We congratulate Professor Eastwood and the Carbon Control and Cardiac Arrest (CCC) trial group on an informative and thought-provoking pilot study, which provides an insight into further potential research.Increases in PaCO2 are well known to produce an increase in cerebral blood flow (CBF), mediated by its vasodilatory actions on cerebral vascular smooth muscle. Outside of extremes of PaCO2 levels this relationship is linear with a 1?mmHg increase in PaCO2 causing an increase in CBF of 1–2?ml/100?g/min.Further to improved cerebral perfusion, carbon dioxide also has anti-convulsant and anti-inflammatory properties. It has been postulated that following cardiac arrest, the cerebral vasodilatation associated with mild hypercapnia will improve reperfusion and consequently neurological outcomes. Previous observational studies have suggested that patients admitted to intensive care units (ICU) following cardiac arrest have improved short and long-term outcomes if their arterial carbon dioxide levels had been recorded above 45?mmHg within the first 24?h of their ICU admission., 

Neuron-specific enolase (NSE) is a glycolytic intra-cellular enzyme released into serum following damage to neuronal cells and disruption of the blood–brain barrier. Its use as a surrogate endpoint for neurological outcomes has been validated in previous studies and included in clinical guidelines by the American Academy of Neurology, with values greater than 33?mcg/l predicting poor neurological outcome. As the CCC trial group mention, a previous relationship between therapeutic hypothermia and decreased serum NSE levels has been reported. These concerns may be somewhat alleviated following analysis of data from the Target Temperature Management After Out-of-Hospital Cardiac Arrest (TTM) trial, which suggested that targeted hypothermia did not influence the reliability of serial NSE levels to predict poor neurological outcome.

S100? is a glial-specific protein found in mature astrocytes that enters the bloodstream during the acute phase of brain injury. As with NSE, raised serum levels of S100? have been implicated as a feature of likely poor prognosis following cardiac arrest. A detailed meta-analysis and systematic review looking at the evidence for the use of S100? as a predictor of prognosis following severe traumatic brain injury concluded that there was a significant positive association between serum S100? levels within 24?h and both mortality and poor neurological outcome. This study did however calculate that for unfavourable neurological outcome (Glasgow outcome score ?3) a threshold value of 2.5–3.0?mcg/l was required to provide a specificity of 94% (95% confidence interval 85–98%). This is significantly higher than the values in the CCC trial group pilot study, where values were below 1.5?mcg/l throughout the 72?h in both groups.

Reporting of outcomes following cardiac arrest is known to be inconsistent and heterogeneous with a growing need for a core outcome set that combines physician assessed and patient-reported measures. Evidence for biomarker use is growing and recently conducted pilot studies, awaiting publication, are aiming to ascertain if utilising a combination of biomarkers may provide even more improved short and long-term outcome prediction., 

The CCC trial group pilot study carefully details the patient characteristics, cardiac arrest characteristics and ICU processes of care within each group. While there were no differences that established statistical significance, there were trends to a greater proportion of PEA arrests, higher first recorded minute ventilation volumes and use of active cooling in the targeted normocapnia (TN) group. It should also be noted that despite there not being a statistical difference in mean P/F ratio between the groups, the TN group had a mean P/F ratio of 288, compared to 336 in the targeted therapeutic mild hypercapnia (TTMH) group, indicating the potential for increased rates of mild acute respiratory distress syndrome (ARDS). Not only could this affect outcome between the groups, it could be postulated that striving for normocapnia may directly conflict with lung protective ventilation strategies.

It should also be noted that by their own definitions, the CCC trial group failed to achieve ‘mild hypercapnia’ (PaCO2 50–55?mmHg), with the TTMH group having a mean PaCO2 of 49?mmHg, defined by the authors as ‘slight hypercapnia’. The challenge of defining grades of hypercapnia is evident. One study suggested poorer outcomes with increased PaCO2, where hypercapnia was defined as ?50?mmHg, while previous trials showed improved outcomes when hypercapnia was defined as ?45?mmHg., 

It is evident from this pilot study that reliable randomisation, without imbalances in the baseline characteristics of the groups, and administration of alternative ventilation strategies can be performed in a challenging group of patients. The speed with which randomisation was performed is also to be commended. There is a clear requirement for further investigation into the effects of mild hypercapnia following cardiac arrest, however, we would have preferred to see an a priori description of clear progression criteria to apply to this feasibility study.

We advocate the use of clear definitions for the desired level of hypercapnia, the standardisation of management as far as possible and an attempt to utilise validated neurological outcome assessment tools as well as biomarker surrogates to assess outcomes.

Conflict of interest statement

References