Year in review 2010: Critical Care - neurocritical care

In 2010 Critical Care published a large number of articles on critical care aspects of neurologic and neurosurgical conditions. These aspects included investigation of diagnostic criteria for bacterial meningitis, critical illness myopathy and their relationship to systemic inflammation. A number of studies investigated the biology of sepsis-related delirium, its biomarkers, its relationship to inflammation and its impact on outcome. Other teams reported on the use of magnetic resonance imaging, biomarkers and electroencephalogram to predict outcome in patients who were comatose following cardiac arrest. Our understanding of the pathophysiology as well as management of subarachnoid hemorrhage was addressed in several papers. Topics included the effect of hemodynamic treatment of delayed cerebral ischemia, pulmonary edema and the impact of subarachnoid hemorrhage on endocrine function. Finally, outcome from neurocritical care and patients' retrospective willingness to consent to the treatment they received were reported.

therapy [1], yet distinguishing between bacterial and non bacterial meningitis can sometimes prove diffi cult. Th e use of cerebrospinal fl uid (CSF) lactate, as opposed to conventional tests (such as CSF glucose, CSF/plasma glucose ratio, CSF protein concentration, and CSF leukocyte count), has been investigated in a number of studies to distinguish between bacterial and nonbacterial meningitis.
Huy and coworkers performed a literature review and meta-analysis to evaluate the usefulness of CSF lactate concentration for this purpose [2]. From the 25 studies they identifi ed, the authors concluded that CSF lactate alone had a high degree of accuracy in distinguishing between bacterial and nonbacterial meningitis and performs better than the conventional tests routinely used. CSF lactate was found to be less useful if its concentration was low, but when elevated it was helpful, especially if the diagnosis was otherwise inconclusive. Th is suggests that any elevation in CSF lactate concentration above normal for the assay used could be employed as a diagnostic marker despite the diff erence in cut-off values caused by variance in methods, instruments and hospital laboratories. While the authors conclude CSF lactate is a useful marker to distinguish between bacterial and nonbacterial meningitis, it is not meant to replace conventional tests as they are necessary to diagnose meningitis. Rather, interpretation of lactate alone is a better discriminator between bacterial and nonbacterial meningitis than conventional tests. As a result, measurements of CSF lactate may be considered part of standard practice to distinguish between bacterial and nonbacterial meningitis.

Critical illness myopathy
Critical illness myopathy (CIM) frequently complicates serious illness and often results in prolonged mechanical ventilation. Th is complication is characterized by the atrophy of type II muscle fi bers and thick fi laments, as well as nonexcitable muscle membranes [3]. Risk factors for developing CIM are not well understood.
Weber-Carstens and colleagues performed a pros pective observational study of 40 ICU patients with a Simplifi ed Acute Physiology Score II ≥20 to identify risk

Abstract
In 2010 Critical Care published a large number of articles on critical care aspects of neurologic and neurosurgical conditions. These aspects included investigation of diagnostic criteria for bacterial meningitis, critical illness myopathy and their relationship to systemic infl ammation. A number of studies investigated the biology of sepsis-related delirium, its biomarkers, its relationship to infl ammation and its impact on outcome. Other teams reported on the use of magnetic resonance imaging, biomarkers and electroencephalogram to predict outcome in patients who were comatose following cardiac arrest. Our understanding of the pathophysiology as well as management of subarachnoid hemorrhage was addressed in several papers. Topics included the eff ect of hemodynamic treatment of delayed cerebral ischemia, pulmonary edema and the impact of subarachnoid hemorrhage on endocrine function. Finally, outcome from neurocritical care and patients' retrospective willingness to consent to the treatment they received were reported. factors for CIM. Patients were evaluated on three consecutive days within 1 week of ICU admission [4]. Th e authors investigated demographics, infl ammatory, metabolic and hemodynamic parameters, and medical treatment prior to the fi rst nonelectromyography signs of CIM. On univariate analysis, infl ammation, disease severity, decreased insulin sensitivity, catecholamine use and sedation requirements were all signifi cantly associated with CIM. Low-dose hydrocortisone treatment in septic shock and adminis tration of neuromuscular blocking agents or aminoglyco sides, however, were not related to CIM. In a multivariate Cox proportional hazard regression analysis, IL-6 was the only independent risk factor for CIM (95% confi dence interval (CI) = 1.00 to 1.01; P = 0.005); however, the overall eff ect was small. Th e authors hypothesize that impairment of growthfactor-mediated intracellular signal ing induced by systemic infl ammation is important in the patho physiology of CIM. A weakness of this study was that subject blood samples were only taken at two separate time points, which makes it impossible to determine the daily course of the infl ammatory parameters.

Delirium/encephalopathy
In hospitalized patients, delirium is associated with increased mortality, length of stay and cost [5,6]. Th e epidemiology of delirium in ICUs is poorly understood, however, in part because its evaluation has been limited to single-center studies. Salluh and coworkers performed a 1-day point-prevalence study in 104 ICUs [7]. Patients were excluded if they had Glasgow Coma Scale (GCS) <14 from a primary neurologic diagnosis, were legally blind or deaf, could not speak the language of the country where the ICU was located, or were moribund. Of the 975 patients screened, only 497 were enrolled. Because of the use of sedative drugs, only 232 (46.6%) of the enrolled patients could be evaluated. Of the patients evaluated, 75 met the diagnostic criteria for delirium using the Confusion Assessment Method for the ICU. Th e authors found delirium to be associated with higher mortality, and longer ICU and hospital length of stays. Th is suggests that sicker patients have a higher risk of developing delirium, as well as a higher risk of staying longer or dying in the ICU. Th e main modifi able risk factors were the use of invasive devices and sedatives (midazolam). Th is study confi rmed fi ndings from single-center studies indicating that delirium is common in ICU patients and is associated with longer length of stay and adverse outcomes. Important weaknesses of this study are the high number of patients excluded and that so many patients could not be assessed due to sedation. Hence, only 24% of those patients originally screened were evaluated for delirium. Th is raises an important question of whether we are sedating too many patients in the ICU.
In addition, the true incidence of delirium in the ICU is diffi cult to determine from this study due to the exclusion of the majority of patients.
Another important insight into delirium was reported by Mu and colleagues in a prospective cohort study [8]. In postoperative patients who had undergone coronary artery bypass graft surgery they measured serum cortisol levels and assessed their relationship to delirium. A total of 243 patients were assessed for delirium twice daily during the fi rst fi ve postoperative days using the Confusion Assess ment Method for the ICU. Th e presence of delirium was compared with serum cortisol measured on the morning of the fi rst postoperative day. Postoperative delirium was identifi ed in 50% of patients (n = 123). A high serum cortisol level (odds ratio (OR) = 3.09, 95% CI = 1.76 to 5.41, P <0.001), occurrence of complications during the fi rst day after surgery (OR = 2.49, 95% CI = 1.18 to 5.21, P = 0.016), a history of diabetes mellitus (OR = 1.91, 95% CI = 1.00 to 3.62, P = 0.049), prolonged duration of surgery (OR = 1.36, 95% CI = 1.01 to 1.83, P = 0.043) and increasing age (OR = 1.11, 95% CI = 1.06 to 1.16, P <0.001) were all associated with increased risk of postoperative delirium. Postoperative delirium was also associated with increased length of stay and with worse outcome.
Th is study has some limitations. Serum cortisol levels were not measured at baseline and were measured only once after surgery. As a result the data could potentially be skewed prior to surgery by including patients who had elevated cortisol and who might have been at greater risk of developing postoperative delirium due to preexisting conditions. In addition, the investigators did not perform baseline psychiatric or cognitive screening tests. Any mental disorders present prior to surgery could infl uence the development of postoperative delirium.
Systemic infections can cause serious cognitive dysfunction in ICU patients, which may be related to proinfl ammatory cytokines. IL-1 is a likely cytokine candi date due to its infl uence on both the immune system as well as the central nervous system. To address this question, Terrando and colleagues investigated the development of LPS-induced cognitive dysfunction in wild-type and IL-1 knockout mice [9]. After lipo polysaccharide (LPS) challenge, performance on behavioral tests of memory and response to novel environments deteriorated and reactive microgliosis was seen in the hippocampus. In addition, plasma TNFα, IL-1β, IL-6, and high-mobility group box-1, IL-6 and IL-1β mRNA all increased. Administration of IL-1 receptor antagonist signi fi cantly reduced plasma cytokines and hippocampal micro gliosis, and ameliorated cognitive dysfunction; similar results were seen in IL-1 knockout mice without receiving IL-1 receptor antagonist. Th ese data suggest that behavioral dysfunction could be prevented by blocking the IL-1 signal, attenuating the infl ammatory cascade. Future studies would need to be carried out to determine whether responses are seen in patients.
Sepsis-associated encephalopathy (SAE) is defi ned as diff use cerebral dysfunction that is caused by a systemic response to infection without any evidence of a central nervous system infection [10,11]. However, the pathophysiology of SAE is not well understood. In a prospective cohort study, Berg and colleagues found that brain function might be impaired during systemic infl am mation due to a reduction in the plasma ratio between branched-chain amino acids and aromatic amino acids (BCAA/AAA ratio), which alters their uptake into the brain [12]. Twelve healthy adult male subjects (aged 20 to 33) were enrolled. After an overnight fast, all subjects underwent measurements of cerebral blood fl ow (CBF), plasma BCAA/AAA ratio, cerebral delivery, net exchange of large amino acids and ammonia level. Paired arteriojugular venous blood samples were taken at baseline and after a 4-hour continuous intravenous infu sion of purifi ed Escherichia coli LPS. Th is LPS infusion resulted in an increase in plasma phenylalanine and a signifi cant decrease in the concentration of all other large neutral amino acids except for isoleucine. As the BCAA/AAA ratio decreased, there was an increase in the levels of the neurotoxic AAA phenylalanine -suggesting a relationship to encephalopathy. In this study, however, the investigators did not perform cognitive tests on the subjects. Th ere fore, while systemic infl ammation may aff ect brain function by reducing the BCAA/AAA ratio, it is diffi cult to draw any conclusions regarding the role of the BCAA/AAA ratio in SAE until cognitive function is assessed.
In another study investigating the pathophysiology of SAE, Szatmári and colleagues assessed cerebrovascular reactivity using an acetazolamide challenge in SAE patients [13]. In a prospective cohort, patients with clinical sepsis and disturbances of consciousness of any degree were studied. In addition, 20 age-matched and sex-matched controls without previous diseases aff ecting cerebral vasoreactivity were also included. Transcranial Doppler blood fl ow velocities were then measured at baseline and 5, 10, 15 and 20 minutes after intravenous administration of 15 mg/kg acetazolamide. Th e time course of the CBF velocity response to acetazolamide was measured and cerebrovascular reactivity was calculated (the percentage increase in the mean blood fl ow velocity). In addition, the maximum vasodilatory eff ect (cerebrovascular reserve capacity) was determined. Th e authors found that the time course of the vasomotor reaction to acetazolamide was delayed and the maximal vasodilatory response was reduced in SAE patients. Th e authors propose that cerebral hemodynamic changes might be involved in early pathogenesis of SAE. SAE is common in patients with septic shock and has been associated with elevated levels of brain-specifi c S100β protein. In some studies, treatment of septic patients with drotrecogin alfa (DrotAA) reduces mortality in patients with severe sepsis-induced organ failure. Spapen and coworkers sought to determine whether DrotAA would reduce S100β levels and sepsis-associated encephalopathy [14]. Patients with pre-existing or acute neurological disorders were excluded. Th ose enrolled were classifi ed into two groups based on their level of consciousness: GCS ≥13 and GCS <13. DrotAA was given as a continuous infusion of 24 μg/kg/hour for 96 hours. Th e level of S100β was measured before sedation and serially after the start of DrotAA infusion. Of the 54 patients in the study, one-half had elevated S100β levels. In patients with GCS <13 receiving DrotAA, S100β levels fell from baseline at 64 and 96 hours (both P <0.05). In addition, those who received DrotAA had signifi cantly lower S100β levels at 64 and 96 hours compared with those who did not receive treatment. In patients with GCS ≥13, however, DrotAA had no eff ect on S100β levels. Th ese data suggest an association between the severity of SAE and serum levels of S100β, but the measure of encephalopathy was limited.

Cardiac arrest
Our ability to predict neurological outcome of patients who remain comatose following cardiac arrest is limited. In addition, the existing data must be re-evaluated due to advent of treatment with hypothermia. Choi and colleagues sought to determine whether diff usion-weighted magnetic resonance imaging (MRI) could improve prediction of outcome [15]. Th ey studied 39 patients who suff ered outof-hospital cardiac arrest, performing diff usion-weighted MRI within 5 days of cardiac arrest and then comparing functional outcome at 3 months. Patients were treated with hypothermia in the latter half of the study, limiting interpretation. Diff usion-weighted MRI fi ndings were categorized into four patterns: normal, isolated cortical injury, isolated deep gray nuclei injury, and mixed injury (cortex and deep gray nuclei). Twenty-three subjects with normal MRI scans served as controls. Th e authors found that a mixed pattern of brain injury involving cortex and deep gray nuclei on MRI predicted an unfavorable outcome, defi ned as a Glasgow Outcome Scale (GOS) score of 1 to 3.
Another approach to predicting neurological outcome of cardiac arrest patients involved the use of serum biomarkers. Steff en and colleagues investigated the ability of neuron-specifi c enolase to predict outcome from cardiac arrest treated with hypothermia [16]. Ninety-seven subjects were compared with 133 historical cardiac arrest patients not treated with hypothermia. In this observational prospective cohort study, serum neuron-specifi c enolase was measured 72 hours after ICU admission and the neurological outcome (Pittsburgh cerebral performance category) was recorded at ICU discharge. In univariate analysis, neuron-specifi c enolase serum levels were lower in patients treated with hypothermia. Th e 72-hour neuron-specifi c enolase cut-off levels, however, did not reliably predict poor neurological outcome. Th e authors also note that 1-year survival was signifi cantly higher for the hypothermia group compared with the historical nonhypothermic controls.
Rossetti and coworkers focused on the use of continuous electroencephalogram (EEG) to predict outcome in cardiac arrest patients treated with hypothermia [17]. In a cohort of 34 patients, continuous EEG demonstrating nonreactive or discontinuous background during hypothermia was strongly associated with unfavorable outcome in their patients. In addition, nonreactive background EEG was 100% predictive of mortality (95% CI = 74 to 100%) with a false-positive rate of 0 (95% CI = 0 to 18%). Th ese data suggest that early continuous EEG fi nd ings might be a better predictor of outcome than biomarkers or imaging.

Spontaneous aneurysmal subarachnoid hemorrhage
Th e most common and potentially treatable cause of secondary neurological injury after aneurysmal SAH is delayed cerebral ischemia (DCI). DCI occurs several days after hemorrhage and is considered to result from reductions in CBF below critical ischemic thresholds. Several factors are thought to contribute to DCI including large artery vasospasm, impaired cerebral autoregulation and intravascular volume depletion [18]. Triple-H therapy (hypervolemia, hypertension, hemodilu tion) is the mainstay treatment of DCI with the goal of increasing cerebral perfusion pressure. Dankbaar and colleagues performed a systematic review of the literature address ing the eff ectiveness of triple-H therapy and its com ponents on clinical outcome [19]. Eleven studies of triple-H therapy were identifi ed and included in their meta-analysis. Th e authors concluded that hemodilution did not change CBF and that only one of the seven studies on hypervolemia showed a signifi cant increase in CBF. Th ere was a signifi cant increase in CBF in two of the four studies applying hypertension and in one study using triple-H therapy. None of these studies used control groups or randomi zation. Th e authors conclude that there is no evidence from controlled studies for a benefi cial eff ect of triple-H therapy or its separate components on CBF in SAH patients.
Th e conclusion that there is no evidence to support triple-H therapy is misleading, however, as the treatment has not been properly studied to date. Th is systematic review points out the lack of well-designed studies.
Owing to the amount of underpowered and poorly designed studies, it is diffi cult to come to any conclusions regarding eff ectiveness.
In a prospective cohort study, Bendel and coworkers evaluated neuroendocrine status after SAH and its impact on clinical outcome [20]. Th ey enrolled 30 subjects and 16 controls. Th e study focused on insulin-like growth factor-1 (IGF-1), which is associated with neuronal growth, cell death and apoptosis after ischemic stroke [21]. Th e authors found that serum IGF-1 concen trations were signi fi cantly lower in SAH patients on days 1 to 5 compared with controls (P = 0.01). Th ere was no signifi cant diff erence at 3 months between the groups. In addition, there was no relationship between the severity of SAH and the IGF-1 concentration. Th e data showed patients with GOS ≤4 had lower IGF-1 concen trations and lower health-related quality of life compared with patients with higher GOS scores. Th is is the fi rst pub lished study to evaluate IGF-1 in the acute phase of aneurysmal SAH. Th e authors suggest that low IGF-1 following SAH might be associated with poor outcome or death.
Pulmonary edema is common after SAH [22], but its management is complicated by the need to rigorously avoid hypovolemia in these patients. Hoff and colleagues prospectively studied intravascular volume (using pulse dye densitometry) in 102 SAH patients, comparing those with and without pulmonary edema [23]. Surprisingly, those patients who developed pulmonary edema (17 out of 102) had lower mean circulating blood volume than those without. As a result, fl uid management in SAH patients that develop pulmonary edema is more diffi cult. Th e authors concluded that treatment for pulmonary edema with diuresis must be weighed against the risk of worsening hypovolemia in SAH patients and thus increasing the risk of DCI.

Outcome from neurocritical care
Despite advances in diagnosis and treatment, patients admitted to a neurocritical care unit have poor functional outcome and high mortality [24]. In a retrospective review of 796 consecutive patients admitted to a nonsurgical neurocritical care unit, Kiphuth and colleagues found an in-hospital mortality of 22.5%. Good long-term functional outcome (determined by the modifi ed Rankin Scale) was seen in only 28.4% [25]. Variables found to be independently associated with outcome were admission diagnosis, age, duration of mechanical ventilation and Th erapeutic Intervention Scoring System-28 on day 1. When diagnoses were compared, the authors found that patients with Guillain-Barré syndrome and myasthenia gravis had a good long-term outcome more often than those with other diagnoses.
In another study, Kiphuth and colleagues investigated the relationship between functional outcome and retro spective agreement to neurocritical care [26]. Th e purpose of this study was to determine whether retrospective consent to neurocritical care was infl uenced by func tional outcome. In this study of 704 consecutive patients admitted to a nonsurgical neurocritical care unit, 91% of patients who were independent at 1 year would again agree to the care they received. Only 19% of those who were dependent for activities of daily living, however, would have agreed retrospectively to neurocritical care. Th e authors conclude that eff orts must be taken to increase public awareness regarding the importance of advanced directives, and further research must focus on determining highly specifi c and sensitive predictors for outcome in these patients.
In patients suff ering from nontraumatic supratentorial intracerebral hemorrhage, the infl uence of anemia on functional outcome and mortality was investigated by Diedler and colleagues. In a retrospective cohort study of 196 patients, 23.5% had a favorable outcome (moderate disability or better) [27]. Th e mean hemoglobin concentration was signifi cantly lower (12.3 versus 13.7 g/dl, P <0.001) in those with poor outcome. On multivariate analysis, the independent predictors of poor outcome at 90 days were mean hemoglobin, age and NIH Stroke Scale score on admission. Th e authors conclude, however, that they could not determine whether anemia causes further brain injury or is a marker of severe injury. Despite the diff erence in mean hemoglobin in those patients with poor outcome, this is of uncertain clinical signifi cance.
Outcome after traumatic brain injury is diffi cult to predict. Recent investigations have looked into a variety of biomarkers to better predict and monitor outcome after brain trauma. Biomarkers investigated include glial fi brillary acidic protein [28], neurofi lament light chain [29], S100 [30], tissue transglutaminases [31], ubiquitin [32], β-amyloid precursor proteins [33] and tau [34]. Dong and colleagues investigated the use of the biomarker resistin predicting outcome after traumatic brain injury. Th is group previously reported that elevated levels of resistin are found in blood of patients with intracerebral hemorrhage and that higher levels were asso ciated with poor outcome. Th e authors subsequently studied a cohort of 94 patients with severe traumatic brain injury and compared their resistin levels with those of 50 healthy controls [35]. Plasma resistin levels were elevated during the initial 6 hours after injury and peaked within 24 hours; the patients' levels were substantially higher than those of the healthy controls. Plasma resistin levels were negatively correlated with GCS (P <0.001) and were an independent predictor of 1-month survival. Th e data suggest that the plasma resistin level might refl ect the initial brain injury and could become a useful biomarker to predict survival in these patients.

Conclusion
In 2010 the fi eld of neurocritical care received increased attention, with an increasing number of articles being published in Critical Care. Th e issues of diagnostic criteria for meningitis and critical illness myopathy were further investigated. In addition, studies of the epidemiology, risk factors, and treatment for delirium and encephalopathy in the ICU were published. Aspects of cardiac arrest, aneurysmal SAH and outcomes in neurocritical care were re-addressed using new approaches. Overall, there were many quality articles published in Critical Care during 2010 that have added to our scientifi c and clinical knowledge in the fi eld of neurocritical care.