Discussion:
In this single-center experience, we demonstrated that neither physician adjudicated arrest rhythm (VT/VF versus PEA) nor underlying origin of the cardiac arrest (cardiac versus non-cardiac) were significantly associated with the primary outcome of survival to discharge following ECPR, either alone or in combination. Of the multiple baseline, peri-arrest, and on-circuit characteristics assessed, only younger age and arresting in the emergency department were significantly associated with survival following ECPR, and of the range of arrest etiologies described, only pulmonary embolism approached significance. We found that while ECPR therapy was associated with a high complication burden, with the majority of patients experiencing major bleeding or requiring de novo renal replacement therapy, of the patients who survived to discharge, nearly all experienced myocardial recovery and left the hospital with a good functional neurological status. Of the patients who did not survive, the most common cause of death was the decision to pursue comfort measures only in the setting of multiorgan failure.
The initiation of ECMO at the time of in-hospital cardiac arrest with active CPR remains an evolving treatment strategy without a robust evidence base to discern who benefits most from its deployment. Our study had the advantage of detailed physician review of cardiac arrest origin, etiology, and rhythm and highlights the heterogeneity in adjudicated causes of cardiac arrest preceding ECPR initiation. In doing so, it adds to the existing literature that underscores the salient challenge in predicting survival following ECPR: the complex and heterogeneous epidemiology of cardiac arrest as a whole.
The most parsimonious approach to predicting survival in ECPR may center on a clinical assessment of the reversibility of either the proximal cause of arrest or the clinical characteristics surrounding it. Historically, factors associated with survival in CCPR have included findings such as shockable arrest rhythm29 and cardiac origin of cardiac arrest1. We found no such relationship with survival in ECPR. This study adds to the growing body of literature that suggests these traditional risk stratifiers may not be reliable indicators of survival in ECPR and thus may not deserve a prominent role in the decision to provide extracorporeal support during CPR.
As the utilization of ECPR as a salvage therapy for patients with complete cardiopulmonary collapse expands, more work is needed to identify the patients for whom this resource intensive therapy with high morbidity and mortality would provide a durable benefit. Our results suggest that younger age, pulmonary embolism, and arresting in the emergency department may be associated with survival in ECPR. This does not imply that these should be entertained as possible initiation criteria, but rather that they should serve as hypothesis generating metrics for prospective evaluation of the potential for survival that this population represents.
Younger age may speak to the physiologic reserve required to tolerate not just the arrest event itself and attendant ischemic risks, but also the subsequent ECMO therapy and its high complication burden. Arrest location in the emergency department may be related to the heightened staff awareness regarding potential for decompensation, allowing for swifter response, though the intensive care unit will have as much or more monitoring potential. Therefore, it may also be that compared to patients who present and arrest in the emergency department, those who are already admitted and then arrest as an inpatient may have a poorer prognosis due to their relative comorbid state secondary to their initial, non-arrest presenting diagnosis. Nevertheless, the overall high functional status of those who survive to discharge, despite the high frequency of significant underlying conditions as described in Table 1, can justify the use of this therapy even in patients with multiple pre-existing medical problems.
This study has several important limitations. First, it was subject to limitations inherent in a retrospective, observation design. Second, our findings reflected a small set of ECPR patients at a single center, limiting its generalizability. Third, we were lacking what may arguably be the most important prognostic factors in ECPR: the time from arrest onset to cannulation, and the duration and quality of CPR provided prior to ECMO initiation. It may be that arrest origin, etiology, and rhythm gain prognostic significance when considered alongside these resuscitation metrics. For example, there is evidence that taking into account initial rhythm along with low flow time could help predict neurological outcomes1,31. Ultimately, the potential for ECMO as a rescue strategy may be limited at its onset by the extent of ischemic damage suffered prior to it its initiation. Future prospective studies of ECPR should focus on variables which might capture the depth and severity of ischemic insults sustained in the period after arrest but before ECMO deployment, especially those variables which have the potential to be easily incorporated into future advanced cardiovascular life support protocols. Finally, this study does not account for potential improvements in ECPR initiation and post-ECPR care over time, nor do we provide the total number of patients placed on venoarterial ECMO over the duration of the study period. The current registry also does not include cardiac arrest patients for whom ECMO support was considered but ultimately not provided. While emblematic of the challenges with ECPR studies, our findings should be viewed as primarily hypothesis generating and complimentary to the existing evidence base with similar limitations.