New rules for predicting recovery after cardiac arrest

Therapeutic hypothermia has changed the field of neurology after cardiac arrest.


Neurological examination can help answer the biggest questions after a cardiac arrest, Vanja Douglas, MD, told attendees at the annual meeting of the American Academy of Neurology (AAN), held in Philadelphia in May.

“For those that don't wake up, we have to figure out are they going to do well or are they not going to do well? And what can we tell families?” said Dr. Douglas, an associate professor of clinical neurology and chief of the neurohospitalist division at the University of California, San Francisco.

Image by Getty Images
Image by Getty Images

There used to be an official tool for this task, a practice parameter published by the AAN in 2006. “It goes through these exam maneuvers and tests that you can do to decide whether or not your patient is going to have a good prognosis. The problem with this is that all of the data that went into that practice parameter was derived from patients before the routine use of targeted temperature management,” he said.

It was back in 2002 that temperature management began to dramatically change care, he explained. “There was one large and one small randomized trial looking at therapeutic hypothermia after cardiac arrest in patients who were still unresponsive after their arrest.” The trials, published together in the Feb. 21, 2002, New England Journal of Medicine, found that survival and good neurological outcomes were significantly more likely if patients' body temperature was kept at 32 °C to 34 °C for 24 hours after arrest.

“This was a pretty dramatic result,” said Dr. Douglas. “The number needed to treat here was around 6, so a lot of centers started adopting this therapeutic approach.”

About a decade later, the approach was revised again, after another study, published by the New England Journal of Medicine on Dec. 5, 2013, showed that keeping patients' temperatures to 36 °C worked as well as 33 °C. Current guidelines are based on this finding, Dr. Douglas noted.

“In patients who have an out-of-hospital cardiac arrest with a shockable rhythm, the recommendation is a targeted temperature between 32 °C and 36 °C for 24 hours,” he said. “That's in patients who aren't awake and following commands following their resuscitation.” Because the benefits of targeted temperature management have been seen almost entirely in patients with ventricular fibrillation and pulseless ventricular tachycardia, guidelines only suggest, not recommend, this protocol for patients with nonshockable rhythms, he noted.

The positive result of these changes has been an improvement in the overall low survival rates from arrests. “About 80% die and never make it to the hospital,” said Dr. Douglas, citing a 2014 study published in Circulation. Of those who make it to the hospital, only a fraction will survive, but targeted temperature management shifted that percentage from 39% to 55%, according to his report of pooled data from the three studies.

The challenge is in how the new protocols affect traditional prognostic indicators. “Both hypothermia and the sedation that goes along with treating patients with hypothermia can confound the neurologic examination and cause delay in the recovery of neurologic function. So the neurological exam at 24 hours does not provide reliable information anymore,” said Dr. Douglas.

But although it might involve waiting a bit longer, there are still neurological findings that offer a lot of prognostic information for these patients and families. He reviewed the recent evidence and added some statistical calculations of his own.

“The literature talks about false-positive rates,” he explained. “I would argue that patients and families don't really care about false-positive rates; they care more about the positive predictive value of each examination finding or test.”

He offered the example of status myoclonus after arrest. “Families want to know the chance of recovery in similar patients, who have also shown status myoclonus after arrest,” said Dr. Douglas. Research on the question has found a positive predictive value of 98.3% with status myoclonus. “You can tell the family that there's a 1.7% chance that Granddad is going to recover based on similar patients,” said Dr. Douglas.

Another important consideration when talking to families is what it means for a patient to recover. The studies in this area consider good outcomes to be a cerebral performance category (CPC) of 1 (minimal disability) or 2 (moderate disability, but still independent in activities of daily life).

“A CPC of 3 [conscious, but dependent on others for daily support because of impaired brain function] is acceptable to some families,” he said. “You have to be a little bit careful when you're talking to families and emphasize that the poor outcome we are trying to predict ranges from severe disability with dependence to a vegetative state.”

It's impossible to estimate probability among those worse outcomes, because of how the research has been structured. “When the outcome was thought to be likely a CPC of 3, 4, or 5, in many of these studies families were counseled about the likelihood of this poor outcome, and then support was withdrawn and the patient died. We don't really know the six-month outcome in many of the patients who have these negative prognostic signs because they weren't kept alive long enough to find out,” said Dr. Douglas.

With those caveats, some simple neurologic tests can predict likely outcomes for unconscious patients, beginning with the pupillary light reflex. “If the pupillary light reflex is absent at 72 to 108 hours after cardiac arrest, there's a 98% chance your patient is going to have a bad outcome,” he reported.

At the same timepoint, an absent corneal reflex is associated with a 97% chance of a bad outcome. “Still a pretty reliable prognostic sign, but some people might say, ‘Hey, I'll take that 3% chance for Dad or Granddad,’” said Dr. Douglas.

Motor response to pain is one of the tests that has been most affected by the changes in cardiac arrest care. “This is something that before targeted temperature management, we considered a pretty reliable prognostic sign. You give pain in the extremities and if patients have extensor posturing or no response at 72 hours, that was considered a very reliable sign of poor prognosis,” he said. However, according to the latest data, such patients have a 7% chance of a good outcome.

Instead, you might want to rely on somatosensory evoked potentials (SSEPs). “This is probably the best test for prognosis. There are a lot of patients in whom this has been studied,” said Dr. Douglas. Among patients with bilaterally absent median SSEPs, only 1% survived with a CPC of 1 or 2.

There's one currently available blood test, neuron-specific enolase, that can help, although the cutpoints for it vary in different studies.

“In the original practice parameter, the cutpoint for neuron-specific enolase was 33 [ng/mL]. If the value was over 33, the prognosis was considered poor,” said Dr. Douglas. However, research in the target temperature management era has identified cutpoints of 76, 68, and 45 ng/mL at 24, 48, and 72 hours, respectively.

Another blood test, neurofilament light chain, has recently showed promise, in a study published in the January 2019 JAMA Neurology. “It looks promising as a prognosis biomarker, but it's only been in that one very large study,” Dr. Douglas said. “It has to be validated in another study, probably prospectively, and then also become commercially available. Stay tuned.”

As previously mentioned, status myoclonus is a predictor of bad outcomes. “But not universally bad. When the first papers were published saying that this is a universally bad prognostic sign, then people started publishing case reports saying ‘Well, actually I had a patient who had this and survived,’” said Dr. Douglas.

Status myoclonus is defined clinically as continuous, prolonged, repetitive, unrelenting, generalized multifocal myoclonus involving the face, limbs, or axial musculature, he noted. It doesn't require an electroencephalogram (EEG), but studies that did perform EEGs on such patients have had interesting prognostic findings.

The patients' EEGs generally fit into certain patterns. Most of the patients in a study published in Annals of Neurology in July 2016 had an EEG phenotype with bursts and suppression of the background, and they died. But a few had a different pattern with more variable background frequencies, and half of those had good outcomes.

“It's interesting, and it suggests that the EEG might be a way that we can stratify patients, especially those with status myoclonus, into a good outcome group and a poor outcome group,” he said.

EEG may be similarly useful in the broader population of patients unconscious after cardiac arrest. A study of patients who underwent targeted temperature management, published in Neurology on April 19, 2016, identified a number of EEG features, such as suppressed background, that were categorized as malignant, highly malignant, or benign.

“Having any malignant feature on the EEG captured almost all the patients with a bad outcome,” said Dr. Douglas. “On the flip side of that, if you had a benign EEG, it was very likely you'd have a good outcome.”

He noted that this makes EEG relatively unusual among the prognostic signs, in that it offers the ability to predict a positive outcome, not just a negative one. “If patients are not going to have a bad outcome, we often can't tell their family that they're going to have a good outcome. We often have to say it's an indeterminate situation,” he said. “Now, if there's a benign EEG, we can be pretty confident that patients are going to do well, which is a new aspect of prognostication.” In order to indicate a good prognosis, the EEG has to be reactive, he added.

Obviously, a physician trying to offer a prognosis will likely apply more than one of these tests, so researchers have also looked at how they work in combination, in studies published in Annals of Neurology in 2010 and Neurocritical Care in 2018. SSEPs, EEG, myoclonus, brainstem reflexes, and, in the later study, neuron-specific enolase were included.

“If you have two or more poor prognostic signs, then there's virtually no chance of a good recovery. And if you have only one or zero poor prognostic signs, then patients actually have a pretty good chance of a good recovery,” said Dr. Douglas.

The latter will be easier to tell families, but remember not to hurry things if the prognosis is bad, he advised in conclusion. “This process doesn't have to be done within 24 hours. There's never a rush to withdraw care. There's always time. You can always wait until tomorrow to get more information before you have that family meeting.”