Targeted temperature management guided by the severity of hyperlactatemia for out-of-hospital cardiac arrest patients: a post hoc analysis of a nationwide, multicenter prospective registry

Background The International Liaison Committee on Resuscitation guidelines recommend target temperature management (TTM) between 32 and 36 °C for patients after out-of-hospital cardiac arrest, but did not indicate patient-specific temperatures. The association of serum lactate concentration and neurological outcome in out-of-hospital cardiac arrest patient has been reported. The study aim was to investigate the benefit of 32–34 °C in patients with various degrees of hyperlactatemia compared to 35–36 °C. Methods This study was a post hoc analysis of the Japanese Association for Acute Medicine out-of-hospital cardiac arrest registry between June 2014 and December 2015. Patients with complete targeted temperature management and lactate data were eligible. Patients were stratified to mild (< 7 mmol/l), moderate (< 12 mmol/l), or severe (≥ 12 mmol/l) hyperlactatemia group based on lactate concentration after return of spontaneous circulation. They were subdivided into 32–34 °C or 35–36 °C groups. The primary endpoint was an adjusted predicted probability of 30-day favorable neurological outcome, defined as a cerebral performance category score of 1 or 2. Result Of 435 patients, 139 had mild, 182 had moderate, and 114 had severe hyperlactatemia. One hundred and eight (78%) with mild, 128 with moderate (70%), and 83 with severe hyperlactatemia (73%) received TTM at 32–34 °C. The adjusted predicted probability of a 30-day favorable neurological outcome following severe hyperlactatemia was significantly greater with 32–34 °C (27.4%, 95% confidence interval: 22.0–32.8%) than 35–36 °C (12.4%, 95% CI 3.5–21.2%; p = 0.005). The differences in outcomes in those with mild and moderate hyperlactatemia were not significant. Conclusions In OHCA patients with severe hyperlactatemia, the adjusted predicted probability of 30-day favorable neurological outcome was greater with TTM at 32–34 °C than with TTM at 35–36 °C. Further evaluation is needed to determine whether TTM at 32–34 °C can improve neurological outcomes in patients with severe hyperlactatemia after out-of-hospital cardiac arrest.

Background Out-of-hospital cardiac arrest (OHCA) is associated with high mortality and poor neurological outcomes [1]. Targeted temperature management (TTM) is often used to treat comatose OHCA patients after the return of spontaneous circulation (ROSC) [2][3][4]. The International Liaison Committee on Resuscitation guidelines strongly recommend TTM between 32 and 36 °C [5,6] but did not provide patient-specific target temperatures. Precision medicine allows mild therapeutic hypothermia or normothermia to be chosen for individual patients, based on the initial brain damage [7]. Specific OHCA patient characteristics that can predict the benefits of 32-34 °C compared with 35-36 °C [8][9][10][11][12] have not been identified.
Serum lactate concentration is a nonspecific marker of tissue hypoxia, aerobic glycolysis, and lactate clearance [13]. The association of serum lactate and neurological outcome has been described in OHCA [14][15][16][17] but the differential benefits of TTM at 32-34 °C and TTM at 35-36 °C in patients with different initial serum lactate concentrations have not been reported.
We hypothesized that the effects of TTM at 32-34 °C had different influence according to the severity of hyperlactatemia after ROSC. The study aim was to investigate the benefit of TTM at 32-34 °C in OHCA patients with various degrees of hyperlactatemia compared to TTM at 35-36 °C.

Study design, setting, and patient selection
This study was a post hoc analysis of patients who were included in the Japanese Association for Acute Medicine out-of-hospital cardiac arrest (JAAM-OHCA) registry between June 2014 and December 2015. The registry is a nationwide, multicenter prospective registry that included 56 institutions in 2014 and 73 in 2015. The JAAM-OHCA registry collected both pre-and post-hospitalization data. Prehospitalization data were obtained from the All Japan Utstein Registry of the Fire and Disaster Management Agency as previously reported [18]. In-hospital data were collected via an Internet-based system by physicians or medical staff at each institution. The JAAM-OHCA registry committee integrated the prehospital and in-hospital data, as previously described [1]. The protocol was approved by the Institutional Review Board of each participating hospital. Comatose adult OHCA patients with TTM treatment after ROSC were included and patients with withdrawn TTM were excluded.

Data collection
Patient age, sex, witness status, presence of dispatcher instruction, presence of a bystander who performed cardiopulmonary resuscitation (CPR), etiology of cardiac arrest (cardiac or not), initial cardiac rhythm, prehospital epinephrine administration, prehospital airway management, prehospital automated external defibrillator (AED) use, response time (time from call to contact with a patient), time from call to hospital arrival, prehospital ROSC, Glasgow Coma Scale score, coronary angiography, use of a mechanical circulatory device (extracorporeal membrane oxygen and/or intra-aortic balloon pumping), door-to-first ABG measurement after ROSC, arterial blood gas measurement data including serum lactate concentration after ROSC, door-to-TTM initiation, target temperature during TTM, and neurological outcome 30 days after cardiac arrest were included in the analysis. The responsibility for setting target temperature was entrusted entirely to each physician or each institution. There were no specific protocols of TTM for the JAAM-OHCA registry.

Primary exposure
Patients were divided by serum lactate concentration after ROSC into mild (< 7 mmol/l), moderate (< 12 mmol/l), or severe (≥ 12 mmol/l) hyperlactatemia group. The cutoff values were chosen for reasons that were previously described [15,16]. Then, patients in each hyperlactatemia group were subdivided into 32-34 °C and 35-36 °C groups based on their target temperature.

Outcome measures
The primary endpoint was an adjusted predicted probability of 30-day favorable neurological outcome. A favorable neurological outcome was defined as a cerebral performance category (CPC) score of 1 or 2 [19]. The CPC score includes five outcomes, including 1, good cerebral recovery; 2, moderate cerebral disability; 3, severe cerebral disability; 4, coma or vegetative state; and 5, death/brain death [20]. The secondary outcome was an adjusted predicted probability of 30-day survival.
We firstly examined the comparison between patients finally analyzed and those excluded. Then, we assessed the association of serum lactate after ROSC and both crude favorable neurological outcomes and the adjusted predicted probability of favorable neurological outcomes in the patients finally analyzed. Furthermore, study participants were divided into 32-34 °C and 35-36 °C groups in each hyperlactatemia category based on primary exposure. We compared the adjusted predicted probabilities of favorable neurological outcomes and survival between 32-34 and 35-36 °C groups in each hyperlactatemia category. We calculated p value for interaction between hyperlactatemia group and TTM (32-34 °C or 35-36 °C) for the adjusted predicted probabilities of 30-day favorable neurological outcome and 30-day survival.
Statistical analysis was performed with EZR software (Saitama Medical Center, Jichi Medical University, Saitama, Japan) [43]. Continuous variables were compared with Mann-Whitney U or Kruskal-Wallis tests. Binary variables were compared with Fisher's exact test. p values < 0.05 were considered statistically significant. Missing data were not replaced or estimated.

Baseline patient characteristics
A total of 12,024 OHCA patients were included in the OHCA registry during the study period. Of 3971 adult patients with ROSC, there were 3262 patients without TTM treatment, 162 with withdrawn TTM, and 112 without lactate concentration after ROSC. Then, 435 were included in the final analysis ( Fig. 1). Favorable neurological outcomes were observed in 41.8% of the study patients. Baseline characteristics between the patients excluded and those included are shown in Table 1.

Association of serum lactate concentrations after ROSC and favorable neurological outcomes
The crude and adjusted predicted probability of a favorable 30-day neurological outcome across the range of serum lactate concentrations after ROSC is shown in Additional file 1: Figure S1. We found a trend toward decreases of both crude and adjusted predicted probability of a favorable 30-day neurological outcome with increasing serum lactate concentration. There were 139 patients in the mild, 182 in the moderate, and 114 in the severe hyperlactatemia groups. Table 2 shows the characteristics of the patients in each hyperlactatemia group. There were significant differences among the three groups in witness, initial shockable rhythm, prehospital epinephrine administration, prehospital advanced airway management, prehospital AED, prehospital ROSC, Glasgow coma scale score, ECMO and/or IABP, door-to-first arterial blood gas measurement after ROSC, pH, PaCO 2 , HCO 3 , and base excess. The percentage of patients with favorable neurological outcomes decreased with the worsening of hyperlactatemia: 60.4% with mild, 41.8% with moderate, and 19.3% with severe hyperlactatemia had favorable neurological outcomes.

Baseline characteristics of patients stratified by hyperlactatemia and treated with 32-34 °C or 35-36 °C
Comparison of the baseline characteristics of patients with mild, moderate, or severe hyperlactatemia and treated with either 32-34 °C or 35-36 °C are summarized in Table 3. The analysis revealed significant differences in initial shockable rhythm rates and door-to-TTM initiation time among the mild hyperlactatemia group, in ages and door-to-TTM initiation time among the moderate hyperlactatemia group, and door-to-TTM initiation time among the severe hyperlactatemia group.

Favorable neurological outcomes and survival following 32-34 °C or 35-36 °C in the three hyperlactatemia groups
Multiple logistic regression models to obtain the adjusted predicted probabilities of 30-day favorable neurological outcome showed that age older than 65 years, initial shockable rhythm, prehospital epinephrine administration, prehospital advanced airway management, time from call to hospital arrival, prehospital ROSC, Glasgow Coma Scale score, and PaCO 2 30-50 mm Hg were significantly associated with neurological outcome (Additional file 2: Table S1). There were no significant differences in the rates of crude and adjusted predicted 30-day favorable neurological outcomes following 32 (Table 3, Additional file 4: Figure S2).

Discussion
This post hoc analysis of a nationwide, multicenter prospective registry found no significant differences in the adjusted predicted probability of favorable neurological outcomes with TTM at 32-34 °C or 35-36 °C in OHCA patients with mild (< 7 mmol/l) or moderate (< 12 mmol/l) hyperlactatemia after ROSC. TTM at 32-34 °C was associated with a greater adjusted predicted probability of a favorable outcome in patients with severe hyperlactatemia than TTM at 35-36 °C. Two previous retrospective observational studies also reported that TTM at 32-34 °C was beneficial in patients with severe OHCA patients [8,44]. One was a singlecenter cohort study including 1200 witnessed OHCA TTM targeted temperature management, CPR cardiopulmonary resuscitation, AED automated external defibrillator, ROSC return of spontaneous circulation, ECMO extracorporeal membrane oxygenation, IABP intra-aortic balloon pumping, ABG arterial blood gas a There were 1160 patients excluded due to death at the emergency department and 89 patients excluded because they got Glasgow Coma Scale motor score of 6 after return of spontaneous circulation b There was one patient excluded due to death at the emergency department patients [8]. It was found that TTM at 32-34 °C was associated with improved neurological outcomes in patients with relatively longer times from collapse to the start of resuscitation attempts. The second study evaluated 871 witnessed, shockable adult patients prospectively included in a population-based registry in Toronto, Canada [44]. The beneficial effect of TTM at 32-34 °C was more apparent in patients with a prolonged time from collapse to initial defibrillation. Two retrospective studies reported that lower target temperature improved outcome in OHCA patients with less severe baseline characteristics [11,12]. One evaluated the outcomes in a cohort of Japanese patients treated with either low (32.0-33.5 °C) or moderate (34.0-35 °C) target temperatures and stratified to a short (< 30 min) or long interval to resuscitation [11]. Improved neurological outcomes were observed only in the short-interval group. However, a target temperature of 34 °C was classified as moderate target temperature group in this study, and more than 75% of patients in the moderate target temperature group had a target temperature of 34 °C. In case of classifying target temperature of 34 °C into low temperature group as in our current study or other studies [8,12,44], patient characteristics   [6,9] 8 [6,10] 0.749 7 [6,9] 8 [6,9] 0.381 8 [6,9] 8 [7,11] 0.329 Time from call to hospital arrival, min 32 [25,40] 31 [23,33] 0.283 30 [24,34] 31 [24,36] 0.542 29 [25,37] 28 [24,39]  Door-to-first ABG measurement after ROSC, min 11 [5,32] 9 [5,22] 0.300 10 [5,17] 9 [6,18] 0.784 13 [6,30] 19 [11,35]  CPR cardiopulmonary resuscitation, AED automated external defibrillator, ROSC return of spontaneous circulation, ECMO extracorporeal membrane oxygenation, IABP intra-aortic balloon pumping, ABG arterial blood gas and main results may be subject to a major change. Thus, it is very difficult to compare this study and our current study at present. Another retrospective observational study included patients with or without early evidence of hypoxic encephalopathy provided by brain computed tomography [12]. Compared to TTM at 35-36 °C, the benefit of TTM at 34 °C was evident only in patients without hypoxic encephalopathy. In that study, the brain damage resulting from hypoxic encephalopathy was considered to be too severe to permit a favorable response to TTM. The limited available data suggest that patients with severe OHCA benefit from TTM at 32-34 °C, as shown in this study, but the effect may not be seen in extremely severe cases. Serum lactate concentrations reflect not only tissue hypoxia but also aerobic glycolysis and lactate clearance in critically ill patients [13]. Several observational studies have shown that serum lactate concentration in OHCA patients was influenced by factors in addition to the duration of cardiac arrest, including vasopressors, high arterial oxygen partial pressure, and decreased renal function [14,45]. Serum lactate seems to be a comprehensive marker of severity in OHCA patients [46], and the association of serum lactate concentrations and neurological outcome has been described in many previous studies [14-17, 47, 48]. In this study, both crude and predicted probability of 30-day favorable neurological outcomes decreased with increasing serum lactate concentration after ROSC (Fig. 2). Although serum lactate in OHCA patients does not necessarily reflect brain metabolites [49], it does reflect OHCA severity after ROSC.
The finding that TTM at 32-34 °C was not associated with predicted neurological outcomes in patients with mild or moderate hyperlactatemia was consistent with the results of the TTM trial, which did not show a significant benefit of TTM at 33 °C compared with 36 °C [4]. The lactate concentration of 6.7 ± 4.5 mmol/l in the TTM study population was thought to account for the lack of demonstrated benefit of TTM at 33 °C. The mean lactate concentration in both temperature groups evaluated in the TTM trial was similar to the concentration in the mild to moderate hyperlactatemia patients in our current study. Thus, post hoc analysis of the TTM trial results did not identify a patient subset that benefited from TTM at 33 °C compared with 36 °C [9,10].
The present study had several limitations. First, this was a post hoc analysis. Although we calculated