Amyloid Imaging With Carbon 11 Labeled Pittsburgh Compound B for Traumatic Brain Injury

Transcription

1 Research Original Investigation Amyloid Imaging With Carbon 11 Labeled Pittsburgh Compound B for Traumatic Brain Injury Young T. Hong, PhD; Tonny Veenith, FRCA; Deborah Dewar, PhD; Joanne G. Outtrim, MSc; Vaithianadan Mani, FRCA; Claire Williams, FRCA; Sally Pimlott, PhD; Peter J. A. Hutchinson, FRCS, PhD; Adriana Tavares, PhD; Roberto Canales, PhD; Chester A. Mathis, PhD; William E. Klunk, MD, PhD; Franklin I. Aigbirhio, DPhil; Jonathan P. Coles, FRCA, PhD; Jean-Claude Baron, ScD, FMedSci; John D. Pickard, FMedSci; Tim D. Fryer, PhD; William Stewart, FRCPath, PhD; David K. Menon, PhD, FMedSci OBJECTIVES To image amyloid deposition in patients with traumatic brain injury (TBI) using carbon 11 labeled Pittsburgh Compound B ([ 11 C]PiB) positron emission tomography (PET) and to validate these findings using tritium-labeled PiB ([ 3 H]PiB) autoradiography and immunocytochemistry in autopsy-acquired tissue. CME Quiz at jamanetworkcme.com and CME Questions page 124 Supplemental content at jamaneurology.com DESIGN, SETTING, AND PARTICIPANTS In vivo PET at tertiary neuroscience referral center and ex vivo immunocytochemistry of autopsy-acquired brain tissue from a neuropathology archive. [ 11 C]PiB PET was used to image amyloid deposition in 11 controls (median [range] age, 35 [24-60] years) and in 15 patients (median [range] age, 33 [21-50] years) between 1 and 361 days after a TBI. [ 3 H]PiB autoradiography and immunocytochemistry for β-amyloid (Aβ) and β-amyloid precursor protein in brain tissue were obtained from separate cohorts of 16 patients (median [range] age, 46 [21-70] years) who died between 3 hours and 56 days after a TBI and 7 controls (median [range] age, 61 [29-71] years) who died of other causes. MAIN OUTCOMES AND MEASURES We quantified the [ 11 C]PiB distribution volume ratio and standardized uptake value ratio in PET images. The distribution volume ratio and the standardized uptake value ratio were measured in cortical gray matter, white matter, and multiple cortical and white matter regions of interest, as well as in striatal and thalamic regions of interest. We examined [ 3 H]PiB binding and Aβ and β-amyloid precursor protein immunocytochemistry in autopsy-acquired brain tissue. RESULTS Compared with the controls, the patients with TBI showed significantly increased [ 11 C]PiB distribution volume ratios in cortical gray matter and the striatum (corrected P <.05 for both), but not in the thalamus or white matter. Increases in [ 11 C]PiB distribution volume ratios in patients with TBI were seen across most cortical subregions, were replicated using comparisons of standardized uptake value ratios, and could not be accounted for by methodological confounders. Autoradiography revealed [ 3 H]PiB binding in neocortical gray matter, in regions where amyloid deposition was demonstrated by immunocytochemistry; white matter showed Aβ and β-amyloid precursor protein by immunocytochemistry, but no [ 3 H]PiB binding. No plaque-associated amyloid immunoreactivity or [ 3 H]PiB binding was seen in cerebellar gray matter in autopsy-acquired tissue from either controls or patients with TBI, although 1 sample of cerebellar tissue from a patient with TBI showed amyloid angiopathy in meningeal vessels. CONCLUSIONS AND RELEVANCE [ 11 C]PiB shows increased binding following TBI. The specificity of this binding is supported by neocortical [ 3 H]PiB binding in regions of amyloid deposition in the postmortem tissue of patients with TBI. [ 11 C]PiB PET could be valuable in imaging amyloid deposition following TBI. JAMA Neurol. 2014;71(1): doi: /jamaneurol Published online November 11, Author Affiliations: Author affiliations are listed at the end of this article. Corresponding Author: David K. Menon, PhD, FMedSci, Division of Anaesthesia, University of Cambridge, Box 93, Addenbrooke s Hospital, Cambridge CB2 2QQ, England (dkm13@wbic.cam.ac.uk). 23

2 Research Original Investigation Amyloid Imaging With [ 11 C]PiB for TBI There is increasing acceptance of epidemiological and pathophysiological links between traumatic brain injury (TBI) and Alzheimer disease (AD). 1-3 β-amyloid (Aβ) plaques, a hallmark of AD, are found in up to 30% of patients who die in the acute phase following TBI 4-7 ; they appear within hours of injury and are present in all age groups. In contrast, in individuals dying of nonneurological causes, Aβ plaques tend to be confined to elderly individuals. 5 At autopsy, Aβ plaques in patients with TBI are predominantly found in gray matter but have also been reported in white matter. 8 The dominant Aβ species found in TBIassociated plaques and oligomers is Aβ42, 7,9,10 which is the aggregation-prone species also found in AD. Autopsy studies conducted at varying intervals after TBI suggest that these deposits are cleared over a period of weeks following injury. 11 Recent postmortem evidence suggests that, following TBI, the Aβ deposition associated with normal aging may be subsequently accelerated, 12 but our inability to quantify amyloid binding in vivo limits a broader understanding of the temporal profile and outcome of amyloid deposition in TBI. Positron emission tomography (PET) may provide one solution to this problem. Several carbon 11 labeled and fluorine 18 labeled PET ligands for amyloid imaging have been developed and used in AD. 13 The most widely studied of these, Pittsburgh compound B (PiB), 14 is widely validated as a marker of cerebral amyloid deposition. In individuals with AD, and in the general population, the relative distribution of Aβ as measured by carbon 11 labeled PiB ([ 11 C]PiB) PET correlates well with neuropathology, with a predilection for high frontal, temporoparietal, and striatal binding and relatively low but still significant mesial temporal binding. 13 However, documented amyloid deposition may be unassociatedwith[ 11 C]PiB binding seen on PET scans, or it may have distinct patterns of binding across the brain, depending on the genetic abnormality predisposing to amyloid deposition, 18 suggesting molecular heterogeneity in amyloid species in relation to [ 11 C]PiB binding. Therefore, before using serial [ 11 C]PiB PET to examine amyloid deposition in patients with TBI, we need to be certain of the correspondence between [ 11 C]PiB PET imaging and pathology. Furthermore, we have no idea of the time points following injury when changes in amyloid deposition might be best detected. Given these considerations, we have undertaken a cross-sectional pilot study of [ 11 C]PiB PET for patients with moderate to severe TBI, imaged at a range of times after TBI. To validate our imaging findings, we also undertook a comparison of in vitro tritiumlabeled PiB ([ 3 H]PiB) binding against immunocytochemistry for Aβ in autopsy-acquired brain tissue obtained from a separate cohort of patients. Methods This is a cross-sectional study of a convenience sample of patients with TBI and age-matched controls, using [ 11 C]PiB PET. Patients with TBI were recruited up to a maximum of 1 year after injury. Validation of the PET findings was undertaken in postmortem tissue from a separate series of individuals, obtained from a neuropathology archive. Cohorts [ 11 C]PiB PET was performed for 15 participants (median [range] age, 33 [21-50] years) who had a moderate to severe TBI (median [range] Glasgow Coma Scale score, 8 [3-12]) (Table 1). Participants were selected based on a study plan that aimed to recruit patients at different times (up to 1 year) following TBI. Seven of these patients were still receiving critical care at the time of imaging, and they were selected as they were sufficiently stable to undergo PET imaging at times when access to the PET facility was available. The imaging characteristics of injury were assessed at admission by computed tomography using the Marshall score, 19 and outcome was defined using the Glasgow Outcome Scale. 20 All studies were undertaken following informed consent from participants or from a patient representative if the patient did not have the capacity to provide consent. [ 11 C]PiB PET was also performed for 11 healthy controls (median [range] age, 35 [24-60] years), none of whom had any symptoms, signs, or diagnoses of neurological disease or showed any abnormalities on magnetic resonance imaging (MRI) scans of the brain, and all of whom had a normal score on the Mini-Mental State Examination. [ 11 C]PiB PET studies were approved by a regional research ethics committee (National Research Ethics Service Committee East of England Cambridge Central; Imaging Cerebral Amyloid Deposition Using Positron Emission Tomography; study 05/Q0108/ 350) and by the UK Administration of Radioactive Substances Advisory Committee (reference RPC 83/3387/20183). Autoradiography and immunocytochemistry for validation of PiB binding were performed on autopsy-acquired human brain tissue from the unique Glasgow TBI Archive held at the Southern General Hospital, in Glasgow, Scotland. The material examined was obtained from 16 individuals (median age, 46 [21-70] years) who had sustained a TBI and had died at intervals of 3 hours to 56 days after injury. We also examined brain tissue from 7 controls (median age, 61 [29-71] years) with nonneurological cause of death. Use of this material was approved by the appropriate research ethics committee (West of Scotland Research Ethics Service; reference 10/ S0709/58). In each case, matched coronal sections from the level of the lateral geniculate nucleus (including the corpus callosum and adjacent parasagittal white matter, the thalamus and adjacent internal capsule, the hippocampus and the medial temporal lobe, and the insular cortex and adjacent subcortical white matter) were selected for analysis, together with sections of the cerebellar cortex. [ 11 C]PiB PET [ 11 C]PiB PET data were acquired as described in a previous study 21 (eappendix and efigures 1 and 2 in Supplement). For 11 participants (including 4 controls), we obtained arterial blood samples to provide a metabolite-corrected plasma input function for kinetic analysis. To facilitate segmentation into tissue classes and delineation of anatomical regions of interest (ROIs), participants also underwent an anatomical MRI scan with the 3-T Siemens Tim-Trio (Siemens Medical Solutions). 24 JAMA Neurology January 2014 Volume 71, Number 1 jamaneurology.com

3 Amyloid Imaging With [ 11 C]PiB for TBI Original Investigation Research Table 1. Clinical Characteristics of the Participants Who Underwent [ 11 C]PiB PET Studies a Participant No. Patients with TBI Sex Age at PET, y Age at TBI, y Mechanism of Injury Preintubation GCS Score Marshall Score, First CT Scan Lesions b Midline Shift, mm 1 M RTA (car) 3 EML SDH, contusions 2 M Assault 7 EML EDH, contusions 3 F RTA 4 M RTA DAI/TAI on MRI c Injury to PET Interval, d Glasgow Outcome Scale 10 DAI 339 Good recovery 8 DAI 10 Good recovery 3 DI II Contusion 0 DAI 11 Severe disability 7 DI II Contusion 0 DAI 14 Good recovery 5 M RTA (car) 8 DI I Contusion 0 DAI 2 Good recovery 6 M RTA 3 DI II Contusion 0 DAI 54 Severe disability 7 M RTA (cyclist) 8 DI I None 0 DAI 7 Moderate disability 8 M RTA (pedestrian) 7 EML SDH, contusion Severe disability 9 M RTA (car) 7 DI I None 0 DAI 10 Good recovery 10 M RTA 11 M RTA 8 DI I None 0 TAI 2 Moderate disability 5 DI I None 0 7 Vegetative state 12 M RTA (car) 10 DI I None 0 DAI 3 Good recovery 13 M Fall 10 NEML Bilateral contusions Good recovery 14 M RTA (cyclist) 5 DI I None 0 DAI 71 Good recovery 15 M RTA 5 NEML Contusions 0 DAI 109 Vegetative state Abbreviations: [ 11 C]PiB PET, carbon 11 labeled Pittsburgh Compound B positron emission tomography; CT, computed tomography; DAI, diffuse axonal injury (defined as 4 foci of TAI, in conformance with National Institute of Neurological Disorders and Stroke Common Data Elements terminology); DI, diffuse injury; EDH, extradural hematoma; EML, evacuated mass lesion; GCS, Glasgow Coma Scale; MRI, magnetic resonance imaging; NEML, nonevacuated mass lesion; RTA, road traffic accident; SDH, subdural hematoma; TAI, traumatic axonal injury; TBI, traumatic brain injury. a [ 11 C]PiB PET was used to image amyloid deposition in 11 controls (median [range] age, 35 [24-60] years [only 1 female control]). b Only some with significant mass effect. c The MRI scan was the first available, but in some instances this was the one that accompanied the PET study and was hence weeks after the original injury. Images were acquired using a 3-dimensional magnetizationprepared rapid gradient-echo sequence (repetition time/ echo time/inversion time = 2300 milliseconds/2.98 milliseconds/900 milliseconds; flip angle, 9 ; 1 average; 176 slices; matrix size; mmvoxelsize). Statistical Parametric Mapping version 8 (SPM8; http: // was used to align the PET images, coregister a mean-aligned PET image to the MR image from the same participant, and segment the MR images. The gray and white matter probability maps produced by SPM8 segmentation were smoothed with an isotropic gaussian function to approximate the PET resolution 5 cm off-axis (6.8-mm full-width at half maximum [FWHM]). An ROI was delineated in the superior gray matter of the cerebellum using a 90% threshold on the smoothed gray matter probability map and was applied to the coregistered PET images to provide a reference tissue time-activity curve. The regional and voxelwise distribution volume ratios (DVRs) were determined with the reference tissue input Logan graphical method 22 (fitting time >40 minutes). The regional standardized uptake value ratios (SUVR) were also determined from data acquired 50 to 70 minutes after injection, normalized by the cerebellum ROI signal. The DVRs and the SUVRs were quantified in 4 ROIs delineated using the smoothed probability maps: white matter and 3 gray matter ROIs (cortical gray matter, the striatum [caudate nucleus and putamen], and the thalamus) (eappendix and efigure 1 in Supplement). The cortical gray matter ROI approximated a previous quantification technique used for [ 11 C]PiB PET in AD by Morris et al, 23 which quantified the mean nondisplaceable binding potential (BP ND ) in cortical ROIs (the prefrontal cortex, gyrus rectus, lateral temporal cortex, and precuneus). 24 However, we did not wish to make any a priori assumption about where binding would be seen in TBI, and hence defined an ROI that included all cortical gray matter. To verify this approach and explore the regional distribution of [ 11 C]PiB binding, we also quantified the [ 11 C]PiB DVR and SUVR in multiple cortical and white matter ROIs (eappendix and etable in the Supplement). To reduce partial volume error, especially between gray and white matter signals, 5 iterations of the Lucy-Richardson deconvolution algorithm 25,26 with a 6.8-mm FWHM gaussian point-spread function were applied to the emission data prior to estimation of the DVR and the SUVR (eappendix and efigure 1 in Supplement). jamaneurology.com JAMA Neurology January 2014 Volume 71, Number 1 25

4 Research Original Investigation Amyloid Imaging With [ 11 C]PiB for TBI Figure 1. Carbon 11 Labeled Pittsburgh Compound B ([ 11 C]PiB) Distribution Volume Ratio (DVR) Maps From a Control and 3 Patients With Traumatic Brain Injury (TBI) Healthy control [ 11 C]PiB DVR hours after TBI Autoradiography and Immunocytochemistry Acute TBI cases and noninjured controls were identified within the Glasgow TBI Archive. Details on tissue collection and processing were as described previously. 27 Immunostaining for Aβ and β-amyloid precursor protein (βapp) was performed with a monoclonal antibody directed to the N-terminal epitope amino acid residues Aβ/ 8 17 (clone 6F/3D; Dako; 1:75) and a monoclonal antibody specific to APP (clone 22C11; Millipore; 1:50 000). For autoradiography, slides from sections adjacent to those used for immunocytochemistry were incubated in phosphate-buffered saline containing 10% ethanol and 0.5nM N-methyl-[ 3 H]2-(4 -methylaminophenyl)-6-hydroxybenzothiazole ([ 3 H]PiB; PerkinElmer LAS). For both immunocytochemistry and autoradiography, plaque density was assessed in sections or film images by an investigator blind to whether the brain tissue was from a patient with TBI or a control. Furthermore, in the immunocytochemistry-stained sections, plaques were assessed as either fibrillary or diffuse in nature. The extent and distribution of axonal pathology were assessed in the APP-stained sections, again by an investigator blind to whether the brain tissue was from a patient with TBI or a control, and its nature as either vascular axonal injury (arising as a consequence of raised intracranial pressure) or diffuse traumatic axonal injury recorded, with the extent and distribution of Aβ-positive axonal profiles recorded in adjacent sections. Finally, corresponding autoradiographic images were reviewed for evidence of white matter amyloid binding. The tissue blocks available primarily provided access to cerebral and cerebellar gray and white matter. We do not have data on immunocytochemistry or [ 3 H]PiB binding in the striatum. Transverse, coronal, and sagittal 54 days after TBI sections of DVR maps from a healthy control and 3 patients at varying time intervals after TBI, overlaid on coregistered T1-weighted magnetic resonance imaging scans. All maps have the same intensity range (DVRs between 1.00 and 1.30) to aid 339 days after TBI comparability. The interval between TBI and imaging is shown on the right. Statistical Analysis No sample size estimates were obtained prior to this exploratory study because no pilot data were available. Statistical analysis was performed using Statview version 5 (SAS Institute). Data are expressed as medians and ranges, unless otherwise stated. Measures of PiB binding were compared using the Mann-Whitney U test, with P <.05 after correction for multiple comparisons considered statistically significant for definitive analyses. Comparisons of [ 11 C]PiB binding in cortical and white matter subregions (eappendix and etable in the Supplement) were primarily undertaken to explore the regional distribution of [ 11 C]PiB binding; these exploratory analyses are reported with uncorrected P values. Results The characteristics of the participants who underwent [ 11 C]PiB PET are listed in Table 1. There were no significant differences in age at PET imaging between patients with TBI and controls (P =.29). Plasma-based kinetic analysis indicated that the validity of reference tissue modeling was not affected by TBI (eappendix and efigure 2 in the Supplement). Controls showed relatively low [ 11 C]PiB binding, predominantly in the central white matter and deep gray matter structures (Figure 1). Patients with TBI showed greater binding in both these areas (Figure 1), but there was substantial spatial variation and temporal variation among participants. Lesional and perilesional regions in and around contusions (n = 9) or underlying subdural hematomas showed very little excessive [ 11 C]PIB binding com- 26 JAMA Neurology January 2014 Volume 71, Number 1 jamaneurology.com

5 Amyloid Imaging With [ 11 C]PiB for TBI Original Investigation Research Figure 2. Box and Whisker Plots of Regional Carbon 11 Labeled Pittsburgh Compound B Distribution Volume Ratios (DVRs) for Controls and Patients With Traumatic Brain Injury (TBI) DVR DVR DVR DVR Controls Patients With TBI Gray Matter 1.0 Controls Patients With TBI White Matter 1.00 Controls Patients With TBI Thalamus 0.85 Controls Patients With TBI Striatum Plots of DVRs in cortical gray matter, white matter, striatal, and thalamic regions of interest from controls and patients with TBI patients are shown. The horizontal line inside each box represents the 50th percentile; the top and bottom borders of each box represent the 75th and 25th percentiles, respectively; the whiskers above and below each box represent the 90th and 10th percentiles, respectively; and the circles beyond the whiskers represent outliers. Significant differences between the 2 groups were found for cortical gray matter (P =.009[P =.04 when corrected for multiple comparisons]) and the striatum (P =.01[P =.048 when corrected for multiple comparisons]). pared with normal tissue (efigure 3 in Supplement). In particular, regions that showed a high signal on fluid-attenuated inversion recovery and T2-weighted MRI scans and were also sites of maximal vasogenic edema showed no particular increase in the [ 11 C]PiB DVR. Compartmental modeling with metabolite-corrected plasma input, available for 6 patients with TBI, imaged at early time points after injury, and 4 controls, showed no difference in the kinetic parameters characterizing transport of the tracer across the blood-brain barrier between the 2 groups (eappendix in the Supplement). The [ 11 C]PiB DVRs in the patients with TBI compared with the controls (Figure 2) were significantly higher in cortical gray matter (P =.009 [P =.04 after correction]) and the striatum (P =.01[P =.048 after correction]), but not in white matter (P =.29) or the thalamus (P =.59). The data suggest temporal variations in binding (efigure 4 in Supplement), but our sample size and numbers at later time points were too small to draw robust conclusions. Analysis using the [ 11 C]PiB SUVRs resulted in identical results, with significantly higher SUVRs in cortical gray matter and the striatum in the patients with TBI (P <.01 and <.05, respectively, corrected for multiple comparisons) but not in the thalamus or white matter (P =.22 and.70, respectively; etable in Supplement). Analysis of the [ 11 C]PiB DVRs in cortical subregions showed significantly higher values for patients with TBI than for controls (uncorrected P <.05; etable in Supplement) in all regions except the medial temporal/hippocampal ROI. The [ 11 C]PiB DVRs in ROIs defined within white matter were not significantly different between patients with TBI and controls. The regional distribution of differences in [ 11 C]PIB SUVRs was similar to that of [ 11 C]PIB DVRs (etable in Supplement). Data from immunocytochemistry and [ 3 H]PiB binding in autopsy-acquired sections from controls and patients with TBI are shown in Table 2. No plaque-associated [ 3 H]PiB binding or Aβ immunoreactivity was detected in any of the sections of the cerebellar cortex from controls or from patients with TBI (Figure 3). However, cerebellar tissue from a 61-year-old woman who survived 12 hours after a severe TBI showed amyloid angiopathy in meningeal vessels, both by immunocytochemistry and autoradiography (Figure 3). The sections from patients with TBI had more frequent and denser plaques in neocortical gray matter than did the sections from controls, both by immunocytochemistry and [ 3 H]PiB binding, with immunocytochemistry being the more sensitive of the 2 techniques. In participants who showed [ 3 H]PiB binding to neocortical gray matter, there was good correspondence between the 2 techniques in identifying regions with Aβ plaques. However, although fibrillar and diffuse plaques were identified using both techniques, [ 3 H]PiB binding was greatest in participants with fibrillar plaque and was less evident or absent in participants with only diffuse plaque (Figure 4). [ 3 H]PiB binding was seen in 4 of the 6 patients in whom cortical plaque was identified by immunocytochemistry, and no instances of falsepositive [ 3 H]PiB binding were seen in regions that did not show plaque by immunocytochemistry. Axonal Aβ and βapp were detected by immunocytochemistry exclusively in patients with TBI, where they were found in 87.5% and 100% of participants, respectively. However, we were unable to demonstrate [ 3 H]PiB binding in white matter regions that showed substantial and extensive axonal Aβ and βapp immunohistochemistry. The tissue blocks that were available for analysis did not allow us to explore the immunocytochemistry or [ 3 H]PiB binding in the striatum. Discussion We demonstrate, for the first time, that [ 11 C]PiB shows increased binding in cortical gray matter and the striatum fol- jamaneurology.com JAMA Neurology January 2014 Volume 71, Number 1 27

6 Research Original Investigation Amyloid Imaging With [ 11 C]PiB for TBI lowing TBI. Correspondence between immunocytochemistry and [ 3 H]PiB binding in separate samples of autopsyacquired tissues supports the inference that [ 11 C]PiB binding Table 2. Data on Participants From the Glasgow TBI Archive Whose Brains Were Used for ICC and [ 3 H]PiB Autoradiographic Studies Participants, No. % Characteristic Controls (n=7) Patients With TBI (n = 16) Age, median (range), y 61 (29-71) 46 (21-70) Post-TBI survival, range, d 3-56 Neocortical plaque detected by 2 (29) 6 (38) ICC Neocortical plaque detected by 1 (14) 4 (25) [ 3 H]PiB binding Axonal Aβ detected by ICC 0 (0) 14 (88) Axonal βapp detected by ICC 0 (0) 16 (100) White matter detected by [ 3 H]PiB 0(0) 0(0) binding Cerebellar axonal βapp detected 0 (0) 11 (69) by ICC Cerebellar plaque detected by ICC 0 (0) 0 (0) Cerebellar plaque detected by 0(0) 0(0) [ 3 H]PiB binding Cerebellar meningeal amyloid angiopathy 0(0) 1(6) Abbreviations: Aβ, β-amyloid; βapp, β-amyloid precursor protein; [ 3 H]PiB, hydrogen 3 labeled Pittsburgh Compound B; ICC, immunocytochemistry; TBI, traumatic brain injury. is a marker of Aβ plaque in gray matter with regard to TBI. Tissue binding of [ 3 H]PiB was observed in neocortical regions in 4 of the 6 tissue samples in which plaque was identified by immunocytochemistry, suggesting that PiB binding is moderately sensitive at detecting Aβ plaque. However, there were no instances of false-positive [ 3 H]PiB binding in tissue in which no plaque was present, suggesting that PiB binding is a relatively specific marker of cortical Aβ plaque. Although individual patients showed increases in [ 11 C]PiB binding in white matter, we were unable to show significant group differences between patients and controls. Immunocytochemical examination of autopsy-acquired brain tissue from patients with TBI showed that these white matter regions exhibit substantial amounts of Aβ staining, but we were unable to directly correlate this with [ 3 H]PiB binding in the same regions. We could not demonstrate any significant increase in thalamic [ 11 C]PiB binding in vivo by PET imaging or in thalamic [ 3 H]PiB binding in postmortem tissue. We could only find one prior study 28 that examined amyloid deposition in cerebellar tissue in patients with TBI. This study included 14 cases and found cerebellar amyloid in just 2 samples, for which it was present as diffuse deposits rather than as the fibrillar plaque that was associated with [ 3 H]PiB binding in our autoradiographic studies. In our series of 16 patients with TBI and 7 controls whose tissue samples were examined postmortem, no parenchymal Aβ plaque in cerebellar tissue was found. Furthermore, we were unable to show Figure 3. Images of β-amyloid (Aβ) in Cerebellar Tissue From Tritium-Labeled Pittsburgh Compound B ([ 3 H]PiB) Autoradiography and Immunocytochemistry A B C D E F [ 3 H]PiB autoradiography and immunocytochemistry were performed for 2 patients who died 30 hours (A-C) and 12 hours (D-F) after a traumatic brain injury (TBI). The 2 red boxes shown in the low magnification regions (B and E, respectively) were selected for higher magnification (C and F, respectively). [ 3 H]PiB binding is shown for the patient who died 12 hours after a TBI; this binding corresponds to vascular amyloid in meningeal vessels (D [arrowheads]), which are revealed by immunocytochemistry (E [original magnification 3] and F [original magnification 50]) and best seen at high magnification (F [arrowheads]). The neocortical tissue of the patient who died 30 hours after a TBI (A-C) is also shown in Figure 4A-C, where the neocortical tissue shows clear amyloid plaque and [ 3 H]PiB binding, whereas the cerebellar tissue does not. 28 JAMA Neurology January 2014 Volume 71, Number 1 jamaneurology.com

7 Amyloid Imaging With [ 11 C]PiB for TBI Original Investigation Research Figure 4. Images of β-amyloid (Aβ) in Adjacent Tissue Sections From the Parasagittal Cortex From Tritium-Labeled Pittsburgh Compound B ([ 3 H]PiB) Autoradiography and Immunocytochemistry A B C D E F [ 3 H]PiB autoradiography and immunocytochemistry were performed for 2 patients who died 30 hours (A-C) and 5 days (D-F) after a traumatic brain injury (TBI). Note the regional correspondence in fibrillary Aβ plaque detection with the 2 techniques. However, the predominantly diffuse plaque in the patient who died 30 hours after a TBI is more evident in the stained sections (B and C). The higher magnification fields (C and F [scale bars, 100 μm]) correspond to the boxes (B and E, respectively [scale bars, 1 mm]) in the low magnification regions. plaque-related [ 3 H]PiB binding in cerebellar tissue in any of the participants. However, amyloid deposition in cerebellar tissue was found in 1 patient with TBI (6% of the TBI tissue examined), in the form of amyloid angiopathy. Although these data suggest that [ 11 C]PiB binding in the cerebellum is uncommon, it remains possible that amyloid angiopathy may have contributed to cerebellar cortical [ 11 C]PiB binding in PET studies. However, it is important to note that the presence of such binding in the cerebellar reference region used for [ 11 C]PiB PET would lead to a reduction in the calculated DVR and SUVR in target ROIs (eg, cortical gray matter and the striatum), rather than the increase we observed in patients with TBI. Together with a range of analyses (eappendix and efigure 2 in the Supplement), this provided reassurance that our demonstration of increased [ 11 C]PiB PET binding in cortical regions and the striatum was unlikely to be a methodological artifact. The cortical [ 11 C]PiB binding that we demonstrate was quantified using techniques similar to those used in the AD literature. For example, Morris et al 23 used the mean cortical binding potential from 4 regions (the prefrontal cortex, the gyrus rectus, the lateral temporal region, and the precuneus) to differentiate patients with AD from controls, based on knowledge of brain areas that show a predilection for Aβ deposition in AD, 29,30 with a mean cortical binding potential threshold value of 0.18 (equivalent to a DVR of 1.18 [DVR = BP ND + 1]) to define abnormality. 23 Although we used a wider range of cortical regions (all of the cortex), we were able to demonstrate a significant difference in cortical [ 11 C]PiB binding between patients and controls. Although intercenter comparisons are imperfect, it is worth noting that some of the cortical [ 11 C]PiB DVRs that we found in early TBI were well within the spread of corresponding BP ND values reported for amnestic mild cognitive impairment. 30 Indeed, in one of our patients (Figure 1 [second row]), the cortical [ 11 C]PiB DVR widely exceeded corresponding BP ND levels suggested as diagnostic thresholds for AD, 23 which have been reported to show high sensitivity and specificity for the diagnosis of AD. We showed that there was no significant increase in [ 11 C]PiB binding in white matter in the patients with TBI as a group. However, visual inspection of images suggested increased [ 11 C]PiB binding in the white matter of individual patients (particularly early after TBI; Figure 1 and efigure 4 in Supplement). Binding of [ 11 C]PiB in white matter in AD is thought to be nonspecific 31 because of negative in vitro binding assays and because Aβ plaques are not found in white matter in AD. One study 32 has suggested that binding of PiB in white matter is to a separate myelin epitope and that such binding might be used to detect demyelination. However, in our patients, the binding of [ 11 C]PIB occurred in regions of white matter that are known to show abundant axonal Aβ deposition following TBI and to contain damaged axon terminals that are known to bind thioflavin (which is closely chemically related to PiB). 33 Finally, the binding of PiBtomyelinshownbyStankoffetal 32 was decreased, rather than increased, in the presence of white matter inflammation and injury in the setting of multiple sclerosis, suggesting that jamaneurology.com JAMA Neurology January 2014 Volume 71, Number 1 29

8 Research Original Investigation Amyloid Imaging With [ 11 C]PiB for TBI white matter inflammation and injury, on their own, would not readily account for the increased binding that we demonstrate in our patients. Given these considerations and our relatively small sample size, we cannot exclude the possibility that a larger PET study, or one focused on early TBI, might demonstrate significant increases in white matter binding of [ 11 C]PiB. However, even if subsequent imaging studies do demonstrate increased binding of the ligand in white matter following TBI, a further investigation is needed to determine why we were unable to replicate this with [ 3 H]PiB binding. The early binding of [ 11 C]PiB in the striatum in our patients broadly replicates the atypical early binding demonstrated in patients who carry mutations in the presenelin-1 (PS1) gene. 34 As a proposed mechanism of amyloid deposition, 35 such PS1 mutations, and other genetically driven early-onset forms of AD, may all overproduce Aβ (particularly Aβ42), in contrast to typical late-onset AD, for which diminished clearance may play a larger part. 36 The topography of [ 11 C]PiB binding observed in our study would therefore be concordant with currently proposed mechanisms of overproduction leading to amyloid deposition in TBI. 3 Our study has significant limitations, the most important of which are its small sample size and the absence of serial imaging. Perhaps most importantly, although our imaging data are entirely concordant with previous postmortem studies that have examined amyloid deposition in TBI, the small proportion of late studies in our 15-patient cohort limits inferences regarding the temporal pattern of [ 11 C]PiB binding in vivo and requires confirmation in a larger cohort. By performing kinetic analysis with a metabolite-corrected plasma input function for 6 patients with TBI and 4 controls, we explored the possibility that the increased cortical and striatal [ 11 C]PiB binding might have been an artifact of the disruption of the bloodbrain barrier following TBI; however, we found no differences in key kinetic parameters between these groups (eappendix in the Supplement). Three of the patients with the highest cortical gray matter [ 11 C]PiB DVRs had such data available (efigure 4 in Supplement), providing some reassurance that blood-brain barrier abnormalities were unlikely to be an important cause of elevated [ 11 C]PiB DVRs. However, although we found no increase in [ 11 C]PiB DVRs in areas of T2- weighted hyperintensity on MRI scans, we did not perform contrast-enhanced MRI or computed tomography, and hence we have no direct imaging of blood-brain barrier dysfunction in our patients. Future studies will need to more fully address this potential confounder. Finally, the autopsy-derived brain tissue samples that we used to assess [ 3 H]PiB binding were all obtained within the first 70 days after TBI. Tissue-based validation of PiB binding at later time points will require additional data. The use of [ 11 C]PiB PET for amyloid imaging following TBI provides us with the potential for understanding the pathophysiology of TBI, for characterizing the mechanistic drivers of disease progression or suboptimal recovery in the subacute phase of TBI, for identifying patients at high risk of accelerated AD, and for evaluating the potential of antiamyloid therapies. Future studies could use [ 11 C]PiB PET to serially follow the course of amyloid deposition and clearance following TBI, in a way that is self-evidently impossible with postmortem studies, and to perhaps document the reemergence of Aβ decades after TBI. 12 Such an approach would also allow for the analysis of Aβ deposition in TBI in the context of cognitive function and host genotype (particularly APOE), 37 which were not addressed in the present study. [ 11 C]PiB PET could also help us in identifying the proportion of patients with moderate to severe TBI who do show amyloid deposition, and in exploring whether any patients with mild TBI show such deposition because postmortem evidence is unavailable for this latter group. Additional studies are also needed to further explore the white matter binding that we demonstrate in PET studies, and to determine whether this is an artifact or whether it represents a useful marker of white matter injury. ARTICLE INFORMATION Accepted for Publication: August 21, Published Online: November 11, doi: /jamaneurol Author Affiliations: Wolfson Brain Imaging Centre, University of Cambridge, Cambridge, England (Hong, Canales, Aigbirhio, Baron, Pickard, Fryer, Menon); Division of Anaesthesia, University of Cambridge, Addenbrooke s Hospital, Cambridge, England (Veenith, Outtrim, Mani, Williams, Coles, Menon); Institute of Neuroscience and Psychology, University of Glasgow, Scotland (Dewar, Tavares); University of Glasgow and Southern General Hospital, Glasgow, Scotland (Pimlott, Stewart); Department of Clinical Neurosciences, University of Cambridge, Cambridge, England (Hong, Hutchinson, Canales, Aigbirhio, Baron, Pickard, Fryer); Departments of Radiology and Pharmaceutical Sciences, University of Pittsburgh, Pittsburgh, Pennsylvania (Mathis); Departments of Psychiatry and Neurology, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania (Klunk); INSERM U894, Université Paris Descartes, Sorbonne Paris Cité, Paris, France (Baron). Author Contributions: Dr Menon had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Drs Fryer, Stewart, and Menon contributed equally to the study. Study concept and design: Hong, Hutchinson, Aigbirhio, Coles, Baron, Pickard, Fryer, Stewart, Menon. Acquisition of data: Veenith, Dewar, Outtrim, Mani, Williams, Pimlott, Tavares, Canales, Aigbirhio, Coles, Baron, Pickard, Fryer, Stewart, Menon. Analysis and interpretation of data: Hong, Veenith, Dewar, Pimlott, Tavares, Canales, Mathis, Klunk, Fryer, Stewart, Menon. Drafting of the manuscript: Hong, Fryer, Stewart, Menon. Critical revision of the manuscript for important intellectual content: All authors. Statistical analysis: Hong, Menon. Obtained funding: Baron, Pickard, Menon. Administrative, technical, or material support: Dewar, Outtrim, Pimlott, Tavares, Canales, Coles, Menon. Study supervision: Hutchinson, Aigbirhio, Coles, Baron, Pickard, Menon. Conflict of Interest Disclosures: GE Healthcare holds a license agreement with the University of Pittsburgh based on the PiB technology described in this article. Drs Klunk and Mathis are coinventors of PiB and, as such, have a financial interest in this license agreement. No other disclosures were reported. Funding/Support: This study was supported by the Neuroscience Theme of the National Institute for Health Research Cambridge Biomedical Research Centre. Drs Menon and Pickard are supported by National Institute for Health Research Senior Investigator awards. Dr Hutchinson is supported by funding from the Academy of Medical Sciences. Radiochemistry development of [ 11 C]PiB was supported by grant RG46503 from the Medical Research Council (United Kingdom). Data collection in controls was undertaken by funding awarded to Dr Baron by the National Institute for Health Research Cambridge Biomedical Research Centre. Dr Klunk is supported by the following grants: P50 AG005133, R37 AG025516, P01 AG JAMA Neurology January 2014 Volume 71, Number 1 jamaneurology.com

9 Amyloid Imaging With [ 11 C]PiB for TBI Original Investigation Research Role of the Sponsor: GE Healthcare provided no grant support for this study and had no role in the design or interpretation of the results or in the preparation of the manuscript. REFERENCES 1. DeKosky ST, Ikonomovic MD, Gandy S. Traumatic brain injury football, warfare, and long-term effects. N Engl J Med. 2010;363(14): Fleminger S, Oliver DL, Lovestone S, Rabe-Hesketh S, Giora A. Head injury as a risk factor for Alzheimer s disease: the evidence 10 years on; a partial replication. J Neurol Neurosurg Psychiatry. 2003;74(7): Johnson VE, Stewart W, Smith DH. Traumatic brain injury and amyloid-β pathology: a link to Alzheimer s disease? Nat Rev Neurosci. 2010;11(5): Roberts GW, Gentleman SM, Lynch A, Graham DI. β A4 amyloid protein deposition in brain after head trauma. Lancet. 1991;338(8780): Roberts GW, Gentleman SM, Lynch A, Murray L, Landon M, Graham DI. Beta amyloid protein deposition in the brain after severe head injury: implications for the pathogenesis of Alzheimer s disease. J Neurol Neurosurg Psychiatry. 1994;57(4): Ikonomovic MD, Uryu K, Abrahamson EE, et al. Alzheimer s pathology in human temporal cortex surgically excised after severe brain injury. Exp Neurol. 2004;190(1): DeKosky ST, Abrahamson EE, Ciallella JR, et al. Association of increased cortical soluble Aβ 42 levels with diffuse plaques after severe brain injury in humans. Arch Neurol. 2007;64(4): Smith DH, Chen XH, Iwata A, Graham DI. Amyloid β accumulation in axons after traumatic brain injury in humans. J Neurosurg. 2003;98(5): Gentleman SM, Greenberg BD, Savage MJ, et al. A beta 42 is the predominant form of amyloid beta-protein in the brains of short-term survivors of head injury.neuroreport. 1997;8(6): Horsburgh K, Cole GM, Yang F, et al. β-amyloid (Abeta)42(43), abeta42, abeta40 and apoe immunostaining of plaques in fatal head injury. Neuropathol Appl Neurobiol. 2000;26(2): Chen XH, Johnson VE, Uryu K, Trojanowski JQ, Smith DH. A lack of amyloid β plaques despite persistent accumulation of amyloid β in axons of long-term survivors of traumatic brain injury. Brain Pathol. 2009;19(2): Johnson VE, Stewart W, Smith DH. Widespread τ and amyloid-β pathology many years after a single traumatic brain injury in humans. Brain Pathol. 2012;22(2): Herholz K, Ebmeier K. Clinical amyloid imaging in Alzheimer s disease. Lancet Neurol. 2011;10(7): Klunk WE, Engler H, Nordberg A, et al. Imaging brain amyloid in Alzheimer s disease with Pittsburgh Compound-B. Ann Neurol. 2004;55(3): Ikonomovic MD, Abrahamson EE, Price JC, et al. Early AD pathology in a [C-11]PiB-negative case: a PiB-amyloid imaging, biochemical, and immunohistochemical study. Acta Neuropathol. 2012;123(3): Rosen RF, Ciliax BJ, Wingo TS, et al. Deficient high-affinity binding of Pittsburgh compound B in a case of Alzheimer s disease. Acta Neuropathol. 2010;119(2): Cairns NJ, Ikonomovic MD, Benzinger T, et al. Absence of Pittsburgh compound B detection of cerebral amyloid beta in a patient with clinical, cognitive, and cerebrospinal fluid markers of Alzheimer disease: a case report. Arch Neurol. 2009;66(12): Villemagne VL, Ataka S, Mizuno T, et al. High striatal amyloid beta-peptide deposition across different autosomal Alzheimer disease mutation types.arch Neurol. 2009;66(12): Marshall LF, Bowers Marshall S, Klauber MR, et al. A new classification of head injury based on computerized tomography. J Neurosurg. 1991;75:S14-S20. /sup s.0s14. Accessed October 22, Wilson JT, Pettigrew LE, Teasdale GM. Structured interviews for the Glasgow Outcome Scale and the extended Glasgow Outcome Scale: guidelines for their use. J Neurotrauma. 1998;15(8): Landt J, D Abrera JC, Holland AJ, et al. Using positron emission tomography and carbon 11-labeled Pittsburgh Compound B to image brain fibrillar β-amyloid in adults with down syndrome: safety, acceptability, and feasibility. Arch Neurol. 2011;68(7): Logan J, Fowler JS, Volkow ND, Wang GJ, Ding YS, Alexoff DL. Distribution volume ratios without blood sampling from graphical analysis of PET data. J Cereb Blood Flow Metab. 1996;16(5): Morris JC, Roe CM, Xiong C, et al. APOE predicts amyloid-beta but not tau Alzheimer pathology in cognitively normal aging. Ann Neurol. 2010;67(1): Innis RB, Cunningham VJ, Delforge J, et al. Consensus nomenclature for in vivo imaging of reversibly binding radioligands. J Cereb Blood Flow Metab. 2007;27(9): Lucy LB. An iterative technique for the rectification of observed distributions. Astron J. 1974;79: doi: / Richardson WH. Bayesian-based iterative method of image restoration. J Opt Soc Am. 1972;62(1): doi: /josa Dewar D, Graham DI. Depletion of choline acetyltransferase activity but preservation of M1 and M2 muscarinic receptor binding sites in temporal cortex following head injury: a preliminary human postmortem study. J Neurotrauma. 1996;13(4): Graham DI, Gentleman SM, Lynch A, Roberts GW. Distribution of beta-amyloid protein in the brain following severe head injury. Neuropathol Appl Neurobiol. 1995;21(1): Mintun MA, Larossa GN, Sheline YI, et al. [ 11 C]PIB in a nondemented population: potential antecedent marker of Alzheimer disease. Neurology. 2006;67(3): Devanand DP, Mikhno A, Pelton GH, et al. Pittsburgh compound B ( 11 C-PIB) and fluorodeoxyglucose ( 18 F-FDG) PET in patients with Alzheimer disease, mild cognitive impairment, and healthy controls. J Geriatr Psychiatry Neurol. 2010;23(3): Fodero-Tavoletti MT, Rowe CC, McLean CA, et al. Characterization of PiB binding to white matter in Alzheimer disease and other dementias. J Nucl Med. 2009;50(2): Stankoff B, Freeman L, Aigrot MS, et al. Imaging central nervous system myelin by positron emission tomography in multiple sclerosis using [methyl-¹¹c]-2-(4 -methylaminophenyl)-6- hydroxybenzothiazole. Ann Neurol. 2011;69(4): Chen X-H, Siman R, Iwata A, Meaney DF, Trojanowski JQ, Smith DH. Long-term accumulation of amyloid-beta, beta-secretase, presenilin-1, and caspase-3 in damaged axons following brain trauma. Am J Pathol. 2004;165(2): Klunk WE, Price JC, Mathis CA, et al. Amyloid deposition begins in the striatum of presenilin-1 mutation carriers from two unrelated pedigrees. J Neurosci. 2007;27(23): Younkin SG. The AAP and PS1/2 mutations linked to early onset familial Alzheimer s disease increase the extracellular concentration and A beta 1-42 (43). Rinsho Shinkeigaku. 1997;37(12): Whitaker C, Eckman C, Almeida C, et al. Live discussion: Amyloid-beta degradation: the forgotten half of Alzheimer s disease. 12 September J Alzheimers Dis. 2003;5(6): Zhou W, Xu D, Peng X, Zhang Q, Jia J, Crutcher KA. Meta-analysis of APOE4 allele and outcome after traumatic brain injury. J Neurotrauma. 2008;25(4): jamaneurology.com JAMA Neurology January 2014 Volume 71, Number 1 31