Mild Hypothermia in Cardiogenic Shock Complicating Myocardial Infarction: Randomized SHOCK-COOL Trial

PCI was performed according to current guideline recommendations with procedural details left to the discretion of the interventionalist.^17,18 In all patients, a 3-lumen pulmonary artery catheter was inserted. Baseline hemodynamics were assessed by right-sided heart catheterization, and cardiac index was assessed by the Fick equation using pulmonary and arterial oxygen saturation after PCI. In patients randomized to MTH, cooling was initiated with cooled saline in the catheterization laboratory, and an invasive cooling catheter (ZOLL Medical Corp, Chelmsford, MA) was introduced in the femoral vein during the index procedure.

After PCI and transportation to the intensive care unit, cooling was maintained with a commercially available system (CoolGard, ZOLL Medical Corp) in patients receiving MTH. By protocol, cooling down to the target temperature of 33°C was set at the maximum possible cooling rate. After the target temperature was reached, it was maintained for 24 hours with the automatic temperature regulation function of the CoolGard system by central temperature measurement in the urinary bladder. After 24 hours, rewarming was initiated with a speed of 0.25°C/h to a target temperature of 37.0°C. To avoid shivering in the MTH group, the patients were treated with a protocol including deep sedation and optional muscle relaxation. In patients randomized to control, no specific temperature control was applied.

After the index measurement in the catheterization laboratory, further measurements were repeated every 8±1 hours for at least 48 hours and afterward until sustained hemodynamic stabilization. Measurements included arterial, pulmonary artery, central venous, and pulmonary capillary wedge pressures. Cardiac output was calculated via oxygen saturation from arterial and pulmonary arterial blood by the Fick equation.

The sublingual microcirculation vascular network was monitored as previously described with a sidestream dark-field microcirculation intravital camera (Microscan, Microvision Medical, Amsterdam, the Netherlands).^19–21 The recorded sequences were analyzed offline in a core laboratory by a trained investigator blinded to group allocation and outcome using dedicated software (Automated Vascular Analysis, AVA 3.0, Microvision Medical). Capillaries were defined as microvessels with a diameter ≤20 µm. Perfused capillary density was calculated by measuring the total length of the perfused capillaries divided by the image area. Similarly, total vessel density was calculated for vessels with a diameter of up to 100 µm. Vessels were regarded as perfused if they had one of the following flow classifications obtained by visual inspection: sluggish, continuous, or hyperdynamic. Unperfused capillaries (ie, capillaries with absent or intermittent perfusion) were judged not to take part in circulation.

The primary end point was CPI at 24 hours after randomization, which is the product of mean arterial pressure and cardiac index divided by a constant factor and is expressed in Watts per square meter body surface area²²:

CPI is has been found as strong hemodynamic correlate of mortality in CS.²² Secondary end points included sidestream dark-field imaging as described above, other hemodynamic measurements such as mean arterial and pulmonary arterial blood pressures and arterial lactate measured from blood drawn via an arterial bloodline every 2 hours by point-of-care testing, and duration of mechanical ventilation and intensive care unit stay or critical illness severity measured by the Simplified Acute Physiology II score. Follow-up was performed at 1, 6, 12, and 24 months. All-cause mortality after 30 days, 1 year, and 2 years was defined as a secondary end point.

Predefined safety end points included sepsis, pneumonia, and bleeding.

Additional data on parameters such as creatine kinase release over time as a surrogate for infarct size, vasopressor and inotrope duration and dosing, cumulative parenteral volume administration, and urine output were collected.

On the basis of previous experimental trials in CS, a CPI of 0.28±0.07 W/m² was assumed in the control group.²³ Estimating an absolute difference in the CPI of 0.07 W/m², we calculated a power of 83% at a 2-sided α=0.05 with the inclusion of 2×18 patients. Because of the high early mortality in CS, we anticipated a dropout rate of 10% in the primary end point at 24 hours, leading to 2×20 patients to be randomized. In the primary analysis, patients who died before assessment of the primary end point were counted as dropouts.

Most variables showed a nonnormal distribution; therefore, all continuous variables are presented as median with interquartile range (IQR). Differences between the MTH group and control were analyzed by the Fisher exact test for dichotomous variables or the Mann-Whitney-Wilcoxon test for comparison of continuous variables. The median difference between the groups for the primary end point was calculated by the Hodges-Lehmann estimate of location shift, with the 95% CI derived according to Conover. For the primary end point, a post hoc sensitivity analysis also was performed, setting all patients with missing values resulting from death before the 24-hour measurement to a CPI of 0. Time to all-cause death was analyzed by the Kaplan-Meier method with log-rank testing. For differences over time between patients in MTH and control, longitudinal data were fitted by a mixed model (with treatment group as a factor) with random intercepts with baseline values as a covariate, time as a continuous variable, and the group-time interaction calculating a P value for interaction between the groups. Statistical analysis was performed with commercially available software (MedCalc for Windows, version 17.9.2, MedCalc Software, Ostend, Belgium; and RStudio, version 1.1.383, RStudio, Inc, Boston, MA). A 2-tailed value of P<0.05 was considered statistically significant.

The primary end point of CPI measured at 24 hours was assessable in 32 of 40 patients (80%) because 7 patients died within the first 24 hours and the measurement could not be performed in 1 patient for technical reasons. This was equally distributed between the groups, with 4 patients in each group with missing values for the primary end point. There was no significant difference in the primary end point of CPI at 24 hours (MTH versus control: 0.41 [IQR, 0.31–0.52] versus 0.36 [IQR, 0.31–0.48] W/m²; P=0.50; median difference, −0.025 W/m²; 95% CI, −0.12 to 0.06) or over time (Figure 3A). In the post hoc sensitivity analysis including missing measurements for the primary end point owing to death during the first 24 hours set to 0, there was also no difference (MTH versus control: 0.37 [IQR, 0.23–0.51] versus 0.34 [IQR, 0.29–0.46] W/m²; P=0.78; median difference, −0.01 W/m²; 95% CI, −0.14 to 0.1).

Similarly, no significant differences between MTH and control were observed with respect to other secondary hemodynamic end points over the first 48 hours (Figure 3B–3D).

Arterial lactate decreased over time in both groups, with a slower and flatter decrease in the MTH group (P for interaction=0.002; Figure 3E).

In sidestream dark-field imaging, no differences in median perfused capillary density (P=0.96) and total vessel density (P=0.35) were observed (Figure 4A and 4B).

The early clinical course showed no differences in duration of mechanical ventilation and intensive care unit stay (Table). In addition, Simplified Acute Physiology II scores as a surrogate for critical illness severity were equal in both groups (Table).

Overall, all-cause mortality at 30 days, 1 year, and 2 years was 55%, 65%, and 68%, respectively. No differences were observed between MTH and control (Figure 5A and 5B)

All 40 patients needed catecholamines. The use of vasopressors was comparable in both groups; inotropes had a faster decline in the MTH group (Figure 3F and 3G). The median length of catecholamine support was equal in both groups (MTH versus control: 3 [IQR, 1.5–5] versus 2.5 [IQR, 1–3]; P=0.24; Table).

Patients in the MTH group had higher amount of fluid resuscitation (cumulative parenteral volume administered during the first 48 hours: MTH versus control, 14.3 [IQR, 10.8–18.2] versus 10.8 [IQR, 8.8–13.6] L; P=0.048) and a higher urine output (cumulative urine volume during the first 48 hours: MTH versus control, 6.3 [IQR, 3.8–7.1] versus 3.5 [IQR, 3.3–5.0] L; P=0.009).

Maximum creatine kinase as a surrogate for infarct size did not differ between treatment groups (MTH versus control: 31 [IQR, 12–65] versus 45 [IQR, 23–81] µmol/s·L).

Despite numerically slightly higher rates for pneumonia and bleeding events in the MTH group, these differences were not statistically significant between the groups (Table).

In the first large-scale trials investigating the effects of MTH on outcome after cardiac arrest, patients with CS were excluded because of the belief that MTH may further compromise the hemodynamic situation.^7,8 In the Target Temperature Management trial, patients with severe shock state, defined as a systolic blood pressure <80 mm Hg despite fluid loading, vasopressors, inotropes, and/or treatment with an intra-aortic balloon pump, were also excluded; patients in moderate CS were allowed for randomization.²⁴ Therefore, randomized human data on MTH in severe CS are lacking. Furthermore, patients in CS without prior resuscitation as randomized in the present study may differ from patients in CS after resuscitation with postcardiac arrest syndrome with potentially earlier and higher systemic inflammatory response syndrome.²⁵ The results of the analysis of the moderate CS subgroup including 139 patients were in line with our results. Temperature management with 33°C or 36°C did not show differences in 30- and 180-day survival. In addition, no differences were observed in mean arterial pressure, whereas arterial lactate showed a delayed decline in the 33°C group, similar to our study.²⁴ Hemodynamic parameters beyond mean arterial pressure were not reported, and given the selection of the exclusion of patients with severe CS, a comparison with our cohort is difficult. The slowed lactate clearance in the study by Annborn et al²⁴ and our trial may indicate higher tissue hypoxemia in patients receiving MTH. On the other hand, higher lactate levels during cooling may be explained by some metabolic changes. MTH may cause higher fat metabolism, leading to an increase in the levels of lactate.²⁶ Shivering induced by MTH may also cause higher lactate levels. A strict protocol of deep sedation and muscle relaxation was followed, so shivering was avoided in the present trial. Because arterial lactate levels and lactate clearance are potent prognosticators in CS,^27,28 this finding in both trials supports the lack of a potential beneficial effect of MTH in CS.

Experimental and animal models found potential positive effects of MTH in CS. In 1 canine model of AMI complicated by CS, reductions in heart rate, left ventricular end-diastolic pressure, and systemic and myocardial oxygen consumption were observed, and cardiac output remained unchanged.²⁹ Two porcine models enhanced the evidence of a possible beneficial effect of MTH in CS. The first study by Götberg et al³⁰ randomized 16 of 25 pigs developing CS after inflation of a PCI balloon in the proximal left anterior descending coronary artery for 40 minutes to MTH versus control. No differences in cardiac output were observed between the 2 groups, but blood pressure was significantly higher in the MTH group. The study also showed a significant survival difference in favor of the MTH group.³⁰ In the second study by Schwarzl et al,³¹ AMI in CS was induced in 16 pigs. After randomization to MTH (33°C) and normothermia (38°C), the authors found similar results with an increase in mean aortic pressure but no difference in cardiac output. In contrast to both studies, no differences between MTH and control were detected in our human study for the primary end point of CPI and for all other tested secondary end points, including mean arterial pressure. This may be explained by methodological differences between experimental and clinical studies. In experimental models, the effect of a single intervention is usually tested in a fixed model with no further adjustment of therapies. In contrast, in clinical trials like ours, the intervention is added to the standard of care, which includes several actions such as fluid resuscitation, catecholamine therapy, and other intensive care unit measures that may differ between groups.³² Therefore, only very strong experimental benefits may likely translate into the same effect in clinical studies.

Retrospective human data comparing 20 consecutive patients with CS after resuscitation undergoing MTH with a historical propensity score–matched control group suggested an increase in mean arterial blood pressure and lower cumulative doses of vasopressors but not inotropes.³³ In addition, lactate clearance was better in the MTH group. In contrast, we observed a significantly slower decline of lactate in the MTH group. These differences may be explained by a risk of bias in trials with historical control groups. Parallel to the introduction of MTH, the rates of early revascularization increased steadily after publication of the SHOCK trial (Should We Emergently Revascularize Occluded Coronaries for Cardiogenic Shock),^3,34 which may explain the observed benefit.

Negative results in a trial with a small sample size raise a question of power. Despite the a priori sample power calculation based on previous results and adjustment for potential dropouts, the estimate was still inaccurate. Early mortality in the first 24 hours was slightly higher than anticipated (18% versus 10%), and the observed difference in CPI between the groups was much smaller than anticipated. However, our very low median difference with wide CIs and the lack of any significant effect in secondary end points make detection of a relevant effect in a larger population unlikely.

In addition, the chosen primary end point CPI may be questioned. However, no other measured secondary end point showed any trend for a significant difference, and CPI was found to be a strong correlate to mortality in previous trials.²² Finally, our study was not powered for clinical end points; therefore, the mortality rates presented may at best be considered exploratory.