Identifying Candidates for Advanced Hemodynamic Support after Cardiac Arrest

Out-of-hospital cardiac arrest has a high incidence with approximately 400,000 cases in the United States each year.¹ Between 20 and 40 percent of these patients achieve return of spontaneous circulation (ROSC) resulting in hospital admission.^{2, 3} While ROSC is strongly associated with survival, only 25–40% of admitted patients survive to hospital discharge. The most common cause of death for patients admitted after cardiac arrest is neurologic injury. However, shock accounts for most deaths within the first three days⁴. Given these competing risks, the selection of candidates for advanced hemodynamic therapies should consider both neurologic and hemodynamic prognosis.

While the timing of coronary angiography following cardiac arrest without signs of STEMI remains controversial, there has been a dramatic rise in the use of mechanical circulatory support which can be deployed quickly and safely in the cardiac catheterization laboratory to support patients suffering cardiac arrest. These devices include extra-corporeal membrane oxygenation (ECMO), Impella, and Tandem Heart. Multiple small observational studies have demonstrated that post-cardiac arrest shock can be effectively treated with these devices.^{5, 6} However, the logistical and financial burdens of these advanced therapies, risk of complications to the patient and the absence of definitive randomized trials have led some clinicians to a nihilistic resistance to the use of these support technologies in patients after cardiac arrest. Wider adoption of hemodynamic support technologies will require accurate identification of neurologically viable patients likely to develop shock such that they would benefit from advanced hemodynamic support. To date, no such identification system exists.

In this issue of Circulation, Bascom et al describe the development and validation of a simple scoring system for the prediction of refractory hemodynamic compromise for patients who survived to ICU admission after out-of-hospital cardiac arrest.⁷ The CREST score is comprised of the following factors: 1) known history of coronary artery disease, 2) non-shockable rhythm upon presentation to EMS, 3) left ventricular ejection fraction less than 30% on admission, 4) shock at the time of admission to the intensive care unit (including systolic blood pressure less than 90 despite fluids, inotropes, pressors and/or need for an intra-aortic balloon pump), and 5) ischemic time greater than 25 minutes (from time arrest to ROSC). The primary outcome was a composite of death from repeat cardiopulmonary arrest, progressive refractory shock, refractory arrhythmia, or progressive lactic acidosis and multi-organ system failure as determined by the treating physician.

The CREST score attempts to predict refractory cardiovascular compromise in patients treated with current standards of practice. It is intended as an adjunct to neurologic assessment to fully assess the patient’s risk for the two primary causes of death after cardiac arrest. Higher CREST scores were associated with increased risk for the primary outcome of refractory cardiovascular compromise (50% risk for those with a score of 5). While the primary outcome occurred in 19% of the derivation and validation cohorts evaluated by Bascom et al, multiple studies have shown improved survival when more advanced hemodynamic support options are utilized.⁶ Therefore, the CREST score may be best used as an impetus to consider more aggressive hemodynamic support. Importantly, however, this hypothesis was not tested in the current study.

Ideally, the decision to escalate care would be made in concert with neurologic prognostication such that only neurologically viable patients would be considered for advanced support. However, rapid neurologic prognostication is difficult, particularly early in the patient’s course. The American Heart Association 2015 guidelines for post-cardiac arrest care support use of neurological exam findings no earlier than 72 hours after return to normothermia with a typical time for prognostication at 4–5 days.⁸ Brain electroencephalograms are similarly best assessed at 72 hours after return to normothermia. Brain imaging, including marked reduction of gray white ratio (GWR) on CT performed within 2 hours of the cardiac arrest or restriction of diffusion on MRI performed between days 2 and 6 after the cardiac arrest, may also be used to predict a poor neurologic outcome. Bilateral absence of the N20 somatosensory evoked potential may be considered a predictor of poor outcome at 24 to 72 hours after return to normothermia. Unfortunately, no single test is accurate for neurological prognostication, and results are best assessed days into the hospital course. Therefore, the ultimate objective for advanced mechanical support programs - accurately identifying patients that are both neurologically viable and at risk of cardiovascular collapse - is not achievable at this time and requires further study and innovation.

The CREST score does not include patients with ST-elevation myocardial infarction on EKG as the recommended treatment pathway is more established in this population. Interestingly, these patients tend to have high survival rates (>70%).⁹ Thus, excluding these patients is likely to provide a more accurate assessment of risk for the primary outcome by avoiding the dilutional effects of these more survivable patients on the score. It may be interesting to apply a prognostic score to this population when considering advanced support, but the authors instead focus on patients without ST elevation in an effort to focus on patients whose treatment pathway is less clear. Patients who suffered an in-hospital cardiac arrest were also excluded, but would likely confound the results due to the presence of multiple comorbid conditions. Patients with ongoing CPR and refractory cardiac arrest were also excluded from this data set. While this population could technically be categorized as having shock, their expected mortality is substantially higher than the group included in this study. Even in this high risk group, a significant benefit with advanced hemodynamic support may exist.^{3, 6} Therefore, patients with ongoing CPR should be included in future prognostic scores if supported with advanced mechanical support.⁶

The CREST score predicts refractory hemodynamic compromise upon admission to the intensive care unit based on the data available in the International Cardiac Arrest Registry. Patients surviving to the emergency department but not to the intensive care unit were excluded from this study. Future versions of the CREST score will hopefully address patients in the emergency department where the CREST criteria could easily be implemented including completion of a bedside echocardiogram. This is a time-critical period for advanced mechanical hemodynamic support as the benefits of initiating this support is higher when delivered as early as possible. In addition, the logistic and patient safety issues associated with transport to the cardiac catheterization laboratory and the intensive care unit is best addressed with a cohesive plan established early in the emergency department. Other unstudied variables such as lactic acid, a marker of accrued hypoperfusion,^{10, 11} or a clinical history of end-stage renal failure requiring hemodialysis¹² may also be readily accessible in the emergency department and may further enhance the prognostic capabilities of the score.

We should be cautious generalizing the CREST score to all patients that suffer out-of-hospital cardiac arrest given our evolving understanding of resuscitation outcomes. Although patients presenting with shockable rhythms are the minority of all OHCA (25–30%) they represent more than 85% of all survivors, likely due to a high prevalence of reversible causes of arrest^{6, 13, 14}. Acute contractile dysfunction, present for the first 3–5 days after prolonged CPR and resuscitation, appears to be mostly reversible. Early implementation of ECMO to support patients through those 3–5 days may lead to near complete recovery of LV function¹¹. Thus, despite the resistance of clinicians to more invasive life support, post-cardiac arrest myocardial dysfunction appears treatable with hemodynamic and pulmonary support during post arrest shock and acute lung injury^{6, 11}. Given the dramatic differences in survival rates between shockable and non-shockable rhythms, these groups should be differentiated in an effort to predict and identify viable patients¹⁵.

In conclusion, we congratulate Bascom et al for this important effort to provide prognostication for cardiovascular death in post resuscitation patients using the current standards of care. As we accrue data regarding the role of early implementation of new technologies to facilitate recovery for patients suffering from cardiac arrest, development of accurate prognostic tools to differentiate the risks of poor neurologic or hemodynamic outcomes may be the most important task in the near future. The field of cardiopulmonary resuscitation is facing rapid evolution with newly defined capabilities. However, many challenges lie ahead to appropriately provide these capabilities to improve outcomes, minimize complications, limit cost, and simplify implementation.