Hemodynamic variations and pitfalls during venoarterial extracorporeal membrane oxygenation and left ventricular apical unloading as bridge to heart transplantation

Abstract

Despite the exponential increase in venoarterial extracorporeal membrane oxygenation (VA-ECMO) use during the past decade, adult cardiac ECMO is still accompanied by a high mortality rate. Moreover, although left ventricular distension is now a well-known drawback of VA-ECMO, there seems to be great variability in the hemodynamic management strategies and in the results reported among the various centers. Hemodynamic management of VA-ECMO can be even more challenging when complex configurations are deployed. Here we present and discuss an interesting case of a modified VA-ECMO that although it occurred a few years ago it is instructive for its hemodynamic implications and pitfalls. VA-ECMO can either save the patient or catalyze the deterioration of a compromised clinical condition and thus a close multiparametric monitoring is mandatory especially with complex ECMO arrangements. A thorough understanding of the hemodynamic changes and problems that may occur during these cases is necessary too. Ultimately, critical thinking along with a proactive approach for early referral to more specialized centers and immediate unloading of the left ventricle whenever it is deemed necessary, together may contribute to reduce the relatively high mortality rate with this type of support.

<Learning objective: Venoarterial extracorporeal membrane oxygenation (VA-ECMO) can either save the patient or catalyze the deterioration of a compromised clinical condition if support-related drawbacks are not correctly identified and promptly adjusted. Management of complex VA-ECMO configurations can be challenging and thus a thorough understanding and close multiparametric monitoring of the hemodynamic implications and pitfalls are necessary in order to prevent negative outcomes.>

Introduction

Despite the exponential increase in venoarterial extracorporeal membrane oxygenation (VA-ECMO) use during the past decade, adult cardiac ECMO has still a high mortality rate (41–56%) due to device- and patient-related factors [1,2]. Besides, a recent survey showed great variability in the hemodynamic management strategies and in the results reported around the world [3]. Moreover, different VA-ECMO arrangements have different hemodynamic implications and pitfalls that may contribute to negative outcomes if not promptly identified and adjusted [4,5]. Here we focus on a particular femoro-femoral (ff) VA-ECMO configuration with left ventricular (LV) apical drainage and bridge to heart transplantation (HTx) that occurred a few years ago when a proactive LV unloading was unusual.

Case report

In 2013, a 35-year-old male was admitted to our hospital for acute decompensated heart failure. In 2008, he had undergone mechanical aortic valve (AoV) replacement of a bicuspid valve and mitral valve repair in dilated cardiomyopathy with reduced ejection fraction (EF 35–40%) and no coronary artery disease. In 2012, his functional status started deteriorating despite optimal medical therapy. Εchocardiography revealed a dilated (end-diastolic diameter, EDD 74–80 mm) and diffusely hypokinetic LV with depressed EF (20–25%), severe mitral and tricuspid valve regurgitation, right ventricular (RV) EDD 40 mm, tricuspid annular plane systolic excursion 15 mm, pulmonary artery systolic pressure 35–45 mmHg, and inferior vena cava 25–30 mm non-collapsible during inspiration. Right heart catheterizations confirmed post-capillary pulmonary hypertension, low cardiac output (2.5 l/min) and low cardiac index (1.45 l/min/m²). He had a pacemaker and implantable cardioverter defibrillator (ICD) (Medtronic Protecta, Medtronic, Minneapolis, MN, USA) fitted and started the evaluation for HTx. Coming to his last hospitalization in 2013, after an initial good response to intravenous diuretics he then started “sliding on inotropes” (INTERMACS profile 2) with pulmonary congestion, peripheral edema, and ascites. Dopamine and levosimendan were initially added followed by intra-aortic balloon pump (IABP) insertion, yet he remained hypotensive. Kidney and liver function were preserved with serum creatinine reaching 1.5 mg/dl, while total serum bilirubin, aspartate, and alanine transaminase reached 3 mg/dL, 45 IU/L, and 35 IU/L respectively. Lastly, adrenaline was added. Persistent hemodynamic instability, increasing lactate levels, and the patient now in INTERMACS profile 1, together dictated the need of VA-ECMO which was immediately undertaken with the patient conscious and on spontaneous breathing. We used a Maquet PLS Bioline Coating circuit (MAQUET Cardiopulmonary AG, Hirrlingen, Germany), a Maquet Quadrox oxygenator, an Edwards FemFlex (Edwards Lifesciences LLC, Irvine, CA, USA) 20 Fr femoral arterial cannula (FAC), an Edwards FemTrak 24 Fr femoral venous cannula (FVC) and a distal limb perfusion catheter. The National Transplant Organization was then alerted for an emergency transplantation.

The hemodynamic modifications registered during the support are summarized in Table 1 and Fig. 1.

Table 1.

Hemodynamic variations during VA-ECMO and LV venting with pressures and flows reported. Interesting spontaneous change in flows 36 h after the LV venting.

	Before ECMO	Beginning of VA-ECMO	6 h after LV venting	36 h after LV venting
mSAP (mmHg)	60	80	75	70
CVP (mmHg)	30	14	10	9
mPAP (mmHg)	35	25	21	20
PCWP (mmHg)	–	27	21	13
TPG (mmHg)	–	−2	0	+7
Pump flow (l/min)	–	5	–	–
FVC flow (l/min)	–	–	2	0,3
VENT flow (l/min)	–	–	2,3	4,5
FAC flow (l/min)	–	–	5	4,8

mSAP, mean systemic arterial pressure; CVP, central venous pressure; mPAP, mean pulmonary artery pressure; PCWP, pulmonary capillary wedge pressure; TPG, transpulmonary gradient; FVC, femoral venous cannula; VENT, venting cannula; FAC, femoral arterial cannula; VA-ECMO, venoarterial extracorporeal membrane oxygenation; LV, left ventricle.

Fig. 1.

Hemodynamic variations during ff-VA-ECMO and LV venting with pressures and flows reported. TPG changed gradually from phase 2 to phase 4, turning from negative to positive, with concomitant reduction in PCWP after venting the LV. We also noticed a spontaneous change in flows from phase 3 to phase 4 due to a competition between the two inflow cannulas (VENT and FVC) fortunately in this case without signs of RV failure. mSAP, mean systemic arterial pressure; CVP, central venous pressure; mPAP, mean pulmonary artery pressure; PCWP, pulmonary capillary wedge pressure; TPG, transpulmonary gradient; FVC, femoral venous cannula; VENT, venting cannula; FAC, femoral arterial cannula; VA-ECMO, venoarterial extracorporeal membrane oxygenation; LV, left ventricle; RV, right ventricle; LVAD, left ventricular assist device; HTx, heart transplantation.

After the ff-VA-ECMO was commenced, echocardiography revealed a well-drained RV, whereas the LV was distended (EDD 75 mm) and the AoV remained almost always closed with low systemic pulse pressure (PP). Inotropes and vasodilators (dobutamine, nitroglycerin, sodium nitroprusside) did not bring any significant modification and despite ECMO's oxygenator adjustments, gas-exchange started deteriorating with concomitant increase in lactate levels (maximum 2.67 mmol/L). Non-invasive ventilation was attempted but the patient did not tolerate it well.

Two days after the initiation of ECMO, after a sustained ventricular tachycardia (VT) successfully treated by the ICD, the patient eventually needed invasive mechanical ventilation. Chest radiograph (CXR) showed pulmonary edema. Finally, the LV pressure and volume overload with stasis in its cavity, the closed AoV, the arrhythmic event, the gas-exchange deterioration, and the radiologic data, indicated jointly, albeit late, an LV unloading.

At the time, a percutaneous venting (e.g. Impella Abiomed Inc., Danvers, MA, USA) was neither available in our institution nor would we have been able to use it due to the mechanical prosthesis. Considering the previous sternotomy, we opted for a technique described by Massetti et al. and we performed an LV apical drainage with a cannula (VENT) (12 mm Berlin Heart EXCOR apex cannula, Berlin Heart GmbH, Berlin, Germany) inserted through a left anterolateral minithoracotomy and connected to the ECMO's inflow with a Y-connector (Fig. 1) [6]. Although a complete apical venting would not have favored AoV opening and could have further increased the risk of valve clotting, at the time that was the only accessible solution. Moreover, it could have allowed to bridge later to a left ventricular assist device (LVAD) by removing the FVC and the oxygenator. We also applied two flow sensors before the Y-connector, on the VENT and FVC, and a third one on the FAC (Fig. 1).

After 12 h of venting, a CXR demonstrated partial regression of pulmonary edema. Interestingly, 24 h later (36 h of venting) and without any modification we noticed that flows had changed (Fig. 1). Initially, an insufficient venous drainage was supposed and the pump's revolutions-per-minute were increased, yet without attempting any LV-venting reduction due to concern for pulmonary edema. We witnessed immediately the inflow cannula chattering, a rapid decrease in all flows and in the systemic arterial pressure too. Position and size of the FVC were both correct, whereas thrombosis, accidental kinking, or clamping of the cannula were excluded. Volume status was adequate and tension pneumothorax and tamponade were ruled out. Hence, we returned to the previous pump velocity with immediate restoration of hemodynamic stability. Echocardiography, pulmonary PP, and pressure line tracing through pulmonary artery catheter (PAC), all demonstrated an increase in RV pulsatility and after careful reasoning the spontaneous flow modifications that had been registered were attributed to a competition between the two inflow cannulas. Joining both at the Y-connector, the lower impedance of the bigger VENT and the higher LV pressures favored the flow through the VENT and reduced that of the FVC, fortunately without any signs of RV failure. RV recovery was thus obtained with an “auto-weaning” from the right heart support. Pulmonary congestion further decreased and we decided to bridge to an LVAD but our center was informed about an available donor and the patient underwent directly an HTx. The post-operative course was uneventful. He was extubated on post-transplant day 3 and discharged 25 days later. Eight years later he is alive and in New York Heart Association class II.

Discussion

The negative impact of a distended LV during VA-ECMO is now unequivocal and thoroughly described [4,[7], [8], [9]]. Besides, a recent meta-analysis demonstrated that early LV venting enhances ECMO weaning and improves short-term survival [8]. However, a proactive approach apparently is mostly limited to experienced ECMO centers [3]. This might explain the high average mortality rate with this type of support and the different results reported from various institutions. Moreover, although the literature is richer now than in 2013, there has not been as yet sound evidence about when and by which means to drain the LV. There are different unloading techniques with different efficacy, invasiveness, duration, hemodynamics, pitfalls, and last but not least cost [4,5]. In the end, local availability and center's experience guide the choice of the unloading strategy.

The modified ff-VA-ECMO described here, can provide excellent biventricular support and transformed later into an LVAD. However, it bears some configuration-related pitfalls. Apical venting may hinder the opening of the AoV and along with the retrograde flow from the FAC they can increase the risk of clotting in the aortic root. In addition, as happened in our case, there may be a competition between the two inflow cannulas with consequent drainage/support reduction of one of the two ventricles. To overcome these pitfalls, we recommend a thoughtful titration of the VENT flow by using a clamp (e.g. Hoffman screw clamp) and based on echocardiographic, flowmeter, PAC, and CXR parameters. VENT flow should be adjusted on a case-by-case basis and according to the desired result (biventricular vs LV support/unloading, AoV opening, Harlequin syndrome prevention). Finally, surgical LV venting bears a bleeding and infection risk.

It is noteworthy that if we had vented the LV earlier, probably we would have prevented the development of VT, of pulmonary edema and the need for mechanical ventilation. Unfortunately, at the time a proactive LV unloading was unusual. Likewise, it was uncommon to adjust LV afterload by titrating ECMO's outflow, while hemodynamic management would rarely be based on systemic PP. Interestingly, although IABP is often considered an unloading technique, in our case it did not help. This simply confirms the fact that the response to an unloading solution is extremely variable and may depend on the contractile reserve of each ventricle [7].

Finally, although the mathematical sign and the absolute value of the transpulmonary gradient may seem irrelevant in ECMO, we believe that they may give an idea of both the direction and the amount of blood flow as well as of the resistance through the pulmonary vascular bed. Should this hypothesis be confirmed, it could be a useful adjunct to the management of VA-ECMO patients (Table 1) (Fig. 1).

Conclusions

In conclusion, VA-ECMO can either save the patient or catalyze the deterioration of a compromised clinical condition. Close multiparametric monitoring, especially with complex ECMO configurations, and thorough understanding of the hemodynamic changes and pitfalls are mandatory [10]. Ultimately, critical thinking along with a proactive approach for early referral to more specialized centers and immediate unloading of the LV whenever it is deemed necessary, together may contribute to reduce the relatively high mortality rate with this type of support.

Declaration of Competing Interest

The authors declare that there are no conflicts of interest.