 (17.1%)16 (17.0%)2 (18.2%)ABCG2 421 C > AC/C, n (%)50 (47.6%)46 (48.9%)4 (36.4%)0.330C/A, n (%)46 (43.8%)40 (42.6%)6 (54.5%)A/A, n (%)9 (8.6%)8 (8.5%)1 (9.1%)CYP3A5 6986 A > GA/A, n (%)11 (10.5%)11 (11.7%)0 (0.0%)0.427A/G, n (%)36 (34.3%)31 (33.0%)5 (45.5%)G/G, n (%)58 (55.2%)50 (53.2%)6 (54.5%)Abbreviations: DOAC, direct oral anticoagulant; EMR, endoscopic mucosal resection. Polymorphism of DOAC-metabolizing enzymes in patients taking DOAC who undergo EMR. Abbreviations: DOAC, direct oral anticoagulant; EMR, endoscopic mucosal resection. Predictive factors for post-EMR bleeding: multivariate analysis The bleeding risk related to anticoagulant ability was decided for patients with Youden index [the point where (sensitivity + specificity − 1) is at its maximum value in the ROC curve] > 0.161 by the ROC analysis using FXa activity at both trough and Tmax as “High post-EMR bleeding risk” (Supplementary Fig. 2). We first performed univariate analysis using both pharmacological and major risk factors reported as common post-EMR bleeding. In the univariate analysis, the predictive factors were younger age (OR, 0.919; 95% CI, 0.851–-0.991), current alcohol consumption (6.346, 1.300–31.001), current smoking (5.435, 1.485–19.901), and high bleeding risk with high anticoagulant ability (11.250, 1.169–108.280). Because there were 11 events, 1 or 2 explanatory variables were desirable in the multivariate analysis. However, as the four extracted candidate risk factors were all recognized as risks for bleeding based on existing reports and high risk by FXa activity was an important pharmacological factor, predictive factors with a p-value < 0.05 on multivariate analysis and potential confounding factors (clip suturing after EMR and polyp location) were included in the multivariate analysis. The only risk factor was high post-EMR bleeding risk with high anticoagulant ability (adjusted OR; 11.970, 95% CI; 1.140-125.675, p = 0.038) (Table 5). Table 5Univariate and multivariate logistic regression analysis of predictive factors for EMR-associated bleeding at the time of treatment in DOAC users.Univariate analysisMultivariate analysisOR95% CIp-valueOR95% CIp-valueAdjusted OR95% CIp-valueAge0.9190.851–0.9910.0290.9000.779–1.0410.1580.9180.818–1.0310.149Alcoholcurrent drinker (vs. previous or non-drinker)6.3461.300-31.0010.0226.5680.383–109.30.1966.4820.485–82.0100.185Smokingcurrent smoker (vs. previous or non-smoker)5.4351.485–19.9010.0111.7820.160–19.830.6385.8700.663-52.0000.111BMI> 25 kg/m2 (vs. < 25 kg/m2)0.8000.198–3.231Heart failureWith heart failure (vs. without)0.4880.058–4.081HypertensionWith hypertension (vs. without)1.0380.284–3.801AntiplateletsUser (vs. non-user)0.3950.048–3.277Number of tumors1.0710.804–1.427Tumor size0.9130.806–1.034Tumor locationRight side* (vs. left side)3.7500.746–18.840Stalk typeWith stalk type (vs. without)1.5830.382–6.569HistologyCancer (vs. without)0.7550.088–6.476ClipUse (vs. non-use)0.3570.067–1.918ABCB1 1236 C > TT allele carrier (vs. non carrier)0.3530.081–1.535ABCB1 3435 C > TT allele carrier (vs. non carrier)0.4110.115–1.471ABCG2 421 C > AA allele carrier (vs. non carrier)1.6770.460–6.113CYP3A5 6986 A > GG carrier (vs. non carrier)1.0320.294–3.618Factor Xa activity**High post-EMR bleeding risk** (vs. low risk)11.2501.169–108.2800.03612.1611.026–144.10.04712.2111.081-137.9220.043Because EMR clip suturing (clipping for all polyps resected) and polyp location (right side) were considered confounding factors for post-EMR bleeding, multivariate analysis adjusted by both factors were performed and adjusted.* The cecum, ascending and transverse colon were defined as the right side, while the descending, sigmoid colon and rectum as the left side.**High post-EMR bleeding risk: This criterion of post-EMR bleeding risk was determined by the Youden index (> 0.161) analyzed by receiver operating characteristic analysis using Factor Xa activity at both the trough and Tmax.Abbreviations: APTT, activated partial thromboplastin time; BMI, body mass index; eGFR, estimated glomerular filtration rate; DOAC, direct oral anticoagulant; EMR, endoscopic mucosal resection; CI, confidence interval; OR, odds ratio; PT-INR, prothrombin time-international normalized ratio; Tmax, time to maximum plasma concentration. Univariate and multivariate logistic regression analysis of predictive factors for EMR-associated bleeding at the time of treatment in DOAC users. Because EMR clip suturing (clipping for all polyps resected) and polyp location (right side) were considered confounding factors for post-EMR bleeding, multivariate analysis adjusted by both factors were performed and adjusted. * The cecum, ascending and transverse colon were defined as the right side, while the descending, sigmoid colon and rectum as the left side. **High post-EMR bleeding risk: This criterion of post-EMR bleeding risk was determined by the Youden index (> 0.161) analyzed by receiver operating characteristic analysis using Factor Xa activity at both the trough and Tmax. Abbreviations: APTT, activated partial thromboplastin time; BMI, body mass index; eGFR, estimated glomerular filtration rate; DOAC, direct oral anticoagulant; EMR, endoscopic mucosal resection; CI, confidence interval; OR, odds ratio; PT-INR, prothrombin time-international normalized ratio; Tmax, time to maximum plasma concentration. Discussion Although we previously demonstrated a significant association between high anticoagulant ability and post-ESD bleeding in DOAC users33, no studies have investigated whether DOAC users receiving EMR with high anticoagulant activity, as analyzed by pharmacologic parameters, have an increased risk of bleeding. This study, which focused on the association between post-EMR bleeding and the pharmacodynamic parameters of DOAC, was performed using a prospective multicenter design. Although the rate of bleeding depends on the definition of bleeding, the definition of post-EMR bleeding in the current study was established with reference to the clinical guidelines for colorectal ESD/EMR38, and the results showed a post-EMR bleeding rate of 10.5% in DOAC users. This is higher than the 0.6%–3.2% of post-EMR bleeding in patients not taking antithrombotic drugs7–10, and the incidence of post-EMR bleeding in this study was comparable to the incidence of post-ESD bleeding for colorectal tumors in DOAC users in a Japanese multicenter collaborative study (9.32% [95/1019])44. In the present study, we observed wide variations among individuals in plasma DOAC levels, C/D ratios, and FXa activity at both the trough and Tmax. Furthermore, multivariate analysis revealed that the risk factor for bleeding was higher pharmacological anticoagulant efficacy. Post-EMR bleeding in DOAC users Understanding risk factors before treatment is important for managing bleeding complications. Tumor factors include lesion size, number of lesions, and lesion morphology11,15. Treatment-related factors include poor bowel preparation, use of incisional waves, and resection before energization14,15. Multivariate analysis of patient-related factors identified systemic condition, cardiac disease, chronic renal failure, and anticoagulant medication as risk factors, of which anticoagulant medication was still one of the major risk factors (OR: 3.71, 95% CI: 1.05–13.05)11. For EMR in patients receiving anticoagulants, the risk of bleeding must be balanced with that of thromboembolism from DOAC withdrawal. The Japanese guidelines recommend that DOAC be discontinued during high-risk endoscopic treatment for bleeding only on the morning of treatment and resumed the following morning if no bleeding occurs18. The updated version of the Korean guideline recommends withholding DOACs for > 48 h before the procedure with high-risk and ultra-high-risk bleeding and resuming them within 2–3 days after adequate hemostasis has been achieved post-procedure21. The United States guidelines published in 202223 recommend temporarily interruption of DOACs use (approximately 24–48 h), rather than continuation, in DOAC users undergoing procedures with a high risk of bleeding. Thus, guidelines for the management of DOACs differ by country. These differences highlight the importance of examining the risk of bleeding and thromboembolism using pharmacologically objective measures of DOAC to establish treatment guidelines that can be used worldwide. Our prospective study revealed a post-EMR bleeding rate of 10.5%. In addition, given the finding that patients taking DOAC bleed 5–10 times more frequently than non-users, even when guidelines are followed, such patients require an adjustment in patient education/perioperative management in addition to post-treatment bleeding control measures. Management based on Pharmacological characteristics of DOACs The next focus related to the factors associated with post-treatment bleeding in DOAC users. The identification of factors that further increase the risk among DOAC users and those who are already at high risk could be an effective tool for bleeding risk management in these complicated cases. In this study, the rate of post-EMR delayed bleeding in patients with FXa activity at both trough < 98% and Tmax < 25% was 26.7%, which was higher than that in the other groups (0.0%–2.9%). Patients with strong anticoagulant activity are at risk of post-EMR bleeding. Previous studies have shown that the plasma DOAC levels are associated with the risk of major intracerebral and gastrointestinal bleeding27,45,46. In this study, among the pharmacological parameters of DOAC, FXa activity at Tmax may be more important in EMR than FXa activity at the trough. The pathophysiology of bleeding is mainly ulcerative bleeding caused by endoscopic treatment. Therefore, the pathophysiology of major bleeding complications may differ in patients taking DOACs. Although the interval from EMR to post-bleeding is commonly within 14 days47, the median time in patients receiving anticoagulants was 2.0 days (1–9 days)9, as observed in the current study (2.0 [1–5] days), indicating a shorter time to bleeding than in non-DOAC users. The finding that FXa activity at the trough had already returned to baseline and that patients restarted DOAC on the day following EMR suggests that FXa activity at Tmax may be associated with bleeding. FXa activity at Tmax may be important in patients with injury-related bleeding, whereas FXa activity at the trough may be important in patients without injury (e.g., intracerebral bleeding). Because the plasma level of DOACs is related to FXa activity, it is possible that the difference in the number of doses administered (apixaban: od, and edoxaban and rivaroxaban: bid) may have influenced the results. In this study, we evaluated individuals at high risk of bleeding using only trough or Tmax as indicators, but no significant results were obtained to determine the cutoff value. Therefore, by combining both to define a high-risk bleeding group, it is possible to demonstrate that FXa activity is a significant risk factor for post-EMR bleeding. Determining the period of preoperative drug withdrawal and the timing of postoperative resumption to balance drug levels and post-endoscopic resection ulcer healing is important. In this study, when EMR was performed in accordance with Japanese guidelines, the incidence of thromboembolism was 0%. The AGSE guidelines state that if DOACs cannot be resumed within 24 h post-procedure, thromboprophylaxis should be considered to reduce the risk of thromboembolism in patients at high risk for thromboembolism23. Therefore, when EMR is performed in patients taking DOAC, the timing of drug resumption should be considered to minimize the risk of bleeding without increasing the risk of thrombosis. At a minimum, patients with high DOAC levels and low FⅩa activity at Tmax may need to postpone postoperative DOAC. However, if DOAC levels are low and FXa activity is high, DOAC may need to be restarted as soon as possible, or may not require withdrawal in the first place. To test this hypothesis, it is necessary to investigate the efficacy of tailored treatment methods that control the timing of resuming DOAC according to the anticoagulation ability of FXa. Recently, the benefit of cold-snare polypectomy was reported to be the prevention of post-procedure bleeding, and compared with hot-snare polypectomy, cold-snare polypectomy may have a lower rate of bleeding complication48. This study aimed to clarify the factors that cause post-EMR bleeding after hot polypectomy in DOAC users; however, it would be beneficial to investigate the association between post-cold-snare polypectomy bleeding and pharmacological parameters. Limitations This study has certain limitations. First, we did not include patients who did not receive DOAC. Second, we followed the Japanese guidelines for DOAC withdrawal. Whether the rate of post-EMR bleeding in DOAC users increases or decreases when guidelines in other regions are followed remains unclear. Third, although the bleeding rate may vary depending on the type of DOAC used, the sample size was calculated by determining the association between FXa activity and post-EMR bleeding, irrespective of the type of DOAC used, and the influence of DOACs on bleeding could not be calculated. Conclusion To investigate whether the pharmacodynamic parameters for post-ESD and EMR delayed bleeding were similar between both post-endoscopic bleeding procedures, we conducted the study in the same way as the study design used in our previous clinical trials33. We demonstrate here, for the first time, that DOAC users with high anticoagulant effects using FXa activity have an increased risk of post-EMR bleeding and accordingly recommend the measurement of FⅩa activity before endoscopic treatment to predict post-EMR bleeding. We consider this to be a preliminary study and envisage enrolling numerous patients receiving DOAC for colorectal neoplasms; future prospective studies are planned to investigate the association between FXa activity and post-EMR bleeding, involving the cost-benefit of practicality or drug-specific risk assessment. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 2 Supplementary Material 3 Author contributions Conception and design: MM and MS; acquisition of data: MM, MS, SU, YN, MO, YF, TM, MS, NY, KI, YM, TT, ST, TY, TY, YT, NY, SO, YY, SO, SM, DH, TT, TK, MF, and TK; analysis and interpretation of data: MM and MS; drafting of the manuscript: MM and MS; critical revision of the manuscript: MM, MS, SU, YN, MO, YF, TM, MS, NY, KI, YM, TT, ST, TY, TY, YT, NY, SO, YY, SO, SM, DH, TT, TK, MF, and TK; statistical analysis: MS, MM, and SO; and study supervision: TT, TK, MF, and TK. All listed authors contributed substantially to the design, data collection, analysis, and editing of the manuscript. Funding This study was supported in part by a Grant-in-Aid from Scientific Research (C) (21K07949) and from the JFE (Japanese Foundation for Research and Promotion of Endoscopy) grant. Data availability Data based on the results of the current study were obtained and are accessible to the corresponding authors upon reasonable request. Declarations Competing interests The authors declare no competing interests. Ethics approval and consent to participate The study protocol conformed to the ethical guidelines of the Declaration of Helsinki and was approved by the Institutional Review Board of each institution (approved number in Tokyo Medical University: T2020-0025). Written informed consent was obtained from all patients. The registry number for this clinical trial was UMIN000042576 at the University Hospital Medical Information Network. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Material 1 Supplementary Material 2 Supplementary Material 3 Data Availability Statement Data based on the results of the current study were obtained and are accessible to the corresponding authors upon reasonable request.