Baseline anemia is associated with transfusions and increased risk of mortality in patients who underwent cardiac surgery and percutaneous coronary intervention. The impact of blood transfusions in anemic patients who underwent transcatheter aortic valve implantation (TAVI) remains unclear. All patients who underwent transfemoral TAVI at our institution were retrospectively included. We determined the effect of blood transfusions on short- and long-term mortality and its interaction with baseline hemoglobin levels and bleeding complications. Additionally, we evaluated baseline hemoglobin effect on mortality. A total of 332 patients were included. All patients (99%) except 2 (1%) met the definition for anemia. Of the 332 patients, 124 (37%) received a blood transfusion and 208 (63%) did not. Blood transfusions were associated with increased in-hospital (p <0.001), 30-day (p <0.001), and 1-year (28% vs 13%, p = 0.001) mortality; however, after a landmark analysis, the effect of blood transfusion on mortality was only seen within 30 days of the procedure (p = 0.001). The increased risk of mortality associated with blood transfusion was present after multivariate adjustment at 30 days (hazard ratio 3.59 [1.04 to 12.4]; p = 0.04). Major vascular complications were a correlate for short- and long-term death. In contrast, baseline hemoglobin level and bleeding complications were not independently associated with mortality. The p value for interaction was not significant between transfusion and hemoglobin level and for transfusion and bleeding complication and transfusion and major vascular complication. In conclusion, the presence of anemia in patients who underwent TAVI does not affect mortality. Transfusion is a correlate of all-cause mortality in this patient population and should be used with caution.
Baseline preprocedural anemia is often treated with transfusions and is associated with increased risk of mortality in patients who underwent cardiac surgery and percutaneous coronary intervention. The role of baseline anemia in patients who underwent transcatheter aortic valve implantation (TAVI) has been recently investigated. However, the relation between anemia, bleeding complications, and blood transfusion and its impact on short- and long-term mortality has not been thoroughly investigated. We, therefore, sought to determine the prevalence of anemia and the impact of blood transfusions on short- and long-term mortality.
Methods
From May 2007 to November 2013, 332 consecutive patients with severe symptomatic aortic stenosis who underwent TAVI through the transfemoral route were included in this retrospective study. All patients were evaluated according to protocol by our institution’s multidisciplinary heart team composed of interventional cardiologists, general cardiologists, and cardiac surgeons who determined the eligibility of each patient.
The pre-TAVI evaluation was performed with coronary angiography to detect significant coronary stenoses, contrast computed tomography with standard postprocessing reconstructions for aortoiliac sizing evaluation, and transthoracic echocardiography for detailed left ventricular outflow tract measurement, aortic valve peak velocity, mean gradient, aortic valve area, dimensionless index, left ventricular function, and estimation of systolic pulmonary pressures.
All procedures were performed under conscious sedation or general anesthesia provided by a cardiac anesthesiologist in our hybrid catheterization laboratory. Anesthesia type was chosen on a case-by-case basis. The femoral artery was accessed percutaneously or by surgical cutdown. In the first case, the arteriotomy site was preclosed with either a Prostar XL device (Abbott Vascular, Redwood City, CA) or 2 perclose ProGlide 6Fr suture devices (Abbott Vascular). After successful TAVI, antiplatelet therapy consisted of clopidogrel 75 mg and aspirin 81 mg for 3 to 6 months.
Demographic clinical imaging and follow-up data were prospectively collected and entered into a registry by an independent cardiologist from the MedStar Cardiovascular Research Network at MedStar Washington Hospital Center. “Anemia” was defined according to the criteria of the World Health Organization as a preoperative hemoglobin level <12 g dL in women and <13 g/dl in men. “Blood transfusion” was defined as any red cell product given after the procedure. Preoperative serum creatinine was used to calculate the estimated glomerular filtration rate. The population was divided into a transfusion group and a no transfusion group and subsequently subgroups according to the occurrence of a bleeding complication. A separate analysis was performed to explore the impact of baseline anemia on mortality. All definitions of clinical end points used (in-hospital death, 30-day and 1-year all-cause mortality, stroke, vascular complications, bleeding complications, and acute kidney injury) were in concordance with the definitions by the Valve Academic Research Consortium-2 (VARC-2).
The study complied with the principles of the Declaration of Helsinki regarding investigation in humans and was approved by the Institutional Review Board of the MedStar Washington Hospital Center (Washington, DC).
Statistical analyses were performed using SAS 9.2 (SAS Institute, Cary, NC). Continuous variables are expressed as mean ± SD for normally distributed variables. Categorical variables are expressed as percentages. Analyses of the differences between the 2 groups were performed using the chi-square or Fisher’s exact test for categorical variables. Cox proportional hazard analysis was performed to detect predictors of 30-day and 1-year all-cause mortality. Variables were selected on the basis of overall clinical relevance. Variables included were left ventricular ejection fraction, history of chronic obstructive lung disease, age, gender, Society of Thoracic Surgeons (STS) score, previous coronary artery bypass surgery, history of peripheral arterial disease, planned surgical closure, VARC-2 bleeding complication, VARC-2 major vascular complication, blood transfusion, and hemoglobin level. After univariate analysis, variables with a p value <0.1 were incorporated into the multivariate analysis. The results are presented as adjusted hazard ratios with their 95% confidence intervals and p values. Death-free survival rates were calculated using the Kaplan-Meier method. A 30-day landmark analysis was performed. The goal of this landmark analysis was to estimate in an unbiased way the time-to-event probability of death 30 days after the procedure and to estimate the effect of blood transfusion on mortality after the 30-day mark. The log-rank test was used to compare the differences in curves among groups. A p value of <0.05 was considered to be statistically significant.
Results
A total of 332 patients were included. All patients (99%) except 2 (1%) met the definition for anemia. Of the 332 patients, 124 (37%) received a blood transfusion and 208 (63%) did not. The patients’ baseline characteristics and pre- and postprocedural echocardiographic assessments are listed in Table 1 . The average age of the entire population was 83 ± 8 years. Patients who did not get transfused had a higher prevalence of diabetes mellitus (40% vs 29%, p = 0.04) and peripheral arterial disease (38% vs 20%, p = 0.001). In contrast, patients who received a blood transfusion had higher mean STS scores (10.3 ± 4.2 vs 8.9 ± 4.5, p = 0.008) and a significantly higher prevalence of chronic kidney disease (65% vs 46%, p = 0.001). No significant differences were found in echocardiographic parameters pre- and postprocedure ( Table 1 ).
| Variable | Transfusion | p value | |
|---|---|---|---|
| No (n=208) | Yes (n=124) | ||
| Age (years) | 82 ± 7.9 | 83± 8 | 0.50 |
| Men | 116 (55%) | 54 (43%) | 0.02 |
| African American | 23 (12%) | 18 (15%) | 0.47 |
| Mean Society of Thoracic Surgeons score | 8.9 ± 4.5 | 10.3 ± 4.2 | 0.008 |
| BMI (m 2 ) | 1.88± 0.2 | 1.81 ± 0.2 | 0.01 |
| Chronic obstructive lung disease | 70 (35%) | 31 (25%) | 0.07 |
| Hypertension | 183 (91%) | 120 (96%) | 0.06 |
| Diabetes mellitus | 80 (40%) | 36 (29%) | 0.04 |
| Hyperlipidemia | 166 (83%) | 94 (75%) | 0.07 |
| Peripheral arterial disease | 74 (38%) | 23(20%) | 0.001 |
| Prior coronary artery bypass surgery | 68 (34%) | 37 (30%) | 0.45 |
| Prior percutaneous coronary intervention | 65 (33%) | 34 (28%) | 0.36 |
| Atrial fibrillation/atrial flutter | 84 (42%) | 52 (41%) | 0.99 |
| Prior myocardial infarction | 35 (18%) | 21 (17%) | 0.95 |
| Chronic kidney disease | 91 (46%) | 79 (65%) | 0.001 |
| Hemodialysis | 6 (3%) | 2 (2%) | 0.71 |
| Pre-procedural hemoglobin (mg/dL) | 11.7 ± 1.46 | 11 ± 1.53 | <0.001 |
| Pre-procedural hematocrit (%) | 35.8 ± 4 | 33.6 ± 4 | <0.001 |
| Pre-procedural echocardiogram | |||
| Mean left ventricular ejection fraction (%) | 52 ± 13 | 53 ± 13 | 0.67 |
| Left ventricular ejection fraction <40% | 41 (20%) | 26 (21%) | 0.92 |
| Mean transvalvular gradient (mm Hg) | 46 ± 12 | 47 ± 12 | 0.60 |
| Peak transvalvular gradient (mm Hg) | 71.05 ± 15.57 | 72.98 ± 18.75 | 0.36 |
| Aortic valve area (cm 2 ) | 0.67 ± 0.1 | 0.66 ± 0.1 | 0.40 |
| Peak aortic velocity (m/s) | 4.32 ± 0.5 | 4.38 ± 0.5 | 0.33 |
| Mean pulmonary artery systolic pressure | 47 ± 17 | 48 ± 15 | 0.80 |
| Post-procedural echocardiogram | |||
| Mean left ventricular ejection fraction (%) | 55 ± 12 | 55 ± 12 | 0.53 |
| Left ventricular ejection fraction <40% | 33 (16%) | 19 (16%) | 0.97 |
| Mean transvalvular gradient (mmHg) | 11. 2± 4 | 12± 5 | 0.57 |
| Peak aortic velocity (m/s) | 2.2 ± 0.4 | 2.2 ± 0.4 | 0.85 |
| Dimensionless index | 0.52 ± 0.09 | 0.53 ± 0.11 | 0.82 |
Procedural characteristics are listed in Table 2 . Most patients underwent TAVI using conscious sedation (83% vs 80%, p = 0.50). There were significant differences in the vascular closure technique used. In patients who received a blood transfusion, half (56%) underwent percutaneous closure compared with 82% in the group without blood transfusion requirement (p <0.001). Similarly, patients who required a blood transfusion had higher rates of surgical cutdown and planned surgical closure (23% vs 5%, p <0.001).
| Variable | Transfusion | p value | |
|---|---|---|---|
| No (n=208) | Yes (n=124) | ||
| Intraprocedural TEE | 169 (83%) | 115 (92%) | 0.02 |
| Conscious sedation | 174 (83%) | 101 (80%) | 0.50 |
| General anesthesia | 34 (16%) | 25 (20%) | 0.39 |
| Vascular closure | |||
| Planned surgical closure | 11 (5%) | 29 (23%) | <0.001 |
| Perclose closure device | 171 (82%) | 71 (57%) | <0.001 |
| Other closure device | 28 (13%) | 19 (16%) | 0.65 |
| Procedure length (min) | 78 ± 33 | 67 ± 166 | 0.79 |
| Fluoroscopy time (min) | 21 ± 11 | 27 ± 26 | 0.009 |
| Contrast volume (ml) | 128 ± 61 | 129 ± 101 | 0.90 |
| Delivery success | 197 (98%) | 117 (95%) | 0.03 |
| Position success | 194 (94%) | 114 (91%) | 0.30 |
In-hospital outcomes are listed in Table 3 . In-hospital mortality because of all-cause death and cardiac death was significantly higher in patients who received a blood transfusion (all-cause death 9% vs 1%, p <0.001, and cardiac death 8% vs 0.5%, p <0.001). Thirty-day mortality was also increased in the transfusion group (11% vs 2%, p <0.001). At 1 year, this difference in mortality was sustained (28% vs 13%, p = 0.001). Figure 1 depicts a Kaplan-Meier survival curve with a landmark analysis at 30 days. The log-rank p value for the landmark analysis was p = 0.001, and the p value from 30 days to 1 year was 0.1.
| Variable | Transfusion | p value | |
|---|---|---|---|
| No (n=208) | Yes (n=124) | ||
| In-hospital mortality | |||
| All-cause death | 2 (1%) | 12 (9%) | <0.001 |
| Cardiac death | 1 (0.5%) | 11 (8%) | <0.001 |
| 30-day mortality | 4 (2%) | 14 (11%) | <0.001 |
| 1-year mortality | 28 (13) | 35 (28%) | 0.001 |
| Stroke | |||
| All stroke | 10 (5%) | 11 (8%) | 0.14 |
| Ischemic stroke | 9 (4%) | 11 (8%) | 0.09 |
| Hemorrhagic stroke | 1 (0.5%) | 0 | >0.99 |
| VARC-2 minor vascular complication | 2 (1%) | 1 (0.5%) | >0.99 |
| VARC-2 major vascular complication | 6 (3%) | 29 (23%) | <0.001 |
| VARC-2 minor bleed | 31 (15%) | 37 (30%) | <0.001 |
| VARC-2 major bleed | 1 (0.5%) | 3 (2%) | 0.14 |
| VARC-2 life threatening bleed | 0 | 15 (12%) | <0.001 |
| VARC-2 AKI 1 | 9 (4%) | 5 (4%) | 0.83 |
| VARC-2 AKI 2 | 12 (6%) | 13 (10%) | 0.14 |
| VARC-2 AKI 3 | 6 (3%) | 3 (2%) | >0.99 |
| Hemoglobin nadir (mg/dL) | 9.69 ± 1.23 | 8.26 ± 1.25 | <0.001 |
| Hematocrit nadir (%) | 29.4 ± 3.6 | 25.04 ± 3.8 | <0.001 |
| Hemoglobin at discharge (mg/dL) | 10.36 ± 1.28 | 10 ± 1.21 | 0.02 |
| Hematocrit at discharge (%) | 31.66 ± 3.80 | 30.80 ± 3.54 | 0.04 |
| Hemoglobin change admission to nadir (mg/dL) | 2.09 ± 1.02 | 2.74 ± 1.87 | <0.001 |
| Hematocrit change admission to nadir (%) | 6.39 ± 3.07 | 8.55 ± 5.49 | <0.001 |
| Length of hospital admission | 7.2 ± 4.8 | 10.8 ± 7.2 | <0.001 |
| Post-procedure length of stay | 5.5 ± 4.1 | 8.3 ± 6.2 | <0.001 |
| Intensive care unit length of stay | 1.7 ± 2 | 3.1 ± 3.8 | <0.001 |
VARC-2 minor and life-threatening bleeding complications were higher in the transfusion group (VARC-2 minor 30% vs 15%, p <0.001, and VARC-2 life-threatening bleed 12% vs 0%, p <0.001). When comparing the need for a blood transfusion because of a bleeding complication with the need for transfusion without a bleeding complication, patient who had transfusion because of bleeding had higher 30-day mortality (18% vs 6%, p = 0.03); however, no difference in mortality was found at 1 year (p = 0.07; Figure 2 ). In contrast, in patients without a blood transfusion, mortality was similar at 30 days (p >0.99) and 1 year (p = 0.08; Figure 2 ). Figure 3 shows the survival probability in patients who did not receive a transfusion with patients whose transfusion was because of bleeding and those whose transfusion was not because of bleeding. The patients who required a transfusion because of bleed had significantly lower survival (p <0.001). Length of hospital admission, postprocedure length of stay, and intensive care length of stay were all significantly longer in the transfusion group (p <0.001 for all). The proportion of patients leaving the hospital with anticoagulant and antiplatelet therapy was similar between without a transfusion versus patients with transfusion (aspirin 91.8% vs 91.5%, p = 0.93, clopidogrel 94.2% vs 94.5%, p = 0.93, and Coumadin 39% vs 31%, p = 0.33).
Additional analyses were conducted to analyze the impact of baseline hemoglobin on 1-year mortality. The entire population was divided into tertiles according to pre-TAVI hemoglobin values (lower tertile: 7.8 to 10.8 mg/dl, intermediate tertile: 10.8 to 12.2 mg/dl, and higher tertile: 12.2 to 16.6 mg/dl). Mean hemoglobin values among groups were 9.9 ± 0.7 mg/dl, 11.5 ± 0.4 mg d/L, and 13.2 ± 0.8 mg/dl, respectively, for the low, intermediate, and high tertiles. Renal insufficiency was most prevalent in the low tertile (70% vs 47.5% vs 45.3%, p <0.001). There were statistically significant differences in renal insufficiency rates between the low and intermediate tertiles (70% vs 45.3%, p = 0.002) and the low and high tertiles (p <0.001); however, no difference was found between the intermediate and high tertiles (p = 0.87). There was a significant difference between the rates of blood transfusion between the low and high tertiles (50% vs 26%, p <0.001). No difference was found when comparing the low tertile with the intermediate tertile (50% vs 38%, p = 0.09) and when comparing the intermediate tertile with the higher tertile (38% vs 26%, p = 0.10). There were no differences in the rates of in-hospital mortality (p >0.99), 30-day mortality (p = 0.97), and 1-year mortality (p = 0.68) among tertiles ( Figure 4 ).
