Therapeutic and prophylactic hypothermia for traumatic brain injury

Therapeutic and prophylactic hypothermia are effective methods of controlling number-induced panic in the depressing setting of severe traumatic brain injury. To some extent ICP is regulated by the cerebral metabolic rate, and metabolic rate is affected by temperature: ergo, with the reduction of temperature one should be able to reduce the ICP. There are also seductive theoretical benefits, suspected influence on the cellular damage, inflammation,  oxudative injury, etc. More importantly, the intensivist feels better. With control of temperature comes a reduction in the ICP, and the anxiety-provoking monitor begins to display more pleasing values. Also, the use of muscle relaxant to reduce shivering has a calming effect on the nursing staff (a paralysed sedated patient is easy to look after). However, there may be no benefit in this technique as far as patient-centered outcomes are concerned.

Both a generic question and a specific question have been asked about this topic.  Question 9 from the first paper of 2015 touched on the issue by exploring hypothermia in a broader sense   (i.e. asking the candidates "what indications can you think of for therapeutic hypothermia?"). Then, Question 27 from the first paper of 2017 asked the trainees to "critically evaluate the role of induced hypothermia in the management of traumatic brain injury". It was an easily expected move, with the typical lag of SAQ development following in the 18-24 months after publication of  the EUROTHERM 3235 trial (2015) and the recent publication of the new (2016) BTF guidelines.

History and evolution of this technique

Hypothermia has become very popular for the management of cardiac arrest patients, and these days it is mainly known from that context. However the use of this technique for brain injury is older. Back in the day before post-arrest hypothermia, all the injured heads ended up getting cooled. This was driven by animal studies and physiological concepts rather than anything like what we now call "Level 1 evidence". 

The first application of the techniques to humans actually pre-dated the abovementioned animal experiments. Fay (1943) published his observations on "generalised refrigeration" in severe brain injury. He cooled a series of patients to as low as 28°C for 4–7 days and reported results which (in his view) were better than what he might have otherwise expected from such patients.

Detailed studies of hypothermia in brain injury followed. For instance, in 1959 Hubert Rosomoff found that mongrel dogs with brain injury survived much longer if they were cooled to 25° C for up to 18 hours. On the basis of this, Sedizmir (1959) recommended its use in brain injured humans should become more widespread. Lundberg et al (1956) were already using it for neuroprotection in neurosurgical patients (14 cases were cooled to as low as 24.8°C) and so the transition of this practice into neurotrauma seemed very reasonable.

The technique seemed to have become very popular in the 1960s, in spite of disappointing real-life efficacy. For instance, Drake and Jory (1962) presented their experience with it in 21 patients. "Six of the 12 survivors are chronic nursing problems with severe dementia", the authors complained. The others died of multiple complications, some of which can be associated with the hypothermia itself (and the rest with the grim reality of critical care in this heroic age). The authors present wonderful graphs of the patients' temperature, punctuated with significant events (eg. "tracheotomy", "talked to wife", "GI bleeding", and ultimately "DEATH"). The technique lost popularity, presumably because many others were disappointed with the results. Jennett et al (1980) described it as a passing fad in their review of management strategies.

Rationale for hypothermia in traumatic brain injury

 Most of these points have been scavenged from the excellent articles by Kees H. Polderman (2001) and Schmutzhard et al (2012). Additionally, the a priori trial protocol for Eurotherm 3235 goes into some significant detail about it. If the exam question invites one to "Critically evaluate therapeutic hypothermia", one would be expected to list a few of the points mentioned below. 

Decreased cellular injury

• Decreased neuronal apoptosis (Xu et al, 2002)
• Decreased production of oxygen free radicals (Globus et al, 1995)
• Reduced excitotoxicity (Busto et al, 1989)
• Suppression of epileptic activity also decreases the risk of excitotoxicity

Decreased inflammatory response and oedema

• Impairment of neutrophil and macrophage function should decrease the size of the injury and mitigate the oedema (Siesjö et al, 1989)
• Decreased permeability of the blood-brain barrier, also decreasing the oedema (Fischer et al, 1999)

Decreased cerebral metabolic rate, with many benefits:

• Decrease oxygen and glucose consumption by the oedematous brain tissue (by 6-10% per 1°C - Polderman, 2001 )
• Decreased intracranial pressure (thus, improved perfusion of the healthy brain)

Arguments against the use of hypothermia

• The studies exploring its benefits have been either human case series or animal data
• The technique is not without risk (i.e. even mild hypothermia has a rate of significant complications)
• The rate of complications is relatively high, eg. pneumonia
• The deeper the hypothermia, the greater the risk of complications
• Haemodynamic instability in hypothermia may decrease CPP
• Catecholamine responsiveness is diminished at low temperature, making CPP more difficult to achieve with noradrenaline
• Decreased platelet function may lead to expansion of conservatively managed haemorrhages or contusions

Evidence for and against hypothermia in traumatic brain injury

In their answer to Question 27 from the first paper of 2017, the college made an attempt to link their answer to published literature by alluding to "Cochrane Systematic Review 2009" and "two paediatric studies and a Japanese study". If this were a university paper turned in for grading, the lecturer would have sent it back with stern remarks regarding appropriate academic referencing standards. All we can surmise is that these studies must have been published after the 2007 BTF guidelines came out, because that's implied in the college answer. That would mean that at least one of the "two paediatric studies" is Hutchison et al (2008),  who concluded that prophylactic hypothermia "does not improve the neurologic outcome and may increase mortality". The other paediatric trial may be Li et al (2009), which came to totally opposite conclusions ( "moderate hypothermia provided neuronal protection for children with severe TBI, and maintaining the intracranial temperature at 34.5°C for 72 h was safe").

Looking at the list of trials undergoing meta-analysis in Lewis et al (2017), the "Japanese study" is probably Hifumi et al (2016), whose B-HYPO trial did not find any significant difference in mortality or neurological outcome among 135 adult patients.

Pre-Eurotherm era

Georgiou et al (2013): the meta-analysis mentioned by the college in Question 9 from the first paper of 2015. Eighteen trials were found, with 1851 patients in total. There was no benefit in mortality when only high quality trials were included. A risk of pneumonia was noted among a series of six low quality trials.  

Eurotherm 3235 trial (2015):

This was probably the most interesting thing to happen in this area in recent history. Among the trial collaborators was Kees H. Polderman whose brilliant reviews of hypothermia are an excellent resource for the CICM trainee. St Emlyn's Blog has an excellent article by Dan Horner describing the trial. Anyway, the following analysis probably has some exam importance. In summary:

• Enrolled 387 patients;
• Hypothermia was used as a second-line therapy to reduce ICP. 
• No survival benefit was observed.
• Recruitment was suspended early owing to safety concerns.
• ICP control was in fact better in the hypothermia group (they required rescue therapies less frequently) but this seems to have made no difference to mortality.

This trial may be criticised for using hypothermia as a mid-tier therapy (whereas in many places it is used as a rescue therapy for otherwise uncontrollable ICP, as an alternative to decompressive craniectomy). Stage 1 therapies for raised ICP were composed of very basic manoeuvres: elevating the head of the bed, boluses of propofol, adequate analgesia, opening the EVD and keeping a MAP of 80 mmHg. Then, if these simple things failed to reduce the ICP, they graduated to Stage 2 and the Eurothermians either started cooling them or use some sort of osmotherapy (for the non-hypothermia group). Barbiturates and decompressive craniectomy were reserved for refractory cases as Stage 3 therapies. This differs from local practice, and some might argue that hypothermia belongs alongside decompressive craniectomy as a last-ditch measure. In their response to such criticism, the Eurotherm authors pointed out that it is implausible to expect a mortality benefit from hypothermia when it is used in super-severe cases after all other therapies had failed

Moreover, the trial power calculation yielded 1800 patients, which was reduced to 600 (somehow preserving the same power) but they only ended up enrolling 387 patients before the trial was terminated ("futility", etc). The enrolled patients ended up being cooled late (on average, 12 hours late) and were given 2000ml of cold saline (which probably did not help the brain oedema).

The trial was terminated early "after the steering committee concluded that there were signs of harm with the treatment being evaluated"; there was an almost 10% difference (25.7% vs 36.5%) in disability outcome, and risk of death strongly favoured the control group (OR was 1.45). 

Systematic reviews of pre- and peri-Eurotherm data

"Cochrane Systematic Review 2009" from the college answer to probably refers to Sydenham et al (2009), who found that among a total of 1409 patients, 1382 came from twenty small trials which were methodologically flawed. The authors found "some statistically non-significant benefit of hypothermia which could be due to the play of chance" and concluded that "hypothermia should only be given to patients taking part in a randomised controlled trial", essentially saying that we should not be promoting this strategy to families as a well-proven therapy.

What comes up as the first Google search result for "hypothermia AND Cochrane AND traumatic brain injury" is actually the 2017 update to this systematic review by Lewis et al, which was published in September, some months after this college paper was marked. Since the first review, nine new studies had come out. The addition of these trials did nothing to improve the attractiveness of hypothermia or increase the quality of the data; in fact the authors complained that "differences in study data remained such that we did not perform meta-analysis" and that "studies were generally poorly reported."Unsurprisingly, the conclusion reached after trawling through this low-quality data swamp was that "further research, which is methodologically robust, is required".

Post-Eurotherm era

Following neatly from the conclusions of Lewis et al, ANZICS (the very temple of "further research, which is methodologically robust")  had published the POLAR trial (2018). ANZICS enrolled  511 patients with an average GCS of 6, and cooled half of them to 33 C for three days. The rewarming was titrated to ICP control (i.e. if their ICP increases, the rewarming stops and cooling is reintroduced). The "P" in POLAR stands for "Prophylactic"; this trial looked at the use of hypothermia to prevent the neurological badness of severe TBI. During the intervention period, both the cooled and the "normothermed" group had roughly the same average ICP, around 15 mmHg. Unlike EUROTHERM, this trial did not demonstrate an increased mortality affecting the cold group, but with the cooling they appeared to be virtually guaranteed to develop pneumonia (70.5% in the hypothermia group and 57.1% in the normothermia group). As for the neurological outcome, it was essentially the same for both groups. What this means for the use of hypothermia in refractory raised ICP is still uncertain.

Society recommendations

The Brain Trauma Foundation does not recommend prophylactic hypothermia. The main reason for their doing so is not safety or futility concerns; instead it seems all the good-quality studies about this intervention have clinically significant differences in the way they use hypothermia, which makes it difficult to perform a meta-analysis (and the BTF people in fact gave up on trying to meta-analyse them). Until better data are available, the BTF has cooled towards hypothermia. When they incorporate POLAR into their thinking processes, one can be sure that the attitude will be even more chilly.