Wednesday, May 5, 2010

Hypothermia, SSEPs, and Prognosis

As a follow up for today's top-post, some more floundering about prognosis in brain injuries, and yes this one is for the progno-wonks out there. 

Neurology also has an article looking at the prognostic importance of bilateral absence of the N20 response of median nerve somatosensory evoked potentials for patients with hypoxic-ischemic encephalopathy (anoxic brain injury) who have undergone therapeutic hypothermia. 

Some background (although this NEJM review is the best place to go for a quick summary) on why I'm even mentioning this.  Absence of the N20 response of median nerve SSEPs more than 24 hours after a cardiac arrest has been reported in multiple studies to indicate a uniformly poor neurologic prognosis (comatose patients after cardiac arrest who have these SSEP findings have uniformly poor outcomes), and it's become a well-established marker (among several) to identify patients with essentially no chance of a good neurologic recovery (see today's other post for more about 'good neurologic recovery'). 

In the last several years therapeutic hypothermia has been widely adopted for patients who survive initial resuscitative efforts after an arrest (essentially patients are cooled to ~32 deg. C for 24 hours - this is thought to attenuate some of the brain damage that occurs during/after an arrest and has been shown to improve neurologic recovery).  Most of the research on prognosis for comatose patients after cardiac arrest was done before hypothermia was adopted, so there's some concern that previously established prognostic criteria are no longer valid (or at least not as powerfully predictive).   See for instance this study, which attempts to investigate whether absent SSEPs still invariably predicts a poor outcome for patients who undergo hypothermia (it indicates they do).

The current article, which is a retrospective case series from a single German ICU in which they looked at all patients who underwent hypothermia and had SSEP measurement performed, presents data about two patients who suggest this isn't necessarily the case.  Of 36 patients with bilateral absent N20 responses, a 43 year old (who had absent N20 responses at day 3 after his arrest) went on to have a normal cognitive recovery (so much so that apparently he went on to resume drinking and they were able to measure SSEPs 18 months later when he was hospitalized and in alcohol withdrawal - a very sad detail they shared - he had recovered N20 responses at that measurement).  (They also report on a patient who had abnormal/barely detectable N20 responses who went on to have a normal cognitive recovery).

35 out of 36 patients with absent N20s did not recover, despite hypothermia.  (While that makes 97%, one should be cautious; this is a retrospective case series and it's inappropriate to establish event rates from this study design.)  One did.

My suspicion is that we will be seeing similar findings for other established 'uniformly bad' prognostic indicators in anoxic injuries.  It always seemed, in a perverse way, too good to be true - that there were not just one but several findings which implied, based on all the available research, that if a patient had one or more of them they have essentially no chance of recovery.  It will not be that these markers are worthless - they still will be (and this series supports it) - evidence of devastating injury with slim chance of recovery, but not some sort of 100%/clean cut/no questions asked evidence.  The world's too messy for that, especially when it comes to brain injuries.

I'm curious as to whether anyone is seeing this used in clinical practice, and having conversations with colleagues about the effect of  hypothermia protocols on how and when to examine patients and perform tests (for prognostic purposes), and how and when to talk with families about what we are concluding?

ResearchBlogging.orgLeithner C, Ploner CJ, Hasper D, & Storm C (2010). Does hypothermia influence the predictive value of bilateral absent N20 after cardiac arrest? Neurology, 74 (12), 965-9 PMID: 20308680

Pallimed | Blogger Template adapted from Mash2 by Bloggermint