Chapter 10: Cardiac Arrest and Anoxic Brain Injury

Many patients and their surrogates would consider limiting care if there is no hope for recovery. Therefore, it is important for medical providers to be aware of the prognostic possibilities at all stages of care so that accurate information can be provided to family and informed decisions can be made. Ideally, variables used for prognostication should produce few or no instances of “false positives.” In other words, we want to make sure that close to 100% of those labeled as having a poor prognosis do in fact have no chance of recovery.

Unfortunately, most elements of the neurologic examination immediately after cardiac arrest (CA) lack sufficient predictive value to provide accurate outcome prediction.^1-3 Prognostication based solely on a “poor examination” in the hours after CA is inadvisable.

After a complete cessation of cerebral perfusion, cerebral oxygen is depleted within 10 seconds and unconsciousness ensues.⁴ Within 20 seconds, electroencephalography (EEG) shows electrocerebral silence.^5,6 Cerebral adenosine triphosphate (ATP) and glucose stores are consumed within 5 minutes, resulting in dysfunction of ion pumps and channels, and loss of transmembrane sodium, potassium, and calcium gradients.^6-8 Membrane depolarization results in excessive release of excitatory neurotransmitters, which in turn leads to further accumulation of intracellular calcium. Calcium overload results in the activation of lipases, proteases, nucleases, and other destructive enzymes.⁹ Bulk inflow of ions brings with it water, which results in cell swelling. Free-radical production dramatically increases.⁹ Systemic and local anaerobic metabolism results in neuronal acidosis, leading to deranged function of a wide variety of proteins.¹⁰ Hyperglycemia can worsen this acidosis.^11,12 The process of cell death can begin as little as 1.5 to 5 minutes after complete cessation of cerebral blood flow (CBF), particularly within sensitive cell populations such as the CA1 hippocampal pyramidal neurons, cerebellar Purkinje neurons, medium spiny striatal neurons, and pyramidal neurons in layers 3, 5, and 6 of the neocortex.^13-15

Reperfusion leads to a rapid increase in ATP levels¹⁶ and reestablishment of ionic gradients- and is essential for the continued survival of neurons. Unfortunately, reperfusion also results in further damage. Arachidonic acid and other free fatty acids, which had been released by lipases during ischemia, undergo rapid oxidation. This leads to the production of superoxide radicals.^17-19 Xanthine dehydrogenase is converted to xanthine oxidase, ²⁰ which results in additional superoxide radical production. Free iron released from damaged proteins^21,22 catalyzes the destructive peroxidation of membrane lipids by superoxide radicals. Superoxide also reacts with nitric oxide to form peroxynitrite, which is not a free radical but is a potent oxidizer. The damaging effects of peroxynitrite and xanthine oxidase may be particularly important in vascular endothelium,^23,24 leading to increased permeability of the blood-brain barrier and impairment of vascular reactivity.

Numerous interventions have been studied to block the neurotoxic cascade after both global and local ischemic cerebral insults. Most of these have targeted specific steps in the cascade; for example, free- radical scavenging or calcium channel blockade. Despite promising animal studies, all drugs studied in humans have failed to show any clear benefit, including calcium channel blockers, barbiturates, benzodiazepines, magnesium, and glucocorticoids.^25-30 To date, only mild systemic hypothermia has been shown to improve outcome after CA.

Hypothermia after cardiac arrest (HACA) was initially studied in humans in the late 1950s, with some degree of possible success.^31,32 Peter Safar documents his use of HACA to 30°C in routine clinical practice during the 1950s and 1960s (Figure 10-1).³³ Despite this, additional human trials with HACA were not performed until the 1990s, when 3 small pilot studies showed mild HACA to be safe and feasible.^34-36

In 2002, an Australian group published the results of a 77-patient controlled trial of hypothermia versus normothermia after out-of-hospital cardiac arrest (OHCA).³⁷ They included men older than 18 years and women older than 50 years whose initial rhythm was VF, and who remained comatose after spontaneous circulation was successfully reestablished. Coma was not defined. Patients were excluded if they had a systolic pressure less than 90 mm Hg despite pressors or if there was another possible cause of coma. There was no true randomization—on odd-numbered days, patients were assigned to hypothermia. Paramedics applied ice packs to the head and torso in the field. On arrival at the ED, all patients received initial doses of midazolam and vecuronium as well as lidocaine infusion to prevent recurrent ventricular arrhythmia. They maintained a mean arterial pressure (MAP) of 90 to 100 mm Hg, Pao₂ of 100 mm Hg, Paco₂ of 40 mm Hg, and glucose less than 180 mg/dL. Hypothermia patients had additional ice packs applied and maintained until core temperature reached 33°C. Additional midazolam and vecuronium were given as needed for shivering. At 18 hours, the patients began to be actively rewarmed with heated-air blankets. Normothermia patients were maintained at 37°C throughout this period.

The primary outcome measure was good neurologic outcome, defined as discharge home or to an acute rehabilitation facility. Significantly fewer patients in the hypothermia group received bystander CPR. Despite this, 49% of hypothermia patients had a good outcome versus 26% of normothermia patients (P = .046). Mortality (51% versus 68%) was not significantly different. There were no meaningful, deleterious laboratory or clinical consequences of hypothermia.

Holzer et al carried out a larger, more rigorous 275-patient trial of HACA conducted through nine centers in five European countries.³⁸ Patients 18 to 75 years old who had a witnessed OHCA were included if they had VF or nonperfusing ventricular tachycardia (VT) on paramedic arrival, 5 to 15 minutes of downtime before commencement of CPR, and no more than 60 minutes from CA to ROSC. Patents were excluded if they had a temperature below 30°C, sustained hypoxia or hypotension before randomization, or known impediments to long-term follow-up. No specific treatment was initiated before ED arrival. Randomization was via envelope-concealed, computer-generated treatment assignments. All patients received pancuronium, fentanyl, and midazolam for 32 hours. Hypothermia patients were placed in an air-cooling device that enveloped their body, with a goal temperature of 32°C to 34°C. Ice packs were added if needed. Hypothermia was maintained for 24 hours from initiation, and then patients were allowed to passively rewarm. Control patients had normothermia maintained.

The primary outcome measure was good neurologic outcome at 6 months, as defined by a Cerebral Performance Category (CPC) (Table 10-1) of 1 or 2. CPC scoring was determined by a blinded examiner. Just as in the Australian HACA trial, bystander CPR was performed on significantly fewer hypothermia patients. Good outcome was attained by 55% of hypothermia patients versus 39% of normothermia patients (P = .009). Fewer hypothermia patients were dead at 6 months (41% versus 55%; P = .02). There was no significant difference in any of the tracked complications.

After the publication of these two controlled trials, the International Liaison Committee on Resuscitation (ILCOR) Advanced Life Support Task Force published the following recommendations³⁹:

Notice the ILCOR recommendation statement does include, “Such cooling may also be beneficial for other rhythms or in-hospital CA.” Similar recommendations remain in the 2010 ILCOR CPR guidelines.⁴⁰ A Cochrane review pooling individual data from the above two RCTs, plus data from a 30-patient RCT of helmet cooling,⁴¹ found an odds ratio of 1.55 (95% confidence interval 1.22 to 1.96) for good outcome at hospital discharge, with no significant difference in complications.⁴²

Hypothermia has multiple effects on the brain and the toxic cascades after hypoxia-ischemia, which differentiates it from prior failed therapies. Polderman eloquently describes the numerous, interrelated effects of hypothermia on the injured brain, observed primarily in animal studies.⁴³ These are divided into early and late mechanisms of action, as summarized in Figure 10-2. At the target temperature for HACA, cerebral metabolic rate (CMR) is reduced by nearly 50%. The reduction in CMR leads to decreases in the CBF threshold for ischemia, cerebral volume, and free-radical production. More energy is available for restoration and maintenance of neuronal ionic gradients, in turn reducing calcium overload, intracellular acidosis, and continued accumulation of glutamate. Oxidation from free radicals and peroxynitrite proceed at a slower rate. Hypothermia reduces disruption of the blood-brain barrier and improves endothelial function, thereby attenuating cerebral edema and, in turn, intracranial hypertension. Thromboxane A₂ and endothelin-1 levels are decreased, lessening vasoconstriction. A systemic inflammatory response⁴⁴ and activation of coagulation⁴⁵ normally occur after ROSC, and these are both attenuated by hypothermia. This may reduce cerebral microvascular thrombosis, in turn reducing ongoing ischemia. Activation of caspases is reduced, resulting in improved mitochondrial function and sparing some neurons from apoptosis. Finally, convulsive and nonconvulsive seizures occur commonly after CA, and hypothermia has been observed to have an antiepileptic effect in both animals and humans.⁴⁶

Table 10-2 lists some methods of cooling. Some modalities are suitable only for induction (eg ice-cold saline) or maintenance (eg, conventional cooling blankets). Some modalities are more likely to result in overcooling (eg, ice packs), which can lead to treatment complications such as bleeding and arrhythmias.⁴⁷ Cost, convenience, and effectiveness of each method need to be taken into consideration. Automated systems with patient temperature feedback are ideal from a convenience and effectiveness standpoint, but cost more.

Shivering leads to an increase in metabolic rate, including CMR, which is counterproductive.^48-51 Severe shivering will prevent core body temperature from falling. The pivotal trials used sedation and paralysis throughout the hypothermia period to prevent shivering.^37,38 Paralysis is the most effective means of preventing or treating shivering, but obscures the detection of seizures.⁵² EEG can be applied, but EEG would require frequent review by appropriately skilled professionals to detect seizures in a timely fashion. As a compromise, paralytics can be used solely in the induction phase to prevent shivering, and thereby speed cooling. Most patients do not shiver once at 33°C. Skin counter warming, buspirone, opioids, centrally acting α₂-agonists, and/or sedatives and anesthetics may be used to control shivering.⁴⁸

Levels of hypothermia are typically described as mild (32°C to 35.5 °C), moderate (26°C to 31°C), deep (20°C to 25 °C), and profound (less than 20°C). Experiments in cats and monkeys during the late 1970s found 48 hours of moderate hypothermia (29°C) to be neurologically detrimental in animals with focal cerebral ischemia and even in control animals.^53,54 It was speculated that increased blood viscosity might have impaired CBF. Using a dog model, Safar's team found 3 hours of moderate HACA to 32°C and 28°C resulted in less brain histologic damage compared with normothermia.⁵⁵ The 28°C group had more myocardial histologic damage, although this was not sufficient to impair cardiac function before sacrifice at 96 hours. In a separate study, they confirmed myocardial histologic damage in dogs after 1 hour of hypothermia to 15°C and 30°C compared with 34°C and 37.5°C.⁵⁶ Thereafter, the Safar group conducted all HACA studies using mild hypothermia.^57-59

Twenty-four hours of hypothermia is reasonable. The Safar group attained its best brain histology results with a 12-hour protocol of mild hypothermia.⁵⁹ Colbourne and Corbett conducted a series of experiments with gerbils, in which the best long-term histologic and functional outcomes were attained with 24 hours of mild hypothermia.^60,61 Prolonged hypothermia was also safe and efficacious in rats.⁶² The aforementioned human pilot studies and subsequent phase III trials were based on these temperatures and durations.^34-38 The European HACA trial,³⁸ the larger and more rigorously conducted of the two pivotal trials, maintained mild hypothermia for 24 hours.

RCTs of mild HACA have not shown any significant complications.^37,38 Hypokalemia, hyperglycemia, and cold diuresis are known to occur with mild hypothermia. Hypothermia less than 32°C can lead to bradycardia and ventricular arrhythmias, low cardiac function, immunocompromise, coagulopathy, and increased blood viscosity.⁶³ See Chapter 19 on therapeutic temperature modulation for further information.

Until further data are available, the best strategy may be to, at least, offer HACA to those PEA/asystole patients who are judged to have some chance of a recovery (eg, short time to ROSC, readily reversible cause of CA, younger or healthier patients).

PEA and asystole are the initial rhythms in 60% of all OHCA^64,65 although the proportion has been as high as 93% in some series.⁶⁶ The two major trials of HACA included only VF and/or VT patients, essentially excluding the majority of OHCA patients.^37,38 The large trials focused on VF/VT because it has a better outcome than PEA/asystole, and therefore smaller sample sizes would suffice to show a treatment effect.

A word of caution before moving forward: One must be careful in interpreting outcome rates for OHCA, as this is heavily dependent on the choice of denominator.⁶⁷ Did the study in question include all patients encountered by EMS, most of whom did not obtain ROSC? This would create the largest possible denominator, and yield the poorest outcome rates. Other studies look only at those who attained ROSC and remained alive by ED arrival. Still others focus only on those who survived to ICU admission. Any data on neuroprotective strategies in PEA/asystole patients should be compared against the outcomes of this last subset of PEA/asystole patients, as other patients would not be amenable to neuroprotective treatment. In the pre-HACA era, 17% to 26% of these patients survived to discharge^68-71 (Table 10-3). In addition, one must be cognizant of the types of patients included in these series. Some studies include only patients whose arrest was cardiac in origin (ie, no drug overdoses, choking victims) or whose time to ROSC was within a certain limit.

Oddo et al prospectively collected data after the institution of a HACA program.⁷² Of 74 consecutive patients (excluded age ≥80 years or known terminal disease), 36 had PEA/asystole, and all patients underwent HACA. Seventeen percent of PEA/asystole patients survived to discharge, which is in line with survival estimates in the pre-HACA era discussed above. Using stepwise logistic regression, only time to ROSC and lactate levels were found to be associated with poor outcome, whereas initial rhythm, shock, age, and gender were not. The investigators concluded that PEA/asystole carries a poorer chance of recovery compared with VF/VT because PEA/asystole generally has a longer time to ROSC, not because of anything inherent to the rhythms of PEA/asystole.

Nielsen et al published data from an international HACA registry.⁷³ PEA/asystole was the initial rhythm in 283 of 986 patients, and 30% of these survived to discharge (73% of these survivors with good function). The exclusion criteria were left up to individual centers, and these criteria were not disclosed, so these results may reflect a selection bias.

Don et al published their single-center experience before and after institution of a HACA protocol.⁶⁹ They found no difference in outcomes for PEA/asystole after they began cooling patients. However, there are several issues with the analysis. Patient ascertainment and data collection were done retrospectively, and all patients who presented to the ED were included even if they did not survive to ICU admission or receive HACA. The latter would certainly dilute any effect of hypothermia for PEA/asystole patients. It is stated that an analysis restricted to patients admitted to the ICU failed to show any difference for PEA/asystole patients, but further details were not provided.

Cronberg et al cooled consecutive CA patients⁷⁴ unless there was a comorbidity which assured a poor outcome or time to ROSC was long/unknown (T. Cronberg, MD, personal communication via email, January 17, 2011). PEA/asystole was the initial rhythm in 29 of 94 patients. Although outcome at discharge is not provided, 31% of PEA/asystole patients had good functional outcome at approximately 6 months. This is substantially better than even survival to discharge in the pre-HACA era.^68-71

As it stands, there is no clear data to advocate one way or the other for HACA in PEA/asystole. Based on the above studies and others,^75-79 there is no hint that HACA is less safe in PEA/asystole patients compared with VF/VT. The 2010 ILCOR guidelines endorse HACA as an option for PEA/asystole.⁴⁰

The European HACA RCT did not enroll patients with persistent shock. The Australian RCT did enroll such patients, but did not report results separately. Both trials included patients who received thrombolytics for acute myocardial infarction without apparent sequelae.^37,38 Subsequent data have not shown any significant adverse effects of HACA for patients requiring percutaneous coronary intervention, receiving thrombolytics, or in persistent shock.^{69,72-74,80-86}

Cerebral autoregulation is known to be impaired after CA,^87,88 therefore higher MAPs may be needed to ensure adequate CBF and avoid secondary ischemia. Higher MAPs have been repeatedly correlated with better outcome, but no prospective data exists to demonstrate a causal association.^89-93 To avoid excessive strain on an already injured myocardium, the minimal MAP necessary to support CBF would be ideal. In this case, measuring surrogates of CBF, such as oxygen saturation at the jugular bulb (Sjvo₂)^88,94,95 or partial pressure of oxygen in brain tissue (Pbto₂) may aid in determining the appropriate MAP.

Many physiologic parameters other than MAP are known to affect CBF and cerebral oxygen supply and demand (Table 10-4). Hypocapnia has been found to cause cerebral ischemia in a number of disease states, including after the global cerebral insult of CA.^53,94-96 Intracranial pressure has been looked at prospectively after CA, and values > 25 mm Hg occur in at least 26% of patients.⁹⁷ Seizures and status epileptcus dramatically increase CMR, and are common after CA, occurring in up to 36%⁹⁸ and 12%^52,99 of patients, respectively. In the past, status epilepticus after CA was felt to guarantee a poor outcome, but anecdotal evidence and case reports¹⁰⁰ suggest that this may not be true after hypothermia, provided that status epilepticus is aggressively treated. Fever also increases CMR and is very common after CA.^101-103

Hyperoxemia has been shown to exacerbate neurologic injury after CA in animals,^104-106 and has been correlated with worse outcome in humans. Kilgannon et al found even small increases in Pao₂ to be associated with poor functional outcome and mortality.¹⁰⁷ Each 25–mm Hg increment in Pao₂ was associated with a 6% increase in mortality rate.

Figure 10-3 shows the current American Academy of Neurology (AAN) evidence-based algorithm for prognosticating after CA.¹⁰⁸ It is important to note that this is based on data from the pre-HACA era, and these guidelines are NOT applicable to patients who were cooled. The presence of the following essentially guarantees poor functional outcome or death if the patient was not cooled:

1. Absent pupillary or corneal reflexes on day 3 (but not before)

2. Bilateral extensor posturing or absent motor responses on day 3 (but not before)

3. Myoclonus status epilepticus within 24 hours (Note: The AAN refers to multifocal jerking, not an electrographic phenomenon, to define this)

4. Serum neuron-specific enolase greater than 33 μg/L on days 1-3

5. Somatosensory evoked potentials (SSEPs) showing bilaterally absent N₂₀ responses

For patients who underwent HACA, the following independent features CANNOT provide foolproof prognostication:

1. Motor examination^109,110

2. Serum neuron-specific enolase¹¹¹

3. Early myoclonus¹⁰⁹

4. Pupillary or corneal reflexes¹⁰⁹

Rossetti et al prospectively looked at what factors can serve as accurate guides after HACA.¹⁰⁹ Bilaterally absent N₂₀ responses on SSEPs, an unreactive EEG, or the combination of early myoclonus and absent pupillary or corneal reflexes provided certainty of a poor outcome with no false positives. However, a larger retrospective study did have 1 patient with bilaterally absent N₂₀ responses who made an excellent functional recovery.¹¹² Until more prospective data is available on prognosticating in HACA patients, it would be prudent to proceed with caution.