Pathophysiologic Mechanisms of Cerebral Ischemia and Diffusion Hypoxia in Traumatic Brain Injury

Transcription

1 Research Original Investigation Pathophysiologic Mechanisms of Cerebral Ischemia and Diffusion Hypoxia in Traumatic Brain Injury Tonny V. Veenith, FRCA; Eleanor L. Carter, FRCA; Thomas Geeraerts, PhD; Julia Grossac, MD; Virginia F. J. Newcombe, PhD; Joanne Outtrim, MSc; Gloria S. Gee, AS; Victoria Lupson, BSc; Rob Smith, PhD; Franklin I. Aigbirhio, PhD; Tim D. Fryer, PhD; Young T. Hong, PhD; David K. Menon, PhD; Jonathan P. Coles, PhD IMPORTANCE Combined oxygen 15 labeled positron emission tomography ( 15 O PET) and brain tissue oximetry have demonstrated increased oxygen diffusion gradients in hypoxic regions after traumatic brain injury (TBI). These data are consistent with microvascular ischemia and are supported by pathologic studies showing widespread microvascular collapse, perivascular edema, and microthrombosis associated with selective neuronal loss. Fluorine 18 labeled fluoromisonidazole ([ 18 F]FMISO), a PET tracer that undergoes irreversible selective bioreduction within hypoxic cells, could confirm these findings. Editorial page 54 Supplemental content at jamaneurology.com OBJECTIVE To combine [ 18 F]FMISO and 15 O PET to demonstrate the relative burden, distribution, and physiologic signatures of conventional macrovascular and microvascular ischemia in early TBI. DESIGN, SETTING, AND PARTICIPANTS This case-control study included 1 patients who underwent [ 18 F]FMISO and 15 O PET within 1 to 8 days of severe or moderate TBI. Two cohorts of 1 healthy volunteers underwent [ 18 F]FMISO or 15 O PET. The study was performed at the Wolfson Brain Imaging Centre of Addenbrooke s Hospital. Cerebral blood flow, cerebral blood volume, cerebral oxygen metabolism (CMRO 2 ), oxygen extraction fraction, and brain tissue oximetry were measured in patients during [ 18 F]FMISO and 15 O PET imaging. Similar data were obtained from control cohorts. Data were collected from November 23, 27, to May 22, 212, and analyzed from December 3, 212, to January 6, 216. MAIN OUTCOMES AND MEASURES Estimated ischemic brain volume (IBV) and hypoxic brain volume (HBV) and a comparison of their spatial distribution and physiologic signatures. RESULTS The 1 patients with TBI (9 men and 1 woman) had a median age of 59 (range, 3-68) years; the 2 control cohorts (8 men and 2 women each) had median ages of 53 (range, 41-76) and 45 (range, 29-59) years. Compared with controls, patients with TBI had a higher median IBV (56 [range, 9-281] vs 1 [range, -11] ml; P <.1) and a higher median HBV (29 [range, -16] vs 9 [range, 1-24] ml; P =.2). Although both pathophysiologic tissue classes were present within injured and normal appearing brains, their spatial distributions were poorly matched. When compared with tissue within the IBV compartment, the HBV compartment showed similar median cerebral blood flow (17 [range, 11-4] vs 14 [range, 6-22] ml/1 ml/min), cerebral blood volume (2.4 [range, ] vs 3.9 [range, ] ml/1 ml), and CMRO 2 (44 [range, 27-67] vs 71 [range, 34-88] μmol/1 ml/min) but a lower oxygen extraction fraction (38% [range, 29%-5%] vs 89% [range, 75%-1%]; P <.1), and more frequently showed CMRO 2 values consistent with irreversible injury. Comparison with brain tissue oximetry monitoring suggested that the threshold for increased [ 18 F]FMISO trapping is probably 15 mm Hg or lower. CONCLUSIONS AND RELEVANCE Tissue hypoxia after TBI is not confined to regions with structural abnormality and can occur in the absence of conventional macrovascular ischemia. This physiologic signature is consistent with microvascular ischemia and is a target for novel neuroprotective strategies. JAMA Neurol. 216;73(5): doi:1.11/jamaneurol Published online March 28, 216. Author Affiliations: Division of Anaesthesia, University of Cambridge, Addenbrooke s Hospital, Cambridge, England (Veenith, Carter, Geeraerts, Grossac, Newcombe, Outtrim, Menon, Coles); Department of Critical Care Medicine, University Hospital of Birmingham National Health Service Trust, Queen Elizabeth Medical Centre, Birmingham, England (Veenith); Department of Anesthesiology and Critical Care, University Hospital of Toulouse, Toulouse, France (Geeraerts, Grossac); Wolfson Brain Imaging Centre, Department of Clinical Neurosciences, University of Cambridge, Addenbrooke s Hospital, Cambridge, England (Gee, Lupson, Smith, Aigbirhio, Fryer, Hong). Corresponding Author: Jonathan P. Coles, PhD, Division of Anaesthesia, University of Cambridge, Addenbrooke s Hospital, Hills Road, PO Box 93, Cambridge CB2 QQ, England (jpc44@wbic.cam.ac.uk). 542 (Reprinted) jamaneurology.com

2 Cerebral Ischemia and Diffusion Hypoxia in Traumatic Brain Injury Original Investigation Research Previous studies 1,2 have used oxygen 15 labeled positron emission tomography ( 15 O PET) to define evidence of cerebral ischemia after early traumatic brain injury (TBI). Although other 15 O PET studies 3,4 have found less convincing evidence of ischemia, they typically demonstrate evidence of metabolic dysfunction that correlates with focal microdialysis. The relevance of such early physiologic derangements are evidenced by their association with local tissue fate, 5,6 chronic brain atrophy, 4 and clinical outcome. 1 Although nonischemic derangements after TBI 7 have been attributed to mitochondrial dysfunction, 3,8,9 studies have shown reductions in brain tissue PO 2 (PbtO 2 ) These PbtO 2 reductions are associated with worse outcome, 1-12 and interventions aimed at optimizing oxygen delivery show promise. 1,13,14 These discordant findings are best explained by an increased gradient for oxygen diffusion from the cerebral microvasculature to the interstitium. A previous study 15 used PbtO 2 monitoring and 15 O PET to demonstrate increased gradients for oxygen diffusion within hypoxic regions in the absence of macrovascular ischemia. Further, several investigators have shown that diffusion barrier ischemia can be explained by microvascular failure due to endothelial swelling, perivascular edema, and microthrombosis, particularly surrounding focal lesions. Fluorine 18 labeled fluoromisonidazole ([ 18 F]FMISO) is a hypoxia PET tracer that undergoes irreversible selective bioreduction within hypoxic but viable cells 18 and has been used after stroke and subarachnoid hemorrhage. 22 This study is the first clinical investigation, to our knowledge, to combine [ 18 F]FMISO with 15 O PET measurement of cerebral blood flow (CBF), cerebral blood volume (CBV), cerebral oxygen metabolism (CMRO 2 ), and oxygen extraction fraction (OEF) to interrogate pathophysiologic derangements. Imaging tissue hypoxia in the absence of macrovascular ischemia would confirm evidence of microvascular ischemia. We aimed to use [ 18 F]FMISO and 15 O PET to demonstrate the burden, distribution, and physiologic signatures of conventional macrovascular and microvascular ischemia after TBI. Methods Key Points Question Do cerebral ischemia and diffusion hypoxia have distinct pathophysiologic mechanisms in traumatic brain injury (TBI)? Findings In this case-control study using oxygen 15 labeled and fluorine 18 labeled fluoromisonidazole positron emission tomography in 1 patients with TBI and 2 controls, tissue hypoxia after TBI was not confined to regions with structural abnormality and could occur in the absence of conventional ischemia. Meaning This physiologic signature is consistent with microvascular ischemia and is a target for novel neuroprotective strategies. Participants Ten patients with TBI (9 men and 1 woman) with a median age of 59 (range, 3-68) years were recruited from Addenbrooke s Hospital. The patients presented with a median postresuscitation Glasgow Coma Scale score of 7 (range, 3-12), but required sedation and ventilation for control of intracranial pressure (Table). Patient management included protocoldriven care aiming for an intracranial pressure of less than 2 mm Hg, cerebral perfusion pressure of greater than 65 mm Hg, and, where available, PbtO 2 values of greater than 15 mm Hg. 14,24 Patients who received surgical intervention or second-tier medical therapies (barbiturate coma or moderate hypothermia [33 C-35 C]) are specified in the Table. The outcome was evaluated using the Glasgow Coma Scale at 6 months. Ten healthy volunteers (8 men and 2 women) with a median age of 45 (range, 29-59) years underwent 15 O PET and another 1 healthy volunteers (8 men and 2 women) with a median age of 53 (range, 41-76) years underwent [ 18 F]FMISO PET (control groups). The Cambridge Central Research Ethics Committee and Administration of Radioactive Substances Advisory Committee approved this study. Written informed consent or consultee agreement from the next-of-kin was obtained in accordance with the Declaration of Helsinki. 25 Imaging Data were collected from November 23, 27, to May 22, 212. Participants underwent magnetic resonance imaging using a 3-T scanner (Magnetom Verio; Siemens AG) within the Wolfson Brain Imaging Centre at the University of Cambridge. Sequences included a 3-dimensional T1-weighted magnetization prepared rapid gradient echo (MPRAGE), fluid-attenuated inversion recovery (FLAIR), gradient echo, and susceptibility-weighted imaging. Patients underwent 15 O PET, followed by [ 18 F]FMISO PET in the same session and scanner (Advance; GE Medical Systems), with FMISO injection at least 15 minutes after 15 O delivery. 15 O Positron Emission Tomography Emission data were acquired in 3-dimensional mode for the last 1 minutes of a 2-minute steady-state infusion of 8 MBq of 15 O water, in 3-dimensional mode for 5 minutes after a 1-minute inhalation of 75 MBq of 15 O carbon monoxide, and in 2-dimensional mode for the last 1 minutes of a 2-minute steady-state inhalation of 72 MBq of 15 O oxygen (to convert to megacuries, multiply by ). Parametric maps of CBF, CBV, CMRO 2, and OEF were calculated by inputting simultaneous PET and arterial tracer radioactivity concentration measurements into standard models. 1 [ 18 F]FMISO Positron Emission Tomography After [ 18 F]FMISO injection (3 MBq), PET data were acquired in 3-dimensional mode for 2.5 hours. Arterial plasma samples provided the input function for kinetic analysis. Voxelwise compartmental modeling used the irreversible version of BAFPIC, a basis function approach to 2-tissue compartmental modeling with a plasma input function. 26 Hypoxia was mapped using the parameter k 3 determined from BAFPIC, which denotes the tissue-trapping rate of [ 18 F]FMISO. 27 jamaneurology.com (Reprinted) JAMA Neurology May 216 Volume 73, Number 5 543

3 Research Original Investigation Cerebral Ischemia and Diffusion Hypoxia in Traumatic Brain Injury Table. Patient Characteristics Patient No./Age, y/sex Mechanism of Injury Injury Marshall Type GCS a CT Score b APACHE Classification c ISS d Neurosurgery Second-Tier Medical Therapy Time to Imaging, d GOS e 1/47/M Assault Contusions 8 EML DC NA 5 SD 2/59/M Fall Contusions 4 EML DC NA 1 D and SDH 3/34/M Fall Contusions, 6 EML EVD, DC NA 5 MD SDH, and DAI 4/6/F RTC EDH and TSAH 4 EML 25 2 C Hypothermia 2 GR 5/3/M RTC Contusions, 3 EML DC NA 1 MD SDH, and DAI 6/65/M Assault Contusions, SDH, and TSAH 12 NEML NA NA 8 GR 7/58/M Fall Contusions, SDH, TSAH, and DAI 8/66/M RTC Contusions, SDH, and DAI 9/31/M Fall Contusions, SDH, and DAI 1/68/M Fall Contusions, EDH, SDH, and TSAH Abbreviations: APACHE, Acute Physiology and Chronic Health Evaluation; C, craniotomy; D, death; DAI, diffuse axonal injury; DC, decompressive craniectomy; EDH, extradural hemorrhage; EML, evacuated mass legion; EVD, external ventricular drain; GCS, Glasgow Coma Scale; GOS, Glasgow Outcome Scale; GR, good recovery; ISS, Injury Severity Score; MD, moderate disability; NA, not applicable; NEML, nonevacuated mass lesion; RTC, road traffic crash; SD, severe disability; SDH, subdural hemorrhage; TSAH, traumatic subarachnoid hemorrhage. a Ranges from 3 to 15, with higher scores indicating less severe degrees of coma. 8 NEML 2 34 NA NA 8 GR 1 NEML NA Hypothermia 3 MD 5 NEML EVD, DC Hypothermia 5 SD 11 NEML NA NA 5 GR b Described by Marshall et al. 23 c Ranges from to 71, with higher scores indicating more severe disease and a higher risk of death. d Ranges from 1 to 75, with higher scores indicating more extensive and severe injury. e Ranges from good recovery to death. Image Processing We processed the PET data using an integrated image analysis gateway (PETAn) 28 that incorporates Statistical Parametric Mapping (SPM; Wellcome Department of Imaging Neuroscience, University College London), Matlab (MathWorks, Inc), and Analyze (AnalyzeDirect, Inc). Using the brain extraction tool in the FMRIB Software Library, 29 the skull and extracranial soft tissue were stripped from the MPRAGE and the extracted brain registered to the summed H 2 15 O and [ 18 F]FMISO images. The cerebrospinal fluid segment and CBF voxels of less than 2.36 ml/1 ml/ min consistent with the core lesion in PET studies were removed based on a positive predictive value of.95 for nonviable tissue. 5 To avoid artifacts within regions of obvious injury, analyses were conducted in native PET space, rather than using spatial transformation to a standard template. 2,3 Imaging Analysis Region of Interest Analysis Focal contusions were defined on FLAIR and segregated into core, contusion, and pericontusion using MPRAGE, gradient echo, and susceptibility-weighted imaging. 3 The lesion core was identified as a region of mixed-signal intensity consistent with hemorrhage and necrotic tissue and excluded from subsequent analyses. Contusion was identified as an area of high FLAIR signal consistent with edema; pericontusion, as a 1-cm border zone of normal-appearing tissue surrounding a contusion. The FLAIR images were coregistered to PET space using SPM8, and coregistration parameters were applied to lesion regions of interest. For comparison, a region of normalappearing mixed gray and white matter was defined in the patients with TBI. Ischemic Brain Volume We used OEF to assess the burden of ischemia with a technique that was previously validated in TBI. 1,2,31 We estimated an individualized critical OEF threshold (OEF crit ) (which equated to a cerebral venous oxygen content [CvO 2 ] of 3.5 ml/ 1 ml) for each participant using OEF crit =(CaO 2 3.5)/ CaO 2, where CaO 2 = 1.34 HbSaO PaO 2.CaO 2 is arterial oxygen content, Hb is the hemoglobin concentration in grams per 1 ml, SaO 2 is the fractional arterial oxygen saturation, and PaO 2 is the arterial PO 2. Application of these patientspecific OEF thresholds allowed calculation of the volume of voxels with CvO 2 values below this threshold, and hence estimation of the ischemic brain volume (IBV). Hypoxic Brain Volume We used k 3 values to assess the burden of tissue hypoxia because k 3 represents the rate constant for FMISO trapping under hypoxic conditions. Using the mean and SD of voxel k 3 values within the whole brain of each control participant, we calculated the upper 99% CI threshold using the mean plus 3 SDs. We used the brain volume with k 3 values above this threshold to calculate the hypoxic brain volume (HBV). We examined the volume, spatial location, and mismatch between the HBV and IBV. 544 JAMA Neurology May 216 Volume 73, Number 5 (Reprinted) jamaneurology.com

4 Cerebral Ischemia and Diffusion Hypoxia in Traumatic Brain Injury Original Investigation Research Figure 1. Evidence of Cerebral Ischemia Using Oxygen 15 Labeled Positron Emission Tomography After Head Injury 4 Cerebral Blood Flow, ml/1 ml/min 1.1 Oxygen Extraction Fraction, % [ 18 F] FMISO Trapping Rate, 1/min Fluid-Attenuated Inversion Recovery 15 Cerebral Oxygen Metabolism, μmol/1 ml/min Ischemic Brain Volume Hypoxic Brain Volume Fluid-attenuated inversion recovery (FLAIR), cerebral blood flow (CBF), cerebral oxygen metabolism (CMRO 2 ), oxygen extraction fraction (OEF), ischemic brain volume (IBV), fluorine 18 labeled fluoromisonidazole ([ 18 F]FMISO) trapping rate (k 3 ), and hypoxic brain volume (HBV) are shown in patient 1, who sustained a head injury after a fall. During imaging, cerebral perfusion pressure was 82 mm Hg, and intracranial pressure was 12 mm Hg. The FLAIR image demonstrates bilateral contusions within the temporal and parietal lobes on the right and the temporal lobe on the left. Cerebral blood flow is low in these regions, particularly on the right side. Cerebral oxygen metabolism is mildly reduced within the right temporal region, but a large increase in the OEF is seen, particularly within the right but also within the left temporal and parietal cortices. Increased k 3 values are found within the right temporal region but also across other injured and normal-appearing regions. The region with a critical increase in OEF above the individually calculated ischemic threshold (IBV) and the HBV are both shown in red overlying the FLAIR image. Within the total IBV of 131 ml in this patient, the mean CBF was 14 ml/1 ml/min; CBV, 3.4 ml/1 ml; CMRO 2, 84 μmol/1 ml/min; and OEF, 9%. The total HBV in this patient was 7 ml, with a mean CBF of 13 ml/1 ml/min; CBV, 2.2 ml/1 ml; CMRO 2, 47 μmol/1 ml/min; and OEF, 49%. The volume of overlap between these 2 tissues classes in this patient was 6 ml. Comparison With PbtO 2 For participants who underwent PbtO 2 monitoring, monitoring was continuous with values recorded every 5 to 1 minutes throughout PET. A 2-mm-diameter region of interest 15 was drawn around the sensor tip (LICOX; Integra Neurosciences Corp), and the [ 18 F]FMISO k 3 values were compared with PbtO 2 values. Statistical Analysis Data were analyzed from December 3, 212, to January 6, 216. Statistical analyses were conducted using Statview (version 5; SAS Institute Inc). Data are expressed as median (range), unless otherwise stated. Data were compared using Mann- Whitney and Spearman rank correlation tests and P values quoted after Bonferroni correction (where appropriate), with corrected P <.5 considered significant. The Dice similarity coefficient 32 was used to measure the degree of spatial overlap between IBV and HBV. Results Regional Physiologic Findings Physiologic findings were highly variable even within regions that appeared structurally normal (efigure in the Supplement). When compared with data from controls, contusions showed lower CBF and CMRO 2 (P <.1, Mann-Whitney test with Bonferroni correction), whereas CBV and OEF were variable but similar to those of the controls. Pericontusional tissue and regions that appeared structurally normal had lower CMRO 2 than tissue in controls (P <.1, Mann-Whitney test with Bonferroni correction), whereas CBF, CBV, and OEF were similar to those of controls. Ischemic Brain Volume When compared with controls, the IBV in the patient group was significantly higher (56 [9-281] ml vs 1 [-11] ml; P <.1, Mann-Whitney test). Although much of the IBV was close to visible lesions, with 23% (4%-65%) found within contusional and pericontusional regions, the IBV was also distributed across normal-appearing brain tissue (Figure 1). Physiologic features within the IBV are shown in Figure 2. We found no association between the IBV and days since injury or age (P =.49 and P =.34, respectively, Spearman correlation rank test). Hypoxic Brain Volume The HBV was variable but significantly higher in patients compared with controls, with a median of 29 (-16) ml vs 9 (1-24) ml (P =.2, Mann-Whitney test). A group change toward correlation between the IBV and HBV failed to jamaneurology.com (Reprinted) JAMA Neurology May 216 Volume 73, Number 5 545

5 Research Original Investigation Cerebral Ischemia and Diffusion Hypoxia in Traumatic Brain Injury Figure 2. Comparison of Physiologic Features Within the Ischemic Brain Volume (IBV) and Hypoxic Brain Volume (HBV) A Cerebral Blood Flow, ml/1 ml/min C Cerebral Oxygen Metabolism, μmol/1 ml/min Cerebral blood flow HBV a b, c IBV Normal Control Appearance Cerebral oxygen metabolism HBV d b, c IBV Normal Control Appearance achieve significance (ρ =.61 and P =.7, Spearman rank correlation test). The overlap volume between these 2 pathophysiologic tissue classes was 1 (-19) ml, and we found substantial spatial mismatch (Dice similarity coefficient, [-.1]) (Figure 1 and Figure 3). Although the HBV was often related to visible lesions, with 37% (21%-57%) found within contusional and pericontusional regions, HBV was also seen within normal-appearing brain tissue (Figures 1 and 3). Figure 2 compares summary physiologic data from the IBV and HBV tissue classes, tissue that B Cerebral Blood Volume, ml/1 ml D Oxygen Extraction Fraction, % Cerebral blood volume HBV a IBV Normal Control Appearance Oxygen extraction fraction HBV a IBV Normal Control Appearance Box and whisker plots of cerebral blood flow, cerebral blood volume, cerebral oxygen metabolism, and oxygen extraction fraction in brain tissue constituting the HBV, IBV, brain tissue that appeared to be structurally normal (normal appearance), and healthy volunteers (control). The central lines in each box denote median values; lower and upper boundaries, the 25th and 75th percentiles, respectively; error bars, the 1th and 9th percentiles, and solid circles, outliers. For all comparisons, Mann-Whitney tests with Bonferroni correction were used. a P <.1, IBV vs normal appearance. b P <.1, HBV vs control. c P <.1, IBV vs control. d P <.5, HBV vs normal appearance. c appeared structurally normal and was in neither of these classes, and tissue from controls. When tissue constituting the HBV was compared with tissue within the IBV (Figure 2), they showed similar median CBF (17 [range, 11-4] vs 14 [range, 6-22] ml; P =.22), CBV (2.4 [range, ) vs 3.9 [range, ] ml/1 ml; P =.9), and CMRO 2 (44 [range, 27-67] vs 71 [range, 34-88] μmol/1 ml/min; P =.14), but lower OEF (38% [range, 29%-5%] vs 89% [range, 75%- 1%]; P <.1, Mann-Whitney tests with Bonferroni correction). Cerebral metabolism below published 15 O PET thresholds for irreversible injury (37.6 μmol/1 ml/min) in TBI 5 was observed in 3 of 1 patients in the HBV tissue class compared with 1 within the IBV tissue class. We found no relationship between the HBV and the days since injury or age (P =.7 and P =.9, respectively, Spearman rank correlation test). Comparison With PbtO 2 Measurements Brain tissue PO 2 was available in 5 participants, and measurements during PET were 34 (16-55) mm Hg. The region of interest around the probe tip showed no [ 18 F]FMISO k 3 or OEF values that exceeded our HBV and IBV thresholds. Discussion By combining 15 O and [ 18 F]FMISO PET, we demonstrate evidence of conventional macrovascular cerebral ischemia and tissue hypoxia as long as 1 week after TBI. Spatial matching of these 2 tissue classes was poor, with voxels contributing to the HBV more frequently found within the vicinity of lesions. The IBV and HBV voxels showed comparable reductions in CBF, but the HBV tissue class showed a tendency toward lower CBV and CMRO 2 and significantly lower OEF. Further, the HBV more frequently exhibited CMRO 2 values within the range of irreversible injury. Although the IBV identifies conventional macrovascular ischemia, the coexistence of normal OEF (identified by 15 O PET) and low tissue PO 2 (identified by high [ 18 F]FMISO trapping) in the HBV is the typical signature of diffusion barrier hypoxia that, along with lower CBV, implies microvascular collapse and ischemia as an underlying mechanism. These findings confirm the existence of diffusion hypoxia, characterize its pathophysiologic signature as distinct from macrovascular ischemia, and show that they have incomplete spatial concordance. Diffusion hypoxia is a potential target for future novel neuroprotective strategies. Although 15 O and [ 18 F]FMISO PET have been used separately to identify ischemia, 19-21,33 we combined both tracers to interrogate pathophysiologic derangements after TBI. For [ 18 F]FMISO PET, we used kinetic analysis to calculate k 3 as a measure of hypoxia. Although the influx rate constant (K i ) 34 is often used to quantify irreversible trapping or metabolism of a tracer in tissue, it is sensitive to changes in tracer delivery, which is CBF dependent. This confounding issue is obviated through estimation of k 3, the rate constant for tissue [ 18 F]FMISO trapping, and in the context of low CBF in the vicinity of contusions, 35,36 k 3 is more suited to represent 546 JAMA Neurology May 216 Volume 73, Number 5 (Reprinted) jamaneurology.com

6 Cerebral Ischemia and Diffusion Hypoxia in Traumatic Brain Injury Original Investigation Research Figure 3. Evidence of Tissue Hypoxia Using Fluorine 18 Labeled Fluoromisonidazole ([ 18 F]FMISO) Positron Emission Tomography Fluid-Attenuated Inversion Recovery Ischemic Brain Volume Hypoxic Brain Volume Cerebral Blood Flow, ml/1 ml/min Oxygen Extraction Fraction, % [ 18 F] FMISO Trapping Rate, 1/min Fluid-attenuated inversion recovery (FLAIR), cerebral blood flow (CBF), oxygen extraction fraction (OEF), ischemic brain volume (IBV), 18 F-FMISO trapping rate (k 3 ), and hypoxic brain volume (HBV) are shown in patient 9, who sustained a head injury after a fall. During imaging, cerebral perfusion pressure was 8 mm Hg and intracranial pressure was 21 mm Hg. The FLAIR image demonstrates hemorrhagic contusions with surrounding vasogenic edema within bilateral frontal and right temporal regions. Additional areas of high signal consistent with injury are evident within the left thalamus and bilateral occipital regions. Thin subdural hematomas are seen over the right cortex and left frontal region. Cerebral blood flow is low within the frontal regions and is associated with increased k 3 values in the absence of an increase in OEF consistent with conventional macrovascular ischemia. The HBV (1 ml) in this patient had a mean CBF of 14 ml/1 ml/min; cerebral blood volume (CBV), 2.1 ml/1 ml; cerebral oxygen metabolism, 27 μmol/1 ml/min; and OEF, 35%. These values did not match the region of brain within the IBV (149 ml), with a mean CBF of 15 ml/1 ml/min; CBV, 3.4 ml/1 ml; CMRO 2, 63 μmol/1 ml/min; and OEF, 88%. The volume of overlap between these 2 tissues classes in this patient was1ml. trapping of [ 18 F]FMISO within hypoxic brain tissue. 27,37 The HBV was calculated from the total volume of voxels with k 3 values larger than the upper 99% CI value from control data. Because derangements are common across the whole brain after TBI, 1,6,35 we cannot use a similar approach to that used after ischemic stroke that defined increased [ 18 F]FMISO trapping greater than the upper 99% CI value from contralateral brain. 38 Other studies 2,21 used spatial normalization and voxelwise statistical testing to compare with controls. These approaches are less applicable to TBI because structural distortions are usually larger, making spatial normalization less dependable. We sought to identify [ 18 F]FMISO trapping within areas of obvious injury and normal-appearing regions, and therefore, used native PET space analyses to avoid artifacts from such processing techniques. 3,39 The volumes of hypoxic brain with increased [ 18 F]FMISO trapping in our participants were similar to those seen in ischemic stroke, 19,4 with a mean HBV of 47 ml in our patients and 4 patients with an HBV of at least 5% of brain volume. We removed lesion core using magnetic resonance imaging and excluding voxels where CBF was less than 2.36 ml/1 ml/min based on a positive predictive value of.95 for nonviable tissue in TBI. 5 After TBI, increased FLAIR signal can disappear on sequential imaging 41 and is not predictive of pan necrosis for all lesion voxels. 5 Because derangements are often found in normal-appearing regions, 6 we examined the whole brain, but highlighted when these regions were found within the vicinity of lesions using standard magnetic resonance imaging sequences. The mean HBV found outside contusion and pericontusion areas was 3 (-75) ml, and 4 participants had greater than 5 ml of such tissue. These results have implications for our understanding of oxygen delivery and use in clinical TBI. Although the HBV and IBV showed some overlap, most IBV voxels did not show jamaneurology.com (Reprinted) JAMA Neurology May 216 Volume 73, Number 5 547

7 Research Original Investigation Cerebral Ischemia and Diffusion Hypoxia in Traumatic Brain Injury Figure 4. Association Between Positron Emission Tomography Variables [ 18 F] FMISO Trapping Rate, 1/min Oxygen Extraction Fraction, % The association between the oxygen extraction fraction (OEF) and fluorine 18 labeled fluoromisonidazole ([ 18 F]FMISO) trapping rate (k 3 ) within voxels across the whole brain of individual patients (n = 1) is plotted using locally weighted scatterplot smoothing (LOWESS) with 66% tension 45 in Statview software (version 5; SAS Institute Inc). LOWESS is an outlier-resistant method based on local polynomial fits. 46 For this comparison, voxels with cerebral oxygen metabolism of less than 37.6 μmol/1 ml/min were excluded based on the lower 95% CI for nonlesion voxels after head injury. 5 significant [ 18 F]FMISO trapping. This finding suggests that at least some of the voxels with OEF in excess of our threshold CvO 2 values could maintain tissue PO 2 levels above those that result in irreversible bioreduction of [ 18 F]FMISO. The PO 2 at which bioreduction occurs is unclear, but in vitro data show that, although FMISO bioreduction shows some enhancement with PO 2 of less than 6 mm Hg, it rises steeply at a PO 2 of less than 1 mm Hg. 42 Given that the normal PbtO 2 is approximately 25 mm Hg, 43,44 many IBV voxels may have had PbtO 2 values in the range of 1 to 2 mm Hg, and our HBV threshold may simply be a more stringent physiologic marker of tissue hypoxia. In comparison with the IBV, CMRO 2 below thresholds previously identified for survival 5 was more common in the HBV. Although [ 18 F]FMISO trapping does not occur within necrotic tissue, 18 a proportion of these voxels may be destined for infarction, and the HBV may provide a more specific marker of tissue on the cusp of survival. To explore the relationship between OEF and [ 18 F]FMISO k 3 within viable brain, we excluded voxels at high risk for infarction based on the lower 95% CI for nonlesion brain after TBI, using a CMRO 2 threshold of 37.6 μmol/ 1 ml/min. 5 In Figure 4, the subsequent relationship between OEF and k 3 values within individual patients is shown using locally weighted scatterplot smoothing. 6,45,46 This association shows variability between patients, with a subset showing increased [ 18 F]FMISO trapping above an OEF threshold of approximately 6% to 7%, broadly consistent with conventional ischemia. We must emphasize that levels of [ 18 F]FMISO trapping were, in most cases, below our HBV thresholds and equivalent to tissue PO 2 levels of about 1 to 25 mm Hg. 42 Given that these data provide a transition zone between normal tissue and tissue that shows increased [ 18 F]FMISO trapping, physiologic characterization of tissue that shows pathologic levels of [ 18 F]FMISO trapping would be useful. Although participants showed increased [ 18 F]FMISO trapping across the whole brain, none was found within the vicinity of focal PbtO 2 monitoring probes. Clinically significant reductions in PbtO 2 are typically reported below 1 to 15 mm Hg, 47 and our local treatment protocol aims to maintain values of greater than 15 mm Hg, which were achieved in all participants. The lowest PbtO 2 level recorded during PET was 16 mm Hg. In terms of the threshold value at which [ 18 F]FMISO trapping occurs in TBI, we can conclude that PO 2 values within the HBV may have been no greater than 1 to 15 mm Hg. The proportion of HBV voxels that had high OEF meeting the criteria for macrovascular ischemia (IBV) was small (8%). This finding may relate to the fact that our patients underwent imaging 1 to 8 days after injury, at a time when conventional ischemia is less prevalent, 1,48 but lesions remain at risk for expansion. 3,39 This timing may have meant that many tissue regions showed complex and varying mixtures of macrovascular ischemia and diffusion hypoxia, making detection of clear OEF thresholds for [ 18 F]FMISO trapping challenging. Our characterization of prominent [ 18 F]FMISO trapping in perilesional regions is worth highlighting. Previous PET studies have shown severe derangements within and around cerebral contusions, but an increase in OEF consistent with cerebral ischemia is not always identified. 5,36 We found that tissue within the vicinity of such lesions is hypoxic but does not fulfil the criteria for conventional macrovascular ischemia. An explanation for these findings comes from studies showing widespread microvascular occlusion and perivascular edema after TBI, 16,49 associated with selective neuronal loss. 17 Conventional physiology dictates that to maintain CMRO 2 inthefaceoflowcbf,oef must be increased. 1,33 However, hypoxic regions may be less able to increase OEF owing to an increased gradient for oxygen diffusion, 15 which could explain our findings of low OEF and CBV despite evidence of low CBF and tissue hypoxia. Other studies have used diffusion-tensor imaging to demonstrate contusion expansion and that a rim of low apparent diffusion coefficient consistent with cytotoxic edema is often found surrounding contusions. 3 This finding may characterize a region of microvascular failure and represent a traumatic penumbra that may be rescued by effective therapy such as hyperoxia 14,39 or may be subsumed as the contusion enlarges. Conclusions These findings confirm the existence of diffusion hypoxia, characterize its pathophysiologic signature as distinct from conventional macrovascular ischemia, and show that diffusion 548 JAMA Neurology May 216 Volume 73, Number 5 (Reprinted) jamaneurology.com

8 Cerebral Ischemia and Diffusion Hypoxia in Traumatic Brain Injury Original Investigation Research hypoxia and macrovascular ischemia have incomplete spatial concordance. This physiologic signature is consistent with microvascular ischemia, and, importantly, this mechanism is also found within regions that appear structurally normal. Such findings require further scrutiny and are relevant to the development of future neuroprotective strategies. ARTICLE INFORMATION Accepted for Publication: January 1, 216. Published Online: March 28, 216. doi:1.11/jamaneurol Author Contributions: Drs Menon and Coles are joint senior authors and contributed equally to the study. Dr Coles had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Veenith, Fryer, Menon, Coles. Acquisition, analysis, or interpretation of data: All authors. Drafting of the manuscript: Veenith, Newcombe, Gee, Aigbirhio, Fryer, Menon, Coles. Critical revision of the manuscript for important intellectual content: Veenith, Carter, Geeraerts, Grossac, Newcombe, Outtrim, Lupson, Smith, Fryer, Hong, Menon, Coles. Statistical analysis: Veenith, Geeraerts, Newcombe, Menon, Coles. Obtained funding: Veenith, Newcombe, Fryer, Menon, Coles. Administrative, technical, or material support: Veenith, Outtrim, Lupson, Smith, Aigbirhio, Fryer, Hong, Menon, Coles. Study supervision: Fryer, Menon, Coles. Conflict of Interest Disclosures: None reported. Funding/Support: This study was supported by a clinical research training fellowship from National Institute of Academic Anaesthesia and Raymond Beverly Sackler studentship (Dr Veenith), a clinical research training fellowship from the Société Française d Anesthésie et de Réanimation (Dr Geeraerts), a Clinician Scientist Fellowship from the Health Foundation/Academy of Medical Sciences (Dr Newcombe), a Senior Investigator Award from the National Institute for Health Research (Dr Menon), grant WT93267 from the Wellcome Trust Project, program grant G ID from the Medical Research Council (Acute Brain Injury: Heterogeneity of Mechanisms, Therapeutic Targets and Outcome Effects), the UK National Institute of Health Research Biomedical Research Centre at Cambridge, and the UK Department of Health (Drs Aigbirhio, Fryer, Menon, and Coles, for the Technology Platform). Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication. Additional Contributions: Josef Alawneh, PhD, Addenbrooke s Hospital, Ramez Moustafa, PhD, Ain Shams University, Cairo, Egypt, and J. C. Baron, ScD, Institut National de la Santé et de la Recherche Medicale U894, Centre de Psychiatrie et Neurosciences, Hôpital Sainte-Anne, Université Paris, helped with recruitment and acquisition of fluorine 18 labeled fluoromisonidazole positron emission tomography among the controls. No compensation was received for these contributions. REFERENCES 1. Coles JP, Fryer TD, Smielewski P, et al. Incidence and mechanisms of cerebral ischemia in early clinical head injury. J Cereb Blood Flow Metab. 24;24(2): Coles JP, Fryer TD, Coleman MR, et al. Hyperventilation following head injury: effect on ischemic burden and cerebral oxidative metabolism. Crit Care Med. 27;35(2): Vespa P, Bergsneider M, Hattori N, et al. Metabolic crisis without brain ischemia is common after traumatic brain injury: a combined microdialysis and positron emission tomography study. J Cereb Blood Flow Metab. 25;25(6): Xu Y, McArthur DL, Alger JR, et al. Early nonischemic oxidative metabolic dysfunction leads to chronic brain atrophy in traumatic brain injury. J Cereb Blood Flow Metab. 21;3(4): Cunningham AS, Salvador R, Coles JP, et al. Physiological thresholds for irreversible tissue damage in contusional regions following traumatic brain injury. Brain. 25;128(pt 8): Coles JP, Cunningham AS, Salvador R, et al. Early metabolic characteristics of lesion and nonlesion tissue after head injury. J Cereb Blood Flow Metab. 29;29(5): Hutchinson PJ, Jalloh I, Helmy A, et al. Consensus statement from the 214 International Microdialysis Forum. Intensive Care Med. 215;41 (9): Dienel GA. Lactate shuttling and lactate use as fuel after traumatic brain injury: metabolic considerations. J Cereb Blood Flow Metab. 214;34 (11): Verweij BH, Muizelaar JP, Vinas FC, Peterson PL, Xiong Y, Lee CP. Impaired cerebral mitochondrial function after traumatic brain injury in humans. J Neurosurg. 2;93(5): Beynon C, Kiening KL, Orakcioglu B, Unterberg AW, Sakowitz OW. Brain tissue oxygen monitoring and hyperoxic treatment in patients with traumatic brain injury. J Neurotrauma. 212;29(12): Ponce LL, Pillai S, Cruz J, et al. Position of probe determines prognostic information of brain tissue PO 2 in severe traumatic brain injury. Neurosurgery. 212;7(6): Spiotta AM, Stiefel MF, Gracias VH, et al. Brain tissue oxygen-directed management and outcome in patients with severe traumatic brain injury. J Neurosurg. 21;113(3): Rockswold SB, Rockswold GL, Zaun DA, Liu J. A prospective, randomized phase II clinical trial to evaluate the effect of combined hyperbaric and normobaric hyperoxia on cerebral metabolism, intracranial pressure, oxygen toxicity, and clinical outcome in severe traumatic brain injury. J Neurosurg. 213;118(6): Nortje J, Coles JP, Timofeev I, et al. Effect of hyperoxia on regional oxygenation and metabolism after severe traumatic brain injury: preliminary findings. Crit Care Med. 28;36(1): Menon DK, Coles JP, Gupta AK, et al. Diffusion limited oxygen delivery following head injury. Crit Care Med. 24;32(6): Bullock R, Maxwell WL, Graham DI, Teasdale GM, Adams JH. Glial swelling following human cerebral contusion: an ultrastructural study. J Neurol Neurosurg Psychiatry. 1991;54(5): Stein SC, Graham DI, Chen XH, Smith DH. Association between intravascular microthrombosis and cerebral ischemia in traumatic brain injury. Neurosurgery. 24;54(3): Takasawa M, Moustafa RR, Baron JC. Applications of nitroimidazole in vivo hypoxia imaging in ischemic stroke. Stroke. 28;39(5): Alawneh JA, Moustafa RR, Marrapu ST, et al. Diffusion and perfusion correlates of the 18 F-MISO PET lesion in acute stroke: pilot study. Eur J Nucl Med Mol Imaging. 214;41(4): Markus R, Donnan GA, Kazui S, et al. Statistical parametric mapping of hypoxic tissue identified by [(18)F]fluoromisonidazole and positron emission tomography following acute ischemic stroke. Neuroimage. 22;16(2): Markus R, Reutens DC, Kazui S, et al. Hypoxic tissue in ischaemic stroke: persistence and clinical consequences of spontaneous survival. Brain. 24;127(pt 6): Sarrafzadeh AS, Nagel A, Czabanka M, Denecke T, Vajkoczy P, Plotkin M. Imaging of hypoxic-ischemic penumbra with (18)F-fluoromisonidazole PET/CT and measurement of related cerebral metabolism in aneurysmal subarachnoid hemorrhage. J Cereb Blood Flow Metab. 21;3(1): Marshall LF, Marshall SB, Klauber MR, et al. The diagnosis of head injury requires a classification based on computed axial tomography. J Neurotrauma. 1992;9(suppl 1):S287-S Brain Trauma Foundation; American Association of Neurological Surgeons. Congress of Neurological Surgeons. Guidelines for the management of severe traumatic brain injury. J Neurotrauma. 27;24(suppl 1):S1-S World Medical Association. World Medical Association Declaration of Helsinki: ethical principles for medical research involving human subjects. JAMA. 213;31(2): doi:1.11/jama Hong YT, Fryer TD. Kinetic modelling using basis functions derived from two-tissue compartmental models with a plasma input function: general principle and application to [18F]fluorodeoxyglucose positron emission tomography. Neuroimage. 21;51(1): Hong YT, Beech JS, Smith R, Baron JC, Fryer TD. Parametric mapping of [18F]fluoromisonidazole positron emission tomography using basis functions. J Cereb Blood Flow Metab. 211;31(2): Smielewski P, Coles JP, Fryer TD, Minhas PS, Menon DK, Pickard JD. Integrated image analysis jamaneurology.com (Reprinted) JAMA Neurology May 216 Volume 73, Number 5 549

9 Research Original Investigation Cerebral Ischemia and Diffusion Hypoxia in Traumatic Brain Injury solutions for PET datasets in damaged brain. J Clin Monit Comput. 22;17(7-8): Smith SM. Fast robust automated brain extraction. Hum Brain Mapp. 22;17(3): Newcombe VF, Williams GB, Outtrim JG, et al. Microstructural basis of contusion expansion in traumatic brain injury: insights from diffusion tensor imaging. J Cereb Blood Flow Metab. 213;33 (6): Coles JP, Fryer TD, Smielewski P, et al. Defining ischemic burden after traumatic brain injury using 15 O PET imaging of cerebral physiology. J Cereb Blood Flow Metab. 24;24(2): Peeters SG, Zegers CM, Lieuwes NG, et al. A comparative study of the hypoxia PET tracers [ 18 F]HX4, [ 18 F]FAZA, and [ 18 F]FMISO in a preclinical tumor model. Int J Radiat Oncol Biol Phys. 215;91 (2): Moustafa RR, Baron JC. Clinical review: imaging in ischaemic stroke implications for acute management. Crit Care. 27;11(5): Patlak CS, Blasberg RG, Fenstermacher JD. Graphical evaluation of blood-to-brain transfer constants from multiple-time uptake data. J Cereb Blood Flow Metab. 1983;3(1): Coles JP, Steiner LA, Johnston AJ, et al. Does induced hypertension reduce cerebral ischaemia within the traumatized human brain? Brain. 24; 127(pt 11): Wu HM, Huang SC, Vespa P, Hovda DA, Bergsneider M. Redefining the pericontusional penumbra following traumatic brain injury: evidence of deteriorating metabolic derangements based on positron emission tomography. J Neurotrauma. 213;3(5): Bartlett RM, Beattie BJ, Naryanan M, et al. Image-guided PO 2 probe measurements correlated with parametric images derived from 18F-fluoromisonidazole small-animal PET data in rats. J Nucl Med. 212;53(1): Read SJ, Hirano T, Abbott DF, et al. Identifying hypoxic tissue after acute ischemic stroke using PET and 18F-fluoromisonidazole. Neurology. 1998; 51(6): Veenith TV, Carter EL, Grossac J, et al. Use of diffusion tensor imaging to assess the impact of normobaric hyperoxia within at-risk pericontusional tissue after traumatic brain injury. J Cereb Blood Flow Metab. 214;34(1): Markus R, Reutens DC, Kazui S, et al. Topography and temporal evolution of hypoxic viable tissue identified by 18F-fluoromisonidazole positron emission tomography in humans after ischemic stroke. Stroke. 23;34(11): Moen KG, Skandsen T, Folvik M, et al. A longitudinal MRI study of traumatic axonal injury in patients with moderate and severe traumatic brain injury. J Neurol Neurosurg Psychiatry. 212;83 (12): Gross MW, Karbach U, Groebe K, Franko AJ, Mueller-Klieser W. Calibration of misonidazole labeling by simultaneous measurement of oxygen tension and labeling density in multicellular spheroids. Int J Cancer. 1995;61(4): Pennings FA, Schuurman PR, van den Munckhof P, Bouma GJ. Brain tissue oxygen pressure monitoring in awake patients during functional neurosurgery: the assessment of normal values. J Neurotrauma. 28;25(1): Meixensberger J, Dings J, Kuhnigk H, Roosen K. Studies of tissue PO 2 in normal and pathological human brain cortex. Acta Neurochir Suppl (Wien). 1993;59: Cleveland WS, Devlin SJ. Locally weighted regression: an approach to regression analysis by local fitting. J Am Stat Assoc. 1988;83: Alian AA, Rafferty T. The best fit function for the tee short axis left ventricular ejection fraction and radionuclear gold standard relationship is curvilinear. J Clin Monit Comput. 28;22(3): Valadka AB, Gopinath SP, Contant CF, Uzura M, Robertson CS. Relationship of brain tissue PO 2 to outcome after severe head injury. Crit Care Med. 1998;26(9): Bouma GJ, Muizelaar JP, Choi SC, Newlon PG, Young HF. Cerebral circulation and metabolism after severe traumatic brain injury: the elusive role of ischemia. J Neurosurg. 1991;75(5): Schwarzmaier SM, Kim SW, Trabold R, Plesnila N. Temporal profile of thrombogenesis in the cerebral microcirculation after traumatic brain injury in mice. J Neurotrauma. 21;27(1): JAMA Neurology May 216 Volume 73, Number 5 (Reprinted) jamaneurology.com