Epinephrine, inodilator, or no inotrope in venoarterial extracorporeal membrane oxygenation implantation: a single-center experience

The study includes all adult patients after VA-ECMO implantation due to cardiogenic shock or extracorporeal cardiopulmonary resuscitation (eCPR). All data were collected retrospectively from a tertiary referral university hospital between October 2010 and December 2018. Patients after non-traumatic out-of-hospital cardiac arrest (OHCA) as well as patients after non-postoperative intra-hospital cardiac arrest (IHCA) were included. Patients who died within the first 24 h after cannulation were excluded in this research. For OHCA, eCPR cannulation was performed in-hospital after transport with ongoing manual CPR or using a mechanical chest compression device (LUCAS2, Physio Control, Neuss, Germany). The decision to cannulate was driven by team decision including at least one ECMO specialist (intensivist or cardiologist) at the bedside. Implantation was performed either in the catheterization laboratory, in the emergency room, or in the intensive care unit. By local standard, patients after OHCA with shockable primary rhythm, ST-segment elevation myocardial infarction (STEMI), or other clinical signs like chest pain before collapse (indicating a cardiac cause for collapse) are routed directly in the catheterization laboratory. Patients without non-shockable rhythm were routed to the emergency room. After VA-ECMO implantation, further diagnostic steps, including a CT scan in most patients, were directed by the responsible intensivists following current guidelines written for patients with return of spontaneous circulation (ROSC) without ECMO.

The VA-ECMO therapy was initiated in patients with prolonged ongoing cardiopulmonary resuscitation without return of spontaneous circulation or in patients with severe therapy-resistant shock as indicated by the ELSO guidelines for adult cardiac failure. The indication for ECLS was stratified by the underlying disease. Cannulation for VA-ECMO was performed predominantly bi-femoral in Seldinger’s technique without primary surgical cut down by two experienced intensivists and one perfusionist. Typical venous (draining) cannulas were 21–23 Fr (French = Charrière) in diameter while arterial (returning) cannulas were 15–17 Fr. All components of the extracorporeal oxygenation system were coated with heparin. For patients without life-threatening bleeding, anticoagulation was provided by intravenous unfractionated heparin aiming at a partial thromboplastin time of 50–60 s. Mechanical ventilation was reduced during ECMO support. Peak airway pressures were aimed below 25 cmH₂O; respiratory tidal volumes were adjusted between 4 and 6 ml/kg optimal body weight aiming at an oxygen partial pressures of 60–80 mmHg and a fractional carbon partial pressures of 35–45 mmHg as described earlier [22].

The management of vasopressors (and fluid therapy) was driven by clinical judgment of the ECMO-experienced intensivist in charge. An indication for administration of inotropic agents was to secure LV ejection thereby decreasing the risk of intra-cardial stasis. While treatment of VA-ECMO patients is strongly guided by standard operation procedures at our institution, no recommendation on positive inotropic therapy could be made. The decision to administer dobutamine, epinephrine, levosimendan, or a combination of these was driven by clinical judgment and decision-making of the experienced intensivist in charge. All patients in the levosimendan group received levosimendan 12.5 mg in 500 ml glucose 5%, given as a continuous infusion over less than 24 h without an initial bolus according to the local standard protocol.

Data presented derives from a single-center retrospective registry analysis and was blinded to patient identity and covered by an ethics approval (Ethics Committee of Albert-Ludwigs University of Freiburg, file numbers 525/17 and 151/14). For data analysis, SPSS (version 23, IBM Statistics), Prism (version 5, GraphPad), and Stata (version 15.1, StataCorp) were employed. For statistical analysis, unpaired t test, Fisher’s exact test, Gray test, and Wald test were used as applicable, and a p value of ≤ 0.05 was considered statistically significant. Data are given as [mean ± standard deviation] or [odds ratio (OR), 95% confidence interval (CI)] if not stated otherwise.

Groups were formed according to inotropic therapy given within the first 24 h after cannulation for VA-ECMO. All patients with continuous epinephrine infusion (with or without dobutamine or levosimendan) were grouped in the “epinephrine group” (group C). All patients without epinephrine but with either dobutamine, levosimendan, or a combination of both were grouped in the “dobutamine/levosimendan or inodilator group” (group B). Patients without epinephrine, dobutamine, or levosimendan within the first 24 h were grouped in the “no inotropy group” (group A). Patients were excluded when survival was below 24 h and when patients could not be reliably stratified into one of the three groups. As for 30-day survival, all patients dismissed from our hospital alive before reaching 30 days of hospitalization were considered 30-day survivors. Mode of death has been categorized according to Witten et al. [23]. Propensity score matching was performed using SPSS with a nearest neighbor matching algorithm and a caliper of 0.1. Matching was performed for predictors of hospital survival available during the first 24 h (as detected by the multivariate logistic regression analysis of hospital survival in the whole cohort (age, eCPR as indication for VA-ECMO, and gender, see Fig. 4) as well as for known predictors of survival (a shockable first rhythm and lactate levels 24 h after cannulation). All factors were known at the time of treatment and might have influenced the physician’s decision to treat a patient with different inotropic agents. Cumulative incidence curves were calculated using competing risk regression according to the Fine and Gray method [24] with discharge alive from the hospital as a competing event.

Between October 2010 and December 2018, 332 patients were treated with a VA-ECMO due to a cardiogenic shock or resuscitation without return of spontaneous circulation. After exclusion of 101 patients, a total of 231 patients were evaluated in this research (for reasons of exclusion, see Fig. 1). Patients included were at a median age of 58.58 ± 14.27 years, and 29.9% were female. Reasons for VA-ECMO implantations were extracorporeal cardiopulmonary resuscitation (eCPR) in 58% or cardiogenic shock (mostly due to STEMI or NSTEMI in 25.9% and 25.5%, respectively). Patient characteristics are given in Table 1 and in Additional file 1.

Average VA-ECMO blood flow was similar between the groups (p = 0.240, Table 1). The 30-day survival was evaluated in different subgroups stratified by early inotropic therapy used. When considering no early inotropic therapy as baseline, there was no significant difference between patients with dobutamine, levosimendan, or the combination of both and patients without early inotropic therapy. The worst outcome was detected in patients treated with epinephrine with or without a combination of dobutamine or levosimendan (Fig. 2a). This difference in survival was confirmed by the cumulative incidence of hospital death curve as given in Fig. 3. When evaluating the concomitant vasopressor therapy, there was a significant difference in norepinephrine dose between the groups with the highest vasopressor doses being detected in the epinephrine group (Fig. 2b).

Factors that were associated with survival to 30 days were tested in a multivariate logistic regression analysis. In the whole cohort, implantation during eCPR, age, and epinephrine use were significant and independent predictors of poor outcome as given in Fig. 4a. After propensity score matching for items given in the “Methods” section, we were able to match 49 patients in the epinephrine group (group C) with 49 patients without epinephrine. In a multivariate logistic regression analysis of the matched cohort, age and epinephrine were strong independent predictors of survival alongside female gender and lactate, as demonstrated in Fig. 4b.