Risk factors for cannula-associated arterial thrombosis following extracorporeal membrane oxygenation support: a retrospective study

Article information

Acute Crit Care. 2023;38(3):315-324

Publication date (electronic) : 2023 August 23

doi : https://doi.org/10.4266/acc.2023.00500

Ngan Hoang Kim Trieu^,1,2, Xuan Thi Phan³, Linh Thanh Tran², Huy Minh Pham², Dai Quang Huynh^1,2, Tuan Manh Nguyen², Anh Tuan Mai^1,2, Quan Quoc Minh Du², Bach Xuan Nguyen^1,2, Thao Thi Ngoc Pham^1,2

¹Department of Emergency and Critical Care Medicine, University of Medicine and Pharmacy at Ho Chi Minh City, Ho Chi Minh City, Vietnam

²Intensive Care Unit, Cho Ray Hospital, Ho Chi Minh City, Vietnam

³Intensive Care Unit, Tam Anh Hospital, Ho Chi Minh City, Vietnam

Corresponding author: Ngan Hoang Kim Trieu Intensive Care Unit, Cho Ray Hospital, 201B Nguyen Chi Thanh St, District 5, Ho Chi Minh City, Vietnam Tel: +84-35-202-4204 E-mail: drtrieungan12@gmail.com

Received 2023 March 21; Revised 2023 July 13; Accepted 2023 July 13.

Abstract

INTRODUCTION

Venoarterial extracorporeal membrane oxygenation (VA-ECMO) is being used increasingly to support patients with cardiogenic shock refractory to conventional therapies [1]. The use of ECMO can result in hemostatic dysfunction through direct exposure of the patient's blood to non-biomaterials inherent in the ECMO system. This exposure can activate endogenous and exogenous coagulation cascades, platelets, leukocytes, and complements. Additionally, the effects of shear stress may promote the formation of thrombi, particularly in low blood flow positions [2,3]. Moreover, increasing stage of extracorporeal circulation increases the risk of thrombosis in critically ill patients with multiple disease-related thrombotic risk factors [2].

Limb ischemia complications with multifactorial etiologies can develop at any stage of ECMO support, including during cannulation, during support, and after decannulation, resulting in increased morbidity and mortality [4-6]. Several studies have reported a high prevalence of limb ischemia complications during ECMO support. The primary risk factors identified for such complications are female gender, younger age, and preexisting peripheral vascular disease. Additionally, using large cannulas during ECMO can lead to flow obstruction [5-8]. However, limb ischemia due to thrombosis can be high even after decannulation. The formation of arterial thrombosis in VA-ECMO is potentially more frequent in the low-flow space segment because of the low blood flow between the arterial cannula and the distal perfusion cannula [4,9,10] (Figure 1). If left unnoticed and inappropriately managed, arterial lumen occlusion could result in acute limb ischemia [4].

Figure 1.

Position of arterial thrombus formation in the low-flow space. Created with BioRender.com.

The risk factors for venous thrombosis following cannula removal have been well documented [11-14]. Conversely, there has been limited research into cannula-associated arterial thrombosis (CaAT) after decannulation. Reported incidence rates of CaAT in patients undergoing VA-ECMO following decannulation have varied widely, ranging from 5% to 41% across medical centers. Furthermore, research exploring the prevalence of CaAT in Asian countries is scarce [15-18]. To the best of our knowledge, there is a lack of studies that have identified risk factors associated with development of CaAT after ECMO post-decannulation [15]. Therefore, we conducted this study to determine the incidence of CaAT and to identify the risk factors associated with it in patients requiring VA-ECMO.

MATERIALS AND METHODS

RESULTS

Baseline Characteristics of the Patients

Among the 85 patients who were supported by VA-ECMO and hybrid modes from January 2019 to February 2020, 38 were eliminated based on exclusion criteria (Figure 2), and 47 patients (39 receiving VA-ECMO and 8 receiving hybrid ECMO) were included in the final analysis. The median age was 33 years (26–46 years). Fewer patients were male (16 patients, 34%) than female. Acute myocarditis accounted for 68.1% of the main indications for ECMO. The median SOFA and APACHE II scores were 11 (IQR, 8–13) and 18 (IQR, 12.5–25), respectively. Three patients had a history of diabetes, and one patient had peripheral vascular disease. The median cannula–BSA ratio was 10.1 Fr/m², and the difference between the two groups was not significant. The ECMO system characteristics and outcomes between the CaAT and non-CaAT groups are presented in Tables 1 and 2.

Figure 2.

Flowchart of patient inclusion. ECMO: extracorporeal membrane oxygenation; PE: pulmonary embolism; VA: arteriovenous; CaAT: cannula-associated arterial thrombosis.

Table 1.

Characteristics of the study population

Table 2.

Outcomes of patients in two groups

CaAT and Outcomes

Forty-seven patients with arterial cannulas were successfully weaned from ECMO. CaAT occurred in 29 patients (61.7%); among them, superficial femoral artery thrombosis was reported in 15 (51.7%), and common femoral artery thrombosis was reported in 5 (20.6%). The distribution of patients with CaAT is shown in Figure 3. Additionally, 21 of 47 patients (44.7%) underwent surgical vascular closure in the operating room, and 55.3% of patients underwent percutaneous vascular closure at the bedside. No significant difference was observed in the incidence of CaAT between the two groups, with rates of 51.7% and 48.3%, respectively. After cannula removal, 41 patients (87.2%) were administered UFH or enoxaparin to prevent or treat thromboembolism.

Figure 3.

The distribution of cannula-associated arterial thrombosis.

In our study, limb ischemia occurred in 5 of 29 patients with CaAT (17.2%), and none occurred in the other group. Arterial thrombosis was detected in all patients with limb ischemia. None of them had peripheral arterial disease. Four patients were effectively treated with anticoagulant, and one patient with progressive limb ischemia required surgical thrombectomy. All five patients survived and were discharged. We observed other vascular complications, such as infection at the removal site or pseudoaneurysm, as in Table 2. The hospital mortality rates were not different between the two groups.

Risk Factors for CaAT

There were no significant differences in age or sex between the two groups, and arterial cannula and distal perfusion cannula sizes were not significantly different. The number of blood products transfused during ECMO was higher in the CaAT group, but the difference was not significant. At our center, cannula removal is performed using two methods: (1) surgical vascular closure and (2) percutaneous vascular closure. There is no significant difference in the thrombosis rate between the two methods. The duration of ECMO support was similar in the groups with and without CaAT (6 days [IQR, 4–8] vs. 6 days [IQR, 4–7], P=0.991, respectively). The ICU length of stay was 9.0 days (IQR, 7.0–12.0) vs. 9.50 days (IQR, 8.0–21.2) (P=0.582). Risk factors among the characteristics of patients and ECMO systems are presented in Table 1.

The hemostasis parameters are presented in Table 3. We found that the minimum and maximum fibrinogen concentrations were significantly lower in the group with CaAT (P<0.05). In addition, the number of days with platelets <150 g/L as the highest count was also significantly different between the two groups (P<0.05).

Table 3.

Results of the blood coagulation parameters

Univariate logistic regression showed that the maximum fibrinogen concentration, number of days with fibrinogen >4 g/dl, and ECMO flow rate–BSA ratio were associated with CaAT. A multivariable logistic regression model for CaAT prediction revealed that only the mean ECMO flow rate–BSA ratio was an independent risk factor for CaAT, with an OR of 0.79 (95% CI, 0.66–0.95; P=0.014) (Table 4).

Table 4.

Multivariable analysis of factors associated with CaAT

DISCUSSION

VA-ECMO use for refractory cardiogenic shock has markedly increased over the last decade, but hemostatic dysfunction in ECMO-supported patients remains a challenge for clinicians [9,23]. In particular, thrombotic complications after ECMO decannulation are underdiagnosed [24]. The cannula-associated deep vein thrombosis (CaDVT) rate following ECMO cannula removal ranges from 18% to 85%, and risk factors for CaDVT have been reported [11-14]. However, the incidence of CaAT after VA-ECMO decannulation has not been clearly defined. Recent studies have shown that the incidence of CaAT ranges from 5% to 41.0% [15,16,18]. The different incidence rates of CaAT among studies depend on two main causes: (1) the difference in diagnostic techniques among studies, including Doppler vascular ultrasound, vascular computed tomography (CT) scan, clinically demonstrated hypoperfusion, and surgical protocol records at the time of ECMO cannula removal and (2) the different rate of thrombosis screening post-ECMO weaning among centers. Some centers investigate CaAT based on suggestive clinical signs of limb hypoperfusion after cannula removal. Thrombosis is identified less frequently by Doppler vascular ultrasound or clinical signs of hypoperfusion than by surgical/autopsied visualization. The underestimation of arterial thromboembolic events can occur in more than 25% of patients [9,24]. In recent reports, Doppler vascular ultrasound or CT scanning of blood vessels was used to screen for CaAT when symptoms and signs of limb ischemia appeared, leading to a relatively low CaAT rate of 5%–12.5% [18,25]. In contrast, Cartwright et al. [16] reported that routine arterial thrombosis screening by Doppler vascular ultrasound could detect CaAT in up to 41% of patients after cannula removal. The present study found a higher rate of CaAT (61.7%) than in other studies [15-18]. Surgical vascular closure was performed in the operating room for up to 44.6% of patients undergoing cannula removal. Vascular surgeons routinely used a Fogarty catheter to screen for thrombosis during the operation, which may explain the higher rate of thrombosis detection. All patients whose cannulas were removed by percutaneous vascular closure with Proglide devices (Abbott Vascular) were routinely screened by vascular Doppler ultrasound for thrombosis, which also explains the high rate of CaAT in this study (Figure 4). In addition, a distal perfusion cannula, inserted in a common site of thrombus formation, was routinely inserted at our ECMO center.

Figure 4.

Detection of thrombus on bedside Doppler ultrasound on a 42-year-old male patient after supported extracorporeal membrane oxygenation due to acute myocarditis. (A) The transverse view depicts the color Doppler spectrum in the femoral vein. (B) The longitudinal view revealed nearly complete occlusion of the common femoral artery and superficial femoral artery and the absence of arterial flow by pulsed Doppler, indicating cannula-associated arterial thrombus (red arrow).

The arterial cannula inserted from the common femoral artery and the distal perfusion cannula inserted from the superficial femoral artery can create a low-flow or absent-flow between them (Figure 1). Low or absent flow at this space of the femoral artery leads to an extremely high risk of thrombus formation. This may explain why some clinicians favor surgical decannulation over percutaneous decannulation to remove thrombi [9]. In our center, a distal perfusion cannula is routinely inserted in patients supported by VA-ECMO. Therefore, the superficial and common femoral arteries were the most common sites of CaAT, with thrombosis rates of 51.7% and 20.6%, respectively. Although the incidence of CaAT was high, only 17.2% of this group had limb ischemia complications. The routine use of a Fogarty catheter at surgical vascular closure, daily Doppler vascular ultrasound, and appropriate heparin dosing after decannulation can explain the low incidence of limb ischemia. Furthermore, based on the principle of "less is more," our center tends to choose arterial cannulas with a small size of 15–17 Fr (92.5%) to minimize vascular wall damage and thrombus formation [5,6]. Additionally, BSA is an essential factor to consider, as it can influence the selection of arterial cannula size and flow rate in VA-ECMO supports [6,26]. A higher cannula–BSA ratio (>11 Fr/m²) is associated with an increased incidence of lower limb ischemia [6]. In our study, the median cannula–BSA ratio was 10.1. This ratio was less than 11 Fr/m² and was similar in the two groups, suggesting that BSA may not affect the discriminative ability of ECMO flow rate for arterial thrombosis. We suggest screening for arterial thrombosis following VA-ECMO support for early detection of complications, optimal utilization of anticoagulation after decannulation, and early thrombectomy when necessary. However, further research is needed to determine the risks and benefits of such screening.

Patients undergoing VA-ECMO are susceptible to multiple complications that can arise at different stages, with post-decannulation complications being underdiagnosed [9,24]. Bidar et al. [15] observed a post-decannulation CaAT rate of 14% in patients receiving VA-ECMO who were examined using Doppler vascular ultrasonography, despite the absence of risk factors. To the best of our knowledge, the current literature reports no risk factor for CaAT [15,16,18]. We analyzed risk factors for CaAT to minimize the possibility of thrombosis formation. Univariate logistic regression analysis showed that the maximum fibrinogen concentration, number of days of fibrinogen >4 g/dl, and ECMO flow rate–BSA ratio were associated with CaAT. In multivariate logistic regression analysis, the ECMO flow rate–BSA ratio was an independent risk factor for CaAT. For every 100 ml/min/m² increase in ECMO flow rate–BSA ratio, the odds of developing CaAT decrease by 21%. The ECMO flow rate is the velocity at which the ECMO system pumps blood into the circulation of the patient through the arterial cannula. The higher is the ECMO flow rate, the higher is the blood flow through the low-flow space between the arterial cannula and the distal perfusion cannula, decreasing the possibility of blood clots. Moreover, the lower is the ECMO flow rate, the more slowly is the blood pumped through the arterial cannula, resulting in reduced shear pressure and longer residence time around the artery cannula, leading to increased platelet adhesion and thrombus formation [2,27]. These changes may explain the increased cannula arterial thrombus formation in the low-flow state and warrant adequate anticoagulation during the phases of low ECMO flow rates and weaning trials.

Several limitations should be noted in our investigation. First, it was a single-center retrospective study with a limited sample size. Second, some anti-Xa data were missing, so the complete assessment of anticoagulation during ECMO may not have been accurate. Third, arterial cannulas were removed either by the percutaneous vascular closure technique using ProGlide devices (Abbott) at the bedside or by surgical vascular closure in the operating room. We did not have clear selection criteria for these procedures. The preferred method was at the discretion of the attending physician; therefore, the rate of arterial thrombosis could be underestimated in the percutaneous vascular closure group because of the lower sensitivity of vascular ultrasound in evaluating the femoral to iliac artery segment. Finally, the presence of these incidental small linear thrombi, detectable through surgical exploration or vascular ultrasound, may not have a substantial influence on patient outcomes or warrant specific interventions. Although limb ischemia can be devastating and cause disability, the criteria of thrombus that result in complications remains uncertain, warranting further investigation.

Our study showed that the incidence of CaAT after ECMO decannulation was 61.7%. Moreover, the ECMO flow rate–BSA ratio was an independent risk factor for CaAT. We suggest that screening for arterial thrombosis following VA-ECMO support is necessary, and that further research is needed to determine the risks and benefits of such screening.

KEY MESSAGES

Article information Continued