The present study is the first nationwide multicentre study of outcomes in patients receiving ECPR for refractory OHCA in Denmark. The main findings showed that one in four ECPR patients survive to hospital discharge with a good neurological outcome. Outcome was significantly associated with initial presenting rhythm, signs of life during CPR, transient ROSC, pre-hospital low-flow time, initial pH and lactate levels. By offering ECPR only to patients meeting the strictest criteria, survival exceeded 48%, but this occurred at the cost of withholding life-saving treatment in the majority of patients saved by ECPR in the cohort. Thus, a continuing need exists for optimization of the ECPR selection criteria.
ECPR has emerged as a salvage therapy for patients suffering from refractory OHCA. Although resource demanding, ECPR has been shown to be both feasible and cost effective(11). Several previous observational studies have demonstrated encouraging survival rates and a favourable neurological outcome from applying strict inclusion and exclusion criteria in distinctive patient populations(12–16). The first randomized clinical trial recently published by Yannopoulos et al. (the ARREST trial) revealed superiority of ECMO-facilitated resuscitation with a survival rate of 43% compared with 7% with standard advanced life support(7). Survival at six months was also greater in the ECMO group (hazard ratio (HR) 0.16, 95% CI 0.06–0.41, p = 0.0001). Whereas the trial was well-designed and supports use of ECPR in refractory OHCA, apparent limitations are present as this was an open-labelled single-centre study with a relatively small and highly selected patient population. Despite growing interest in and a growing body of literature on ECPR for refractory OHCA, robust evidence on patient eligibility is still lacking.
In Denmark, a number of initiatives to improve pre-hospital quality of care for OHCA patients have produced a remarkable increase in 30-day survival rates from 3.9% in 2001 to 16% in 2018(17). Concurrently, ECPR has been evolving steadily and is now an established treatment for selected patients with refractory OHCA in all Danish regions. However, given the formal inclusion and exclusion criteria defined in the Danish national ECPR consensus, one might assume that the results of our study would reflect a more homogeneous population. Despite our intention to select qualified candidates, violation of the national consensus was seen in a substantial proportion of patients, as patients with an initial presenting rhythm with asystole, unwitnessed arrests and no-flow time > 10 minutes were among those who were treated. In the present study, ECPR was more often reserved for younger patients. This may explain why no significant difference was detected in age between survivors and non-survivors. For patients of younger age, ECPR was initiated despite more frequent failure to meet the criteria, whereas patients with a more advanced age were considered for ECPR only if they were optimal candidates according to the remaining selection criteria. Previous studies have proposed advanced age as a predictor of a poor outcome in patients with ECPR(18), and some studies have even suggested an age of > 75 years as a contraindication for ECPR(19).
The predictive value of pH and lactate levels in patients with cardiac arrest is well established. Controversy still exists regarding the ECPR population. In our study, initial arterial pH and lactate levels were found to be associated with mortality. This finding is consistent with previous findings(12, 20). Jung et al. retrospectively reviewed 93 patients with cardiac arrest undergoing ECPR and found results similar to our results(21). On the contrary, Leick et al. found no association between elevated lactate levels and mortality(22). Acidosis and high lactate levels indicate prolonged phases of anaerobic metabolism due to tissue oxygen deficiency, advocating that pH and serum lactates may serve as alternate indicators of low-flow times and CPR quality. Conversely, no well-defined cut-off levels exists to predict a poor prognosis. Our results support the inclusion of pH and lactate into our decision-making when considering patients for ECPR, whereas specific cutoffs still need conformation in other cohorts. Importantly, a stringent use of pH > 6.8 and lactate < 15 mmol/L as selection criteria, may result in denying life-saving therapy to a considerable number of the survivors present in this cohort.
In contrast to previously published results(14, 23), known prognostic factors such as witnessed arrest, bystander CPR and end-tidal CO2 were not significantly correlated with survival in our patient population. The lack of statistical significance may be explained by the highly selected population in the present study. The high frequency of patients with witnessed arrest receiving bystander CPR and the low frequency of the number of patients with a low initial end-tidal CO2 in our study hampers evaluation of the prognostic value of these factors. Although no association was found in our study, these factors remain of paramount importance.
Pre-hospital parameters are of great assistance the selection of appropriate candidates for ECPR. Initial shockable rhythm, transient ROSC and signs of life during CPR are considered favourable prognostic factors in ECPR(24–26). In a prospective registry study, Bougouin et al. compared conventional CPR with ECPR in 13,191 consecutive patients with OHCA(27). Prognostic factors in the ECPR group comprised an initial shockable rhythm and transient ROSC prior to ECPR implementation. One of the main findings in our study was that patients with signs of life during CPR had a threefold higher survival rate than patients without signs of life during CPR. This is in correlation with a study recently published by Debaty et al.(28) The authors found, that any signs of life before or during CPR substantially improved 30-day survival with favourable neurological outcome in a multivariable prognostic model (OR 7.35, 95% CI 2.71–19.97). Moreover, the absence of signs of life more likely precluded survival with a good neurological outcome in patients with non-shockable rhythm. This observation supports the evidence of incorporating signs of life as an important factor in the selection of patients for ECPR. The present findings do not allow us to determine whether any of the patients would have survived without ECPR. However, the long total low-flow times observed makes this unlikely.
In the present cohort, a pre-hospital low-flow time < 60 minutes was associated with an increased survival rate; however, patients exceeding pre-hospital low-flow times > 80 minutes also had an advantageous outcome. Patients with long pre-hospital low-flow times had a higher rate of signs of life during CPR than patients with a pre-hospital low-flow time of 60–80 minutes, indicating the highly selected nature of this population. Our results suggest that prolonged resuscitation efforts in the field may not be futile, especially in patients showing favourable circumstances and where ECPR can be established within a reasonable timeframe.
Historically, the arrest-to-perfusion time has been linked to survival(14, 16, 29, 30). Wengenmayer et al. reported that among 133 patients with cardiac arrest treated with ECPR, low-flow time was an independent predictor of mortality(30). Bartos et al. demonstrated a significant association between time from arrest to sufficient ECPR flow and neurological outcome in a cohort of 160 patients(16). These results are similar to ours. In the present study, hospitals serving patients in remote and rural areas had longer arrest-to-perfusion time due to longer distances to the invasive centre. Implementation of systematic pre-hospital ECPR calls, more rapid allocation of helicopter-mediated transport and direct triage to the catheterization laboratory may improve the performance and facilitate a reduction in system delay for these patients.
In the present study, we assessed pre-hospital and in-hospital factors in relation to outcome, which may come in benefit for clinicians in the selection of appropriate candidates for ECPR. Our results suggest that a more refined assessment of the inclusion criteria, comprising additional criteria such as signs of life during CPR and lactate levels, may improve selection of patients for ECPR. Nevertheless, one must recognize that limiting patient selection to strict pre-defined criteria may exclude some patients in whom ECPR would have bought valuable time until the reversible cause could be treated. The fact that the national consensus was violated in 52% of patients, of whom some survived to hospital discharge with a good neurological outcome in 94% of the cases, indicates that there is likely room for individualized decision-making, especially in the young patients. Patient selection for ECPR continues to be a challenging part of real-world clinical practice and further randomized clinical trials are warranted.
Limitations
The present study has several limitations. Its retrospective nature makes it subject to patient selection bias. The national consensus was available and adopted to some extent in all centres. This produces risk of bias in the evaluation of the associations with outcome. Although we conducted a multicentre study using nationwide registry data, the heterogeneity of the study population with a mixed cohort of patients with OHCA hampers generalization of the results. Neurological outcome at hospital discharge is a fairly crude measure; it is, however, broadly used in cardiac arrest studies. Studies assessing long-term survival and neurological outcome are necessary.