A retrospective study of conscious sedation versus general anaesthesia in patients scheduled for transfemoral aortic valve implantation: A single center experience

Abstract Objectives The current 2017 ESC/EACTS guidelines recommend transcatheter aortic valve implantations (TAVIs) as the therapy of choice for inoperable patients with severe symptomatic aortic stenosis. Most of the TAVIs worldwide are performed under general anaesthesia (GA). Although conscious sedation (CS) concepts are increasingly applied in Europe, it is still a matter of debate which concept is associated with highest amount of safety for this high‐risk patient population. The aim of this single center, before‐and‐after study was to investigate feasibility and safety of CS compared with GA with respect to peri‐procedural complications and 30‐day mortality in patients scheduled for transfemoral TAVI (TF‐TAVI). Methods From March 2012 until September 2014, patients scheduled for the TF‐TAVI procedure were included in a prospective, observational manner. From the 200 patients finally included, 107 procedures were performed under GA, using either an endotracheal tube or a laryngeal mask, and balanced anaesthesia. CS was performed in 93 patients using low‐dose propofol and remifentanil. Results Conversion to GA was needed 4 times due to procedural‐related complications (4.3%), in one patient due to ongoing agitation (1.1%). The CS‐group showed significantly shorter key time courses: anaesthesia time (105 [95‐120] minutes vs 115 [105‐140] minutes, P‐value = 0.009, Mann‐Whitney‐U‐test) and length of stay in the intensive care unit (1.6 [1.0‐1.5] d vs 2.1 [1.0‐2.0] d, P‐value = 0.002, Mann‐Whitney‐U‐test). The lowest mean arterial pressure was significantly higher in the CS‐group compared with the GA‐group (74.3 mmHg vs 55.2 mmHg, P‐value <0.0001, t‐test). CS was associated with less requirements of norepinephrine (0.1 μg/kg vs 2.3 μg/kg, P‐value <0.0001, Mann‐Whitney‐U‐test). Conclusions Our single‐center data demonstrate that CS is a feasible and safe alternative, especially with respect to a higher degree of intra‐procedural haemodynamic stability, and a reduced length of stay in the intensive care unit.


| INTRODUCTION
Aortic valve stenosis is one of the most common acquired valvular heart diseases in the elderly and, generally, a major cause of morbidity and mortality. [1][2][3] For many years, the only available options for patients classified to be inoperable were conservative medical treatment and/or balloon aortic valvuloplasty. However, both options must be considered as a palliative approach, since the benefit regarding improvement in outcome is highly limited. 4 In 2002, Cribier et al reported the first experience with transfemoral transcatheter aortic valve implantation (TF-TAVI) in patients with inoperable end-stage aortic stenosis. 5 To date, it is estimated that more than 200 000 TAVI procedures have been performed worldwide, and it has been established as a minimally invasive technique for this patient population. 6 Nearly 70% to 80% of TAVI procedures have been performed using the transfemoral approach (TF-TAVI). 7 Increasing experience, device-specific improvements, and economic aspects coupled with the possibility to generally perform TF-TAVI procedure under conscious sedation (CS) have led to an ongoing debate about the anaesthesiological management associated with the highest degree of safety for this patient population. [8][9][10] However, worldwide, approximately 90% of the TF-TAVIs are operated on general anaesthesia (GA) using transoesophageal echocardiography (TOE) guidance. 11,12 Between 2008 and 2015, more than 48 000 TAVI procedures were documented in Germany, showing a considerable higher amount of TF-TAVIs compared with the transapical and transaortic surgical approach 13 Several investigations have demonstrated that CS is a feasible and safe concept for TF-TAVI procedures. 10,14,15 In consideration of the ongoing high proportion of procedures worldwide performed under GA, the objective of the present retrospective analysis of 200 TF-TAVIs was to evaluate the feasibility and safety of our CS-concept compared with GA, and to report peri-procedural complications, main time courses, and 30-day mortality in this patient population. Since the logEuroSCORE I is used to calculate the predicted perioperative mortality of patients undergoing cardiac surgery, we moreover hypothesized that patients presenting a logEuroSCORE I ≥ 22% might benefit more from a CS-concept than from GA. A perfusion technician was always available to set up emergent extracorporeal circulation, if necessary. With the opening of the new hybrid OR, our Heart Team decided to change the first line anaesthesia management from GA (n = 107) to CS (n = 93), which is the preferred technique in our center up to now. Patients undergoing transapical or transaortic TAVI were excluded from this analysis. In regard to our hypothesis that patients at higher risk of periprocedural mortality, defined by a logEuroScore ≥22%, might benefit more from a CS-concept than from GA, we built subgroups of patients with a logEuroSCORE I </≥22%, which has been the calculated median value of our patient population. We, thereafter, compared the GA-group with a logEuroSCORE I ≥ 22% (n = 52) with the CS-group with a logEuroScore I ≥ 22% (n = 47) in consideration of peri-procedural complications, main time courses, and 30-day mortality.

| Anaesthesia techniques
Two consultant anaesthetists with many years of comprehensive experience in the field of cardiac surgery and TAVI procedures were responsible for the anaesthesiological management. All patients were monitored with a five-channel electrocardiogram, pulse oximetry, invasive radial artery blood pressure, and a central venous catheter. A heating blanket was positioned underneath the patient, and warmed intravenous fluids were given. In both groups, two external adhesive radio-transparent defibrillator pads were attached to the patient for early treatment of procedure-related episodes of ventricular fibrillation. GA, as well as CS, were provided by two qualified cardiothoracic anaesthesiologists.

| General anaesthesia
In the GA-group, patients received low-dose midazolam i.v. (0.01-0.02 mg/kg) prior to the beginning of the anaesthesia preparations. After insertion of a radial artery line under local anaesthesia (LA) for continuous blood pressure measurement, GA was induced with a bolus of propofol 1 to 1.5 mg/kg or etomidate 0.15 to 0.3 mg/kg body weight, followed by a continuous infusion of propofol (3-5 mg/kg/h), along with continuous infusion of remifentanil (0.3-0.4 μg/kg/min) and a bolus of rocuronium (0.5-0.6 mg/kg). Airway management was performed using a single lumen endotracheal tube or a laryngeal mask. Subsequently, a multilumen central venous catheter and a venous sheath were inserted. In the GA-group undergoing endotracheal intubation, transoesophageal echocardiography (TOE) was performed, whereas in patients ventilated via a laryngeal mask, transthoracic echocardiography was applied. Urinary catheterization was performed, and bladder temperature measured. Whenever possible, extubation of the patients at the end of the procedure was performed. After the intervention, all patients were transferred to the intensive care unit (ICU) or to the intermediate care station (IMC), for postoperative monitoring.

| Conscious sedation
In the CS-group, patients received low-dose midazolam i.v.

| Transfemoral transcatheter aortic valve implantation
The current 2017 ESC/EACTS guidelines recommend TAVI as the therapy of choice for inoperable patients with severe symptomatic aortic stenosis. In addition, the ESC/EACTS suggest TAVI as an alternative to surgical aortic valve replacement in patients with high operative risk. 18 The current update of the AHA/ACC guidelines even advises to consider TAVI in patients with intermediate risk. 19 At our center, patients estimated either inoperable or at elevated risk for a conventional surgical procedure were evaluated by an institutional Heart Team, composed of cardiologists, cardiothoracic surgeons, and anaesthetists.
In general, the interventional procedures were carried out by the same two cardiologists, using mostly valves from the balloon-expand-  Information about demographics, co-morbidities, NYHA and ASA classification, concomitant diseases, chronic medication, biochemical data, and laboratory parameters were obtained from the patients records with a follow-up time of 30 days after the intervention, according to internal standards. The STS score and the logEuroSCORE I have been described in detail elsewhere. [20][21][22][23] Procedural data were obtained from the anaesthetic protocols, which were evaluated retrospectively.
Peri-procedural-related parameters were defined as total amount of propofol, remifentanil, norepinephrine, epinephrine, amiodarone, maximum and minimum of the mean arterial pressure (MAP) and heart rate, total amount of administered i.v. fluids and red blood cells, platelets, and fresh frozen plasma transfused. During the time period of valve implantation, the rapid pacing maneuver, as a basic prerequisite, is purposely induced to reduce ventricular stroke volume and consequently blood pressure, independent of the type of anaesthesia performed. Consequently, this time period has been excluded from data analysis.

| Statistical analysis
Statistical analyses were performed using PASW statistics 24 software (SPSS, Chicago, USA). All continuous variables were tested for normal distribution with the Kolmogorov-Smirnov test and according to this result, summarized as mean with standard deviation or median with lower and upper quartile. For the comparison of the mean, the t-test was performed; for not normally distributed data, the Mann-Whitney-U-test was used. Discrete variables were compared using the chi-squared-test. All tests were two-tailed at a significance level of P < 0.05. A log rank test was performed for all time-to-event data, after dichotomizing the continuous parameters at the median.
All variables with significant log-rank tests were used for Cox regression analysis. Based on forward selection (likelihood ratio criteria), independent risk factors were identified, and the type of anaesthesia was then added into the model. The impact of these factors upon the survival time is presented as hazard ratio (HR) with 95% confidence intervals (CI), and assessed by Wald-test. The Kaplan-Meyer curves were plotted using the software R (Version 3.3.2).

| Conversion from CS to GA
A conversion from CS to GA due to procedural-related complications was needed in five cases (5.1%) due to different reasons: in four cases (4.3%), conversion from CS to GA was related to procedural complications, while in the remaining patient (1.1%), this was done due to persistent agitation (Supplementary Table S1).

| Intra-procedural anaesthesia characteristics
The anaesthesia-related procedural variables can be found in Table 2. 3.4 | Peri-procedural complications, main periprocedural time courses, and mortality All peri-procedural complications, main peri-procedural time courses, and mortality variables are shown in Table 3. No significant differences were seen between the GA-group and the CS-group, except for the need for cardiopulmonary resuscitation (CPR), which was significantly higher in the GA-group (CPR, GA 10 [9.3%] vs CS 0 [0%], Pvalue = 0.027, t-test). The analysis of the main post-procedural key     Consortium-2 26 at day seven post-procedural or at discharge (when the stay in the hospital was shorter than 7 days), showed no differences between both groups ( Figure 2). With respect to 30-day mortality, no significant difference was seen between the groups (GA 9.3% vs CS 5.4%, P-value = 0.288, t-test) ( Figure 1A).

| Intra-procedural anaesthesia characteristics in patients with a logEuroSCORE I ≥ 22
In high-risk patients with a logEuroSCORE I ≥ 22 (n = 99), differences between GA and CS in intra-procedural anaesthesia characteristics are shown in the Table 4. In contrast to the unselected population, no differences were seen between the groups regarding anaesthesia time  74.5 (12.9) mmHg, P-value <0.0001, t-test). Additionally, transfusion rate of red blood cells in the GA-group was significantly higher compared with the CS-group.
3.6 | Peri-procedural complications, main periprocedural time courses, and mortality in patients with a logEuroSCORE I ≥ 22 In patients with a logEuroSCORE I ≥ 22, differences between GA (n = 52) and CS (n = 47) in peri-procedural complications, main time courses, and mortality are shown in Table 5. Regarding peri-procedural complications, the only significant difference between the GA and CSgroup was the more frequent necessity of CPR in the GA-group compared with the CS-group (9.6% vs 0%, P-value = 0.039, t-test). LOS in the OR and in the hospital did not differ between groups (LOS OR P- to 30-day mortality, no difference was observed between the groups (GA 11.5% vs CS 6.4%, P-value = 0.492, t-test; Figure 1B).

| Cox regression analysis
A cox regression analysis was utilized to identify a number of variables that were independently related to mortality. In order to assess whether there is any additional effect by type of anaesthesia we added this parameter later to the analysis. Beside CPR (HR 6.85, 95% CI  (Table 6).

| DISCUSSION
The main finding of our registry-derived, single-center, before-andafter study is that conscious sedation for the TF-TAVI procedure is generally feasible, safe, and non-inferior compared with general anaesthesia with respect to peri-procedural complications and 30-day mortality.
CS, in contrast to GA, is characterized by a significantly higher degree of intra-procedural balanced haemodynamics, mirrored by less i.v. fluids and catecholamines administered, and less red blood cells transfused. Moreover, the CS-group stayed significantly shorter in the ICU, which has been shown to be an independent risk factor of death in the first 30-days after the procedure. 27   TABLE 5 Peri-procedural complications, main time courses, and mortality in patients at higher risk, defined by a logEuroScore ≥22   10,14,28,29 With respect to the necessity to convert from CS to GA, we report on a conversion of 5 times (5.1%).
In 4 cases (4.3%), conversion from CS to GA was related to procedural complications, and in only 1 patient (1.1%) due to persistent agitation.
The low conversion rate of 1.1% specifically related to this CS-concept underlines its safety and effectiveness. These data are in good agree- Although this is not the main focus of the present manuscript, we can report that we saw no differences between the groups regarding success of positioning of the valve or a higher percentage of misinterpretation of valve regurgitation in the CS-group ( Figure 2).
Another interesting observation in the context of CS, beyond the improved haemodynamics -evidenced by lower requirements of catecholamines during the procedure-, is the significantly reduced amount of total i.v. fluids needed to achieve this stability, together with a reduced amount of red blood cell transfusion. These findings have also been described previously. 15 In addition, Goren et al found significantly less episodes of sepsis in the CS-group, 10 possibly related to significantly less red blood cell transfusion, a supposable relationship that has been described previously in cardiac and non-cardiac surgery patients. 31,32 Secondly, they found a trend towards reduced post-procedural pulmonary complications, likewise to be related to red blood cell transfusion. 31  With the present study, we were able to show that the implementation of a CS-concept could lead to significantly reduced key time periods like anaesthetic time, and LOS in the OR and in the ICU.
Although not the primary aim of our investigation, reduced time courses have been shown to effectively reduce costs. 34 Although there might be some high level contemporary departments where the LOS in hospital for patients after TF-TAVI is somewhere between 1 and 3 days 35 , most of the available literature reports LOS in hospital between 7 and 13 days, which is in agreement with our data. [36][37][38] The present study has several limitations. This is a single-center observation in a before-after-design with a limited number of patients.
Due to the before-after-design, the two discussed anaesthesia regimes were not randomized, but the result of the departmental decision to change anaesthesia management for TF-TAVI from GA to CS at a given time. This was done in order to further develop our anaesthesia concept, helping to improve patient safety. Another point of criticism relates to the ongoing "learning curve" of the cardiologist, on the one hand, and the ongoing "learning curve" of the anaesthesiologist responsible for the CS-concept, on the other. The finding that every tenth patient in the GA-group was in need of CPR might also be interpreted as an indicator of the ongoing "learning curve." However, continuous improvement of manual, technical, and logistic skills are basic requirements for enhanced patients' safety and, consequently, can never be ruled out. Since there is only one randomized controlled trial available for the moment on this issue, 33

| CONCLUSION
In summary, this study provided evidence that CS in patients scheduled for TF-TAVI is feasible, safe, and non-inferior regarding peri-procedural complications and post-procedural 30-day mortality compared with GA, even in patients at higher surgical risk. Moreover, implementing this CS-concept led to significantly reduced key time periods like anaesthesia time, LOS in the OR, and LOS in the ICU.