Introduction

Although endovascular treatment (EVT) is known to be beneficial in acute ischaemic stroke patients with large vessel occlusion (LVO) in the anterior circulation, more than half of these patients remain functionally dependent, despite the high reperfusion rates.1 2 Indeed, infarct progression despite reperfusion is well documented in the literature3 and is not unexpected given the progression in time. Furthermore, a substantial number of patients have been observed to develop reperfusion injury following successful recanalisation.4

Thus, new strategies such as hypothermia-induced neuroprotection need to be explored, not only to bridge the time to reperfusion but also to attenuate reperfusion injury. Therapeutic hypothermia has been demonstrated as a robust form of neuroprotection in animal ischaemic-reperfusion models.5 6 Nonetheless, randomised trials in acute stroke patients have failed to demonstrate the efficacy of induced hypothermia compared with conventional treatment.7

The reasons for this treatment failure are diverse and include treatment delay, mainly due to elaborate and invasive cooling procedures; the infeasibility of inducing and maintaining hypothermia until the target temperature is reached, mainly in awake patients and the lack of recanalisation in a large proportion of patients.6–8

By attaining high rates of recanalisation using EVT in sedated patients and applying improved methodology for the induction of hypothermia, the failure of hypothermia observed in previous acute stroke studies can potentially be overcome.9

Indeed, fever (>37.5°C) has been shown to have a negative influence on postprocedural infarct growth in acute ischaemic stroke patients undergoing EVT, despite successful reperfusion.10 11 In addition, a higher body temperature during both the prerecanalisation and postrecanalisation phases were associated with poorer clinical outcomes in EVT patients.12 In a recent observational study, the application of a short-duration, intra-arterial, selective-cooling infusion with 4°C saline for 5 min prior to EVT and 10 min after recanalisation was associated with a smaller infarct volume and a trend towards clinical benefit.13

In experimental and clinical efficacy studies with invasive brain temperature measurements, the transnasal cooling device RhinoChill resulted in a fast and uniform reduction of brain temperature in all parts of the anterior circulation.14–16 The present study, therefore, aimed to investigate the feasibility and safety of a rapidly initiated cooling process to a target temperature of 35°C before recanalisation. This was achieved via transnasal cooling by RhinoChill, followed by surface cooling for 6 hour after recanalisation in intubated and sedated patients with LVO undergoing thrombectomy.

MethodsTrial design

The present study is a prospective, single-arm, open-label clinical trial to assess the feasibility and safety of hypothermia induction by transnasal cooling before mechanical recanalisation immediately after EVT-dependent intubation, followed by surface cooling for 6 hours after recanalisation in patients with LVO.

The study was conducted at a single comprehensive stroke centre (CSC). Patients included were either admitted directly to the CSC or secondarily transferred from a primary stroke centre of the stroke network after diagnosis of LVO. The stroke network comprised 12 primary stroke centres and one CSC. Patients were enrolled from December 2021 until May 2022.

Written informed consent was obtained either from the patient or the patient’s legal representative. If this was not possible, an independent physician who was not involved in the study gave informed consent in accordance with the patient’s presumed wishes; subsequent written informed consent was obtained from the patient or the legal representative as soon as possible.

Patients

The inclusion and exclusion criteria are shown in box 1. In summary, acute stroke patients with moderate to severe ischaemic stroke (defined by a National Institutes of Health Stroke Scale (NIHSS) score of >5) caused by LVO, and who underwent thrombectomy within 24 hours of symptom onset according to the current European Stroke Organization guidelines17 were enrolled in the study. Major exclusion criteria were severe coagulopathy, severe infection, brain trauma or neurovascular intervention <3 months prior to LVO.

Box 1 Inclusion and exclusion criteria

Inclusion criteria

Age >18 years.

Prestroke modified Rankin Scale score of 0–2.

Indication for endovascular treatment

Acute ischaemic stroke with an NIHSS score >5.

Intracranial occlusion of the M1 or M2 segment of the middle cerebral artery or internal carotid artery or tandem occlusion on CT/MR-angiography.

Time window:

Time from last known normal-to-groin-puncture <6 hours: native CT or MR-DWI with ASPECTS >5 (optional CT-perfusion or MR-perfusion).

Time from last known normal-to-groin-puncture 6–24 hour or unknown time window: significant imaging mismatch according to the eligibility criteria of

DEFUSE-335: infarct core <70 mL (defined by CBF <30% or by MR-DWI), mismatch volume >15 mL (defined by the difference between Tmax >6 s volumes and infarct core), mismatch ratio >1.8.

DAWN-trial36: infarct core defined by CBF <30% or by MR-DWI.

>80 years and NIHSS >10: infarct core <21 mL.

<80 years and NIHSS >10: infarct core <31 mL.

<80 years and NIHSS >20: infarct core <51 mL.

With or without intravenous rtPA thrombolysis.

Written informed consent.*

Exclusion criteria

Known severe haemorrhagic diathesis (international normalised ratio >3.0, partial thromboplastin time >70 s, platelet count <50.000/μl).

Brain trauma or neurovascular surgery/intervention <3 months prior to LVO.

Severe infection.

Life expectancy <6 months.

Pregnancy.

ASPECTS, Alberta Stroke Program Early CT Score; CBF, cerebral blood flow; LVO, large vessel occlusion; MR-DWI, MR-diffusion-weighted imaging; NIHSS, National Institutes of Health Stroke Scale; Tmax, time to maximum of the residue function; rtPA, recombinant tissue-type plasminogen activator.

*Obtained either from the patient, the patient’s legal representative or an independent physician who was not involved in the study.

Endovascular treatment

The thrombectomy technique was chosen at the discretion of the interventionalist on an individual patient basis and comprised either the use of a stent retriever, direct thrombus aspiration or a combination of both methods. Angioplasty and stenting of the proximal internal carotid artery (ICA) were performed as needed.

Reperfusion grades were assessed according to the modified thrombolysis in cerebral infarction (TICI) classification.

EVT was performed under general anaesthesia in all patients according to our institutional protocol, and based on the findings of recent meta-analyses, which demonstrated that patients who undergo EVT under general anaesthesia have better recanalisation rates and outcomes than those treated with conscious sedation.18–20

Medical management was consistent with the recommendations published by the European Stroke Organization.17

Therapeutic hypothermiaCooling devices

Several studies have used the RhinoChill cooling device (BrainCool) for induction of hypothermia in patients after cardiac arrest21 22 and brain injury,14 where it is described in detail. The RhinoChill system comprises three components: a control unit that is connected to an oxygen supply; a set of tubes with two nasal catheters and a 2L bottle that contains coolant (perfluorohexane, a chemically and biologically inert fluorocarbon). The 10 cm long nasal catheters are fully inserted through the nostrils and feature spray ports on the dorsal surface to distribute the coolant in the nasal cavity. Evaporation of the coolant absorbs heat from the tissue and rapidly cools the nasal cavity to ~2°C, whereby the large nasopharyngeal surface area and extensive adjacent vascularity (carotid arteries and cavernous sinus) mediate the direct conductive and indirect haematogenous cooling of the brain and body.

In the present study, we used an oesophageal temperature probe to monitor cooling, since the temperature measured in the lower oesophagus has been shown to correlate best with the brain temperature.16

The components of the RhinoChill device can be connected within 30–60 s, and the device is ready to use as soon as it has been switched on. Preparation of the RhinoChill-system was performed by the attending neurologist. Coolant delivery is maintained at a constant ratio of oxygen flow, such that the cooling level is controlled by setting the oxygen flow rate between 0 L/min and 60 L/min.

In previous studies, the RhinoChill device has so far only been used to induce hypothermia with a maximum duration of 60–90 min.14–16 21 22 Furthermore, the device does not have an automated control feedback unit, which would allow the target temperature of 35°C to be kept constant and would allow a controlled rewarming. Therefore, the cooling technique was switched to a surface cooling method (BrainCool) with an automated control system to maintain the target temperature of 35°C for a total of 6 hours after recanalisation and to achieve controlled rewarming.

Cooling protocol

Directly following intubation for EVT, an oesophageal temperature probe (Mon-a-Therm, Covidien, Massachusetts) was placed transorally to continuously measure oesophageal temperature (distance from incisors to tip: 30–35 cm, confirmation of correct position in the lower oesophagus by X-ray during subsequent angiography). In addition, tympanic temperature (BraunThermoscan Pro6000, New York) was measured at predefined time points. Then, the RhinoChill nasal catheters were introduced into the nasal cavity. The RhinoChill system was switched on, and oxygen flow was gradually increased to 60L/min within 1 min. After 20–25 min, the empty coolant bottle was changed and oxygen flow was continued at 40L/min. The maximal total duration of RhinoChill use was set to 90 min. Nasopharyngeal cooling was halted if the target oesophageal temperature of 35°C was attained, and, if applicable, restarted when the temperature reincreased to over 35°C.

After thrombectomy, the patient was immediately transferred to the neurological Intensive Care Unit (ICU), where cooling was maintained by surface cooling (BrainCool) for a duration of 6-hour postrecanalisation (or at the end of the EVT procedure when recanalisation was unsuccessful).

Thereafter, patients were actively rewarmed at a rate of 0.2 °C/hour (7 hours) until the oesophageal temperature reached 36.5°C, after which the cooling device was removed.

During the entire cooling and rewarming period up to 36.0°C, all patients were intubated, mechanically ventilated, and sedated, showing no response to vocal or physical stimulation (Richmond Agitation Sedation Scale score = −5).

Analgetic and sedating medications included propofol (mean dosage: 104±47 mg/hour) and remifentanyl (mean dosage: 0.21±0.15 mg/hour) in all patients, and norepinephrine (mean dosage: 0.14±0.25 mg/hour) was used to maintain systolic blood pressure between 140–160 mm Hg during the EVT procedure, and 120–160 mm Hg after recanalisation; the target mean arterial pressure was >75–80 mmHg.

Imaging

According to institutional protocol, patients within the 6-hour time window underwent non-contrast CT and CT-angiography (CTA) as well as optional CT-perfusion (CTP) at admission, whereas patients admitted after 6 hour, or those who had an unknown time window, underwent CT, CTA and CTP (Somatom-Definition-Flash or Somatom-Definition-64AS; Siemens).

Within a post-thrombectomy period of 24±3 hours, a non-contrast CT was performed to determine the potential presence of intracerebral haemorrhage. Further cranial CT or MRI scanning was performed at day 3–7 at the discretion of the treating neurologist, to determine the final infarct volume prior to initiation of oral anticoagulation.

To determine an irreversible initial infarct core and penumbra, the fully automated volumetry approach VEOcore (VEObrain23) was used. Final infarct volume on MRI was measured using VEOcore with an apparent diffusion coefficient threshold of <620 ×10−6 mm2/s for infarct determination. Final infarct volume on CT was measured by manually tracing the hypodense infarct area on each slice and multiplying the traced area with the slice thickness (Syngo.via, Siemens Healthcare).

Final infarct volumes were used to compute relative penumbra consumption, defined as ((final infarction volume-initial CTPcore)/initial CTPpenumbra) and were computed for each patient as previously described.10

OutcomesPrimary outcome

The primary outcome was the time taken to reach the oesophageal target temperature of 35°C from the point of initiation of transnasal cooling.

Secondary outcomes

Secondary outcome measures included the modified Rankin Scale (mRS) at both discharge and 3 months, the NIHSS at 24 hours and at discharge as well as the evolution of ischaemic core and penumbra from admission to day 2–7 after EVT.

Safety outcomes included all serious adverse events occurring within 48 hours of initiation of cooling. This involved all device-associated complications, disturbances of electrolytes and renal function, cardiac arrhythmia, haemodynamic instability, pneumonia, any intracranial bleeding and symptomatic intracranial haemorrhage within 48 hours as well as smell and taste disturbances at day 7 (Sniffin’ sticks, burghart-Messtechnik, Holm, Germany).

Statistics

Variables were tested for normal distribution using the Shapiro-Wilk test. Given that not all data were distributed normally, variables are presented as the median and IQR of distribution. NIHSS scores (admission vs day 7), the initial and final infarct core volumes and differences between patients with good (mRS 0–2) and poor (mRS 3–6) 90-day outcome were compared, using the Wilcoxon test and Mann-Whitney-U-Test. Statistics were performed using the IBM SPSS-Statistics-software V.28.0.1.0.

ResultsPatient characteristics

A total of 22 patients (median age 77y, range 50–94) were enrolled from December 2021 until May 2022. The baseline characteristics are shown in table 1. The median NIHSS score at admission was 15 (12.5–19.75) and the ASPECTS score was 9 (8.25–10). The median time from LSN-to-admission was 265 min (191–490), and 14 patients (64%) received intravenous thrombolysis with recombinant tissue-type plasminogen activator (rtPA). Nine patients (41%) were transferred from primary stroke centres after diagnosis of LVO.

Table 1

Patient characteristics (n=22)

EVT procedure

The characteristics of the EVT procedures are shown in table 2. The median time from LSN-to-groin-puncture and recanalisation was 309 min (228–580) and 355 min (258–645), respectively. Eleven patients were admitted <6 hour after symptom onset, 5 patients >6 hour after onset and 6 patients had an unknown time window.

Table 2

Characteristics of endovascular thrombectomy and hypothermia (n=22)

Twenty patients (91%) were treated with stent retrievers (Solitaire n=16, pReset n=2, Embotrap n=2), of which three additionally received direct aspiration, and two (9%) showed spontaneous recanalisation with angiographic demonstration. Four patients (18%) with tandem occlusion of the proximal ICA and the first segment of the middle cerebral artery received additional ICA-stenting. Successful recanalisation with TICI scores of 2b–3 was obtained in 19/22 patients (91%).

Hypothermia

Transnasal cooling was initiated 58 min (39–79) after admission, 5 min (4–5.25) after EVT-dependent intubation, and 5.5 min (2–10) prior to groin puncture (table 2).

The course of oesophageal and tympanic temperatures within the first 14 hours of hypothermia induction is shown in figure 1. The median oesophageal and tympanic temperature before the start of cooling (is equal to time 0 in figure 1) was 36.35°C (36.0–36.6) and 36.45°C (36.0–36.6), respectively.

Figure 1

Oesophageal (A) and tympanic (B) temperatures in the first 14 hours of cooling induction. (A) After start of transnasal cooling (= time 0), oesophageal temperature decreased from 36.35°C at admission to target temperature of 35°C within 30 min. Despite halting transnasal cooling when target temperature of 35°C was reached, temperature continued to decrease slowly (time 45–60 min). At the time of recanalisation, median oesophageal temperature was 34.9°C. Between end of thrombectomy procedure and arrival to the ICU, temperature further decreased slowly without active cooling (time 75 min). Surface cooling was started 31 min after recanalisation on ICU for 5.6 hour to reach a total duration of steady-state hypothermia of 35°C for 6 hour after recanalisation. (B) Course of tympanic temperatures did not differ significantly when compared with oesophageal temperatures, although tympanic temperatures decreased faster during transnasal cooling and were slightly above the oesophageal values during surface cooling and rewarming phase.

The median time required from the start of RhinoChill to reach the oesophageal target temperature of 35°C was 30 min (23–37), with a cooling rate of 2.6°/hour (range, 1.5–3.6). Despite the interruption of transnasal cooling when the oesophageal target temperature of 35°C was reached, temperature continued to slowly decrease in all but 2 patients (time 45 and 60 min in figure 1). In these two patients, oesophageal temperature reincreased to over 35°C after cessation of transnasal cooling before recanalisation was achieved, and RhinoChill had to be restarted for a duration of 6 min and 8 min, respectively, to reach the target temperature of <35°C again.

The target temperature was reached in all patients (range 13–78 min) and was <35°C in 86% (19/22) of patients at the time of recanalisation. Median oesophageal temperature at recanalisation was 34.9°C (34.7–35.0).

Although no active cooling was carried out during the period between the end of thrombectomy in the angiography room and the arrival at the ICU (median duration 31 min (24–36)), oesophageal temperature continued to slowly decrease to 34.8°C (34.6–35.0). Thus, after arrival in the ICU, surface cooling was used to maintain stable hypothermia of 35°C for a duration of 5.6 hours (5.4–5.7), thus achieving a complete duration of hypothermia of 6 hours after recanalisation (figure 1). The median oesophageal temperature during the surface cooling period was 34.9°C (34.7–35.0).

The course of tympanic temperatures (figure 1B) did not differ significantly when compared with oesophageal temperatures but showed greater dispersion as indicated by wider IQR throughout.

The duration of rewarming was 7 hour (7–9), and patients were intubated for 21 hours (16–25).

Functional outcome

The median NIHSS score was 15 (12.5–19.75) at admission and decreased to 7 (3–12.5) at 24 hours, and 2 (1–8) at day 7 (p<0.001). The respective lengths of stay in the ICU and the hospital were 3 (2–4) and 8 (7–10) days. Nine patients (41%) were discharged home. The distribution of mRS at discharge (median day 8 (7–10)) and after 3 months (day 89, (86–92)) is shown in figure 2. A good neurological outcome with an mRS score of 0–2 was seen in 14 out of 22 patients (64%) at discharge, and in 15 patients (68%) after 3 months. Eleven patients (50%) had reached an excellent functional outcome by 3 months post-treatment, with an mRS score of 0–1. In the subgroup of patients who additionally received intravenous thrombolysis (n=14), 90-day outcome was good (mRS 0–2) in nine patients (71%).

Figure 2

Distribution of modified Rankin Scale (mRS) score at discharge (median day 8) and after 3 months. A good neurological outcome with a mRS score of 0–2 was observed in 14 out of 22 patients (64%) at discharge, and in 15 patients (68%) after 3 months.

Overall, patients with a good 90-day outcome had a significantly higher ASPECTS score at admission (10 (9.25–10) vs 8 (7.5–9), p<0.001) and a shorter ventilation time (18 hours (16–23) vs 38 hours (24–59), p=0.006) compared with patients with poor outcome. Furthermore, patients with good outcome were younger (74 years (69–82) vs 80 years (77–85)), were more often women (47% vs 43%), had lower NIHSS at admission (14 (11–17) vs 18.5 (14–20)) and received intravenous thrombolysis more often (67% vs 57%), although these differences did not reach significance.

Safety outcomes

Procedural and clinical complications are summarised in table 3. The most common side effect during hypothermia was asymptomatic bradycardia not requiring intervention (n=7, 32%). Pneumonia occurred in four (18%) patients. Asymptomatic haemorrhagic transformation on the 24-hour-CT was observed in four (18%) patients. None of the patients showed signs of symptomatic intracranial haemorrhage, disturbances in electrolytes, renal function, coagulation or smell or taste function.

Table 3

Complications and physiological parameters during and after hypothermia

In the subgroup of patients receiving additional intravenous thrombolysis (n=14), asymptomatic haemorrhagic transformation occurred in three patients (21%), and nose bleeding (without the need for intervention) in one patient (7%).

The hospital mortality and 3-month mortality rates were 5% (1/22; withdrawal of care due to a persistent, severe deficit in an 88-year-old patient at day 4) and 9% (2/22; death due to myocardial infarction and multiorgan failure 6 weeks poststroke onset in the rehabilitation hospital), respectively.

Evolution of infarct core and penumbra

VeoCore analysis of initial infarct core and penumbra volumes could be performed in 20 patients (CTP was done in 11 patients with unknown or >6-hour time window and (optionally) in 9 of 11 patients within the 6-hour time window). The initial infarct core volume (19 mL (6–38)) and infarct core on day 2–7 (22.5 mL (8.5–31)) did not differ significantly (p=0.379; final infarct volume was defined by the 24-hour CT in six patients, by CT at day 3–7 in four patients, and by MRI at day 2–7 in 10 patients). The volume of penumbra tissue at risk was 103 mL (85–118) at admission. Relative penumbra consumption was 5% (0–8), meaning that 95% (90–100) of the initial penumbra could be salvaged.

Discussion

This pilot study has demonstrated the feasibility and safety of a new cooling algorithm that covers the prerecanalisation and postrecanalisation phase in thrombectomy patients. This hypothermia-based approach was associated with very good short and 90-day functional outcomes.

Although hypothermia has consistently been associated with robust neuroprotection in preclinical studies, all randomised-controlled trials in acute ischaemic stroke patients have failed to confirm the potential benefits of cooling.6 7 The reasons for this are manifold and include the following major problems: (1) the delay in the initiation of the cooling procedure due to either study protocol limitations, or invasive and time-consuming cooling methods; (2) the difficulty and infeasibility of reaching and then maintaining the target temperature in awake patients due to intolerance and shivering8; (3) the heterogeneity of the patient cohort,7 which usually comprises patients with different stroke severities and can also include those with mild and lacunar stroke, these latter having a small amount of tissue-at-risk; (4) low recanalisation rates of 10%–30%, even in patients receiving rtPA6 7 24 and (5) poor recruitment due to the elaborate and difficult cooling procedures.8

The recent success of thrombectomy in patients with LVO combined with the high proportion of tissue-at-risk and high recanalisation rates in these patients offers new perspectives for the use of hypothermia in acute stroke patients. Consequently, we attempted to address the above-described reasons for hypothermia failure by designing a hypothermia protocol that differed from previous trial protocols in the following five main aspects:

The technique is relatively simple and minimally-invasive, effective means of quickly inducing hypothermia.

Hypothermia was initiated prior to recanalisation in order to modulate reperfusion injury.

Study inclusion was restricted to patients with LVO and EVT with high recanalisation rates and significant penumbra, which is the main target of neuroprotection.

Patient discomfort, shivering and ultimate cessation of the cooling procedure, was avoided by extending peri-interventional intubation and sedation throughout the entire period of cooling and rewarming.

A very mild degree of hypothermia (35°C) was applied for a relatively short period of time to minimise side effects, while still possibly offering a certain degree of neuroprotection during the most vulnerable prerecanalisation and early postreperfusion phases.

Patient selection for hypothermia

The use of hypothermia as a potent neuroprotective agent is aimed at protecting and preserving the ischaemic penumbra. Therefore, the appropriate selection of patients that can benefit from such a therapeutic strategy is essential. However, most clinical trials on hypothermia included a heterogenous patient collective with different stroke severities, and recanalisation could not be achieved in a significant proportion of participants.6 7 Therefore, in the present study, patients were only included if they had an LVO with an indication for thrombectomy (according to present European guidelines). All patients were intubated for EVT according to our established institutional protocol that is based on recent meta-analyses,18–20 and ventilation and sedation were continued for the entire period of hypothermia and rewarming. By using this regimen, only intubated and sedated patients with an assumed large volume of tissue-at-risk were included in the study. Thereby, no direct side effects of hypothermia such as shivering and intolerance were observed, which is otherwise often the case in stroke patients who are awake. Furthermore, the immediate initiation and maintenance of hypothermia was feasible in all patients. As a consequence, the target temperature of 35°C was achieved in all patients and maintenance of the target temperature as well as a steady rise in temperature during rewarming was ensured, without significant fluctuations.

In addition, by exclusively including thrombectomy patients, the vast majority of patients (91%) were successfully recanalised. Preclinical studies consistently showed strong effects of neuroprotection using hypothermia in reperfusion models, but the effects observed in permanent occlusion models were less robust,6 25 suggesting that recanalisation and reperfusion are mandatory for hypothermia to be beneficial.

Timing of hypothermia

In the cerebral ischaemia that follows LVO, the infarct core rapidly grows over time, so that any therapeutic measures that aim to protect the tissue-at-risk need to be initiated as soon as possible. Due to the hypothermia induction methods used in previous studies, the potential beneficial effects of cooling were additionally impaired by the slow-acting treatment methods that ensued. In contrast, RhinoChill is an easy to use transnasal cooling device, which allows the cooling process to be started within just a few minutes.

Using this device, cooling could already be initiated just 5 min after EVT-dependent intubation and never took more than 1 hour postadmission. The target temperature of 35°C was reached after a median of 30 min. Importantly, since the RhinoChill preparation took place simultaneously to the patient preparation for angiography, no time delay in groin puncture for EVT occurred.

Thus, with this approach, a fast induction of hypothermia was feasible. This is important, given that preclinical studies have demonstrated the greater protective effects conferred by prompt intraischaemic hypothermia compared with delayed intraischaemic or postischaemic hypothermia.6 26 By reducing the energy demand of the tissue, hypothermia can limit ischaemic core growth and prevent the penumbra from evolving into an infarction.27

Duration and depth of hypothermia

The ideal extent and duration of hypothermia in acute ischaemic stroke patients are yet to be established, whereas one meta-analysis of preclinical studies found a significant association between lower temperatures and a more prominent reduction in infarct volume,25 a more recent meta-analysis did not find any hypothermic depth level to be particularly superior to the others.28

With respect to hypothermia duration, several experimental findings have suggested that a longer cooling period might provide better neuroprotection.6 However, a recent study using both cell culture and animal stroke models found that a shorter bout of therapeutic hypothermia may be superior and suggested that hypothermia at either 33°C and 35°C effectively preserves all neuronal cell types.29

In addition, prolonged application and lower temperature targets of hypothermia significantly increase the risk of side effects, especially cardiovascular complications and infections, which may weaken or even offset the protective effects of cooling.6 30 Thus, the optimal conditions need to be carefully titrated in order to maximise efficacy while minimising complications.

In summary, the most recent preclinical findings from animal and cell culture models suggest that fast,31 mild (34–35°C29) and non-prolonged29 application of hypothermia can potentially provide the best long-term neuroprotective effects.

Accordingly, we used RhinoChill for a fast initiation of cooling down to the level of mild (35°C) hypothermia in the angiography room before recanalisation and continued hypothermia with surface cooling for a relatively short period of 6-hour postrecanalisation. This coincided with the most vulnerable reperfusion phase within the first hours of recanalisation, while minimising the risk of side effects.

Indeed, the present findings support this new cooling protocol, since 68% of patients gained independency after 3 months, even though 50% of patients had a time window that was either greater than 6 hour or unknown altogether, and 41% were transferred from primary stroke centres. This proportion of patients with a good neurological outcome is superior to that observed in randomised trials with normothermic thrombectomy patients (mean 46%1) as well as to findings from previous studies from our own institution (32%–42%32 33). However, since there was no control group in the present study, we cannot exclude that the good outcome and penumbra salvage observed in this study could reflect the brain tissue saved solely due to timely and successful EVT procedure.

Side effects of hypothermia

The most common side effect in the present study was asymptomatic bradycardia (32% of patients), which was observed during hypothermia but did not need intervention and spontaneously disappeared after rewarming. No other complications that have been described in previous studies on hypothermia in acute stroke were observed. Of note, none of the patients suffered from symptomatic intracranial bleeding despite additional thrombolysis in 64% and ICA stenting, which required additional antiplatelet therapy in 18% of patients.

The pneumonia rate of 18% in the present study is comparable to that reported in prospective trials with normothermic EVT patients under general anaesthesia, with a mean pneumonia rate of 19.4% (range: 13.3%–46%18 20).

Importantly, the median intubation time was 21 hours, which is clearly longer than that usually observed in normothermic EVT patients. However, in the present study, we did not observe any discomfort and shivering representative of physiological reactions that increase metabolic rate, oxygen consumption and respiratory burden. These side effects would obviously be unfavourable for patients with an acute stroke.34 Moreover, the relatively long intubation time was not associated with an increase in complications such as pneumonia rate.