High-risk pulmonary embolism (PE) is defined as acute PE with hemodynamic instability, characterized either as persistent hypotension (systolic blood pressure (SBP) less than 90 mmHg for more than 15 min without signs of organ hypoperfusion), obstructive shock (—SBP less than 90 mmHg with signs of organ hypoperfusion), or cardiac arrest (CA) [1]. While only 4 to 5% of PE are considered high-risk, they account for most PE-related early death [2, 3], with reported mortality rates up to 95% in patients presenting with cardiac arrest [4].

Recently updated guidelines recommend emergent reperfusion therapy for high risk PE, primarily with systemic thrombolysis (class I, level of evidence B), or with surgical embolectomy or catheter-directed therapy (CDT) in case of contraindications to systemic thrombolysis [1]. The use of veno-arterial extracorporeal membrane oxygenation (VA-ECMO) has also been advocated as a life sustaining therapy for patients presenting with refractory shock or cardiac arrest in the setting of high-risk PE [5], while awaiting the resolution of pulmonary artery obstruction. Therefore, and despite the lack of solid level evidence studies, European society of cardiology guidelines indicate that VA-ECMO may be considered (Class IIb, level C) in patients with intractable circulatory collapse related to PE [1].

Due to the current absence of prospective randomized study evaluating VA-ECMO in high risk PE, its potential benefits in this setting have only been presented in case reports and retrospective case series. Although most of these studies reported survival benefits from VA-ECMO, interpretation of these results is hampered by important limitations. The first one is the lack of formal diagnosis of PE as a cause of CA in a proportion of reported cases, where the observation of an acutely dilated right ventricle (RV) was considered as indirect evidence of acute PE. However, the RV may also acutely dilate during CA caused by arrhythmias, hyperkalemia and hypovolemia, hence independently from pulmonary obstruction [6]. The second limitation refers to the indication of VA-ECMO to provide circulatory support in cardiac arrest due to PE. VA-ECMO may indeed provide mechanical support for refractory shock following the return of spontaneous circulation (ROSC) after conventional cardiopulmonary resuscitation (CPR), or may be used for extracorporeal cardiopulmonary resuscitation (ECPR) in refractory cardiac arrest [7]. In the latter, the chances of survival are expected to be low, due to the absence of transpulmonary blood flow during CPR prior to the insertion of ECMO. Results of VA-ECMO for PE-associated cardiac arrest should therefore take into account this important distinction.

In this study, we present a retrospective case series of 18 patients treated in our tertiary-care centre with VA-ECMO for high-risk PE. We report the outcomes, associated therapies and complications, according to the indications of VA-ECMO in this population (ECPR for refractory CA or mechanical support for refractory shock).

The main results of our study are that VA-ECMO in high-risk PE was associated with a ICU survival of 22%, with only 9% survivors when VA-ECMO was inserted during CPR for refractory CA (ECPR), contrasting with 42% survival in patients treated with VA-ECMO for refractory shock, either primary or following resuscitated CA.

High-risk PE (massive PE), characterized by profound hemodynamic instability, is associated with a particularly high mortality, ranging from 47 to 52% in the absence of CA [3, 12] to 84 to 95% when PE is complicated by CA [4, 12]. This ominous prognosis reflects the inability to maintain systemic perfusion due to the obstruction of the pulmonary arteries (PA), leading to acute right ventricle (RV) overload [5]. Hence, treatment of high-risk PE must include a strategy to remove pulmonary obstruction, primarily with systemic thrombolysis (grade I recommendation), or, alternatively, with catheter-directed therapy or surgical embolectomy (grade IIa) [1]. In the presence of contraindications to thrombolysis, or in patients presenting with refractory cardiac arrest or deteriorating in spite thrombolysis, mechanical circulatory support with VA-ECMO may represent the only viable strategy to provide adequate systemic perfusion while awaiting the resolution of pulmonary obstruction (bridge-to-therapy or bridge-to-recovery) [13].

The role of VA-ECMO in high-risk PE has been the matter of several retrospective studies, case reports, reviews and meta-analyses. We reviewed 33 publications presenting retrospective analyses of a total of 2996 high-risk PE patients treated with VA-ECMO, as summarized in Tables 6.1,6.2,6.3 (Ref. [5, 7, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44]). Except from one study totalizing 2197 patients, most studies included a limited number of patients (5–87, average 23 patients), and data regarding the specific indication of ECPR were available in 150 patients from 19 studies. We extracted the proportion of patients with a confirmed diagnosis of PE, the proportion of patients with cardiogenic shock and cardiac arrest, as well as the survival rate in each study. The mean overall survival rate reported was 56%, albeit with considerable variability across studies (20–95% reported survival rates). The 22% ICU survival in our cohort appears therefore much less favourable, but several hypotheses may be advanced to explain such high mortality. The first one is related to the high proportion (89%) of patients experiencing CA in our study, and most significantly, to the high percentage of ECPR (61%, with an ICU survival of 9%, contrasting with 42% survival for non-ECPR indications). It is particularly noteworthy that VA-ECMO in high-risk PE encompasses different indications, with distinct pathophysiology and most probably very different outcomes, which comprise refractory shock, resuscitated CA, and refractory CA [14, 15]. Many published studies did not consider this heterogeneity and did not provide details on the relative mortality of VA-ECMO according to these various conditions. This may represent an important drawback for the correct interpretation of the potential benefits of VA-ECMO in high-risk PE, as emphasized by Karami et al. [45] in their recent meta-analysis.

As a matter of fact, it has been clearly shown that the chances of survival are significantly reduced when VA-ECMO is inserted following CA in the setting of PE, most significantly if it is implemented as ECPR for refractory CA [5, 15]. In a large database of 52 patients, Meneveau et al. [15] reported an overall survival of 38%, but of only 13% in patients experiencing CA and 11% in those treated with ECPR. Also, Corsi et al. [14], showed, in a study on 17 patients, an overall survival of 47%, but 14% in patients treated with ECPR for refractory CA. In one of the largest cohort to date, Giraud et al. [5] recently reported that ECMO after or during CA were both independently associated with increased 30 day mortality among PE patients treated with ECMO. Finally, in a systematic review on VA-ECMO in high-risk PE, Scott et al. [13] reported a six-fold increase in the risk of death when ECMO was inserted during CPR (ECPR).

These observations emphasize the peculiarities of CA in the context of PE, where obstruction to blood flow in the pulmonary circulation precludes both circulation and gas exchange during CPR. Such phenomenon was well-demonstrated in our patients by a high pulmonary obstruction index (POI), which tended to be greater in non-survivors (47% vs 30%). It is here worth to mention that Van der Meer et al. [46] previously reported that a POI greater than 40% was associated with a 11.2-fold increase in mortality in PE patients. In addition, as observed in our study, CA in PE is generally characterized by an initial non-shockable rhythm, known to be correlated with a dismal prognosis in OHCA [47] and IHCA [48].

It is therefore likely that irreversible anoxic end-organ damage may develop during CPR for PE-related CA before the initiation of the artificial circulation, due to poor or absent transpulmonary blood flow. Accordingly, we found that non-survivors displayed a significantly higher value of initial arterial blood lactate (median value 12.4 mmol/L) than survivors, pointing to significant tissue anoxia. In this respect, monitoring the quality of CPR by assessing transpulmonary blood flow using end-tidal CO2 [49] could be particularly useful to determine which patient might or not benefit from VA-ECMO during CPR of PE-related CA, as recently underscored by Giraud et al. [5], but such data were unfortunately not available in our patients.

A second hypothesis to explain the higher mortality in our study refers to the high proportion of patients in whom no reperfusion strategy was implemented. Anticoagulation only was administered to 44% of our patients including 50% of non-survivors. This strategy of stand-alone VA-ECMO may have negatively influenced ECMO outcome, as current recommendations suggest that additional therapies to treat embolic obstruction should be considered for best results [1]. In the largest database of ECMO in high-risk PE to date, Hobohm et al. [7] showed that the lowest mortality was noted in patients treated with VA-ECMO plus surgical thrombectomy (64%) and VA-ECMO plus thrombolysis (69.7%) in comparison to stand-alone ECMO (72.7%). These results are consistent with those of Meneveau et al. [15], who reported a 77.8% mortality of stand-alone VA-ECMO in contrast to only 29.4% of VA-ECMO coupled to surgical embolectomy. These findings support the potential of mechanical embolectomy to improve survival in VA-ECMO for high-risk PE, which should be evaluated in future prospective studies. In our cohort, embolectomy did not appear to show such benefit, with only one surviving patient out of 4 treated with embolectomy.

At variance with the above, Giraud et al. [5] reported a significant survival advantage of stand-alone VA-ECMO (85.5% survival) in comparison to VA-ECMO combined with pre-ECMO thrombolysis or CDT (35.5% survival), in a series of 36 VA-ECMO for high-risk PE. As outlined by the authors, the insertion of VA-ECMO after failed thrombolysis (either systemic or catheter-directed) exposes the patient to major risks of severe bleeding, resulting in significant increases in mortality. Indeed, we found a high incidence of hemorrhagic complications in our cohort (72%), occurring at a comparable rate in survivors and non-survivors. Importantly, massive bleeding was the direct cause of death in 3 patients, all of whom had received pre-ECMO systemic thrombolysis. The findings of Giraud et al. [5] agree with those of Maggio et al. [21], who reported a 77% survival in patients treated with stand-alone VA-ECMO, arguing that in most patients, spontaneous PE lysis generally allows RV recovery within 5 days. Accordingly, Pasrija et al. [23] recently proposed a strategy of VA-ECMO with anticoagulation alone, followed by delayed embolectomy (surgical or percutaneous) only in conditions of persistent pulmonary thrombotic obstruction and RV dysfunction.

An obvious development from the above discussion is that adequate patient selection is crucial to optimize VA-ECMO outcome in high-risk PE patients. In the context of refractory shock, VA-ECMO immediately provides adequate systemic perfusion and reduces right ventricle overload, while giving time for spontaneous or mechanical pulmonary reperfusion to restore RV-PA coupling. Early VA-ECMO support and avoidance of thrombolysis to reduce hemorrhagic risk might therefore represent the most effective therapy in such patients. This is indeed supported by the high survival rate reported in studies using such strategy [5, 21, 23]. In contrast, implementing VA-ECMO for refractory CA due to pulmonary obstruction appears futile, due to the considerable risk of irreversible anoxic injury, unless sufficient transpulmonary blood flow can be demonstrated during CPR by the measurement of end-tidal CO2.

An additional factor to consider for the adequate interpretation of survival data relies in the formal diagnosis of PE, documented in 88% of our patients. In previous studies, such documentation has been variable, ranging from 58 to 100% (see Tables 6.1,6.2). Although current recommendations indicate that echocardiographic evidence of acute RV pressure overload in a haemodynamically compromised patients with suspected PE may justify immediate therapy, without the need of confirmatory CT angiography [1], it must be emphasized that RV dysfunction/dilation during or after CA is not specific for acute PE. Aagaard et al. [6] demonstrated acute RV dilation in experimental CA (porcine model) induced by hypovolemia, hyperkalemia, and primary arrhythmia. Also, in a retrospective study on 59 patients with CA, Wardi et al. [50] found post-CPR RV dysfunction/dilation in the majority of patients, regardless of the aetiology of CA. Accordingly, one may expect that some patients treated with VA-ECMO for suspected PE, solely on the basis of acute RV dysfunction, may have suffered from other causes of CA, with an inherently better prognosis, given the absence of pulmonary vascular obstruction. This might have introduced some bias for the interpretation of survival results in studies without a high level of formal PE confirmation.

Our study has several limitations. First, it was monocentric, retrospective and included a relatively small number of patients, which limited statistical comparisons. It must be noted however that these limitations are shared by many studies in the field, which were all retrospective in design and which reported, with some exceptions, single-center experience of relatively small sample sizes. Obviously, future large-scale prospective studies evaluating VA-ECMO in high-risk PE are warranted to circumvent these limitations. Secondly, we included patients over a 12-year period (2008–2020), with the inherent risk of heterogeneity due to evolving standards of care and practice. Third, the initiation of VA-ECMO in our cohort was decided on a case-by-case basis without a specific protocol, which may have resulted in improper patient’s selection. The implementation of a multidisciplinary pulmonary embolism response team (PERT) [51] might here be helpful to assist the decision making process in the future.

Our single centre experience of VA-ECMO for documented high-risk PE indicates, in spite of several limitations, that this therapeutic strategy was associated with very different outcomes depending on the clinical context. When used as mechanical circulatory support for refractory shock, whether primary or following resuscitated cardiac arrest, VA-ECMO allowed survival in up to 42% of patients. In contrast, when applied as extracorporeal resuscitation (ECPR) for refractory cardiac arrest, VA-ECMO was associated with only 9% ICU survival. Furthermore, VA-ECMO in high-risk PE was associated with significant bleeding complications, with more severe consequences in patients undergoing systemic thrombolysis. Future studies should therefore take into account the distinct clinical presentations of high-risk PE when reporting the effects of VA-ECMO, and should determine the best strategy for reperfusion therapy in such circumstances.

AC, Anticoagulation; BMI, Body Mass Index; CA, cardiac arrest; CDT, Catheter-Directed Therapy; COPD, Chronic Pulmonary Obstructive Disease; CPR, Cardiopulmonary Resuscitation; CS, Cardiogenic Shock; DVT, Deep Vein Thrombosis; ECPR, Extracorporeal Cardiopulmonary Resuscitation; FFP, Fresh-Frozen Plasma; HIT, Heparin-Induced Thrombocytopenia; IHCA, In-Hospital Cardiac Arrest; LOS, Length Of Stay; MV, Mechanical Ventilation; OHCA, Out-Of-Hospital Cardiac Arrest; PE, Pulmonary Embolism; RRT, Renal Replacement Therapy; RV, Right Ventricle; SE, Surgical Embolectomy; VA-ECMO, Veno-Arterial Extracorporeal Membrane Oxygenation; VTE, Venous Thrombo-Embolism.

ZL—Investigation; Data curation; Formal analysis; Writing: original draft. EL—Investigation; Data curation; Formal analysis; Writing: review and editing. JB—Investigation; Data curation; Writing: review and editing. NBH—Investigation; Data curation; Writing: review and editing. VR—Investigation; Data curation; Writing: review and editing. SSK—Methodology, radiological analyses; writing: review and editing. MK—Investigation; Writing: review and editing. JDC—Project administration; Writing: review and editing. LL—Methodology; Project administration; Investigation; Data curation; Formal analysis; Writing: original draft. All authors contributed to editorial changes in the manuscript. All authors read and approved the final manuscript.

This retrospective study was approved by our local ethical committee with waiver of consent (Commission Cantonale d’Ethique de la Recherche sur l’Etre Humain/CER-VD-Nr: 2017–01184).

Not applicable.

This research received no external funding.

The authors declare no conflict of interest.