Author + information
- Received February 10, 2016
- Revision received March 18, 2016
- Accepted March 26, 2016
- Published online September 1, 2016.
- Hug Aubin, MDa,
- George Petrov, MDa,
- Hannan Dalyanoglu, MDa,
- Diyar Saeed, MDa,
- Payam Akhyari, MDa,
- Gerrit Paprotny, Dipl.-Ing.a,
- Maximillian Richter, Cand. Med.a,
- Ralf Westenfeld, MDb,
- Hubert Schelzig, MDc,
- Malte Kelm, MDb,
- Detlef Kindgen-Milles, MDd,
- Artur Lichtenberg, MDa,∗ ( and )
- Alexander Albert, MDa
- aDepartment of Cardiovascular Surgery, Heinrich-Heine University Düsseldorf, Düsseldorf, Germany
- bDepartment of Cardiology, Pulmonology and Vascular Medicine, Heinrich-Heine University Düsseldorf, Düsseldorf, Germany
- cDepartment of Vascular and Endovascular Surgery, Heinrich-Heine University Düsseldorf, Düsseldorf, Germany
- dDepartment of Anesthesiology, Heinrich-Heine University Düsseldorf, Düsseldorf, Germany
- ↵∗Reprint requests and correspondence:
Dr. Artur Lichtenberg, Department of Cardiovascular Surgery, Heinrich Heine University Düsseldorf, Moorenstrasse 5, 40225 Düsseldorf, Germany.
Objectives This study sought to evaluate patient outcome within the Düsseldorf Extracorporeal Life Support (ECLS) Network, a suprainstitutional network for rapid-response remote ECLS and to define survival-based predictors.
Background Mobile venoarterial extracorporeal membrane oxygenation (vaECMO) used for ECLS has become a treatment option for a patient population with an otherwise fatal prognosis. However, outcome data remain scarce and institutional standards required to manage these patients are still poorly defined.
Methods This retrospective cohort study analyzes the outcome of 115 patients consecutively treated between July 2011 and October 2014 within the Düsseldorf ECLS Network due to refractory circulatory failure.
Results Of the 115 patients (56 ± 15 years of age, vaECMO initiation under cardiopulmonary resuscitation [CPR] 77%, CPR duration 45 [range 5 to 90] min), 50 patients (44%) survived to primary discharge and 38 patients (33%) were alive after a median follow-up of 1.5 years (95% confidence interval [CI]: 1.2 to 1.7). Thirty-seven (97%) of the long-term survivors showed a favorable neurological outcome. Risk factors associated with mortality during vaECMO were CPR duration (hazard ratio [HR]: 1.006; 95% CI: 1.00 to 1.01) and ischemic stroke (HR: 2.63; 95% CI: 1.52 to 4.56). Risk factors associated with mortality after vaECMO weaning were renal failure (HR: 6.60; 95% CI: 2.72 to 16.01) and sepsis (HR: 3.6; 95% CI: 1.50 to 8.69). Visceral ischemia had a negative impact (HR: 0.30; 95% CI: 0.11 to 0.84) whereas assist device implantation promoted successful vaECMO weaning (HR: 2.95; 95% CI: 1.65 to 5.25). Further, 3 distinct risk groups with significant differences in survival could be identified, demonstrating that in patients with no or short CPR mortality was not conditioned by age, whereas in patients with prolonged CPR young age was associated with increased survival.
Conclusions This study illustrates the implementation of a suprainstitutional ECLS Network. Further, our data suggest that mobile vaECMO is beneficial for a larger patient population than actually expected, especially regarding young patients presenting with prolonged CPR or patients regardless of age with no or short CPR.
- acute circulatory failure
- cardiogenic shock
- extracorporeal cardiopulmonary resuscitation
- extracorporeal life support
- extracorporeal membrane oxygenation
Venoarterial extracorporeal membrane oxygenation (vaECMO), also referred to as extracorporeal life support (ECLS), has become a valid treatment option for refractory circulatory failure due to a variety of etiologies (1–3), providing temporary circulatory assistance sustaining vital hemodynamics until completion of diagnostics, full recovery from underlying pathology, or implementation of further therapy.
In the past vaECMO therapy has been restricted to patients on site in selected tertiary care centers (4). Today, miniaturization of circulatory assistance systems and incremental clinical experience allow us to provide mobile vaECMO as a rescue therapy in out-of-center emergency situations. However, although registry-derived scores predicting survival for patients receiving vaECMO for refractory cardiogenic shock have already been established (5), decision to treat for out-of-center vaECMO rescue remains challenging, as it has to be taken on the spot by the providing care unit, usually with incomplete information about the patient and his or her condition. Adequate patient selection criteria for such out-of-center emergency situations, that reflect on short- and long-term survival and proportionate cost effectiveness, are still lacking. Further, minimal institutional standards that might be required to manage such patients with all its implications from vaECMO-related complications to complex follow-up therapy still need to be defined.
Because outcome data from randomized controlled prospective trials will rarely be available in the evaluation of last-resort emergency therapy concepts, we have to rely on cohort analysis to guide our patient selection strategies. Therefore, based on the up-to-date largest patient series, the aim of this study was to evaluate the suprainstitutional network for rapid-response remote vaECMO that was started at our institution in 2011 and to define survival-based predictors in order to identify those patients who did benefit from mobile vaECMO. Further, we hypothesized that in a well-defined setting mobile advanced circulatory support is beneficial for a larger patient population than actually expected, as exemplified by the 2 introductory real-life cases in which mobile vaECMO may seem counterintuitive due to prolonged cardiopulmonary resuscitation (CPR) or advanced age (Online Appendix).
The Düsseldorf ECLS Network is a suprainstitutional network for rapid-response remote ECLS based at the Heart Center of Heinrich Heine University in Düsseldorf (Germany). The ECLS Network was launched in July 2011 by the Department of Cardiovascular Surgery as a suprainstitutional effort to offer 24/7 advanced mobile vaECMO for patients with refractory circulatory failure not related to cardiac surgery at a regional level. Indications for mobile vaECMO are refractory circulatory failure nonmanageable on site or cardiac arrest under CPR. vaECMO is declined when life-limiting medical conditions or strong evidence for hypoxic brain damage after prolonged insufficient CPR are present. Decision to treat is undertaken by the cardiovascular surgeon on duty, immediately sending a trained mobile vaECMO team (cardiovascular surgeon and perfusionist) on site upon emergency call.
On site, mobile peripheral vaECMO system (Sorin Lifebox, Sorin Group, Munich, Germany) implantation is performed as previously described (6). After hemodynamic stabilization the patient is immediately transferred to the Department for Cardiovascular Surgery, where a standardized diagnostic and therapeutic algorithm is started by a specially trained interdisciplinary team, including implantation of a distal leg perfusion cannula, whole-body computer tomographic scan, transesophageal echocardiography, peripheral vessel Doppler sonography, cardiac catheterization, mobile vaECMO switch to a stationary system (SPPSC, Sorin Group), point-of-care blood coagulation management, and optimal left ventricular decompressing management. After completion of diagnostics, customized therapy is started depending on the underlying pathology. In cases where predicted neurological prognosis is favorable and vaECMO weaning is not successful or left ventricular decompression cannot be adequately achieved by other measures a permanent left ventricular assist device (VAD) (HeartWare International, Framingham, Massachusetts) is implanted either as bridge to recovery, bridge to transplant, or destination therapy, adhering to current recommendations (7). Patients are treated at our institution until internal or external referral for further recovery after completion of rescue therapy, hospital discharge after completion of follow-up therapy, or death.
This retrospective cohort study includes 115 patients not related to cardiac surgery consecutively treated with mobile vaECMO rescue therapy for refractory circulatory failure within the ECLS Network between July 2011 and October 2014. Refractory circulatory failure was defined as insufficiency to maintain sufficient mean arterial pressure (60 mm Hg) with evidence of end-organ hypoperfusion despite maximal medical therapy under exhaustion of all on-site available supportive measures and/or ongoing CPR, independent from underlying etiology. vaECMO rescue therapy was defined as mobile when an advanced logistic effort was required to reach the patient, only including patients into the study in whom vaECMO was implanted in facilities spatially separated from the Department of Cardiovascular Surgery. Patients in whom vaECMO implantation was declined on site or patients primarily treated with venovenous ECMO were not included in the study. Primary clinical endpoint was survival. CPR duration, age and disease etiology as well as complications and interventions under vaECMO were included in the analysis protocol to account for their potential impact on survival. Median follow-up was 1.5 years (95% confidence interval: 1.2 to 1.7 years; range 0.0 to 3.1 years). In cases where patients were not included in a follow-up routine at our center, follow-up was conducted by personally contacting the patient, relatives, and/or last attending physician. Additionally, a standardized survey was conducted with each patient, allowing for self-assessment of recovery as well as satisfaction about reintegration into work and/or social life. Complete absence of major physical or mental disabilities impeding return to work and/or social life (0 to 3 on the modified Rankin scale) (8) was defined as a favorable neurological outcome.
All data were collected from the ECLS Network registry in which all patients are prospectively registered immediately after vaECMO implantation. Registry data were complemented by an extensive review of the patients’ medical records until primary discharge, including standardized laboratory parameters (neuron-specific enolase, lactate dehydrogenase, aspartate transaminase, high-sensitivity troponin T, creatinine), as well as documentation on diagnostic exams, complication management and follow-up therapy; all data recorded according to the recommendation of the “Association for Clinical Data Management” (9) and institutional quality assurance standards. Treatment costs per patient were assessed by the reimbursement from medical insurance companies according to the German Diagnosis-Related Group system.
Ethics committee approval
This study was approved by the ethics committee of the medical faculty of the Heinrich Heine University, complying with the principles outlined in the Declaration of Helsinki.
This manuscript was prepared according to recommendations of the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) initiative (10). The statistical analysis was performed using IBM SPSS Statistics, version 22 (Armonk, New York), and R, version 2.15.2. Unless otherwise indicated continuous variables are presented as mean ± SD or median (25th to 75th percentile) according to normality of their distribution and compared using unpaired Student t test or Mann-Whitney test as appropriate. Discrete variables are reported as percentages and tested by Pearson’s chi-square test or, when validity conditions were not satisfied, by Fisher’s exact test. Survival was estimated using the Kaplan-Meier method. Comparisons of survival curves were based on the log-rank test. All tests were 2-tailed and p values <0.05 considered statistically significant. In case of multiple comparisons Bonferroni correction was applied.
The impact on the recovery potential from circulatory failure under vaECMO support and subsequently on survival of age, sex, body surface area, etiology for circulatory failure, CPR duration (per minute interval), complications and interventions was assessed by means of multistate Cox-Markov models (11) with the 3 states represented by vaECMO support, successful vaECMO weaning, and death, providing hazard ratio and the corresponding 95% confidence interval.
A decision tree model based on factors available at vaECMO treatment decision was used to stratify the cohort into distinct risk groups with respect to mortality until primary discharge. Chi-square automatic interaction detection was set up as the underlying analytical method. Generalizability was assessed by cross validation with 10 subsamples. Age, sex, body surface area, etiology for circulatory failure, distance to vaECMO implantation site, and duration of CPR were included as covariates in the decision tree model, but only duration of CPR (chi-square = 7.75; p < 0.043) and age (chi-square = 8.89; p < 0.026) remained as independent predictors of mortality. Based on model results 3 distinct risk groups based on following cutoff values were identified:
• no or short CPR (cutoff: CPR ≤45 min, age independent)
• prolonged CPR, adolescent or young adult (cutoff: CPR >45 min, ≤43 years of age)
• prolonged CPR, advanced age (cutoff: CPR >45 min, >43 years of age).
Cost effectiveness analysis was carried out using median relative differences in event-free survival and treatment costs, respectively (12). Significant differences between risk groups were established by 1-way analysis of variance and Tukey’s post hoc test.
Implementation of the mobile ECLS network
Launched in July 2011, today, the Düsseldorf ECLS Network covers a defined area from the medical campus of the Heinrich Heine University to regional hospitals in neighboring counties with 17 participating key centers and a total population of about 800,000 (Figure 1). This includes a core metropolitan area of 217 km2 with 600,000 habitants as well as additional participating centers in urban and sub-urban areas within a maximal radius of 40 km of our institution and additional 200.000 habitants. From July 2011 to October 2014 a total of 130 emergency requests for vaECMO support reached our institution. Out of those, 10 were declined on request and 5 on site. In total 115 patients were consecutively supported with mobile vaECMO on site in 1 of the participating centers and subsequently treated at our institution (Figure 2).
Study population: Baseline characteristics, clinical course, and outcome
Patient characteristics at vaECMO implantation and clinical course during treatment at our institution are outlined in Table 1 and Figure 2. Briefly, mean age at vaECMO implantation was 56 ± 15 years, vaECMO-implantation was performed under ongoing CPR in 77% of the patients, with median CPR duration of 45 (range: 5 to 90) min and median distance to vaECMO implantation site of 2.1 (range: 0.5 to 10) km. Primary etiology for refractory circulatory failure leading to vaECMO support was acute myocardial ischemia in 68% and deteriorated cardiomyopathy in 14% of the cases. In 12% the primary etiology was noncardiac. Median duration of vaECMO support was 4 (range: 2 to 7) days, with successful vaECMO weaning in 54% of the patients. Fifty patients (44%) survived until primary discharge from our institution. Of those patients, 39 showed full cardiac recovery, and 11 patients were discharged with a permanent VAD. Median treatment time was 10 (range: 2 to 26) days, with median treatment costs of 53.933€ (range: 28.859€ to 106.249€) per patient.
Kaplan-Meier survival of the study population is outlined in Figure 3A, reflecting high early-hospital mortality in the cohort and a favorable midterm survival after hospital discharge. Overall survival in the cohort was 33% after follow-up. Survivors were younger with significantly shorter CPR duration at vaECMO implantation. There was no difference between survivors and nonsurvivors in the median distance to the remote vaECMO implantation site (Table 1); 97% of the long-term survivors showed a favorable neurological outcome, with a high-degree of satisfaction concerning reintegration to work and/or social life (71%) (Figure 2).
Determinants of successful vaECMO weaning and mortality during vaECMO rescue therapy
As identified by multistate Cox-Markov model analysis, risk factors associated with mortality during vaECMO support were CPR duration at vaECMO implantation and presence of ischemic stroke. Permanent VAD implantation had a positive effect, whereas renal failure requiring dialysis and visceral ischemia had negative impact on vaECMO weaning possibilities (Figures 4A and 4B). Further, renal failure requiring dialysis and sepsis were associated with increased mortality after successful vaECMO weaning (Figure 4C).
Impact of CPR duration and age on mortality
Baseline characteristics at vaECMO implantation and clinical course at our institution of individual risk groups as identified by decision tree model are summarized in Table 2. There were significant differences in overall mortality between groups: 89% mortality in the prolonged CPR advanced age versus 54% in the no or short CPR and 50% in the prolonged CPR adolescent or young adult group. Kaplan-Meier survival of the risk groups is outlined in Figure 3B, demonstrating comparable in-hospital mortality and midterm survival for the no or short CPR and the prolonged CPR, adolescent or young adult group, as well as a survival benefit of both groups as compared to the prolonged CPR advanced age group. Thus, in patients with no or short CPR mortality was not conditioned by age, whereas in patients with prolonged CPR young age was associated with increased survival.
Cost-effectiveness analysis, comparing survival rates and treatment costs, showed that the survival benefit of the no or short CPR as compared to the prolonged CPR advanced age group was associated with a significant increase in treatment costs. In contrast, the survival benefit of the prolonged CPR adolescent or young adult over the prolonged CPR advanced age group did not translate into a significant increase in treatment costs (Figure 5).
Here we describe the implementation of a suprainstitutional network for rapid-response remote ECLS. In the up-to-date largest series of mobile vaECMO for refractory circulatory failure, covering an area of about 800,000 habitants with 17 participating centers, we observed an overall survival to primary discharge of 44% in a complex and severely compromised patient cohort. Most importantly, the vast majority of long-term survivors showed a favorable neurological outcome.
The feasibility of mobile vaECMO programs providing advanced circulatory support has been demonstrated in small cohorts, though outcome data remain scare. Reported in-hospital survival rates vary between 15% and 53% (13–18), depending on patient and program characteristics. Those encouraging survival rates for patients, who otherwise most likely would have died, indicate the potential of mobile vaECMO However, as this form of highly specialized therapy requires complex logistic organization as well as constant availability of trained medical staff and resources, its application remains limited to selected tertiary-care centers. Nonetheless, as exemplified by the Düsseldorf ECLS Network, suprainstitutional networks can help to provide such highly-specialized patient-centered care within defined metropolitan areas.
Neurological outcome after extracorporeal CPR for out-of-hospital cardiac arrest has been reported to be highly satisfactory. In a review by Morimura et al. (19) 85.7% of the survivors showed complete recovery at hospital discharge, with only 5.2% presenting with mild disability, 3.9% with severe disability, and 5.2% with a vegetative state, which compares to the observed outcome of survivors within the ECLS Network, with 97% of the survivors showing a favorable neurological outcome while expressing high satisfaction about reintegration to work and/or social life after hospital discharge. This does justify the immense resource and personnel burden as well as treatment costs of the program, averaging median 53,933€ per patient and increasing to median 68,981€ for each survivor.
Time to initiation of vaECMO has been shown to be an independent prognostic factor for circulatory rescue therapy (15). Therefore, operating within a defined region in which participating institutions can be reached within a justifiable amount of time may constitute a logistic advantage of the ECLS Network over comparable programs supplying oversized metropolitan areas with more complex logistic challenges (14). Efficient logistic organization within a defined regional setting probably conditioned the observed survival rates in patients with ongoing CPR in our study cohort. This is emphasized by the fact that the distance to the vaECMO implantation site did not reach statistical significance in our model, in contrast to CPR duration, which, however, will always be conditioned by time to initiation of vaECMO in patients under ongoing CPR. Though, whether immediate vaECMO therapy in a pre-hospital or nonspecialized center setting (20,21) will further increase survival remains questionable, as nonspecialized and less experienced teams may not be able to cope with peri- and post-procedural complications.
Complications during vaECMO implementation and therapy are complex and can be life threatening if not handled properly (22). In our patient cohort 49% presented with renal failure requiring dialysis, 33% with ischemic stroke, 15% with sepsis, and 15% with visceral ischemia, all conditions identified to be independent predictors of vaECMO therapy failure and/or mortality. Regardless if the occurrence of such events is secondary to the protracted shock or to the vaECMO rescue therapy itself, optimal supportive therapy is crucial, as it directly influences survival.
One of the major problems of prolonged vaECMO therapy is adequate decompression of the failing left ventricle, which may lead to fatal backward failure with pulmonary edema. Therefore, optimal left ventricular decompressing management is essential. Here, our management included enhanced inotropic support, concomitant intra-aortic balloon pump implantation, and early intervention (percutaneous coronary intervention, coronary artery bypass grafting, or valve surgery) whenever possible. In cases of unclear neurological prognosis temporary Impella (Abiomed, Danvers, Massachusetts) implantation or surgical left ventricular venting were used as additional measures. As soon as the neurological condition could be assessed to have a favorable prognosis, early permanent VAD implantation was performed. In the past there has been some reluctance regarding long-term mechanical support in cases of cardiogenic shock (23,24). However, as just recently shown, severely compromised patients in whom vaECMO was used as a “bridge-to-bridge” strategy before long-term VAD therapy, had no survival disadvantage to patients who could directly be supported by VAD therapy (25), which is consistent with the results of this study. In our cohort, 11 patients were discharged with a permanent VAD. In 3 of those patients the VAD could later be explanted after cardiac recovery, whereas VAD therapy in the remaining 8 patients became bridge-to-transplant therapy. One of those patients underwent heart transplantation during and additional 3 patients outside the study period. Therefore, a more rigorous examination toward VAD indication and early implantation may further increase survival in this population (26).
Nonetheless, despite optimal supportive therapy, in this kind of patient cohort mortality will be conditioned to a large extent by the severity of patient condition at initiation of vaECMO rescue therapy. Therefore, adequate patient selection becomes of utmost importance, as vaECMO treatment requires great amounts of medical resources and goes along with high treatment costs even in case of therapy failure, averaging median 39,905€ per nonsurvivor in our study. Despite registry-derived scores predicting survival for patients receiving vaECMO for refractory cardiogenic shock having already been established (5), they are based on a large subset of physiologic data, which however is unavailable in emergency situations. Usually, mobile vaECMO decision to treat has to be taken on the spot by the providing care unit based on information provided within the emergency call by the referring institution. In the setting of the Düsseldorf ECLS Network, patient age and duration of CPR at vaECMO initiation were the decisive predictors of survival. As those parameters can be directly estimated during the emergency request, this might help in real-life decision making.
Previous reports have concluded that patients after prolonged CPR or with advanced age might not benefit from vaECMO (14,27). However, in our cohort overall mortality in adolescent or young adults with prolonged CPR was no different to patients with no or short CPR. Additionally, the no or short CPR group had an above-average survival, despite an above-average age. This emphasizes the importance of evaluating CPR in the context of age and vice versa. In a recent study analyzing a small, relatively young (52 ± 13 years of age) patient cohort that had received vaECMO for refractory cardiogenic shock Guenther et al. found that neither CPR nor implantation under CPR constituted differences in mortality, however without reporting about the CPR duration (16). Additionally, Mendiratta et al. (28) recently showed that there is no increase in risk for in-hospital mortality after cardiopulmonary failure requiring vaECMO among elderly patients 65 to 75 years of age versus those >75 years of age. Therefore, as shown by this study, in an adequate setting young patients under prolonged CPR and patients regardless of age with no or short CPR still will benefit from mobile vaECMO. This is further exemplified by the reported outcome of the introductory real-life cases in which mobile vaECMO may have seem counterintuitive due to prolonged CPR or advanced age (Online Appendix).
Although this is the up-to-date largest cohort of patients treated with mobile vaECMO, this study is still single center–based with limited data on neurological recovery and quality of life. Additionally, due to the retrospective design there are methodological limitations as compared to randomized trials, which will not be fully eradicated by the employed statistical methodology. Further multicentric studies are warranted to replicate the results in a variety of other settings, in particular regarding patient selection and technical aspects of the program.
In summary, this study illustrates the implementation of a suprainstitutional network for rapid-response remote ECLS for therapy of refractory circulatory failure within a defined regional setting, which might serve as a model for comparable settings. Further, our data suggest that mobile vaECMO is beneficial for a larger patient population than actually expected from available outcome data, especially when individually tailored to young patients presenting with prolonged CPR or patients regardless of age with no or short CPR.
COMPETENCY IN MEDICAL KNOWLEDGE: Mobile vaECMO has become a valid treatment option for patients presenting with therapy refractory circulatory failure providing temporary circulatory assistance sustaining vital hemodynamics until completion of diagnostics, full recovery from underlying pathology, or implementation of further therapy.
TRANSLATIONAL OUTLOOK: Further multicentric studies are warranted to replicate the results in a variety of other settings, in particular regarding patient selection and technical aspects of the program.
For an expanded Methods section, please see the online version of this article.
The authors have reported that they have no relationships relevant to the contents of this paper to disclose. Drs. Lichtenberg and Albert contributed equally to this work.
- Abbreviations and Acronyms
- cardiopulmonary resuscitation
- extracorporeal life support
- venoarterial extracorporeal membrane oxygenation
- ventricular assist device
- Received February 10, 2016.
- Revision received March 18, 2016.
- Accepted March 26, 2016.
- American College of Cardiology Foundation
- Schmidt M.,
- Burrell A.,
- Roberts L.,
- et al.
- Yancy C.W.,
- Jessup M.,
- Bozkurt B.,
- et al.
- Wilson J.L.,
- Hareendran A.,
- Grant M.,
- et al.
- ↵Association for Clinical Data Management. ACDM Guidelines to Facilitate Production of a Data Handling Protocol. 2013. Available at: http://www.acdm.org.uk/files/pubs/DHP%20Guidelines.doc. Accessed February 1, 2016.
- Von Elm E.,
- Altman D.G.,
- Egger M.,
- et al.
- Beurtheret S.,
- Mordant P.,
- Paoletti X.,
- et al.
- Guenther S.,
- Theiss H.D.,
- Fischer M.,
- et al.
- Lebreton G.,
- Pozzi M.,
- Mastroianni C.,
- Léger P.,
- Pavie A.,
- Leprince P.
- Demondion P.,
- Fournel L.,
- Golmard J.L.,
- Niculescu M.,
- Pavie A.,
- Leprince P.