Can we predict patient outcome before extracorporeal membrane oxygenation for refractory cardiac arrest?

Extracorporeal cardiopulmonary resuscitation (E-CPR) is being increasingly used as an effective supportive tool for prolonged cardiac arrest refractory to conventional cardiopulmonary resuscitation (C-CPR) [1‐3]. E-CPR implementation is a complex procedure that requires extensive teamwork for successful cardiopulmonary life support [4, 5]. The procedure can have major complications, such as massive bleeding and brain injury [6, 7], and it has a high financial burden [4, 8]. Despite increased E-CPR use for prolonged cardiac arrest, the survival rate varies from 8.8% [9] to 54% [10], depending on candidate selection, and the overall survival to discharge is 29% [11].

Recent guidelines from the American Heart Association recommend E-CPR use as a class IIb recommendation to improve survival in patients with potentially reversible refractoriness to C-CPR [12]. Although E-CPR may help bridge patients to diagnosis or therapy, it may also bridge patients to nowhere, raising ethical concerns [13]. Because E-CPR is usually performed in the emergency setting when the patient is unconscious, it is difficult to conduct large-scale studies to compare E-CPR and C-CPR results in the real world [14]. Data regarding appropriate patient selection criteria for E-CPR from randomized comparative trials are lacking [15]. However, observational clinical studies aimed at identifying those who may least benefit from such advanced techniques are needed to prevent futile extracorporeal membrane oxygenation (ECMO) use [16, 17]. For example, the Survival After Veno-arterial ECMO (SAVE) score was developed to predict patient survival following cardiogenic shock using clinical parameters [18, 19], and the ENCOURAGE mortality risk score from a French study group was proposed to predict mortality in ECMO-treated patients with acute myocardial infarction [20]. Laboratory data and etiology of cardiac arrest are often unknown prior to E-CPR use. Using such prognostic scores is complex and requires clinical information that is typically only available after ECMO use. Prognostic factors, such as low-flow time and cardiac rhythm, were found to be related to survival and neurological outcome in patients with in-hospital cardiac arrest (IHCA) or out-of-hospital cardiac arrest (OHCA) who received E-CPR [21‐23]. Such prognostic factors have not been combined to predict survival in patients with refractory cardiac arrest receiving E-CPR. In our previous studies, E-CPR was shown to improve patient survival and neurological outcome [3]. In our experience, traditionally poor prognostic factors, such as low-flow time as long as 250 min [1] or nonshockable cardiac rhythm [24], can be appropriately managed with E-CPR to achieve good neurological outcome. In the present study, we aimed to review the clinical outcomes of patients in our institution with refractory cardiac arrest receiving E-CPR and identify potentially useful clinical parameters before ECMO therapy initiation to predict patient outcome in this challenging condition.

In this study, we report that patients with refractory cardiac arrest had an overall survival rate of 40.2, and 30.4% of patients were discharged with good neurological function after E-CPR use. These findings are consistent with those of previous studies of E-CPR for prolonged cardiac arrest [21‐23, 25]. We attempted to stratify our patients into three clinical risk groups according to three clinical parameters (cardiac arrest location, low-flow time, and initial cardiac rhythm) before initiating E-CPR. A high rate (65.8%) and a low rate (0%) of survival to discharge can be discriminated with this clinical classification system. This is the first attempt to use pre-ECMO parameters to predict patient outcome in cases of refractory cardiac arrest. This finding may aid in shared decision-making regarding ECMO use when prolonged cardiac massage occurs in daily practice.

The prognosis for patients with prolonged cardiac arrest is ominous when they are refractory to C-CPR [26], with survival rates of less than 1%. E-CPR has the ability to restore blood flow to prevent metabolic dysfunction in the highly lethal situation of prolonged cardiac arrest [26, 27]. Previous investigations have shown that E-CPR can improve the clinical outcome in patients with IHCA beyond 10 min [19, 28, 29]. Observational studies have also found that early E-CPR use may help to improve the neurological outcome in patients with prolonged cardiac arrest [3, 5, 24, 30]. Early studies found a more favorable outcome of E-CPR among IHCA patients than among OHCA patients [7, 22, 25] because of a shorter interval from cardiac arrest to active ECMO support among IHCA patients. A subsequent study showed that the clinical outcome of E-CPR can be similar between IHCA and OHCA patients with appropriate patient selection [31]. One strength of the present study is that we identified a unique type of cardiac arrest, EDCA. In this series with appropriate patient selection, we found no differences in clinical outcome between the IHCA and OHCA groups. Although E-CPR use in emergency departments was limited in a recent survey in the United States [32], we found that the EDCA group had better clinical outcome than the IHCA and OHCA groups. When patients present to the emergency department with subsequent development of cardiac arrest, these patients are more likely to receive prompt and high-quality CPR than other patients [33]. Even with prolonged cardiac arrest, these patients are most likely to have good prognostic factors and deserve aggressive E-CPR initiation in such a situation [3, 4].

The time interval from CPR initiation to ECMO is correlated with patient outcome. Previous studies showed that low-flow time (from CPR to ECMO) was strongly correlated with survival rate and neurological outcome [34, 35]. In the present study, the low-flow time was longer in nonsurvivors than in survivors (50 vs. 38 min, p < 0.001). Thus, earlier identification of patients with refractory cardiac arrest who are suited for E-CPR treatment may shorten the low-flow time to improve patient outcome [29, 36, 37].

Patient age is an important consideration when initiating E-CPR, taking into account the cost and benefit. Age is an important, but not the only, factor predicting survival after E-CPR for prolonged cardiac arrest [17]. In our study, compared with nonsurvivors, survivors had a higher proportion in the younger age group (< 55 years) (p = 0.018). In some studies, E-CPR use was excluded for patients over 60 years of age [4, 32]. We suggest that such younger-old patients should not be excluded from E-CPR use [38]. We consider 75 years of age an exclusion criterion for E-CPR use in our institution, similar to most other authors [17, 32, 38, 39], because of the low rate of benefits from E-CPR in such old-old patients.

Another important finding of our study is that nonshockable cardiac rhythm should not be the sole contraindication to E-CPR use for prolonged cardiac arrest. Many E-CPR programs exclude patients with nonshockable rhythms due to very limited outcomes after C-CPR [4, 40]. We found that compared with shockable cardiac arrest, nonshockable cardiac arrest had a worse prognosis. However, 10 of 49 patients (20.4%) presenting with nonshockable rhythms survived to discharge with E-CPR use. Our observations, along with those of other recent reports [3, 41, 42], suggest that we need to consider clinical parameters other than rhythm alone as contraindications to E-CPR use.

The current study is novel because it builds the TLR score (low-flow Time, cardiac arrest Location, and initial cardiac arrest Rhythm) with three pre-ECMO factors to stratify our patient outcome after receiving E-CPR with internal validation by bootstrap simulations. As previously mentioned, patient outcome with E-CPR is affected by a complex interplay of multiple factors [17, 21‐23, 43]. Reported predictors of survival and neurological function include age, initial cardiac rhythm, low-flow time, arterial pH value, and serum lactate level [16, 21, 43]. These variable clinical predictors reveal not only the underlying patient characteristics but also the quality of the resuscitation process. For example, initial serum lactate is an indicator of tissue hypoxia [26, 44], but dynamic changes in lactate level also reflect the effect of ECMO therapy [16, 45]. In our study, EDCA was one factor in patient outcome, and such clinical parameters to predict patient outcome are important during E-CPR.

We identified a high-risk group of patients, i.e., those who had no favorable prognostic factors and thus a very low chance (0%) of benefit from E-CPR. Low-risk patients, i.e., those with > 14 points in our scoring system, were most likely to benefit from E-CPR following refractory cardiac arrest, with a 65.8% survival rate at discharge. E-CPR has the potential to put patients who are at the end of their natural life through a very invasive intervention with no meaningful likelihood of benefit. In addition, the cost of E-CPR is substantial [8], and we believe that such a risk stratification is helpful to the physician for identifying appropriate patients for E-CPR. However, a larger study involving more patients with external validation of our observations is required.

Our study has several limitations. First, the study was retrospective, and the number of cases was limited. Second, the quality of CPR performed by personnel in the prehospital setting, the in-hospital setting, and at the emergency department differed and was difficult to evaluate, and it may have been an important cofounding factor influencing patient outcome. Finally, the proposed risk stratification system is used to stratify patient outcome only in our institution, which is equipped to perform E-CPR even in the emergency department. Future prospective studies with larger sample sizes are required to verify these findings and extrapolate them to the wider population.

This study has created a risk stratification system for predicting survival of E-CPR for refractory cardiac arrest. The score identifies a low-risk group of patients with a high rate of survival to discharge (65.8%) and good neurological function at discharge (60.5%) and a high-risk group with a zero rate of survival to discharge. Such risk stratification with pre-E-CPR parameters to predict patient survival after E-CPR may help clinical decision-making for E-CPR implementation.