Bleeding is a known predictor of poor outcome after a number of cardiac interventions. We investigated whether using the new bleeding definition of the Valve Academic Research Consortium predicts a poor outcome after transcatheter aortic valve implantation. In addition, we sought to identify those patient characteristics that predict periprocedural bleeding and investigated the effect of blood transfusion. This was a retrospective study of 101 patients undergoing transapical or transfemoral transcatheter aortic valve implantation at Kings College Hospital from August 2007 to November 2010. The association among bleeding, blood transfusion, and in-hospital and 6-month mortality was examined. Of the 101 patients, 5 (4.9%) had life-threatening periprocedural bleeding related to vascular or apical complications, 17 (17%) had major bleeding, and 79 (78%) had minor or no bleeding. The in-hospital mortality rate for the cohort was 9.9% (n = 10) and the 6-month mortality rate was 18.8% (n = 19). Those patients with life-threatening bleeding and those who were transfused had significantly greater in-hospital mortality rates than the patients without life-threatening bleeding or transfusion (60.0% vs 7.3%, p <0.05; and 14.8% vs 4.3%, p <0.05, respectively). Life-threatening bleeding, a decrease in hemoglobin >5 g/dl, and a blood transfusion of >2 U were associated with increased mortality at 6 months. In a logistic regression model, coexisting vascular disease, diabetes, and preprocedural anemia significantly affected the incidence of life-threatening or major bleeding. In conclusion, consistent with many other cardiac interventions, life-threatening periprocedural bleeding after transcatheter aortic valve implantation is associated with poorer outcomes.
The number of transcatheter aortic valve implantations (TAVIs) performed worldwide has increased dramatically in recent years, and this is likely to continue given the aging population and advancing technology. TAVI is commonly performed in patients with multiple co-morbidities for whom conventional aortic valve replacement is considered too high risk. Bleeding is a common and well-investigated complication after coronary interventions, and in cardiac surgery, it is a well-established predictor of in-hospital mortality. The predictors of bleeding after coronary intervention are well established. Scoring systems such as the Can Rapid risk stratification of Unstable angina patients Suppress ADverse outcomes with Early implementation of the ACC/AHA Guidelines (CRUSADE) score were developed to predict the risk of major bleeding after coronary intervention and are derived from multinational registries. These data have indicated that certain patient characteristics and procedural risk factors predict bleeding. However, to date, no data are available about the effect of bleeding and blood transfusion on TAVI outcomes. The high reported transfusion rate after the procedure, however, suggests that bleeding and pre- and postoperative anemia are important issues. We aimed to establish whether bleeding and blood transfusion after TAVI are associated with increased in-hospital and 6-month mortality. We also investigated the risk factors that predict periprocedural bleeding and blood transfusion in this clinical setting.
Methods
This was a retrospective study involving 101 patients undergoing transapical or transfemoral TAVI at King’s College Hospital from August 2007 to November 2010. Patient characteristics, preoperative investigations, postprocedural complications, and in-hospital and 6-month mortality were collated from a contemporaneous database. Information regarding admission hemoglobin levels, postprocedural hemoglobin nadirs, and whether packed red blood cells were transfused was collected from the patient notes.
We used the Valve Academic Research Consortium (VARC) definition of bleeding complications. This defines life-threatening bleeding as fatal bleeding or bleeding from an organ causing hypovolemic shock or severe hypotension requiring vasopressors or surgery or bleeding from an overt source with a decrease in hemoglobin of ≥5 g/dl or blood transfusion of ≥4 U. Major bleeding is classified as overt bleeding, either associated with a decrease in hemoglobin of ≥3.0 g/dl or that requiring transfusion of 2 or 3 U of blood without meeting “life-threatening” criteria. Minor bleeding is any bleeding of clinical mention (e.g., access site hematoma) that does not qualify as life-threatening or major. A correlation was made between the bleeding events and mortality. We also investigated how periprocedural blood transfusion affected these outcomes. The predictors of bleeding were investigated using univariate and multivariate analyses. The predictors included age (<85 years and ≥85 years), gender, body mass index (<25 and >25 kg/m 2 ), estimated glomerular filtration rate (<60 and ≥60 ml/min), logistic euroSCORE, diabetes mellitus, coexisting vascular disease (defined as previous coronary artery bypass grafting/percutaneous coronary intervention, peripheral vascular disease, significant stenosis in the carotid arteries or previous cerebrovascular accident), or a preprocedural hemoglobin (Hb) level <11.5 g/dl.
Statistical analyses were performed using the Statistical Package for Social Sciences, version 19 (SPSS, Chicago, Illinois). The data are expressed as the mean ± SEM. A comparison of the categorical variables was done using chi-square tests. A comparison of continuous variables between groups was done using an unpaired Student’s t test. Logistic regression analysis was used to assess the effect of multiple variables on periprocedural bleeding. For all statistical tests, a significance level of p <0.05 was used.
Results
A total of 101 patients (50 men, 49.5%) underwent TAVI during the specified period. Of the 101 procedures, 56 were performed transapically and 45 transfemorally. The baseline patient characteristics are listed in Table 1 . Using the VARC definitions, 5 patients (4.9%) had life-threatening periprocedural bleeding related to vascular or apical complications, 17 (17%) had major bleeding, 2 (2%) had minor bleeding, and 77 (76%) had no overt bleeding. In addition, 8 patients (7.8%) developed major vascular injuries such as femoral arteriovenous fistula or pseudoaneurysm, 4 (4.0%) developed late vascular complications such as groin hematoma, and 2 (2.0%) required immediate reintervention for bleeding complications.
| Variable | Value |
|---|---|
| Women | 51% |
| Age (years) | 83.1 ± 0.6 |
| Body mass index (kg/m 2 ) | 25.7 ± 0.6 |
| Estimated glomerular filtration rate (ml/min) | 60.4 ± 1.8 |
| Logistic euroSCORE | 22 ± 0.9 |
| Baseline hemoglobin (g/dl) | 12.1 ± 0.2 |
| Peripheral vascular disease | 32 (32%) |
| Hypertension | 70 (69%) |
| Diabetes mellitus | 22 (22%) |
| Previous cardiovascular disease | 66 (65%) |
The in-hospital mortality rate for the cohort was 9.9% (n = 10), and the 6-month mortality rate was 18.8% (n = 19). The most common cause of death was sepsis, occurring in 12 cases, and was most frequently relating to pneumonia, followed by multiorgan failure in 5 cases. Three of the group who died in-hospital had life-threatening bleeding and 1 had major bleeding. Of those who had died at 6 months, 3 had life-threatening bleeding and 2 major bleeding. Those patients with life-threatening bleeding had significantly greater in-hospital mortality rates than those without such bleeding (60% vs 7.3%, p <0.05). Major bleeding was not associated with significantly greater in-hospital mortality rates (6% vs 8%, p = NS), although it was associated with a significantly longer in-hospital stay (20 ± 4 vs 12 ± 1 days; p <0.05).
The Kaplan-Meier analysis demonstrated that life-threatening bleeding predicts a worse survival for ≤6 months compared to the “major bleeding” group and the “others” group ( Figure 1 ) . The relation between blood transfusion and survival and Hb decrease and survival is shown in Figure 2 . A decrease in Hb >5 g/dl and blood transfusion of >2 U were associated with increased mortality at 6 months ( Figure 2 ).
A total of 54 patients (53.4%) were transfused packed red blood cells; 7 (6.9%) of the patients were transfused ≥4 U, 12 (11%) were transfused 3 U, 26 (26%) were transfused 2 U, and 9 (9%) were transfused 1 U. The mean number of units transfused was 2.5 ± 0.2. Those patients who were transfused not only had a significantly greater in-hospital mortality rate (14.8% vs 4.3%; p <0.05) but also a significantly longer in-hospital stay (17 ± 2 vs 7 ± 1 days; p <0.05). Of the patients who were given ≥4 U, 2 had life-threatening bleeding, 4 had major bleeding, and 1 had no overt bleeding. Of the patients who were given 3 U, 9 had no overt bleeding, and 3 had major bleeding. A total of 13 patients (25%) received ≥2 U of blood without an overt source of bleeding. The indication for blood transfusion was a Hb decrease of <8 g/dl in the postoperative period. The Kaplan-Meier relation between a blood transfusion of >2 U and survival to 6 months is shown in Figure 2 . Of the 19 patients who were transfused >2 U of blood, 2 had life-threatening bleeding and 7 had major bleeding.
We investigated potential risk factors of major or life-threatening bleeding using logistic regression model, including female gender, logistic euroSCORE, diabetes, glomerular filtration rate, baseline Hb, and vascular disease. In this model, diabetes, baseline Hb, and coexisting vascular disease were predictors of life-threatening or major bleeding complications (logistic regression model, p <0.05). No difference was found between the mode of delivery and the occurrence of major bleeding ( Figure 3 ) .
