Selective brain cooling after cardiac arrest
Discussions of therapeutic hypothermia require definitions. We differentiate between mild (34 degrees to 36 degrees C), moderate (28 degrees to 32 degrees C), deep (15 degrees to 25 degrees C), profound (5 degrees to 15 degrees C), and ultraprofound hypothermia (less than 5 degrees C). Only mild and moderate hypothermia are compatible with spontaneous circulation. We also differentiate between hypothermia induced before (protection) and maintained during the insult (preservation), and resuscitative hypothermia (i.e., to help reverse a normothermic insult and support recovery). Cerebral hypothermia, which is noninvasively monitored as tympanic membrane or nasopharyngeal temperature, should be distinguished from core hypothermia, which is monitored as pulmonary artery, central venous, rectal, or urinary bladder temperature.
Between 1950 and 1987, when cerebral protective-preservative moderate hypothermia for open-heart surgery  and for neurosurgery  was introduced, it was believed that only moderate hypothermia, which reduces cerebral oxygen demand to 50% , could be therapeutic . Doubt existed regarding the outcome benefit of moderate resuscitative hypothermia after cardiac arrest in patients [5,6] and animals [7-10], and there were associated management problems with the technique. Side effects of moderate hypothermia include life-threatening dysrythmias, reduction in cardiac output, and, if hypothermia is prolonged, a tendency for coagulation problems and pulmonary infection . Resuscitative hypothermia was dormant for three decades.
In 1987, at a resuscitation researchers' symposium , Hossmann  and Safar  reported that even mild hypothermia, which intensivists identified as safe when found accidentally to be present during global brain ischemia in cats  or during cardiac arrest in dogs , seemed to have a protective-preservative effect on the brain. This finding stimulated the Pittsburgh group to evaluate cerebral resuscitative mild hypothermia after cardiac arrest in five outcome studies in dogs [14-18] and one study in rats . In dogs, there was clearly a resuscitative effect on 96-hr outcome [14-18], particularly if mild hypothermia was achieved within approximate 15 mins of reperfusion . Independently and simultaneously, three other groups found that mild, protective, and resuscitative hypothermia reduced histologic brain damage in rats after incomplete forebrain ischemia [20-22]. Ginsberg's group [23-25] and other groups (discussed in panel []) found, in these models, mitigation by mild hypothermia of various mechanisms underlying the selective vulnerability of neurons after ischemia. The protective-preservative or resuscitative effect of mild hypothermia cannot be explained by the minimal or no reduction in cerebral oxygen uptake after cardiac arrest [27,28], but by synergism of beneficial effects on multiple deleterious cascades which leads to delayed postarrest cell death [29,30]. After cardiac arrest, mild hypothermia is more effective and safer than moderate hypothermia [10,16], and deep hypothermia can make cerebral outcome worse , perhaps because of reduced microcirculation. While protective-preservative hypothermia clearly mitigates postinsult brain damage permanently, there is a possibility that mild early resuscitative hypothermia for 3 hrs in rats might merely delay the ultimate inevitable loss of neurons . This observation remains to be documented in higher animals. Recent studies in gerbils suggested that resuscitative more prolonged mild hypothermia (12 or 24 hrs) will give permanent benefit (at 6 months), even when induction of cooling is delayed by several hours [31,32]. The mere temporary suppression of deleterious chemical cascades could potentiate additional physical  or pharmacologic therapies .
After 11 mins of ventricular fibrillation cardiac arrest in large dogs (20 to 25 kg), resuscitative mild hypothermia by clinically realistic methods was combined with cerebral blood flow promotion, and great improvement of functional and histologic cerebral outcome to 96 hrs was achieved , the best result to date.