Background

Accidental hypothermia is a severe condition with high mortality rate, ranging between 25 and 40% in most studies [1]. In young patients succumbing to accidents at sea or harsh weather conditions, many life-years are lost. It is however possible to survive extreme exposure if correct treatment is provided. Hypothermia lowers metabolism and is neuroprotective, allowing survival after accidental cooling down to a core temperature of 13.7 °C [2]. Hypothermic patients are however at grave risk of developing refractory ventricular fibrillation (VF) and cardiac arrest with little chance of successful defibrillation during evacuation and transport to hospital [3]. Such witnessed hypothermic cardiac arrest is termed “rescue collapse” and Frei et al. found an associated mortality rate of 27% [4]. The pathophysiology has been largely unknown and it is therefore challenging to predict arrhythmias and rescue collapse in hypothermic patients, which is related to movement during extrication, mobilisation or transfer [4].

Main text

In a recent study [5], we found that cooling of rabbit hearts to mild-moderate hypothermia (31 °C) alters ventricular repolarisation while transmural conduction remains relatively unchanged. Rabbits were chosen do to the close resemblance to human cardiac electrophysiology [6]. When provoking arrhythmias by electrical stimulation in the rabbit model, we found that this temperature-dependent combination of effects increased risk for VF (decreased VF threshold) in moderate hypothermia and was pro-arrhythmic. Exposure to severe hypothermia (17 °C), conversely, decreased risk for VF as conduction and repolarisation was equally affected. These changes were reflected in QRS and QT-intervals on the ECG. Correcting the QT interval for heart rate reinforced the association between ECG-findings and pro-arrhythmic activity during hypothermia. We found that relative values of QRS-intervals to the corrected QT-interval, correlates with increased risk for ventricular arrhythmia in moderate hypothermia.

Based on these results, we found two biomarkers that correlate highly (correlation coefficient 0.97–0.98) with risk for cardiac arrest in hypothermic hearts, using Bazett’s (QTc) or Fredericia’s (QTf) correction of QT-interval. We find that QRS/QTc emerge as the most available clinical biomarker. It has a comparable correlation (0.97) with VF threshold and is available for calculation from previously published clinical data (Fig. 1).

Fig. 1
figure 1

a We have found four potential biomarkers that correlate with risk for cardiac arrest in hypothermic rabbit hearts, calculated from QRS interval relative to Bazett’s (QTc) and Fredericia’s (QTf) correction of QT-interval. QRS/QTf is the most promising experimental biomarker. b QRS/QTc data from the included studies plotted against QRS/QTc and VF threshold values from rabbit. Human data showed high correlation (comparing 37 °C, 28–32 °C and < 24 °C) with both rabbit QRS/QTc (correlation coefficient: 0.97) and rabbit VF threshold (correlation coefficient: 1.00). Human values are given as mean (weighted for number of patients in each study) ± SD between study means weighted for number of patients in each study

Full size image

To assess the clinical value of QRS/QTc in hypothermic patients, we extracted data from a recently published systematic review article of hypothermia and cardiac electrophysiology [1]. In a total of 8 studies on accidental and therapeutic hypothermia, QRS- and QTc-intervals were listed from patients that could be categorized (Table 1) into normothermia (35–37 °C) and varying degrees of hypothermia: 32–35 °C (mean 33.1 °C), 28–32 °C (mean 30.6 °C), 24–28 °C (mean 26.7 °C), and < 24 °C (mean 23.3 °C). As expected, at the lowest temperatures we found data from few patients, which must be taken account for when assessing QRS/QTc values from < 24 °C and 24–28 °C.

Table 1 In a total of 8 studies [1] on accidental and therapeutic hypothermia, QRS- and QTc-intervals were published from patients that could be categorized into normothermia and varying degrees of hypothermia. Human values are given as mean (weighted for number of patients)
Full size table

In Fig. 1 we present QRS/QTc data from the included clinical studies plotted against QRS/QTc and VF threshold values from rabbit. Human data showed high correlation with both rabbit QRS/QTc (correlation coefficient: 0.97) and rabbit VF threshold (correlation coefficient: 1.00). This implies that QRS/QTc could predict risk for VF at different temperatures in hypothermic patients. As rescue collapse contributes to the high mortality rate in accidental hypothermia [4], prediction of VF risk could be of high clinical value. This would be relevant also in the in-hospital setting, where QRS/QTc could have a role in assessing safety of therapeutic hypothermia treatment. The inverse calculation; QTc/QRS could be an even more easily accessible clinical marker, relating higher values with higher risk for VF during hypothermia.

Conclusion

We believe that these calculations, as we have applied on preclinical and clinical data, have potential as novel biomarkers for predicting risk of hypothermia-induced cardiac arrest. QTc/QRS or QRS/QTc could easily be tested in the clinic, and have potential to be implemented in guidelines to predict rescue collapse and ease further clinical research into pharmacological prevention of this condition. Our findings apply both to victims of accidental hypothermia and to patients undergoing therapeutic hypothermia during surgery or after e.g. cardiac arrest, where a biomarker used for risk assessment would be of high value.