Neurocrit Care (2014) 21:152–162 DOI 10.1007/s12028-013-9884-4

ESSENTIALS AND BASICS

Cerebral Microdialysis in and Subarachnoid Hemorrhage: State of the Art

Marcelo de Lima Oliveira • Ana Carolina Kairalla • Erich Talamoni Fonoff • Raquel Chacon Ruiz Martinez • Manoel Jacobsen Teixeira • Edson Bor-Seng-Shu

Published online: 26 September 2013 Ó Springer Science+Business Media New York 2013

Abstract Cerebral microdialysis (CMD) is a laboratory Introduction tool that provides on-line analysis of brain biochemistry via a thin, fenestrated, double-lumen dialysis catheter that is Cerebral microdialysis (CMD) is a well-established labo- inserted into the interstitium of the brain. A solute is slowly ratory tool that provides on-line analysis of the brain infused into the catheter at a constant velocity. The fen- biochemistry via a thin fenestrated dialysis catheter that is estrated membranes at the tip of the catheter permit free inserted into interstitium of the brain [1, 2]. Prior to the diffusion of molecules between the brain interstitium and development of CMD in 1970 [1], dextran-filled sacks the perfusate, which is subsequently collected for labora- were inserted into the brains of dogs and the fluid contained tory analysis. The major molecules studied using this there was studied afterwards. The first publication using method are , lactate, pyruvate, glutamate, and microdialysis in humans was a 1987 study investigating glycerol. The collected substances provide insight into the interstitial glucose in adipose tissue. CMD has been used to neurochemical features of secondary injury following study the human brain since 1990, and it has been used in traumatic brain injury (TBI) and subarachnoid hemorrhage neurosurgical intensive care since 1992 [3]. (SAH) and valuable information about changes in brain The catheter acts as a blood capillary in the tissue and is metabolism within a short time frame. In this review, the useful for analyzing the chemical composition of the authors detail the CMD technique and its associated interstitial fluid. The catheter permits the study of the main markers and then describe pertinent findings from the lit- neurochemical features of brains in normal states as well as erature about the clinical application of CMD in TBI and pathological states, such as traumatic brain injury (TBI) [2– SAH. 4] and subarachnoid hemorrhage (SAH) [4]. Markers that are currently studied and are important in the pathophysi- Keywords Microdialysis Á Metabolism Á Brain injury Á ology of brain injury and ischemia include the following: Monitoring Á Subarachnoid hemorrhage Á Critical care energy substrates (glucose), brain glucose metabolism products (lactate, pyruvate), by products of cell membrane degeneration (glycerol), and excitatory (glutamate). Despite improvements in clinical management, SAH and TBI remain significant clinical problems with high M. de Lima Oliveira Á A. C. Kairalla Á E. T. Fonoff Á morbidity and mortality rates. Early aggressive and ade- & M. J. Teixeira Á E. Bor-Seng-Shu ( ) quate treatment has been found to yield long-term Division of Neurological Surgery, Hospital das Clinicas, School of Medicine, University of Sa˜o Paulo, Rua Loefgreen, functional improvements. Cerebral ischemic and nonis- 1.272 – Vila Clementino, Sa˜o Paulo, SP 04040-001, Brazil chemic events are the core mechanisms that lead to e-mail: [email protected] secondary brain damage following TBI and SAH, and they are major causes of increased intracranial pressure (ICP) R. C. R. Martinez Discipline of Surgical Technique, Department of Surgery, and unfavorable prognoses [4–6]. CMD monitoring sys- School of Medicine, University of Sa˜o Paulo, Sa˜o Paulo, Brazil tems permit precocious on-line monitoring of cerebral 123 Neurocrit Care (2014) 21:152–162 153 ischemic markers in patients with TBI and SAH. Within a intraparenchymal pressure device, a microdialysis probe, a short time, early target therapy can be established to reduce multiparameter sensor (for continuous measurement of O2, secondary brain damage in at-risk patients [4, 5]. Jugular CO2, pH and temperature), and a lumen to puncture the venous saturation (SjvO2) and brain tissue oxygen satura- ventricle if necessary [10]. More than one catheter can be tion (PtbO2) can be used to detect secondary insults, used in different areas of the brain, typically the perile- although they are not considered markers of early tissue sional area and an unaffected brain area on the opposite ischemia [5, 7]. In this review, the authors describe the side. The catheter is routinely positioned 2-cm deep in the CMD technique and its associated markers in detail and right frontal region, close to white matter brain areas par- then summarize pertinent findings from the literature ticularly in patients suffering TBI with diffuse injury. If regarding the clinical application of CMD in TBI and SAH. required, the catheter may be positioned within at-risk tissue, such as the penumbral area surrounding the vascular territory that is most likely to be affected by vasospasm or Technique next to parenchymal traumatic lesions. Nevertheless, CMD catheters should not be placed in the contusional tissue [4]. The CMD catheter is thin (0.6 mm) and has a semiperme- Commercially available CMD catheters have a gold tip able membrane that permits the free diffusion of water and whose position can be visualized in computed tomography solutes down the concentration gradient between the sur- scans (CT) [1, 4] and can be used to determine whether the rounding interstitial fluid and the perfusate (Fig. 1)[2, 8]. catheter is located in ‘‘normal’’, pericontusional or contu- Because CMD analyzes the extracellular interstitial fluid sional tissue. ‘‘Normal’’ brain tissue should show no from the specific area where the probe is located [1], macroscopic lesions in CT scans, and pericontusional brain catheter positioning within the brain can influence the areas should occur at a distance of 0.5–1.5 cm from the results. The catheter can be inserted in a triple-lumen core of the lesion in the scans [11]. transcranial bolt alongside cerebral tissue oxygen (PtbO2) After insertion, the CMD is tunneled and sutured to the and ICP probes [9]. Some studies have employed a four- skin of the scalp and connected to a 2.5-ml syringe that is way multiparametric transcranial screw for gripping an placed into a microinfusion pump [12].

Fig. 1 The catheter consists of an inner tube and an outer tube with a semipermeable membrane in the tip. In the inner tube of the catheter, infused solution flows to the outer tube, which allows the exchange of extracellular substances by differences in fluid concentration. Later, the solution is collected at the external tip for biochemical analysis

123 154 Neurocrit Care (2014) 21:152–162

The perfusate is infused through the catheter using a microinfusion pump at a constant rate of 0.3 ll/min, which allows for sampling every 60 min [2]. Extracellular mol- ecules equilibrate across the semipermeable membrane and are eventually collected at the external tip of the CMD catheter for biochemical analysis (Fig. 1). The main markers of brain cell metabolism are collected using a standard dialysis membrane with a molecular weight cut- off of 20 kDa. Larger molecules require a 100 kDa semi- permeable membrane [2]. The recovery of biochemical markers increases with the length of the dialysis membrane that is inserted into the brain tissue. A 10-mm microdi- alysis catheter membrane and a flow rate of 0.3 ll/min permit the collection of 70–80 % of the rou- tinely used biochemical markers. Following these procedures, the concentration of the molecules in the Fig. 2 The cell uses two alternatives for energy synthesis. The dialysate and the extracellular cerebral fluid is approxi- aerobic phase requires oxygen. At this stage, glucose is converted to mately the same [2, 8]. pyruvate; pyruvate is converted to 36 ATP, water, and CO2 in the mitochondria. The anaerobic phase does not require oxygen. At this stage, glucose molecule is converted to pyruvate, resulting in only 2 ATP and lactate in the cytoplasm. This latter stage is associated with Results type 1 LPR increase

Cerebral Microdialysis Markers: Glucose, Lactate, Lactate-Pyruvate Ratio, Glutamate, and Glycerol molecule produces pyruvate, which is in turn converted into lactate with a net yield of only 2 molecules of ATP Under normal circumstances, glucose is the sole source of (Fig. 3). A high concentration of lactate and a concomi- energy utilized by the brain because it is the only molecule tantly high lactate-pyruvate ratio (LPR type 1) are that can be transported across the blood–brain barrier at a subsequently observed. LPR type 1 is associated with a sufficient rate. Once glucose enters the cytoplasm of neu- shift to anaerobic metabolism due to ischemia, hypoxia or a ronal cells, it is converted to pyruvate, which enters the failure of mitochondrial oxidative phosphorylation [13– citrate cycle in the mitochondria. The majority of the net 15]. In reversible ischemia, LPR normalizes within yield of 36 ATP is produced in the mitochondrial phase of 60–90 min of CBF restoration [16–18], which allows the aerobic glucose metabolism, and glucose is ultimately accumulated lactate to be aerobically used and prevents converted to carbon dioxide and water. Alternatively, tissue damage [16, 19, 20]. It has been proposed that the glucose can be anaerobically converted to pyruvate with a brain is capable of using lactate to produce energy under net yield of only 2 ATP per glucose molecule, and part of distress and thus spare glucose [21]. the pyruvate is converted to lactate [6]. In addition to LPR is a nonspecific indicator of brain ischemia [17]. In consuming more glucose to produce less ATP, the anaer- acute brain injury without hypoxia, a reduction in glyco- obic pathway also decreases the interstitial glucose lytic pyruvate production leads to type 2 or ‘‘nonischemic’’ concentration (Fig. 2). elevation of LPR (Fig. 3)[15, 22]. It has been suggested The glucose concentration measured by CMD is deter- that even in the absence of ischemia, acute brain injury can mined by the relationship between the local glucose induce a metabolic crisis that persists for several days [16– utilization rate and the glucose supply. Studies in patients 18]. Some findings also suggest a possible reduction or with TBI and SAH have found that unfavorable outcomes increase in glucose metabolism, both of which are associ- are associated with low cerebral glucose levels. Possible ated with low pyruvate production [20]. The high lactate causes of reduced cerebral glucose levels include increased level associated with LPR type 1 is not observed. Reduced consumption (hyperglycolysis) and decreased delivery due glucose metabolism may be due to the utilization of an to lower cerebral blood flow (CBF) following vasospasm or energy source other than glucose for oxidative metabolism. intracranial hypertension [1, 8, 13]. Experimental evidence suggests that ketones, lactate, and In patients with cerebral oligemia, the interstitial CMD pyruvate may act as alternative oxidative fuels for the brain glucose concentration may drop to low or undetectable under certain conditions [17]. Increased glucose metabo- levels, and brain lactate concentrations increase. These lism may result from increased shunting of glucose to the changes characterize the redox state in which each glucose pentose phosphate pathway (PPP). The posttraumatic brain 123 Neurocrit Care (2014) 21:152–162 155

Fig. 3 Different causes in type 2 LPR increase

may frequently suffer from nonischemic metabolism that immediately or the samples should be frozen (-70 °C) for shifts glucose to the PPP for macromolecular repair and later analysis [20]. defense against oxidative stress, resulting in a shortage of Glutamate is the major excitatory in the brain. pyruvate and type 2 LPR [20, 22]. The PPP produces a Derived from glucose, glutamate is a surrogate measure of reducing equivalent in the form of NADPH, which, through neuronal oxidative metabolism under normal conditions. its action on glutathione, is important in the scavenging of Astrocytic uptake keeps brain interstitial glutamate levels free radicals and plays a protective role in neutralizing low by converting released glutamate into glutamine, which oxygen free radicals [22]. Some researchers have observed is reconverted to glutamate by the neurons [23]. Because this that the levels of glutathione peroxidase, a key antioxidant cycle requires energy, impaired energy metabolism may lead , are elevated as early as 3 h post-injury and peak to low interstitial glutamine/glutamate ratios [15]. Uncon- 7 days post-trauma, suggesting extended periods of free trolled release is observed in TBI and SAH and is mediated radical stress [23]. In vitro studies using cultures of mixed by excessive calcium influx into brain cells via glutamate- primary cerebrocortical cells have demonstrated that the mediated ion channels [1, 15]. Glutamate has been consid- percentage of glucose used by the PPP increases in a dose- ered an early marker of cerebral ischemia [14, 27]. dependent manner from 0.25 % to a maximum of 22 % in Glycerol is the final product of the enzymatic degrada- the presence of H2O2 [24, 25]. This pathway is responsible tion of cell membrane triglycerides, and its presence for approximately 2–5 % of glucose utilization during indicates a loss of cellular structural integrity [8, 15]. normal metabolism but can reach levels exceeding 8–12 % Glycerol is also considered a reliable indicator of tissue after TBI [23]. PPP also produces ribose, which is impor- ischemia that has progressed to a stage that involves further tant for DNA repair and replication and mRNA and cell damage [2, 4]. CMD glycerol increases relatively synthesis [26]. slowly during energy metabolism failure and remains ele- Another cause of pyruvate reduction may be an enzy- vated for some time upon normalization [2]. matic system disturbance at the pyruvate dehydrogenase complex (PDHC) or further downstream in the enzymatic Clinical Application in TBI and SHA Krebs cycle or the electron transport chain. The PDHC system is known to be enzymatically vulnerable, particu- Ischemic and nonischemic injuries play a key role in larly to oxidative stress. Pyruvate may enter the Krebs patients with TBI and SHA. The prevention and early cycle and be converted to intermediate products that exit detection of changes in neuronal cellular distress markers the cycle at various steps [16]. are important aspects of treatment [28]. High lactate and pyruvate levels in the brain without a The main goal of cerebral perfusion pressure (CPP) is to high LPR can be interpreted as cerebral hypermetabolism provide the necessary blood flow for the metabolic needs of that occurs during the recovery phase after ischemia [15]. the injured brain and to avoid exacerbating ischemic To prevent the ‘‘false’’ LPR type 2 phenomenon, cere- insults. CMD can be used routinely to determine the bral pyruvate from collected samples should be analyzed optimal level of CPP in patients with TBI [12]. It is

123 156 Neurocrit Care (2014) 21:152–162 important to note that, in patients with TBI and SHA, the oxygen extraction fraction near ischemic levels, even in the autoregulation of CBF can be impaired, and CBF varies in absence of profound hypoglycemia [35–37]. In patients direct proportion to changes in arterial blood pressure with severe TBI, tight systemic glucose control is associ- [29–31]. ated with an increased prevalence of brain energy crises CMD analyzes can reveal alarming levels of LPR >30 and low cerebral extracellular glucose, which in turn cor- and glucose <0.8 mmol/L during ischemic events [3]. relates with increased mortality [38]. Administration of Ischemia is defined by the combined CMD criteria of insulin in these patients should be carried out with caution LPR >40 and glucose <0.2 mmol/L. In persistent ische- [37]. mia, LPR may exceed 40 and often plateaus at a value High interstitial glucose consumption can occur in cases between 80 and 120. LPR levels as high as 500–1,000 have of central nervous system infection [37]. For example, been documented in cases of impending brain death. meningitis-related glucose depletion might limit the cere- Studies utilizing CMD in normal and pericontusional brain bral energy supply. In patients with SAH, the sensitivity tissue have indicated significantly higher predicted LPR and specificity are 69 and 80 %, respectively, for the values in patients with ICPs >20 mmHg, and the most diagnosis of bacterial meningitis in the presence of a noticeable differences occur when ICP exceeds 30 mmHg. 1 mmol/L decrease in measured CMD glucose and fever These studies have also revealed a significant increase in when cerebrospinal fluid biochemistry analysis fails to LPR in pericontusional brain tissue with decreasing CPP indicate bacterial meningitis. Additionally, CMD glucose due to the loss of CBF autoregulation [11]. LPRs >25 are levels decrease markedly in the days before the meningitis predictive of intracranial pressure elevations in 89 % of diagnosis [39]. TBI cases [2]. An important study showed that changes in CMD has shown an increase in excitatory amino acids, LPR precede the onset of intracranial hypertension and that particularly glutamate, ranging from 500–700 % of base- increased pericontusional LPR is significantly associated line values after TBI. In the acute post-TBI phase, neuronal 2+ with CPP. Low PtbO2 and a high LPR are more frequently depolarization induced by intracellular Ca influx is fol- observed with high ICP and low CPP in poor-grade SAH lowed by an increase in interstitial glutamate, higher patients, although several periods of intracranial hyper- anaerobic glycolytic activity and increased lactate pro- tension and low CPP are not associated with brain hypoxia duction. At the same time, the levels of oxidative [32]. Another study performed on 90 severe TBI patients metabolism are depressed [23, 40]. It has been reported that concluded that CMD cellular distress markers are weakly glutamate induced neurotoxicity occurs at lower concen- related to CPP and ICP. Increased LPR values have been trations when cellular energy production is impaired [21]. linked to different causes of oxygen utilization, such as In several patients, hyperglycolysis and low CMD glucose oxygen diffusion barriers, mitochondrial dysfunction, and have been associated with increased glutamate in over increased metabolism of glucose, rather than oxygen 90 % of cases [40]. delivery [33]. Transient reductions in CPP may be associated with The lactate/glucose ratio has been used as an indicator increases in glutamate, indicating that neuronal populations of increased glycolysis [1]. For example, one study found are as vulnerable to alterations in CPP as they are to that the lactate/glucose ratio is unexpectedly low in both ischemia. In a few cases, glutamate will increase before good and poor outcome groups during the initial 50 post- CPP drops. It is possible that glutamate increases at a injury hours, but subsequent increases are only observed in higher CPP threshold and that its release leads to cellular the poor outcome patients [21]. swelling and increased ICP [21]. Some authors have sug- In patients with TBI, both metabolic deterioration gested that rather than being an unspecific phenomenon (including redox state) and low CMD glucose concentra- that is due to general cell depolarization, the increase in tions can be caused by anemic hypoxia [12]. In patients CMD glutamate may contribute to neuronal depolarization with poor-grade SAH, the incidence of brain hypoxia and and cell damage [12]. A study of patients with TBI and cell energy dysfunction increases significantly at hemo- intractable high ICP who underwent surgical brain globin levels of 9 g/dL, which represents a significant and decompression showed that, despite increased brain tissue independent increase in the risk of cerebral metabolic pO2, individuals with an unfavorable outcomes had per- dysfunction in these patients [34]. sistently high CMD glutamate, LPR and lactate levels that Intracerebral CMD glucose concentration may also were possibly associated with a persistent redox state and decrease after insulin therapy, even if blood glucose levels mitochondrial dysfunction. The favorable outcome group remain unaffected. This observation is supported by data exhibited a large reduction in CMD glutamate levels after from studies of TBI and SAH patients who have shown that surgical brain decompression [11, 14, 27]. An experimental intensive insulin therapy results in a net reduction in CMD study in rats with intact cortices and a clinical study with glucose, an increase in cellular distress markers and an TBI patients both found reduced levels of cerebrospinal 123 Neurocrit Care (2014) 21:152–162 157

fluid and interstitial CMD glutamate after treatment with increased LPRs can precede or be correlated with episodes thiopental sodium [7]. of intracranial hypertension [45]. A PtbO2 catheter is an In SAH, cerebral edema is an independent risk factor for invasive tool that permits continuous assessment of oxygen death and unfavorable prognosis possibly due to glutamate- tension and the early detection of changes in brain tissue mediated excitotoxicity [7]. Glutamate has been identified oxygen that precede ischemic damage [42]. It is important as the earliest marker of the onset of vasospasm, followed to note that higher levels of brain PtbO2 and cellular dis- by lactate, LPR and glycerol [4]. These metabolic changes tress markers in the same area may be indicative of a can be detected by CMD before the onset of symptomatic reduction in brain tissue oxygen utilization. Thus, the vasospasm in 83 % of patients with delayed ischemic monitoring of these types of changes facilitates the diag- neurological deficits [41, 42]. Some authors have reported nosis of mitochondrial phosphorylation failure [14]. It has a decrease in glutamate levels during cooling in SAH been suggested that continuous EEG can identify clinically patients and suggested that this change represents a neu- subtle or undetected seizures and detect cerebral ischemia. roprotective effect [8]. A study of 14 patients with SAH Furthermore, subclinical seizures detected by continuous showed that controlled hyperventilation is associated with EEG in patients with TBI are correlated with more frequent a highly significant decrease in both regional CBF and and prolonged occurrences of elevated LPR, glutamate and transcranial Doppler flow velocity. Therefore, the possi- glycerol [46]. Because the interstitial capillary vessels are bility that hyperventilation can lead to cerebral in close proximity to the neuronal cells, the cerebral flow vasoconstriction, ischemia and metabolism disturbance, velocity immediately reflects changes in neuronal bio- particularly glutamate release, must be considered [43]. In chemical behavior (e.g., redox states and metabolic crises). one report of cerebrovascular surgeries, the samples col- In light of this relationship, TCD ultrasonography can be lected from intraoperatory CMD exhibited high levels of used to assess relative changes in blood flow velocity in the glutamate and other cellular distress markers during brain main intracranial arteries and thus provide a qualitative retraction and temporary brain artery occlusion [10]. analysis of CBF and help to identify the different hemo- LPRs >25 and glycerol levels >100 mmol/L are asso- dynamic phases of TBI and SHA that include ciated with a significantly higher risk of imminent hypoperfusion, hyperemia, and vasospasm [16, 47]. TCD intracranial hypertension [15]. A study of a large cohort of monitoring of both cerebral autoregulation and cerebral 76 patients with severe TBI revealed markedly elevated vasomotor reactivity may help in providing adequate CMD glycerol levels in all patients during the first day after intracranial hemodynamics and, consequently, adequate TBI (these elevated levels most likely reflected the primary CBF in brain-injured patients [48]. injury), followed by a gradual decline over the next 3 days.

PtbO2 levels below 10 mmHg and CPP levels below 70 mmHg are associated with high CMD glycerol concen- Discussion trations. However, isolated episodes in which CMD glycerol levels are not increased when PtbO2 and CPP levels are Glucose Metabolism, Hemodynamic Phases below the same thresholds have repeatedly been observed. of Traumatic Brain Injury (TBI), and Subarachnoid There was no significant difference between the mean CMD Hemorrhage (SAH) glycerol concentrations of patients with favorable and unfavorable outcomes. The authors therefore concluded that In the acute phase of SAH (the first 24 h after injury), CBF CMD glycerol cannot provide significant information about decreases as CPP falls. Simultaneously, there may be acute prognoses [3, 44]. Another study of 16 patients with severe microvascular constriction in areas with low TBI who had undergone decompression surgery did not concentrations [15, 16]. The brain is particularly suscepti- detect any differences in the CMD glycerol concentrations ble to hypotension or increased ICP. Some studies have of the poor outcome group and the favorable outcome group. shown decreases in CMD glucose during the acute phase of Both groups exhibited significant CMD glycerol reductions SAH. Studies utilizing adult animals and humans have after decompression surgery [27]. suggested that decreased oxidative metabolism is due to mitochondrial alterations as opposed to a lack of oxygen and reduced glucose supply caused by decreased cerebral Multimodality Monitoring perfusion to the injured tissue [49]. Glucose can be the sole substrate in the production of energy via the anaerobic Multimodality monitoring is used for the recognition of pathway [7, 16, 50]. Oxidative reduction reaches a level of secondary brain events in neurocritical patients. Monitor- approximately 50 % of the cerebral metabolic rate of ing ICP is recommended for all patients with severe TBI oxygen [24], which leads to Na+/K+ pump impairment and and SAH. High ICP is correlated with poor outcomes, and intracellular Ca2+ influx. Data from animal studies suggest 123 158 Neurocrit Care (2014) 21:152–162 that increased intracellular Ca2+ is associated with mito- cellular injury, in light of the fact that CPP occurs in the chondrial swelling and dysfunction, which reduces ATP context of decreased intracranial compliance and altera- yield and induces cytotoxic brain edema [19, 51]. These tions in the blood–brain barrier. It is therefore possible that changes are more pronounced in TBI and SAH patients an increase in CPP in the context of impaired pressure with unfavorable outcomes [19]. Global cerebral edema autoregulation may lead to hyperemia, and cellular injury has been associated with lower brain glucose in the first may occur despite adequate CPP [32]. 12 h, which may likely be due to higher glucose con- Cortical spreading depression (CSD) was described in sumption via the anaerobic pathway [52]. Maximal 1944 by Lea˜o. This electrical disturbance occurs in 50 % of pulsatility index values observed with transcranial Doppler patients with severe TBI and is an important cause of early have been reported to occur on the day of hemorrhage in ischemia in TBI and SAH patients [15, 43]. Cortical patients with SAH, then decrease until day 10, indicating depolarization waves spread across the cortical surface at that vasoconstriction or cerebral edema occurs during the 2–3 mm/min. Marked glucose depletion occurs during hyperacute period. Persistently, high pulsatility indices are CSD in proportion to the number of depolarizations [3, 15]. more pronounced in patients with unfavorable outcomes Some authors have reported that CSD coincides with a [53]. 200 % increase in glucose utilization in the injured cortex Hyperactive anaerobic metabolism increases lactate and the ipsilateral hippocampus [24, 55]. Another study concentration in the brain interstitium [14], alters CBF, and showed a progressive decline in CMD-measured glucose leads to prolonged hyperemia [49]. Furthermore, the associated with recurrent depolarizations in patients with hyperemic phase may be prolonged by hyperglycolysis head injuries. Under these particular conditions, a vicious caused by a failure in energy production or a greater energy cycle is established; pre-ictal discharges (PID) deplete the demand. The main purposes of brain hyperglycolysis are to residual glucose pool and thereby increase the probability decrease high interstitial glutamate levels, restore ion pump of additional PIDs. Transient glucose depletion appears to homeostasis, enhance the activity of the pentose phosphate be caused by a combination of increased utilization and pathway, provide energy to infiltrated inflammatory cells, reduced vascular delivery as in the case of strict serum and promote glycogen synthesis by reactive astrocytes [24, glucose control that was previously described in this arti- 37]. Other hypermetabolic states of the brain, such as fever, cle. There is a transient increase in lactate that can be seizures, and the development of depression-like phe- attributed to a temporary elevation in the rate of glycolysis nomena, are also associated with hyperglycolysis [24, 27, [56]. In most animal models, CSD is associated with the 52, 54]. transient dilation of superficial arterioles followed by sus- tained constriction. Microvascular vasoconstriction leads to parenchymal hypoperfusion during the period of maximum Seizures, Cortical Spreading Depolarization, and MD metabolic demand. It is more likely that neurovascular coupling is disrupted and that profound neuroglial depo- Posttraumatic seizures have been found to affect 22 % of larization causes swelling that constricts the capillary bed patients suffering TBI and are associated with greater and [57]. In some patients with few initial CSDs followed by prolonged intracranial hypertension and an increased repeated episodes of perinfarct depolarization, depolarizing mortality rate. Electrographic seizures without any clinical episodes may be related to brain tissue hypoperfusion and indicators account for over 50 % of seizures in the inten- expansion of the ischemic penumbra, which potentially sive care unit. LPR, glutamate, and glycerol, as measured expands the core infarct zone and increases the final infarct by CMD [1], increase when electrographic seizures occur. volume [43, 54, 56]. High LPR persists longer and occurs more frequently in patients with electrographic seizures [15, 46]. Increased Cerebral Inflammatory Biomarkers Related to TBI glutamate levels and seizure activity are observed despite adequate CPP in some patients after TBI. Prolonged Inflammation plays a fundamental role in the pathophysi- increases in CMD-measured glutamate levels are associ- ology of TBI. MD studies with molecular weight cut-offs ated with repetitive seizures that occur over many hours. In of 100 kDa have assessed biomarkers that reflect this one study of 17 patients with severe head injuries, gluta- inflammatory process. mate levels at CPPs <70 mmHg were significantly higher Metalloproteinases (MMP) can be associated with the and associated with electroencephalographic seizures [21]. pathophysiology of secondary brain injury. Microdialysate Seizures are accompanied by increases in mean arterial levels of MMP-8 and 9 are elevated after severe TBI, blood pressure and CPP, which may be physiological which can lead to cerebral blood-barrier disruption, cere- responses attempting to meet the brain’s greater metabolic bral edema, and inflammatory cell infiltration. High MMP- demands. Increased CPP may in turn contribute to further 8 levels are related to high ICP and unfavorable outcome. 123 Neurocrit Care (2014) 21:152–162 159

Following 48 h, MMP-8 levels decrease, MMP-3 levels be inserted into specific regions of the brain and used to peak, followed by the gradual increase in MMP-7 levels. analyze brain tumor substances, neurotransmitters, and Both MMP-3 and MMP-7 may have a role in the recovery in the mesencephalic substantia nigra or other from neuronal injury [58]. parts of the brain. Other important molecules associated Tau protein can also be marker of unfavorable progno- with TBI and SAH that can be collected by CMD include ses. A cut-off value of 10,000 pg/ml for tau protein xanthines (markers of reduced ATP production) and gluta- concentration produces 70 % sensitivity and 80 % speci- thione (an important tripeptide that is part of the free radical ficity for predicting poor clinical outcome. Ab protein has scavenging system). Studies of local, tissue-specific effects an inverse relation with tau protein in pericontusional of drugs that have no systemic effects can be performed areas; however, there is no link between this protein and using CMD catheters. CMD can gauge the appropriate clinical prognoses. The diagnosis of diffuse axonal injury concentrations of at target sites during clinical can be supported by high tau and NF-1 protein levels in the treatment, which may elucidate therapeutic failures and the brain interstitial fluid [59]. emergence of drug-resistant bacteria that have been exposed

8-iso-PGF2µ is an important marker of systemic oxi- to noninhibitory drug concentrations in the brain tissue [65]. dative stress. This protein is selectively increased in brain Microdialysis can monitor improvements in brain cell MD samples taken near the injured tissues after TBI and metabolism after TBI treatment with specific drugs such as strongly correlates with MD glycerol and glutamate, which cyclosporin A and Ro5-4864 [66]. Some authors have supports the association between oxidative stress, mem- treated malignant brain tumors with direct and continuous brane phospholipid degradation, and excitotoxicity [60]. administration of substances to the affected tissue using Ubiquitin carboxyl-terminal hidrolyse-L1 (UCH-L1) microdialysis catheter. Thus, this technique has potential and glial fibrillary acidic protein (GFAP) have been con- applications in the field of brain tumor chemotherapy sidered markers of glial and neuronal cell damage. These [1, 67]. are able to predict injury severity and death after severe TBI. A recent study induced traumatic subdural Limits of the Microdialysis Technique hematoma in rats, in which the hematoma was subse- quently removed. This research showed lower MD- Microdialysis is a complex method that is restricted in the measured UCH-L1 and GFAP in rats submitted to preco- following ways: The data from microdialysis are not pro- cious moderate hypothermia before and after surgery vided in real time; the method reflects local neurochemical compared to control groups treated with surgery and nor- changes in the area in which the probe is inserted; micro- mothermia or late hypothermia. Additionally, neuronal dialysis is informative regarding interstitial concentrations degeneration and injury volumes were reduced in the in tissues but not informative regarding intracellular con- precocious hypothermic group. Early peaks in UCH-L1 centrations; microdialysis is highly dependent upon the extracellular concentrations in the normothermic group technique used to collect the samples; and the large number were identified after surgical treatment, indicating that of substances that can be analyzed may complicate the data neuronal damage occurs during the premature reperfusion interpretation. Analyzes should take into account the length phase after removal of subdural hematomas. This peak was of the catheter, membrane characteristics, properties of the lower in the precocious hypothermic treatment group [61]. fluid inside the catheter, the perfusion flow rate, and the Cytokines, especially IL-1b and TNF, can play a pro- velocity of the movement of the substance through the inflammatory role. MD studies have associated IL-1b with extracellular fluid [8, 65]. The insertion of the microdialysis neurodegeneration and poor neurological outcomes. catheter damages a small number of cells and causes a local Increases in IL-1b concentrations can be prolonged in inflammatory reaction. As a result of the inflammatory hypoxic events. TNF has also been associated with delayed reaction and the release of intracellular substances, the neuronal degeneration mediated through biochemical paths collection of the samples is postponed by 1–2 h [64, 65]. that lead to neuronal apoptosis. Both proteins are produced in concert as a result of injury. In contrast, IL-6 and IL-10 have neuroprotective properties after TBI [9, 62–64]. IL-6 appears 24–72 h after TBI, and IL-10 appears for up to Conclusions 6 days [62]. CMD has greatly improved our understanding of dynamic Other Clinical Applications and Perspectives brain energy metabolism in TBI and SAH patients, and it provides early warning signs of impending hypoxia and In addition to collecting samples for analyzes of biochem- ischemia [1, 65]. Combined with other brain monitoring ical markers of cellular metabolism, the microdialysis can techniques, CMD may be used for the delivery of target 123 160 Neurocrit Care (2014) 21:152–162 therapies that are aimed at preventing secondary ischemic 17. Vespa P, Bergsneider M, Hattori N, Wu HM, Huang SC, Martin injury [4] and may also be useful as an indicator of the NA, Glenn TC, McArthur DL, Hovda DA. Metabolic crisis without brain ischemia is common after traumatic brain injury: a effectiveness of therapeutic interventions. CMD has dra- combined microdialysis and positron emission tomography study. matically reduced mortality in patients with severe brain J Cereb Blood Flow Metab. 2005;25:763–74. injuries [38], although it has not yet been found to improve 18. De Fazio M, Rammo R, O’Phelan K, Bullok MR. Alterations in outcomes [68]. More studies are required to assess the cerebral oxidative metabolism following traumatic brain injury. Neurcrit Care. 2011;14:91–6. value of cerebral CMD in the management of neurocritical 19. Soustiel JF, Glenn TC, Shik VA, Boscardin J, Mahamid E, Za- care patients. aroor M. Monitoring of cerebral blood flow and metabolism in traumatic brain injury. J Neurotrauma. 2005;22:955–65. 20. Hillered L, Enblad P. Nonischemic energy metabolic crisis in acute brain injury. Crit Care Med. 2008;36(10):2952–3. References 21. Vespa P, Prins M, Ronne-Engstrom E, Caron M, Shalmon E, Hovda DA, Martin NA, Becker DP. Increase in extracellular 1. Tisdall MM, Smith M. Cerebral microdialysis: research technique glutamate caused by reduced cerebral perfusion pressure and or clinical too. British J Anaesth. 2006;97(1):18–25. seizures after human traumatic brain injury: a microdialysis 2. Nordstron CH. Cerebral energy metabolism and microdialysis in study. J Neurosurg. 1998;89:971–82. neurocritical care. Childs Nerv Syst. 2010;26:465–72. 22. Marcoux J, McArthur DA, Miller C, Glenn TC, Villablanca P, 3. Hillerd L, Persson L, Nilsson P, Engstrom ER, Enblad P. Con- Martin NA, Hovda DA, Alger JR, Vespa PM. Persistent meta- tinuous monitoring of cerebral metabolism in traumatic brain bolic crisis as measured by elevated cerebral microdialysis injury: a focus on cerebral microdialysis. Curr Opin Crit Care. lactate-pyruvate ratio predicts chronic frontal lobe brain atrophy 2006;12:112–8. after traumatic brain injury. Crit Care Med. 2008;36:2871–7. 4. Bellander BM, Cantais E, Nordstrom CH, Robertson C, Sa- 23. Bartinik BL, Sutton RL, Fukushima M, Harris NG, Hovda DA, huquillo J, Smith M, Stocchetti N, Ungerstedt U, Unterberg A, Lee SM. Upregulation of pentose phosphate pathway and pres- Olsen NV. Consensus meeting on microdialysis in neurointensive ervation of tricarboxylic acid cycle flux after experimental brain care. Intensive Care Med. 2004;30:2166–9. injury. J Neurotrauma. 2005;22(10):1052–65. 5. Stocchetti N, Colombo A, Ortoloni F, Videtta W, Marchesi R, 24. Bergsneider M, Hovda DA, Shalmon E, Kelly DF, Vespa PM, Longhi L, Zainer ER. Time course of intracranial hypertension Martin NA, Phelps ME, Mcarthur DL, Caron MJ, Kraus JF, after traumatic brain injury. J Neurotrauma. 2007;24:1339–46. Becker DF. Cerebral hyperglycolysis following severe traumatic 6. Nordstron CH. Cerebral energy metabolism and microdialysis in brain injury in humans: a positron emission tomography study. neurocritical care. Childs Nerv Syst. 2010;26:456–72. J Neurosurg. 1997;86:241–51. 7. Nagel A, Graetz D, Shinck T, Frieler K, Sakowtz O, Vajkoczy P, 25. Ben-Yoseph O, Boxer PA, Ross BD. Assessment of the role of Sarrafzadeh A. Relevance of intracranial hypertension for cere- the glutathione and pentose phosphate pathways in the protection bral metabolism in aneurismal subarachnoid hemorrhage. of primary cerebrocortical cultures from oxidative stress. J Neu- J Neurosurg. 2009;111:94–101. rochem. 1996;66:2329–37. 8. Peerdeman SM, Girbes ARJ, Vandertop WP. Cerebral microdi- 26. Dusick JR, Glenn TC, Lee WNP, Vespa PM, Kelly DF, Lee SM, alysis as a new tool for neurometabolic monitoring. Intensive Car Hovda DA, Martin NA. Increased pentose phosphate pathway Med. 2000;26:662–9. flux after clinical traumatic brain injury: a [1,2-13C2] glucose 9. Hutchinson PJ, O’Connell MT, Nortje J, Smith P, Al-Rawi PG, labeling study in humans. J Cereb Blood Flow Metab. 2007;27: Gupta AK, Menon DK, Pickard JD. Cerebral microdialysis 1593–602. methodology–evaluation of 20 kDa and 100 kDa catheters. 27. Ho CL, Wang CM, Lee KK, Ng I, Ang BT. Cerebral oxygena- Physiol Meas. 2005;26:423–8. tion, vascular reactivity, and neurochemistry following 10. Mendelowitsch A. Microdialysis: intraoperative and posttrau- decompressive craniectomy for severe traumatic brain injury. matic applications in neurosurgery. Methods. 2001;23:73–81. J Neurosurg. 2008;108:943–9. 11. Timofeev I, Czosnyka M, Carpenter KLH, Nortje J, Kirkpatrick 28. Xaredi S, Olivecrona M, Lindgren C, Ostlund AL, Grande PO, PJ, Al-Rawi PG, Menon DK, Pickard JD, Gupta AK, Hutchinson Koskinen LO. An outcome study of severe traumatic head injury PJ. Interaction between brain chemistry and physiology after using the ‘‘Lund therapy’’ with low-dose prostacyclin. Acta traumatic brain injury: impact of autoregulation and microdialy- Anaesthesiol Scan. 2001;45:402–6. sis catheter location. J Neurotrauma. 2011;28:849–60. 29. Ling GSF, Neal JC. Maintaining cerebral perfusion pressure is a 12. Stahl N, Mellergard P, Hasllstrom A, Ungertedt U, Nordstrom worthy clinical goal. Neurocrit Care. 2005;2:75–82. CH. Intracerebral microdialysis and bedside biochemical analysis 30. Bor-Seng-Shu E, Figueiredo E, Fonoff ET, Fujimoto Y, Panerai in patients with fatal traumatic brain lesions. Acta Anaesthsiol RB, Teixeira MJ. Decompressive craniectomy and head injury: Scand. 2013;45:977–85. brain morphometry, ICP, cerebral hemodynamics, cerebral 13. De Andrade AF, Paiva WS, Prudente M, Bernardo L, Teixeira microvascular reactivity and neurochemistry. Neurosurg Review. MJ. Intensive care management in brain contusion with micro- 2012. doi:10.1007/s10143-013-0453-2. dialysis technique. Arq Neuropsiquiatr. 2012;70(8):640–1. 31. Johnson U, Nilsson P, Ronne-Engstrom E, Howells T, Enblad P. 14. Bor-Seng-Shu E, Oliveira ML, Teixeira MJ. Traumatic brain Favorable outcome in traumatic brain injury patients with injury and metabolism. J Neurosurgery. 2010;112:1351–3. impaired cerebral pressure autoregulation when treated at low 15. Goodman JC, Robertson CS. Microdialysis: is it ready for prime cerebral perfusion pressure levels. Neurosurgery. 2011;68: time? Curr Opin Crit Care. 2009;15:110–7. 714–21. 16. Westermaier T, Jauss A, Eriskat J, Kunze E, Roosen A. The 32. Chen HI, Stiefel MF, Oddo M, Mylbi AH, Maloney-Wilensky E, temporal profile of cerebral blood flow and tissue metabolites Frangos S, Levine JM, Kofke WA, LeRoux PD. Detection of indicates sustained metabolic depression after experimental sub- cerebral compromise with multimodality monitoring in patients arachnoid hemorrhage in rats. Neurosurgery. 2011;68:223–30. with subarachnoid hemorrhage. Neurosurgery. 2011;69(1):53–63.

123 Neurocrit Care (2014) 21:152–162 161

33. David WN, Bjo¨rn T, Robert MMC, Harriet N, Anders H, Anders treated by surgical decompression. J Neurosurg. 2006;104: R, Michael W, Bo-Michael B, Eddie W. Analyses of cerebral 93–100. microdialysis in patients with traumatic brain injury: relations to 49. Scafidi S, O’Brien J, Hopikins I, Robertson C, Fiskum G, intracranial pressure, cerebral perfusion pressure and catheter Mackenna M. Delayed cerebral oxidative glucose metabolism placement. BMC Medicine. 2011. doi:10.1186/1741-7015-9-21. after traumatic brain injury in young rats. J Neurochem. 34. Oddo M, Milby A, Chen I, Frangos S, MacMurtrie E, Maloney- 2009;109(Suppl 1):189–97. Wilensky E, Stiefel M, Kofke WA, Levine JM, Le Roux PD. 50. Asgari S, Vespa P, Hu X. Is there any association between Subarachnoid hemorrhage hemoglobin concentration and cere- cerebral vasoconstriction/vasodilatation and microdialysis lactate bral metabolism in patients with aneurysmal. Stroke. 2009;40: to pyruvate ratio increase? Neurocrit Care. 2013. doi:10.1007/ 1275–81. s12028-013-9821-6. 35. Schlenk F, Graetz D, Nagel A, Schmidt M, Sarrafzadeh AS. 51. Kristal BS, Dubinsky JM. Mitochondrial permeability transition Insulin-related decrease in cerebral glucose despite normoglyce- in central nervous system: induction by calcium cycling-depen- mia in aneurysmal subarachnoid hemorrhage. Crit Care. dent and independent pathways. J. Neurochem. 1997;69:524–38. 2008;12:1–7. 52. Helbok R, Ko SB, Schmidt M, Kurtz P, Fernandez L, Choi A, 36. Vespa P, Boonyaputthiku R, MacArthur DL, Miller C, Etchepare Connolly S, Lee K, Bdjatia N, Mayer SA, Claassen J. Global M, Bergsneider M, Glen T, Martin N, Hovda D. Intensive insulin cerebral edema and brain metabolism after subarachnoid hem- therapy reduces microdialysis glucose values without altering orrhage. Stroke. 2011;42:1534–9. glucose utilization or improving the lactate/pyruvate ratio after 53. Soehle M, Chatfield DA, Czousnika M, Kirkpatrick PJ. Predictive traumatic brain injury. Crit Care Med. 2006;34:850–6. value of initial clinical status, intracranial pressure and trans- 37. Zetterling M, Hillered L, Enblad P, Karlsson T, Ronne-Engstrom cranial Doppler pulsatility after subarachnoid haemorrhage. E. Relation between brain interstitial and systemic glucose con- Achita Neurochir. 2007;149:575–83. centrations after subarachnoid hemorrhage. J Neurosurg. 54. Fabricius M, Fuhr S, Bhatia R, Boutelle M, Hashemi P, Strong 2011;115:66–74. AJ, Lauritzen M. Cortical spreading depression and peri-infarct 38. Oddo M, Schmidt JM, Carrera E, Badjatia N, Connolly ES, depolarization in acutely injured human cerebral cortex. Brain. Presciutti M, Ostapkovich ND, Levine JM, Roux PL, Mayer SA. 2006;129:778–90. Impact of tight glycemic control on cerebral glucose metabolism 55. Sunami K, Nakamura T, Kubota M, et al. Spreading depression after severe brain injury: a microdialysis study. Crit Care. following experimental head injury in the rat. Neurol Med Chir. 2008;36(12):3233–8. 1989;29:975–80. 39. Shilenk F, Frieler K, Nagel A, Vajkoczy P, Sarrafzadeh AS. 56. Hopwood SE, Parkin MC, Bezzina EL, Boutelle MG, Strong AJ. Cerebral microdialysis for detection of bacterial meningitis in Transient changes in cortical glucose and lactate levels associated aneurysmal subarachnoid hemorrhage patients: a cohort study. with peri-infarct depolarisations, studied with rapid-sampling Crit Care. 2009. doi:10.1186/cc7689. microdialysis. J Cereb Blood Flow Metab. 2005;25:391–401. 40. Vespa PM, McArthur D, O’Phelan K, Glenn T, Etchepare M, 57. Brennan KC, Beltran-Parrazal L, Lopez-Valdes HE, Theriot J, Kelly D, Bergsneider M, Martin NA, Hovda DA. Persistently low Toga AW, Charles EC. Distinct vascular conduction with cortical extracellular glucose correlates with poor outcome 6 months after spreading depression. J Neurophysiol. 2007;97:4143–51. human traumatic brain injury despite a lack of increased lactate: a 58. Roberts DJ, Jenne CN, Le´ger C, Kramer AH, Gallagher CN, microdialysis study. J Cereb Blood Flow Metab. 2003;23:865–77. Todd S, Parney IF, Doig CJ, Young VW, Kubes P, Zygym DA. A 41. Sarrafzadeh AS, Sakowitz OW, Kiening KL, Benndorf G, prospective evaluation of the temporal matrix metalloproteinase Lanksch WR, Unterberg AW. Bedside microdialysis: a tool to response after severe traumatic brain injury in humans. J Neuro- monitor cerebral metabolism in subarachnoid hemorrhage trauma. 2013. doi:10.1089/neu.2012.2841. patients? Crit Care Med. 2002;30(5):1062–70. 59. Magnoni S, Esparza TJ, Cont V, Carbonara M, Carrabba G, 42. Hanggi D. Monitoring and detection of vasospasm II: EEG and Holtzman DM, Zipefel GJ, Stocchetti N, Brody DL. Tau eleva- invasive monitoring. Neurocrit Care. 2011;15:318–23. tions in the brain extracellular space correlate with reduced 43. Wartenberg KE. Critical care of poor-grade subarachnoid hem- amyloid-b levels and predict adverse clinical outcomes after orrhage. Curr Opin Crit Care. 2011;17:85–93. severe traumatic brain injury. Brain. 2012;135:1268–80. 44. Clausen T, Alves OL, Reinert M. Association between elevated 60. Calusen F, Marklund N, Lewe´n A, Enblad P, Basu S, Hillerd L. brain tissue glycerol levels and poor outcome following sever Interstitial F2-Isoprotstane 8-Iso-PGF2a as a biomarker of oxi- traumatic brain injury. J Neurosurg. 2005;103:233–8. dative stress after severe human traumatic brain injury. 45. Gu¨iza F, Depreitere B, Piper I, Van den Berghe G, Meyfroidt G. J Neurotrauma. 2012;29:766–75. Novel methods to predict increased intracranial pressure during 61. Yokobori S, Hajavelli S, Mondello S, Mo-Seaney J, Bramlett intensive care and long-term neurological outcome after trau- HM, Dietrich D, Bullock MR. Neuroprotective effect of preop- matic brain injury: development and validation in a multicenter eratively induced mild hypothermia as determined by biomarkers dataset. Neurol crit care. 2013. doi:10.1097/CCM.0b013e3182 and histopathological estimation in a rat subdural hematoma 742d0a. decompression model. J Neurosurg. 2013;118:370–80. 46. Vespa PM, Miller C, McArthur D, Eliseo M, Etchepare M, Hirt 62. Helmy A, Antoniades CA, Guilfoyle MR, Carpenter KLH, D, Gleen TC, Martin N, Hovda D. Nonconvulsive electrographic Hutchinson PJ. Principal component analysis of cytokine and seizures after traumatic brain injury result in a delayed, prolonged chemokine response to human traumatic brain injury. PLoS ONE. increase in intracranial pressure and metabolic crisis. Crit Care. 2012;7(6):e39677. 2007;35(12):2830–6. 63. Yan EB, Hellewell SC, Bellander BM, Agyopomaa DA, Morg- 47. Martin NA, Patwardhan RV, Alexander MJ, Africk CZ, Lee JH, anti-Kossmann MC. Post-traumatic hypoxia exacerbates Shalmon E, Hovda DA, Becker DP. Characterization of cerebral neurological deficit, neuroinflammation and cerebral metabolism hemodynamic phases following severe head trauma: hypoperfu- in rats with diffuse traumatic brain injury. J Neuroinflammation. sion, hyperemia, and vasospasm. J Neurosurg. 1997;87(1):9–19. 2011. doi:10.1186/1742-2094-8-147. 48. Bor-Seng-Shu E, Hirsh R, Teixeira MJ, Andrade AF, Junior RM. 64. Bouras T, Gatzonis S, Georgakoulias N, Karatza M, Siatouni A, Cerebral hemodynamic changes gauged by transcranial Doppler Stranjalis G, Boaviatsis E, Vasileiou S, Sakas DE. Neuro- ultrasonography in patients with posttraumatic brain swelling inflammatory sequelae of minimal trauma in the non-traumatized

123 162 Neurocrit Care (2014) 21:152–162

human brain. A microdialysis study. J Neurotrauma. 2011. function after traumatic brain injury in the immature rat and doi:10.1089/neu.2011.1790. piglet. J Neurotrauma. 2011;28:763–74. 65. Rooyackers O, Thorell A, Nygren J, Ljungqvist O. Microdialysis 67. Hugosson R, Sjolander U, Ungerstedt U. Treatment of malignant methods for measuring human metabolism. Curr Opin Clin Nutr glioma by a new therapeutic principle. Acta Neurochir (Wien). Metab Care. 2004;7:515–21. 1992;114:8–11. 66. Kilbaugh TJ, Bhandare S, Lorom DH, Saraswati M, Robertson 68. Klaus S, Heringlake M, Bahlmann L. Bench-to-bedside review: CL, Margulies SS. Cyclosporin a preserves mitochondrial microdialysis in intensive care medicine. Crit Care. 2004;8:363–8.

123