POLAR RCT is probably the most anticipated neurotrauma trial of the year and its findings were finally presented at this year’s ESICM congress in Paris.

Before I comment on what POLAR RCT is all about, it’s worth taking some time to consider what this trial is NOT about. It does not deal specifically with TBI patients with a refractory ICP-elevation, for whom other measures of ICP-control have not been successful. Nor does it say much about the value of moderate interventions such as thermoregulation to 36-37 degrees (which is common practice in many centres) compared to no temperature control at all.

First some background. The largest study thus far is the Eurotherm 32-35 trial (commented here), which was stopped early due to suggested harm from therapeutic hypothermia (TH). This was an landmark trial with many strengths, but it was also limited by a few week points which are addressed in our commentary. Having been discontinued early, the trial fell far short of its intended power. This entails that if just a few patients had ended up with a different outcome the odds ratio for GOSE at 6-months (the trial’s primary outcome) would have lost its statistical significance. Nonetheless, it is a well-wrought and important study with findings that merit our attention.

In POLAR the investigators included 240 severe TBI-patients in each arm, regardless of  ICP on inclusion. In the intervention group patients were cooled to 35 degrees with cold fluids (1-2 litres) and subsequently brought further down to 33 degrees with surface cooling. This therapy was maintained for 3 days to a week. Rewarming was done slowly and under ICP-guidance (a real strength of this study). Median time to reach target temperature was 10,1 hours. In the control group subjects were kept at 36,5 to 37,5 degrees, with active cooling if necessary. All other management strategies were common to both groups and guided by local practice.

There was no significant difference in the proportion of patients with a favourable neurological outcome – defined as GOSE 5-6 at six months (48.8% in the hypothermia group vs 49.1% in the normothermia group). Nor was there any difference in mortality. ICP-values were also largely the same in both arms. Some secondary outcomes differed, though. There was an increased risk of pneumonia in the HT group, as well as marginally longer time on ventilator. Hypothermic patients also seemed to require more vasopressors. In conclusion: early prophylactic hypothermia in patients with TBI does not improve neurological outcomes. But what does that really tell us? And how should that inform future practice?

It makes sense to consider both studies together when pondering those questions. In common for both Eurotherm and POLAR is the use of rather deep hypothermia, 32-35 degrees and 33 degrees respectively. These temperature ranges are not without adverse effects. Another common feature is the use of rather large cold fluid-boluses to help attain early cooling. The latter may not have made much difference for all we know, but it does constitute 1-2 litres of fluids that the control groups in both trials did not receive. One could make the argument that perhaps not all TBI patients will benefit from this otherwise uncalled-for fluid bolus. Where the two studies differ is in their inclusion criteria. While Eurotherm specifically targeted patients with an elevated ICP refractory to lower-tier measures, POLAR sought to evaluate hypothermia as a prophylactic modality, irrespective of ICP upon inclusion. And this is obviously an important difference. Prophylaxis may not benefit all patients, and indeed may be deleterious in some. Patient selection is clearly not straightforward as TBI is phenotypically very heterogenous and we still don’t have any real consensus on how and what to monitor, nor on what should trigger an intervention. The closest we have come is that sustained ICP above 20 (22) mmHg should elicit some kind of response.

The above mentioned factors aside, the most important observation in my opinion is that the control groups in both studies were kept normothermic to a very high degree. In POLAR the protocol specified a target of 37 degrees for the controls, using active cooling if required. In Eurotherm the desired temperature range in the controls isn’t mentioned specifically but their published data clearly reveal that mean temperature barely deviated from normothermia (with narrow confidence intervals). In other words, in both trials the control groups remained normothermic, be it by design or coincidence.

This brings me to my main point: maintenance of normothermia is also an intervention. Thermoregulation and actively avoiding fever is not the same as leaving patients to their own devices. This is perhaps obvious and needs no mentioning to most of us. Nonetheless, following the 2013 TTM trial on TH in out-of-hospital cardiac arrest (OHCA) patients with ROSC, there was seemingly a perception among many clinicians that equipoise between 33 degrees and 36 degrees as targets in post-ROSC management suggests that temperature doesn’t matter. Tellingly, a recent publication by Salter et al indicates that TTM has changed management practice in many centres in such a way as to often permit outright hyperthermia in post-ROSC patients. There is also a trend towards worsening survival in the same data. Even if not intentional, this is a very worrisome development and underscores the impact that certain trials have on clinical practice. While cardiac arrest and TBI clearly are different enitities, I fear that POLAR (especially when added to Eurotherm) will falsely convince clinicians that temperature management of any kind in TBI is futile, analogous to what we have seen in OHCA in the wake of TTM.

All this being said, POLAR is an excellent study. It is well designed for its purpose and meticulously executed. But it is essential that we’re sober about what can be concluded from it, and what still remains unclear. Prophylactic hypothermia to 33 degrees doesn’t seem to confer any survival advantage, nor does it improve neurological function in survivors. That much is now clear. However, POLAR does NOT single out the subset of patients with ICP-elevation refractory to all other measures, arguably the group who could benefit the most from temperature management. Nor does it evaluate the effect of thermoregulation to normothermia/moderate hypothermia as opposed to none at all. In my estimation (to paraphrase Mark Twain) the rumours of the death of thermoregulation in TBI have been greatly exaggerated.

Just as an aside: In the neurocritical care unit where I practise we aim for normothermia (using surface thermoregulation if necessary) in any TBI-patient with a sustained elevation in ICP. In more refractory cases we may consider dropping the temperature to 36 degrees, but very rarely below. We strive to avoid hyperthermia regardless of ICP, especially in the first 1-2 weeks. This strategy is usually well tolerated and associated with few side-effects, most of which are predictable (shivering, cold diuresis etc.). In most cases this will be sufficient as an adjunct to other medical strategies. Where it falls short, we will consider progressing to surgical measures or (occasionally) barbiturate coma.