Acta Neurochir (Wien) (2008) 150: 461–470 DOI 10.1007/s00701-008-1580-3 Printed in The Netherlands
Clinical Article Metabolic failure precedes intracranial pressure rises in traumatic brain injury: a microdialysis study A. Belli1 , J. Sen2 , A. Petzold3 , S. Russo2 , N. Kitchen2 , M. Smith4 1
Division of Clinical Neurosciences, University of Southampton, Southampton, UK Victor Horsley Department of Neurosurgery, The National Hospital for Neurology and Neurosurgery, Queen Square, London, UK 3 Department of Neuroimmunology, Institute of Neurology, Queen Square, London, UK 4 Department of Neuroanaesthesia, The National Hospital for Neurology and Neurosurgery, Queen Square, London, UK 2
Received 10 March 2008; Accepted 14 March 2008; Published online 18 April 2008 # Springer-Verlag 2008
Summary Background: Cerebral microdialysis (MD) is able to detect markers of tissue damage and cerebral ischaemia and can be used to monitor the biochemical changes subsequent to head injury. In this prospective, observational study we analysed the correlation between microdialysis markers of metabolic impairment and intracranial pressure (ICP) and investigated whether changes in biomarker concentration precede rises in ICP. Methods: MD and ICP monitoring was carried out in twenty-five patients with severe TBI in Neurointensive care. MD samples were analysed hourly for lactate:pyruvate (LP) ratio, glutamate and glycerol. Abnormal values of microdialysis variables in presence of normal ICP were used to calculate the risk of intracranial hypertension developing within the next 3 h. Findings: An LP ratio > 25 and glycerol > 100 mmol= L, but not glutamate > 12 mmol=L, were associated with significantly higher risk of imminent intracranial hypertension (odds ratio: 9.8, CI 5.8–16.1; 2.2, CI 1.6–3.8; 1.7, CI 0.6–3, respectively). An abnormal LP ratio could predict an ICP rise above normal levels in 89% of cases, whereas glycerol and glutamate had a poorer predictive value. Correspondence: Antonio Belli, FRCS (Gla), FRCS (SN), MD, Division of Clinical Neurosciences, Southampton University Hospital, Tremona Road, Southampton SO16 6YD, UK. e-mail:
[email protected]
Conclusions: Changes in the compound concentrations in microdialysate are a useful tool to describe molecular events triggered by TBI. These changes can occur before the onset of intracranial hypertension, suggesting that biochemical impairment can be present before low cerebral perfusion pressure is detectable. This early warning could be exploited to expand the window for therapeutic intervention. Keywords: Glutamate; glycerol; ICP; lactate:pyruvate ratio; microdialysis; traumatic brain injury.
Introduction Traumatic brain injury (TBI) is the leading cause of morbidity and mortality in the first four decades of life, accounting for 15% of deaths in the 15–45 year age group and a high prevalence of long-term disability in those who survive [34]. Mortality from TBI has declined over the last two decades thanks to improvements in prehospital care and resuscitation. The importance of targeted management during neurocritical care is now accepted and may further improve outcome [24]. The delivery of targeted therapy after TBI depends upon the ability to monitor pathophysiological changes at and around the site of injury, as well as in remote regions where compensatory changes may occur. Immediately after the primary injury, which occurs at the time of the initial impact and is not amenable to treatment, a complex
462
interplay of pathophysiological mechanisms and compensatory responses are activated [37]. Because of the rapidly evolving nature of TBI, the clinical challenge is to recognise and respond to adverse changes before irreversible damage occurs, whilst avoiding the use of unwarranted medical or surgical interventions. Neurocritical care management of TBI aims at preventing or treating secondary brain damage and relies on the ability to monitor rapidly evolving changes in comatose and anaesthetised patients with highly sensitive and specific techniques. Cerebral microdialysis is a well-established laboratory tool that is increasingly used as a bedside monitor to provide on-line analysis of brain tissue biochemistry after TBI [14, 16, 25]. Microdialysis measures biochemical changes in brain extracellular fluid (ECF) that are known to be early markers of tissue damage and ischaemia and therefore has the potential to monitor the processes of secondary injury after TBI [9, 13]. This may identify patients at risk of the development of secondary injury and guide therapy within clinically useful timescales, thereby potentially reducing the risk of secondary damage and improving outcome [21]. In this prospective, observational study of patients with TBI admitted to the neurocritical care of a University Teaching Hospital, we analysed the correlation between conventional microdialysis variables (lactate:pyruvate ratio; glycerol and glutamate concentrations) and ICP, and investigated whether changes in biomarker concentration are able to predict changes in ICP.
Methods and materials Patient management Following institutional ethics committee approval and written agreement from the next of kin, 25 patients admitted to the neurosurgical intensive care unit (NICU) of the National Hospital for Neurology and Neurosurgery (NHNN) were recruited into the study. Inclusion criteria were: a) traumatic brain injury, b) age 16 years and c) indication for invasive monitoring of intracranial pressure (ICP). NHNN is a tertiary referral centre and all patients were therefore transferred intubated and ventilated after initial resuscitation and stabilisation in the referring Accident and Emergency departments. On the NICU, all patients were sedated with propofol and fentanyl and were mechanically ventilated to maintain PaCO2 between 4.5–5.0 kPa and PaO2 > 13.5 kPa. Routine monitoring requirements included ECG, non-
A. Belli et al.
invasive oxygen saturation, invasive arterial blood pressure (ABP), temperature and end-tidal CO2. All patients with severe head injury (Glasgow Coma Scale 8) or not suitable for immediate weaning of sedation after surgery underwent ICP monitoring using an intraventricular (Spiegelberg GmbH & Co, Hamburg, Germany) or an intraparenchymal microsensor (Codman, Johnson & Johnson, Randolph, MA), which have been shown to produce very similar ICP measurements [43]. All patients received local protocol-guided therapy, based on the Guidelines for the management of severe head injury, Joint Section on Neurotrauma and Critical care of the American Association of Neurological Surgeons [1] and the European Brain Injury Consortium (EBIC) [18] guidelines, to limit rises in ICP and maintain cerebral perfusion pressure (CPP) > 60 mmHg. In the absence of a haematoma amenable to surgical treatment, ICP elevations above 20–25 mmHg and=or the inability to maintain the cerebral perfusion pressure (CPP) above 60 mmHg were treated by an algorithm based on the protocol, with a stepwise utilisation of the following interventions as appropriate: a) volume expansion and vasopressors=inotropes if ABP was insufficient to maintain an adequate CPP, b) increased sedation and analgesia, c) neuromuscular blockade, d) moderate hyperventilation (PaCO2 4.0–4.5 kPa), e) mild hypothermia (target 34.5 C) using a cooling blanket, f) osmotic therapy with mannitol, g) ventricular cerebro-spinal fluid (CSF) drainage and h) decompressive craniectomy. Urgent CT scans were carried out routinely for all unexpected ICP rises in order to rule out hydrocephalus or recollection of a haematoma. Weaning of patients from sedation and ventilation was carried out after the acute phase of treatment, directed by clinical and radiological evidence, according to the NHNN algorithm. This includes: a) normalisation of temperature if the patient was actively cooled ( 36 C), b) discontinuation of paralysis, c) normalisation of PaCO2 (5.0–6.0 kPa), d) gradual discontinuation of sedation and analgesia, e) withdrawal of vasopressors= inotropes as ABP allows and f) weaning of ventilation and tracheal extubation.
Microdialysis interventions Cerebral microdialysis was monitored in all patients using a CMA 600 bedside analyser (CMA Microdialysis, Stockholm, Sweden) and a gold-tipped microdialysis catheter (CMA 70, CMA Microdialysis, Stockholm, Sweden) that has a membrane length of 10 mm, a diam-
463
Metabolic failure precedes intracranial pressure rises in TBI
Data collection and statistical analysis
Fig. 1. CT scan of a patient with diffuse axonal injury who underwent a standard bicoronal decompressive craniectomy. The tip of the microdialysis catheter, visible on the scan as a bright dot, was placed at surgery in the left frontal region, approximately 2 cm from a small contused area identified on the cortical surface
eter of 0.52 mm and a molecular mass cut-off of 20 kDa. The catheters were inserted either intraoperatively, if the patient was undergoing surgery, or via a skull-fixed triple lumen bolt (Technicam Ltd, Newton Abbot, UK) capable of accepting up to two MD catheters at different depth of insertion (cortical and subcortical) and one ICP probe. For the purpose of this study only the cortical catheter was used and this was placed in the ‘penumbra’ surrounding a focal lesion (about 2 cm away from the area of damage) or in the right frontal lobe in those with a diffuse injury (Fig. 1). The intraoperative implantation was carried out under direct vision in undamaged cortex adjacent to the surgical focus (2 cm away). In all cases the position of the probe was verified on subsequent CT scans. An isotonic (NaCl 147 mmol=L, KCl 2.7 mmol= L, CaCl2 1.2 mmol=L, MgCl2 0.85 mmol=L) proprietary perfusion fluid (CMA=Microdialysis, Solna, Sweden) was perfused at a rate of 0.3 mL=min. MD samples were collected and analysed every hour for lactate, pyruvate, glycerol and glutamate by the bedside nurse using the CMA 600 Analyser (CMA=Microdialysis, Solna, Sweden), which utilises an enzymatic reagent and colorimetric technique. For each analysis, the ratio between lactate and pyruvate (LP ratio) was calculated.
ABP, CPP and ICP were collected on a minute-by-minute basis on a central computer station via ad-hoc software (ICU Pilot – CMA=Microdialysis, Solna, Sweden). Data collection did not include the first three hours after catheter insertion in order to reduce the influence of implantation artifact. In order to minimise the effect of transient ICP peaks, e.g. due to coughing, suctioning or other nursing intervention, the moving average over a period of 5 min was calculated for ICP and CPP (avICP and avCPP, respectively). The following analyses were then carried out for microdialysis parameters with respect to such averages: Spearman’s correlation rank, Chi-square test and odds ratio. For the latter two tests the following values were considered abnormal: avICP > 20 mmHg for at least 5 min, LP ratio > 25, glutamate > 12 mmol=L and glycerol >100 mmol=L, in line with previous studies [15, 19]. When correlating real time data (ICP and CPP) with MD values, an allowance of 20 min was made for transit time delay, i.e. the interval it takes the dialysate to reach the vial. For the calculation of specificity, sensitivity, positive predictive value (PPV) and negative predictive value (NPV) intracranial hypertension was defined as the target condition and microdialysis as the test to predict it. Therefore, PPV was the proportion of abnormal MD measurements that were followed within the next 3 h by at least one episode of raised avICP, NPV was the proportion of normal MD measurements that were not followed by episodes of raised avICP for a period of at least 3 h. This analysis was then repeated discounting all MD measurements where the avICP had been abnormal in the hour preceding the measurement (1300 out of 2102 measurements discounted). Results A total of 2102 microdialysis samples were collected and time-matched with ICP and CPP data for this study. MD monitoring was started at a median time of 32 h (range 7–118 h) from injury. The median monitoring time per patient was 3.9 days (range 28 h to 13 days). Demographic data are shown in Table 1. In one patient the MD catheter failed following insertion and had to be replaced, resulting in 5 h of missing data. In another patient the MD analysis was not carried out for the duration of a nursing shift (12 h), therefore the ICP, CPP and ABP data were not included for the same period. Missing data also resulted from the MD
464
A. Belli et al.
Table 1. Patients demographics and interventions Case no.
Main finding on CT scan
Surgical intervention
GCS
GOSy
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
ASDH ASDH, CSDH bilateral contusions ICH, SAH, hydrocephalus small EDH, hydrocephalus ASDH, ICH ASDH, ICH ASDH, traumatic SAH ICH, contusions contusions small ASDH small EDH, contusions ASDH GSW traumatic SAH from stabbing ICH, ASDH and EDH acute on chronic SDH contusion, ASDH diffuse axonal injury contusions ASDH traumatic SAH bilateral contusions ASDH, bilateral contusions diffuse axonal injury
evacuation, decompressive craniectomy evacuation decompressive craniectomy, lobectomy EVD EVD evacuation evacuation, decompressive craniectomy evacuation, lobectomy none none none decompressive craniectomy, lobectomy evacuation none EVD evacuation evacuation evacuation none none evacuation decompressive craniectomy, EVD none evacuation, decompressive craniectomy decompressive craniectomy, lobectomy
3 5 11 11 9 7 6 13 11 3 11 5 6 4 5 8 8 5 7 7 8 15 8 4 5
1 1 1 1 1 1 1 2 4 3 1 4 1 1 1 3 5 1 1 1 1 1 1 1 3
GOSz
Age 40 72 41 60 25 61 19 58 22 25 23 28 69 45 21 37 78 61 38 29 18 35 66 57 16
1 5 3 4
4 5
3
ASDH Acute subdural haematoma, CSDH chronic subdural haematoma, SDH subdural haematoma, ICH intracerebral haematoma, EDH extradural haematoma, GSW gunshot wound, SAH subarachnoid haemorrhage, EVD external ventricular drain. Glasgow Coma Scale after resuscitation and prior to intubation. y Glasgow Outcome Scale at discharge (scores: 1 death, 2 persistent vegetative state, 3 severe disability, 4 moderate disability, 5 good recovery) z Glasgow Outcome Scale at follow up (mean 4.6 months, median 5 months).
hourly monitoring not being carried out during surgery or diagnostic procedures. No infection or dislodgement of the MD catheter or ICP probe was observed. In one patient, in whom the MD catheter had been inserted intraoperatively, an intracerebral haematoma developed when the catheter was removed, which required urgent surgical evacuation.
Table 3. Predictive value of LP ratio >25, glutamate >12mol=L and glycerol >100 mol=L measured in the presence of normal avICP ( 25 and glycerol measurements >100 mmol=L were strongly associated with abnormal avICP values (w2 ¼ 117, p < 0.001, and w2 ¼ 28, p < 0.001, respectively). However, glutamate values
465
Metabolic failure precedes intracranial pressure rises in TBI
>12 mmol=L were not associated with raised intracranial hypertension (w2 ¼ 0.001, p ¼ 0.5). The specificity, sensitivity, PPV and NPV of the microdialysis parameters in respect of avICP are shown in Table 3. Prediction of imminent intracranial hypertension Abnormal microdialysis variables and high ICP values are frequently recorded in a population of severely headinjured patients. In order to exclude a coincidental association between the two phenomena, all samples collected when avICP had been elevated (>20 mmHg) at any point in the preceding hour were excluded (1300 samples). The analysis of the censored 802 samples indicated that the measurement of an abnormal LP ratio (>25) when avICP was normal carried a significant risk of the patient developing intracranial hypertension within the next 3 h (odds ratio 9.8, CI 5.8–16.1, p< 0.001). The same analysis for glycerol >100 mmol=L showed an odds ratio of 2.2 (CI 1.6–3.8, p< 0.001). A high glutamate concentration was not associated with a greater risk of a subsequent avICP rise (odds ratio 1.7, CI 0.6– 3, p ¼ 0.71). As illustrated in Table 3, an abnormal LP ratio in the presence of normal ICP could predict the development of intracranial hypertension within the next 3 h in 89%
of cases with high specificity. Glycerol and glutamate had poorer predictive values. As shown in Table 1, only two patients presented with diffuse axonal injury and in the absence of an identifiable penumbra region had the catheter implanted into the right frontal lobe; therefore a separate analysis for this subgroup of patients was not carried out. Discussion Cerebral microdialysis allows frequent monitoring of brain ECF biochemistry at the bedside. In our experience, although time-consuming to set up, microdialysis was relatively easy to use and well accepted by the Neurointensive Care nursing staff. Lactate:pyruvate ratio Previous studies have demonstrated the presence of abnormal LP ratio, glycerol and glutamate following severe brain injury. Although the mean values recorded in our cohort of patients are higher than those described in other studies [14, 15, 19], wide variations in microdialysis results occur not only between different subjects but also within individual subjects over time [23]. An early study by Persson and Hillered [25] reported 25fold increases in LP ratio and glutamate concentration
Fig. 2. Graph of time-aligned LP ratio, glycerol, glutamate and ICP values of a 19-year old woman admitted with a large acute subdural haematoma and a small intracerabral haematoma following a road traffic accident (patient 7). Evacuation of the subdural haematoma and a decompressive craniectomy were performed immediately after admission to our hospital; at the end of surgery MD catheter and ICP monitor were implanted. The graph shows a steady rise in the MD parameters several hours before a sharp rise in ICP. An urgent CT scan obtained following the elevation in intracranial pressure showed diffuse brain swelling but no recurrent haematoma. Despite medical efforts to control the ICP the patient demised a few hours later
466
after brain injury and Hutchinson et al. [15] subsequently demonstrated that such variations reflect the high and changing levels of metabolic activity within the injured brain. After TBI, a high LP ratio has been found to correlate with the severity of clinical symptoms and fatal outcome [14, 35]. LP ratio is reputed to be a sensitive marker of tissue ischaemia [23, 28, 31]. Following TBI, cellular swelling occurs as a result of an auto-destructive cascade of ionic, metabolic and immunological changes initiated by the primary brain injury [34]. Inside the rigid cranium, this leads to increasing ICP and decreasing CPP. In vulnerable areas, tissue ischaemia results in failing cellular metabolism and subsequent release of excitotoxic amimoacids, production of oxygen free radical and tissue acidosis [37, 44]. If unchecked, these changes will lead to further cellular swelling and a vicious circle of worsening cerebral ischaemia causing increasing tissue acidosis and further excitotoxic and peroxidative damage. In our study we have demonstrated that an abnormal LP ratio is not only associated with intracranial hypertension but often precedes its development. Whilst the LP ratio correlates with the cerebral ischaemia that follows elevations in ICP and reductions in CPP, it may also describe biochemical events occurring at cellular level before the onset of intracranial hypertension. An example of this is illustrated in Fig. 2. A possible explanation is that mitochondrial impairment prevents adequate utilisation of energy substrates even in the presence of a normal cerebral perfusion. In this case, a high LP ratio would signify mitochondrial damage causing decreased oxygen consumption by the electron transport chain and leading to decreased velocity of the Krebs cycle. Even in the presence of a normal glycolyric rate these phenomena should provoke reduction in pyruvate utilisation by the pyruvate dehydrogenase complex with consequent increase in lactate production (resulting in higher lactate release in MD with increase in LP ratio). Mitochondrial failure has been shown to occur following TBI both in animal and human studies [2, 32, 38, 39, 42]. Mitochondria are known to be vulnerable to several pathological mechanisms that can occur following TBI, including incomplete or complete ischaemia and reperfusion damage [26], systemic acidosis [11, 26], hyperglycaemia [17], intracellular Ca2þ accumulation secondary to excitotoxic stimulation [3, 30, 36, 41] and attack by reactive oxygen species generated by a number of different pathways [6–8, 22]. Mitochondrial dysfunction in itself is a recognised trigger of apoptosis and necrosis [6, 22].
A. Belli et al.
The detection of an abnormal LP ratio by microdialysis in the penumbra region, in the presence of normal ICP, should not be interpreted exclusively as an ischaemic change, such as observed during episodes of intracranial hypertension and low CPP, but also as an early manifestation of mitochondrial impairment that precedes rather than follows widespread cellular swelling and intracranial hypertension. A recent study by Vespa’s group [40] showed a 25% incidence of LP ratio > 40 in 19 TBI patients, whereas the incidence of ischaemia as measured by positron emission tomography in the same population was only 2.4%. The authors concluded that TBI leads to a state of persistent altered metabolism that is reflected in abnormal MD LP ratio and that this is not an indicator of ischaemia. Some increments in LP ratio may also be due to a decrease in MD pyruvate secondary to an impairment in the glycolytic pathway, as recently proposed by Hillered et al. [10]. Interestingly, in our cohort we found a weak but significant inverted correlation between MD pyruvate and ICP (r ¼ 0.13, p ¼ 0.001), suggesting that impaired glycolysis may be in part responsible for our findings. In our study some of the patients underwent insertion of an external ventricular drain or a decompressive craniectomy, leading to possible underestimation of intracranial hypertension in relation to abnormal microdialysis parameters, as compared to untreated patients. This in turn may have led to an underestimation of PPV of microdialysis parameters in our study. However, if surgical intervention also results in an amelioration of the brain metabolic state, the systematic effect introduced should simultaneously influence both ICP and microdialysis values, thus PPV should not be significantly affected. The observation in our cohort of patients that low LP ratios are poor negative predictors for ICP rises suggests that not all episodes of intracranial hypertension can be ascribed to biochemical events. Many of these episodes are the result of mechanical events, such as the patient coughing, receiving nursing interventions or physiotherapy, changes in the ventilation settings, etc., that cannot be predicted by brain biochemical monitoring. However, the high PPV and specificity of the microdialysis warning may allow the clinician to deliver targeted therapy in selected patients before irreversible damage occurs.
Glycerol Glycerol ECF concentration rises following the breakdown of cell membranes containing glycerol-phospholip-
467
Metabolic failure precedes intracranial pressure rises in TBI
ids. Although high glycerol values reflect the severity of cell damage and indicate the presence of neuronal death, they should not be interpreted as meaning that there is no scope for intervention to prevent further damage. We hypothesise that if the microdialysis probe is correctly placed in the penumbra, rises in glycerol concentration whilst the ICP is still normal may indicate that the biochemical damage is spreading outward from the epicentre and is affecting the area where the probe is implanted. If unchecked, the damage will progressively extend to neighbouring areas and will eventually result in diffuse swelling detectable with ICP monitoring. Therefore, it is conceivable that rises in glycerol concentration in the penumbra predict diffuse secondary changes later occurring on a global level. In our study, the predictive value of glycerol was much poorer than LP ratio. Several confounding factors may contribute to this, including the actual sources of glycerol in TBI [13]. In theory it is possible that lipolysis of somatic fatty tissue under stress conditions may contaminate ECF measurement. We did not assess plasma levels of this biomarker but other animal and human studies have shown that extracranial contamination is unlikely affect brain tissue glycerol concentrations to any significant degree [12, 20]. In a recent study, where concentrations of this marker were measured alongside brain tissue PO2 in severely head-injured patients, it was observed that glycerol levels tended peak in the first hour after injury, possibly representing degradation of cells damaged by shearing forces [4]. Therefore, glycerol elevations may be interpreted as biochemical residue rather than a secondary event. It was also noted in the same study that glycerol levels were significantly higher only in cases of very low brain tissue PO2 (
