Ultrasound-guided versus conventional femoral venipuncture for catheter ablation of atrial fibrillation: a multicentre randomized efficacy and safety trial (ULTRA-FAST trial)

Abstract

Aims

Complications of catheter ablation for atrial fibrillation (AF) are frequently related to vascular access. We hypothesized that ultrasound-guided (USG) venipuncture may facilitate the procedure and reduce complication rates.

Methods and results

We conducted a multicentre, randomized trial in patients undergoing catheter ablation for AF on uninterrupted anticoagulation therapy. The study enrolled consecutive 320 patients (age: 63 ± 8 years; male: 62%) and were randomized to USG or conventional venipuncture in 1:1 fashion. It was prematurely terminated due to substantially lower-than-expected complication rates, which doubled the population size needed to maintain statistical power. While the complication rates did not differ between two study arms (0.6% vs. 1.9%, P = 0.62), intra-procedural outcome measures were in favour of the USG approach (puncture time, 288 vs. 369 s, P < 0.001; first pass success, 74% vs. 20%, P < 0.001; extra puncture attempts 0.5 vs. 2.1, P < 0.001; inadvertent arterial puncture 0.07 vs. 0.25, P < 0.001; unsuccessful cannulation 0.6% vs. 14%, P < 0.001). Though these measures varied between trainees (49% of procedures) and expert operators, between-arm differences (except for unsuccessful cannulation) were comparably significant in favour of USG approach for both subgroups.

Conclusions

Ultrasound-guided puncture of femoral veins was associated with preferable intra-procedural outcomes, though the major complication rates were not reduced. Both trainees and expert operators benefited from the USG strategy. (www.clinicaltrials.gov ID: NCT02834221).

Introduction

The number of catheter ablation procedures for atrial fibrillation (AF) is increasing world-wide. Due to the complexity of ablation and intensive anticoagulation, the procedure is associated with non-negligible rate of major complications including cardiac tamponade, oesophageal injury, pulmonary vein stenosis, ischaemic stroke, and vascular access site complications.^1–6 Vascular complications which are usually non-lethal but often prolong the hospital stay or require intervention, count as the most frequent ranging from 1 to 13% depending on the definition. Currently, uninterrupted anticoagulant regimen became common in AF ablation procedures with favourable safety outcomes.^3,7 The relationship between uninterrupted anticoagulation and vascular complications remains controversial and the strategies focused on their lowering have not been systematically studied in such setting.

Real-time ultrasound-guided (USG) venipuncture was first introduced for placing central venous catheters. ^8–11 The USG venipuncture was reported to reduce procedure time with decrease in inadvertent artery punctures and puncture attempts overall and, eventually, resulted in reduction of vascular complications compared to conventional (CONV) approach. Compared to these studies with a single, smaller diameter central venous catheter and usually no routine anticoagulants during the procedure,^8–10 catheter ablation for AF requires multiple sheaths with a larger diameter in bilateral femoral veins and higher anticoagulation level during/after the procedure. Hence, a multicentre randomized controlled trial was proposed to evaluate the efficacy and safety of USG venipuncture of femoral veins in patients undergoing catheter ablation for AF without cessation of oral anticoagulants.

Methods

Study design

ULTRAsound-guided Femoral vein Accessiblity, Safety and Time (ULTRA-FAST) trial for AF treatment is a multicentre, randomized controlled prospective study (www.clinicaltrials.gov ID: NCT02834221). All equipment was set up in each centre with no sponsor support. The study protocol was approved by the institutional review board (Ethics Committee) at participating centres, namely at Institute for Clinical and Experimental Medicine, Czech Republic, Liberec General Hospital, Czech Republic, Military University Hospital Prague, Czech Republic, and Kawakita General Hospital, Japan.

Patients older than 18 years scheduled for catheter ablation for AF pulmonary vein isolation while using uninterrupted oral anticoagulation therapy were enrolled. For new oral anticoagulants, the patient skipped the drug on the day of the procedure. The target prothrombin time (international normalized ratio) for those who were taking warfarin was 2.0—3.0. Patients with history of problematic vascular access at groins were excluded. The inclusion to the study was offered to all consecutive patients who were eligible for the study.

Patients who signed informed consent with the study were assigned to either USG or CONV venipuncture in a 1:1 fashion by covariate adaptive randomization method.

Operators

A total of 11 operators participated in the study. Five of them had previous experience with >50 AF ablation procedures (=expert operators) and remaining 6 were trainees. According to self-reported experience, all operators performed <50 procedures with USG venipuncture prior to the study initiation. On the other hand, those USG-naïve were appropriately trained (>20 USG venipuncture procedures) to qualify for the study.

Ultrasonography

An 8 MHz linear array ultrasonography transducer was connected to a portable echocardiograph (Vivid I, GE Health Medical, Horten, Norway). The initial depth setting of ultrasound beam was 4 or 6 cm if the body mass index was lower or higher than 35 kg/m², respectively. The transducer was covered with a sterile sleeve and placed at a 90-degree angle to the course of the vein at the groin to obtain a transverse view (Figure 1A). Femoral vein was differentiated from the artery by compressibility by the transducer downforce.¹² When this was difficult, colour Doppler was used to display arterial/venous flow. Modified Seldinger technique¹³ was used to access the vein under the direct ultrasound visualization of the tip of the needle (Figure 1B).

Ablation procedure

Local anaesthesia was applied with a 23-gauge needle and an 18-gauge needle was used for venipuncture and for passing the guidewire into the veins. Two punctures were performed in each femoral vein to introduce a total of four endovascular sheaths.

One 11-French (Fr) sheath and one 7-Fr sheath were inserted into the left femoral vein in all patients. Instrumentation of the right femoral vein differed according to the ablation technology. Two 8.5-Fr sheaths were inserted into the right femoral vein for the manually-performed procedures. Robotic procedures (Hansen Medical, Inc., Mountain View, CA, USA) required one 14-Fr and one 8.5-Fr sheath insertion and for cryoaballoon ablation, one 12-Fr and one 7-Fr sheath were inserted. Heparin was administered intravenously immediately after obtaining venous access (5000 IU) with subsequent adjustment according to the body weight after transseptal puncture. Target activated clotting time was set between 300–350 s. Adjustment was performed with boluses of heparin according to activated clotting time every 30 min. After the ablation procedure, all sheaths were removed and puncture sites were closed with figure-of-eight sutures. Reversal of anticoagulation with intravenous protamine was permitted at the operator’s discretion. Moderate-level mechanical (sandbag) compression over the groins was left for 6 h followed by overnight bed stay.¹⁴

Venipuncture data collection

Puncture start was recorded at the time of the first placement of the transducer on the groin in the USG group or at the time of the first palpation of the femoral artery to get orientation for the puncture in the CONV group. Puncture finish time was recorded when position of all wires in the inferior vena cava was confirmed with fluoroscopy. The number of venipuncture attempts was incremented for every withdrawal of the needle including test puncture with the anaesthetic needle if performed. Inadvertent arterial punctures were also recorded and when it happened more than once, the operator had a choice to cross-over to alternative venipuncture strategy or hand-over the procedure to another operator. The same applied for the situation when >5 venipuncture attempts were not successful. If the physician had to cross-over or hand-over the procedure, these times were also included in the procedure time.

Follow-up

Vascular access site related pain was self-reported by patients using 11-point numerical rating scale (0 = no pain, 10 = worst) questionnaire on the day after the procedure. This was assisted by a nurse blinded to the study arm allocation. Mandatory clinical visit 3 months after the hospital discharge was scheduled to disclose incidental delayed manifestation of vascular access site complications.

Study endpoints

The primary endpoint was the rate of major vascular access complications defined as (i) haematoma, (ii) arteriovenous fistula, or (iii) pseudoaneurysm which required interventions such as transfusion or surgical repair, and/or resulted in prolongation of the hospital stay or hospital re-admission.¹ These were reviewed by an independent clinical events committee who had access to all medical records.

Intra-procedural secondary objectives were the following: the rate of successful insertion of all four wires with no need for cross-over to the other venipuncture strategy or no need for hand-over the venipuncture to another operator, time for inserting all wires, puncture attempts, inadvertent artery puncture, first pass success, and the use of X-ray to proceed the wire above the inguinal ligament. Prolonged compression after the procedure and proportion of patients who claimed pain scale >3 were also analysed.

Sample size consideration

Based on the previous studies,^15–19 we anticipated the complication rate of 1 and 4% in USG and CONV study arms, respectively. A total of 848 patients (424 patients in each group) would provide 80% power to detect the difference at two-sided alpha level of 5%. Consequently, the study was originally designed to recruit approximately 900 patients.

Statistical analysis

Continuous variables in both study arms are presented as mean ± standard deviation and compared by two-sided t test for independent samples or Mann–Whitney U test, as appropriate. Fisher’s exact test or χ² test was used for categorical variables. A P-value < 0.05 was considered significant. The intention-to-treat analyses were performed in total population and in subgroups of expert operators and trainees separately. All analyses were performed using the SPSS 17.0 statistical package (SPSS, Inc., Chicago, IL, USA).

Results

Study participants

Between March 2016 and November 2016, 323 patients were enrolled into the study. Three patients were excluded before randomization because of known or anticipated vascular access problems (one patient had a femoral artery pseudoaneurysm and two others had an arteriovenous fistula). Eventually, 320 patients were randomized (USG = 160, CONV = 160). At that time, the first interim analysis disclosed the primary endpoint rate lower than expected for each group (USG = 0.6%, CONV = 1.9%). Recalculation of sample size according to this complication rate resulted in doubled number of patients to maintain statistical power, specifically 1896. Because several secondary objectives already showed significant difference between USG and CONV groups, the internal committee recommended to terminate the study prematurely.

From a total number of 320 patients, we excluded one patient in the USG group as femoral vein thrombus was detected at the beginning of the procedure. Four patients did not complete pain assessment; however, since all other parameters were available, they were included for analyses (Figure 2). Baseline patient characteristics were well-matched between both study arms as shown in Table 1. A total of 60% of patients were on warfarin therapy and 40% were prescribed with one of the new oral anticoagulants. Trainees performed 49% of procedures. Cryoballoon and/or robotic ablation procedures requiring endovascular sheath of larger diameter constituted less than 10% of the cases. Protamine was administered in 64% of patients at the end of the procedure with no significant difference between both study groups.

Procedural outcomes

No significant difference in the complication rates between USG and CONV group was observed (0.6% vs. 1.9%, respectively, P = 0.62, Table 2). Two of the observed primary endpoint events included larger haematomas with significant haemoglobin drop and other two comprised arteriovenous fistulas (one requiring surgical intervention and one prolonging hospitalization).

For secondary objectives, USG group showed significantly shorter puncture time, fewer cases of unsuccessful cannulation, extra puncture attempts, inadvertent arterial puncture, and/or use of X-ray (Table 2, Figure 3). In addition, USG group presented with significantly higher first pass success. The necessity to prolong the groin compression did not differ significantly between two groups. There was a trend towards lower number of patients with pain scale >3 in the USG group (2% vs. 6%, P = 0.08). Table 3 shows the details in this subset of patients.

Subgroup analyses

The baseline characteristics did not differ between USG and CONV group in both trainee and expert operator subgroups (Table 4). All intra-procedural measures were significantly in favour of USG venipuncture in the subgroup of trainees (Table 5, Figure 3). Although this was less obvious in subgroup of expert operators, they still had significantly shorter puncture time, higher first pass success, and less extra puncture attempts in USG arm.

Discussion

Major findings

This multicentre, randomized controlled prospective study was prematurely terminated due to lower-than-expected vascular complication rates, which prevented to demonstrate the statistically significant reduction of the primary endpoint when using USG venipuncture, although the complication rates were numerically lower in the USG as compared to CONV study arm. Nevertheless, the main finding of the study was that USG venipuncture significantly improved several secondary intra-procedural objectives, which was particularly evident in subgroup of trainees but also in a group of expert operators.

Primary endpoint

The reported complication rate for femoral venipuncture varies substantially from 1 to 13%, depending on the definition of complication in individual studies.^1–6 For ULTRA-FAST trial, we adopted definition according to the 2012 HRS/EHRA/ECAS Expert Consensus Statement on Catheter and Surgical Ablation of AF.¹ The expected complication rates for USG (1%) and CONV (4%) were estimated from the previous studies describing clearly the puncture method.^15–19 We allowed the operator to cross-over to the other arm if they had difficulty in venipuncture, which resulted in 14 out of 160 (9%) patients crossing over from the CONV to USG group. Twelve of these patients had an inadvertent artery puncture including one patient who later developed arteriovenous fistula. In this case, ultrasonography revealed that the artery was directly above the vein at the puncture site. Overall, this was not a rare finding in our study. In a computed tomography study, 65% of the femoral veins showed an overlap with the artery in the anteroposterior plane.²⁰ Hence, cross-over to USG approach might have prevented complications associated with ongoing CONV venipuncture and diminished the difference in primary endpoint between study arms. Relatively high rate of protamine usage (64%) could be another explanation for the low complication rate compared to previous studies.^15–19 However, they have not reported data on anticoagulation reversal that precludes any comparison.

Secondary endpoints

All intra-procedural measures were improved in the USG compared to the CONV group. This finding is concordant with previous studies.^15–19 Importantly, in 11 out of these 14 cases, the same operator succeeded to cannulate the vein without the need to hand-over the procedure. This implies that benefit from visualization of vessels is more important than skills/experience of individual operators. At the same time, no case of cross-over from the USG to CONV group indirectly indicates that USG approach was appreciated by operators in general and that acceptable/non-confusing imaging of the target area was achieved in majority of cases. Trainees are in general less familiar for CONV approach, and had a significant unsuccessful cannulation rate in this approach. This was not seen in expert operators, indicating they have less understanding of the anatomy of groins. By visualizing the vascular access site with ultrasound imaging, there can be a possibility to improve even their CONV approach in the future by further understanding of the anatomy.

In addition, even expert operators benefitted from USG approach. Although experienced operators are usually reluctant to change their procedural habits, our data suggest that they also could gain an advantage from the USG technique, which can be easily mastered after <10 training cases for both senior and fellow doctors.¹⁶ In our study, none of the operators had previous experience with more than 50 cases before participating in this study, while still receiving the merit.

There were some patients without vascular complication who claimed significant post-procedural pain even if they had no inadvertent arterial puncture, extra puncture attempts, or small haematoma (Table 3). The contributing factor for vascular access site pain is difficult to assess. Nevertheless, fewer patients were suffering from strong pain in the USG vs. CONV group and this difference was close to statistical significance (P = 0.08).

We also analysed the potential of USG approach to reduce fluoroscopy use. As ultrasound visualizes easily even small venous branches in the groin, it may prevent their inadvertent puncture and facilitate the puncture of the main trunk of the femoral vein. This may be the explanation for less use of X-ray to advance the guidewire into the inferior vena cava in the USG group observed in a subgroup of trainees.

Several previous studies compared USG and CONV approach for AF ablation procedures.^15–19 None of the studies allowed crossing over the venipuncture method, which was an option in our study and occurred rather frequently (9%). This cross-over occurred only in patients randomized to CONV approach. Analysis of the trend in frequency of cross-over events revealed that only two cross-overs were within the first 3 months of the recruitment period while 12 remaining occurred thereafter. This suggests that adoption of USG venipuncture method and recognition of associated benefits makes operators more prone to utilize it when they encounter difficulties in CONV venipuncture.

There is only one small study (36 patients) that investigated intra-procedural measures. This study reported less extra puncture attempts and inadvertent arterial puncture with the use of USG approach, though the total time to successful cannulation did not differ between both groups. The time per guidewire insertion was measured excluding the time for anaesthesia, and most importantly, the time spent by palpation to find the arterial pulse, which amounts to considerable proportion of total puncture time in the CONV approach, was not included. This may explain the absence of puncture time reduction in their compared to our study.

In studies that investigated insertion of haemodialysis or central venous catheters, the success rate for USG and CONV approach were 90–100% and 65–90%, respectively.^8–10 Though we had to insert four guidewires/sheaths compared to single one in these studies, our success rate of 99% and 86% was still within this range.

In addition to these benefits, there is a chance to detect vascular abnormalities such as deep venous thrombosis during the ultrasound scanning as in the current study. Since the groin instrumentation in this setting may result in acute pulmonary embolism, ultrasound imaging prior to intervention might be of general clinical value.

Study limitations

In terms of original study design, the major limitation was unexpectedly low vascular complication rate in the trial despite the substantial proportion of venipunctures was performed by trainees. This made the study underpowered to detect the difference in primary endpoint. Cross-over was allowed per study protocol which might also diminish the differences between the study arms that were analysed by intention-to-treat principle.

Conclusions

Because of the lower-than-expected incidence of primary endpoint, this prospective, randomized controlled study did not demonstrate statistically significant benefit from USG venipuncture for the reduction in vascular complication rates. However, USG-guided approach compared to CONV technique significantly facilitated venipuncture as shown by multiple secondary intra-procedural measures. Both trainees and expert operators benefited from the USG strategy. The greater benefit from USG venipuncture was achieved in less experienced operators which highlights utility of ultrasound imaging as a learning tool for electrophysiologists in training.

Acknowledgements

We wish to express our gratitude to the other physicians contributing to the study: Drs Masato Fukunaga, Luka Klemens, Armen Khachatryan, Predrag Stojadinovic, and Mihailo Vukmirovic.

Funding

This study was supported by the project (Ministry of Health, Czech Republic) for development of research organization 00023001 (IKEM, Prague, Czech Republic).

Conflict of interest: J.K. is a scientific advisor and speaker for Biosense Webster, Boston Scientific Corp./EP Technologies, GE Healthcare, Medtronic, St Jude Medical, Siemens Healthcare and speaker for Biotronik GmbH. P.P. is a speaker for St Jude Medical. However, none of those companies provided any support for this study.

References