Incidence, Timing, and Causes of Late Bleeding After TAVR in an Asian Cohort

Background Data regarding the incidence, predictive factors, and clinical outcomes of post-transcatheter aortic valve replacement (TAVR) bleeding is limited in the Asian cohort. Objectives This study sought to assess the predictors and prognostic impact of post-TAVR late bleeding. Methods This study used the Japanese multicenter registry data to analyze 2,518 patients (mean age: 84.3 ± 5.2 years) who underwent TAVR. Late bleeding was defined as any postdischarge bleeding events after TAVR. Baseline characteristics, predictive factors, and clinical outcomes including death and rehospitalization were assessed in patients with and without late bleeding events. Results The cumulative incidence rate of all and major late bleeding and ischemic stroke were 7.4%, 5.2%, and 3.4%, respectively, 3 years after TAVR. The independent predictive factors of late bleeding were low platelet count, high score (≥4) on the clinical frailty scale, and a New York Heart Association functional class III/IV. The cumulative mortality rates up to 3 years were significantly higher in patients with late bleeding than in those without bleeding (P < 0.001). The multivariate Cox regression analysis revealed that late bleeding, included as a time-varying covariate in the model, was associated with an increased risk of mortality following TAVR (HR: 5.63; 95% CI: 4.28-7.41; P < 0.001). Conclusions Late bleeding after TAVR was not a rare complication, and it significantly increased long-term mortality. It should be carefully managed, especially when it is predictable in the high-risk cohort, and efforts should be taken to reduce bleeding complications even after a successful procedure.

W ith the development of clinical devices and evidences, transcatheter aortic valve replacement (TAVR) has been broadly indicated for patients with degenerative severe aortic stenosis. [1][2][3] Considering the widespread use of the TAVR procedure, late adverse clinical events need to be elucidated separately from the procedural complications. Postdischarge bleeding after percutaneous coronary intervention (PCI) is a significant clinical issue because it is found to be related to the increased risk of readmission and mortality during the followup period. [4][5][6] The TAVR cohort is generally composed of patients with specific features such as old age, high frailty, and multiple comorbidities. 7,8 The antithrombotic regimen after the procedure should be carefully decided according to the risk-benefit balance to avoid excessive bleeding and ischemic events, especially in the high-risk subsets of bleeding events. However, there are only a few studies describing late bleeding complications after TAVR. 9,10 Furthermore, no data exist for this investigation in the Asian cohort. Therefore, this study aimed to clarify the incidence, timing, cause, predictive factors, and clinical outcomes related to late bleeding events in patients who had undergone TAVR, using a Japanese multicenter data. Research Consortium (BARC) criteria were additionally adopted to define the severity of late bleeding events. 12  bleeding. The classification and regression tree analysis is an empirical and statistical method to create decision rules based on data rather than speculation and to create the risk stratification model. 13 The classification and regression tree analysis identified the optimal threshold of platelet count. The clinical frailty scale (CFS) reflects the semiquantitative assessment of a patient's frailty, and its cutoff in the present study was $4. Values above this threshold indicated frailty. 8  functional class III/IV, liver disease, and active cancer was significantly higher in the late bleeding group than in the no late bleeding group (all P < 0.05). The classification and regression tree analysis identified that the optimal platelet count threshold was 14.9 Â 10 4 /mL for predicting the risk of late bleeding events after TAVR. The existence of atrial fibrillation (AF)  was observed in 508 patients, while 15.9% of AF patients were not administered OACs (n ¼ 81 of 508).
The patients with AF, who were prescribed OACs, exhibited a trend toward a higher incidence of late bleeding compared with those who were not prescribed OACs (22.1% vs 16.7%; P ¼ 0.098). The prevalence of gastric acid-suppressive agents usage showed no significant differences between the 2 groups (P ¼ 0.42). In addition, the distributions of gastric acid-suppressive agents did not differ in patients with GI bleeding and no GI bleeding (P ¼ 0.38; data not shown). Procedural variables and complications are presented in Table 2. There were no significant differences in the hospital and intensive care unit stay, type of valve, approach route, and periprocedural complications, including any bleeding complications between the 2 groups.  Figure 1B, which include GI bleeding (40.7%); hemorrhagic stroke (25.7%); trauma (15.0%); cancer (7.2%); lung bleeding (2.4%); and bleeding from other organs including the eye, nose, and genitourinary system, as well as unknown causes (9.0%). The cumulative incidence of all, major, and minor late bleeding and ischemic stroke were 7.4%, 5.2%, 2.5%, and 3.4%, respectively, at 3 years (Central Illustration). The all and major bleeding risks were significantly higher than the risk of stroke events (log-rank test; P < 0.001). The incidences of late bleeding events were compared between the subgroup of patients with or without OAC and AF (Figures 2A and 2B). The rates of major bleeding were significantly higher in patients with OACs than in those without OACs, and in patients with AF than in those without AF (P ¼ 0.041 and P ¼ 0.042, respectively). The rates of ischemic stroke did not differ significantly between the OAC and no OAC groups ( Figure 2C) and between the AF and no AF groups ( Figure 2D). The incidence of bleeding complications and ischemic stroke events were also compared in patients with or without OACs and AF (Supplemental Figure 1). The major bleeding complications were significantly higher than those of ischemic stroke events in patients regardless of OAC use and AF (all P < 0.05, raw P value by log-rank test).
The distribution of antithrombotic therapy at the late bleeding event was significantly different from that of baseline therapy ( Figure 3A). The dynamic change of antithrombotic therapy from discharge to the date of the late bleeding event is shown in Figure 3B. The detailed information concerning the dynamic change of antithrombotic therapy is shown in Table 3.

DISCUSSION
In the current study, the incidences of late bleeding and major bleeding after TAVR were 7.4% and 5.2% in the Asian cohort, respectively. The significant predictive factors of late bleeding were low platelet count, CFS $4 (patients with frailty), and NYHA functional class III/IV. The occurrence of late bleeding after TAVR was independently associated with an increased risk of mortality following TAVR. In addition, the rates of major late bleeding were significantly higher than those of ischemic strokes. These  The incidences of all late bleeding events were higher than ischemic stroke events after transcatheter aortic valve replacement (TAVR) (log-rank test; P < 0.001).
The major bleeding events were also significantly higher than ischemic stroke events (log-rank test; P ¼ 0.001  These differences may explain the different results in the multivariate analysis predicting late bleeding after TAVR. Contrarily, patient age was not a risk factor for late bleeding in our TAVR cohort. [17][18][19][20][21] The patients' average age was approximately 85 years, and the majority of patients were octogenarians in this study. In such a specific cohort, age differences might be attenuated to predict the risk of bleeding. Additionally, the frailty status evaluated by CFS was a significant predictive factor of late bleeding. Elderly patients with frailty have fragile tissues and may fall while walking, thereby causing trauma. 22,23 We had previously revealed that a higher CFS grade was significantly related to poor clinical outcomes after TAVR. 8 The CFS assessment allows us to stratify not   Abbreviations as in Table 1.    Abbreviations as in Table 1.
KEY WORDS clinical outcome, late bleeding, transcatheter aortic valve replacement APPENDIX For a supplemental figure and tables, please see the online version of this paper.

PERSPECTIVES COMPETENCY IN MEDICAL KNOWLEDGE:
The cumulative incidence of all and major late bleeding after TAVR in the Asian cohort were similar to those of the Western cohort. In addition, the late bleeding events were significantly associated with an increased risk of mortality after TAVR.
TRANSLATIONAL OUTLOOK: Although it is challenging to reduce the risk of late bleeding after TAVR, further studies are required to establish the optimal management for improving individual patient care after TAVR.