Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2021 Apr 4.
Published in final edited form as: Synapse. 2014 Jul 28;68(11):536–547. doi: 10.1002/syn.21765

Quantitative Imaging of Neuroinflammation in Human White Matter: A Positron Emission Tomography Study with Translocator Protein 18 kDa Radioligand, [18F]-FEPPA

IVONNE SURIDJAN 1,2,#, PABLO M RUSJAN 1,#, MIRAN KENK 1, NICOLAAS PAUL LG VERHOEFF 2,3,4, ARISTOTLE N VOINESKOS 1,2,3, DAVID ROTENBERG 1, ALAN A WILSON 1,2,3, JEFFREY H MEYER 1,2,3, SYLVAIN HOULE 1, ROMINA MIZRAHI 1,2,3,*
PMCID: PMC8019486  NIHMSID: NIHMS926793  PMID: 25043159

Abstract

The ability to quantify translocator protein 18 kDa (TSPO) in white matter (WM) is important to understand the role of neuroinflammation in neurological disorders with WM involvement. This article aims to extend the utility of TSPO imaging in WM using a second-generation radioligand, [18F]-FEPPA, and high-resolution research tomograph (HRRT) positron emission tomography (PET) camera system. Four WM regions of interests (WM-ROI), relevant to the study of aging and neuroinflammatory diseases, were examined. The corpus callosum, cingulum bundle, superior longitudinal fasciculus, and posterior limb of internal capsule were delineated automatically onto subject’s T1-weighted magnetic resonance image using a diffusion tensor imaging-based WM template. The TSPO polymorphism (rs6971) stratified individuals to three genetic groups: high-affinity binders (HAB), mixed-affinity binders (MAB), and low-affinity binders. [18F]-FEPPA PET scans were acquired on 32 healthy subjects and analyzed using a full kinetic compartment analysis. The two-tissue compartment model showed moderate identifiability (coefficient of variation 15–19%) for [18F]-FEPPA total volume distribution (VT) in WM-ROIs. Noise affects VT variability, although its effect on bias was small (6%). In a worst-case scenario, 6% of simulated data did not fit reliably. A simulation of increased TSPO density exposed minimal effect on variability and identifiability of [18F]-FEPPA VT in WM-ROIs. We found no association between age and [18F]-FEPPA VT in WM-ROIs. The VT values were 15% higher in HAB than in MAB, although the difference was not statistically significant. This study provides evidence for the utility and limitations of [18F]-FEPPA PET to measure TSPO expression in WM.

Keywords: translocator protein, neuroinflammation, white matter, positron emission tomography imaging, [18F]-FEPPA

INTRODUCTION

Microglia activation represents one of the main cellular features of neuroinflammation and the hallmark response of the central nervous system to brain insults (Chen and Guilarte, 2008). Microglia express a protein in their outer mitochondria membrane called the translocator protein 18 kDa (TSPO) (Chen and Guilarte, 2008). Increased TSPO expression is recognized as a reliable biomarker of microglial activation both in vitro and in vivo (Chen and Guilarte, 2008; Cosenza-Nashat et al., 2009; Denes et al., 2007; Guilarte et al., 1995; Venneti et al., 2009). Although other cell types express TSPO, activated microglia are considered the dominant cellular source of TSPO expression in both gray and white matter (WM; Banati et al., 2000; Cosenza-Nashat et al., 2009; Venneti et al., 2008). In vitro studies using post-mortem human brain tissue showed that samples containing WM lesions have a three- to fourfold increase in TSPO expression compared to normal WM tissue (Banati et al., 2000; Venneti et al., 2008).

Positron emission tomography (PET) with TSPO-specific radioligands has been used for in vivo evaluation of neuroinflammation in neurological disorders with WM involvement, such as in multiple sclerosis (MS) (Banati et al., 2000; Oh et al., 2011). Together, these findings suggest that focal increases in TSPO expression indicate areas of active inflammation and provide support to the utility of TSPO imaging for quantifying microglia activation/neuroinflammation in WM disease.

Recently, we have shown that [18F]-FEPPA (Wilson et al., 2008) can be quantified in human brain with a two-tissue compartment (2TC) model providing excellent identifiability for the estimation of [18F]-FEPPA total distribution volume (VT) in grey matter (Rusjan et al., 2011). However, [18F]-FEPPA VT estimation in total WM region showed lower identifiability, as indicated by a higher coefficient of variation (COV). A single polymorphism at the TSPO gene (rs6971) accounts for large variations in binding parameters also shown in many second-generation TSPO radioligands (Kreisl et al., 2013; Owen et al., 2012), including [18F]-FEPPA (Mizrahi et al., 2012). Based on this polymorphism, individuals can be classified into: high affinity binders (HAB), mixed affinity binders (MAB), and low affinity binders (LAB) (Kreisl et al., 2010; Owen et al., 2012). However, whether the rs6971 polymorphism also predicts [18F]-FEPPA binding in the WM is unknown (Oh et al., 2011; Thiel et al., 2010).

Previous post-mortem human studies have indicated that microglia undergoes significant age-related alterations in morphology (Sheng et al., 1998; Streit et al., 2004), phenotype, and function (Miller and Streit, 2007). The aging brain shows an increase in the number of activated microglia, characterized by an increase in the expression of surface proteins such as the major histocompatibility complex II protein and a pronounced increase in the expression of several proinflammatory cytokines (DiPatre and Gelman, 1997; Sheng et al., 1998). The increase in the density of activated microglia has also been noted in the WM regions of the cingulate gyrus and corpus callosum (CC) (Sloane et al., 1999). An in vitro autoradiography study using human platelet samples with the prototypical TSPO radioligand, [3H]-PK11195, has found no difference in TSPO density between healthy young and older subjects (Marazziti et al., 1994). However, several in vivo PET studies using both [11C]-PK11195 and second-generation TSPO radioligands have reported mixed findings. While some studies have reported an age-associated increase in TSPO binding throughout cortical grey matter regions (Gulyas et al., 2011; Guo et al., 2013; Kumar et al., 2012; Schuitemaker et al., 2010), others have failed to detect significant associations (Cagnin et al., 2001; Debruyne et al., 2003; Ouchi et al., 2005; Suridjan et al., 2014; Yasuno et al., 2008). Importantly, to our knowledge, there have been no studies that investigated the relationship between age and TSPO expression in the WM regions using second-generation TSPO radioligands.

Our previous observations indicate that [18F]-FEPPA kinetics in the total WM region showed lower uptake and slower wash-out as compared to that in the grey matter regions. The primary aim of this study is to validate the quantification of [18F]-FEPPA using a full kinetic compartment analysis in four specific WM regions that are relatively large in size and that have been previously reported to be affected in the aging process (Kerchner et al., 2012; Salat et al., 2005; Sullivan et al., 2006, 2010; Voineskos et al., 2012) and in disease states (Huang et al., 2012; Sexton et al., 2010). The WM regions studied include the CC, cingulum bundle (CB), superior longitudinal fasciculus (SLF), and posterior limb of internal capsule (PLIC). Furthermore, given the challenges in quantifying [18F]-FEPPA kinetics in the small regions, we investigated the caveats and limitations of [18F]-FEPPA quantification in selected WM regions by performing Monte Carlo simulation to evaluate the effect of noise on the bias, identifiability, and variability of [18F]-FEPPA VT. Furthermore, we also investigated whether differences in WM [18F]-FEPPA VT can be detected between the genetic groups. Finally, we examined age-related changes in [18F]-FEPPA VT across the adult life span to explore whether there is significant association between age and neuroinflammation in these WM tracts.

MATERIALS AND METHODS

Radiopharmaceutical preparation

The details of [18F]-FEPPA synthesis have been described elsewhere (Wilson et al., 2008).

Human subjects

Thirty-two healthy subjects underwent a [18F]-FEPPA PET and a magnetic resonance imaging (MRI) scan. The healthy subjects reported in this study were part of a healthy subject cohort included in a previous study (Suridjan et al., 2014), and 12 of these 32 subjects were also part of the healthy subject’s cohort included in our first study (Rusjan et al., 2011). All subjects underwent interviews for medical history, physical examination, and urinalysis (including toxicological screening for drug use) to rule out current and history of medical and psychiatric illness. Participants were physically healthy and were excluded if they were obese, or if they had history of diabetes, past cardiovascular events, stroke, or other neurological diseases. All participants were deemed cognitively intact and without any memory problems as demonstrated by scores 27 and above on the Mini-Mental State Examination (MMSE) (Folstein et al., 1975). Blood samples were collected during the PET scan for genotyping of the TSPO rs6971 polymorphism (as described in Mizrahi et al., 2012) and for obtaining the arterial input function for the kinetic analysis of [18F]-FEPPA (as described in detail below). All subjects provided written informed consent after all procedures were fully explained. All experimental procedures were approved by the Centre for Addiction and Mental Health Ethics Review Board.

PET scans and arterial blood sampling

The PET images were obtained for 125 min following the injection of [18F]-FEPPA using a 3D high-resolution research tomography (CS/Siemens, Knoxville, TN), which measures radioactivity in 207 slices with an interslice distance of 1.22 mm. All PET images were corrected for attenuation using a 137Cs point source and reconstructed by a filtered back-projection algorithm using a HANN filter at Nyquist cutoff frequency. Images were reconstructed into 34 time frames: 1 frame of variable length until the radioactivity appears in the field of view, 5 × 30 sec, 1 × 45 sec, 2 × 60 sec, 1 × 90 sec, 1 × 120 sec, 1 × 210 sec, and 22 × 300 sec. The reconstructed image has 256 × 256 × 207 cubic voxels measuring 1.22 mm × 1.22 mm × 1.22 mm, and the resulting reconstructed resolution is close to isotropic 4.4 mm, full width at half maximum in plane, and 4.5 mm fill width at half maximum axially, averaged over measurements from the centre of the transaxial field of view to 10 cm off-centre in 1.0 cm increments.

A dose of 185 ± 20 MBq (5 ± 0.5 mCi) of [18F]-FEPPA was administered as a bolus intravenous injection (mass injected: 0.84 ± 0.74 mg; specific activity: 151 ± 122 GBq mmol21; activity injected: 177 ± 13 MBq). An automatic blood sampling system (Model #PBS-101, Veenstra Instruments, The Netherlands) was used to measure arterial blood radioactivity continuously at a rate of 2.5 mL min21 for the first 22.5 min. Manual arterial blood samples were obtained at 2.5, 7, 12, 15, 30, 45, 60, 90, and 120 min to measure whole blood and plasma radioactivity and the plasma metabolite composition. The ratio of whole blood to plasma radioactivity was fitted to a bi-exponential function and applied as a correction factor to the arterial blood radioactivity time-activity curve (TAC) to generate the plasma TAC. Plasma parent radioligand fraction was determined by high-performance liquid chromatography analysis and was fitted with a Hill function. The plasma TAC was multiplied by the fitted plasma parent radioligand concentration and corrected for delay and dispersion to generate a parent compound in the plasma curve to use as an input function for the kinetic analysis (for further details, please see our first article Rusjan et al., 2011).

MRI and regions of interest delineation

For the anatomic delineation of WM tracts, a brain MRI image was acquired for each subject. On 25 of our subjects, T1-weighted and proton density (PD)weighted images were acquired with a General Electric (Milwaukee, WI) Signa 1.5T magnetic resonance image scanner. On the remaining seven subjects, the T1-weighted images were acquired with a 3T General Electric MR750 scanner. For the details of the MRI acquisition parameters, please see our previous study (Suridjan et al., 2014). The T1-weighted images were acquired to use for image coregistration with the PET image (described below). The PD-weighted images were visually inspected for focal and vascular lesions.

Transformation of population-average WM atlas (ICBM-DTI81) to subject T1-MRI and PET space

WM regions of interest (WM-ROIs) were defined with respect to the Johns Hopkins University DTI atlas in ICBM-152 space (ICBM-DTI81) (Mori et al., 2008). The WM-ROIs were automatically delineated onto the PET dynamic images for the generation of regional TACs. The step-by-step procedure implemented for this process is illustrated in Supporting Information Figure S1. First, a nonlinear transformation was calculated to map the ICBM-152 standard template onto each subject’s high-resolution T1-weighted MRI (Ashburner and Friston, 2005) using SPM8 (Wellcome Trust Centre for Neuroimaging Institute of Neurology, UCL, London, UK). This transformation was then applied to the ICBM-DTI81 atlas, to fit the parcellated ROIs to the subject’s native space. The transformed WM-ROIs were refined based on the probability of WM in each voxel. Similar to the procedure used for gray matter delineation (Rusjan et al., 2006), a WM probability map was created with a segmentation algorithm within SPM8 and later denoised by the application of a FMHM=1 mm Gaussian smoothing filter, so that voxels within the WM-ROIs that have low probability of belonging to WM were removed. The individual MR images were then coregistered to each summed [18F]-FEPPA PET image using the normalized mutual information algorithm (Studholme et al., 1997), and the resulting rigid body transformation was applied to the refined WM-ROIs, to mask the PET image and generate the TACs. A frame-by-frame realignment motion correction algorithm was applied to the dynamic PET data to minimize the effect of noise due to motion. Four of the automatically delineated WM-ROIs corresponding to the CC, CB, SLF, and PLIC tracts were used for downstream processing. The mean ± SD volumes for these WM-ROIs were 19 ± 3 cm3, 2 ± 0.8 cm3, 13 ± 2 cm3, and ± 6 0.8 cm3, respectively. Standard uptake values (SUVs) were calculated by correcting the TACs for injected activity and subject weight.

Kinetic analysis

Kinetic analyses were performed using PMOD 3.17 software (PMOD Technologies, Zurich, Switzerland). Rate constants were estimated using a one-tissue compartment model (1TCM: K1 and k2) and a 2TCM (K1, k2, k3, and k4) using [18F]-FEPPA radioactivity in arterial plasma as an input function (as described previously Rusjan et al., 2011). Each model configuration was implemented to account for the contribution of activity from the cerebral blood volume assuming that cerebral blood volume accounts for 2.7% for WM (Leenders et al., 1990). VT was used as an outcome measure (Innis et al., 2007). The COV for VT as calculated in PMOD was used to measure the identifiability of VT. A smaller percentage indicates better identifiability.

Simulation study to assess bias and variability of VT estimates when noise level increases

Simulated TACs with several noise levels were generated to investigate the noise-induced bias and variation of parameters estimated by the 2TCM. Previously reported [18F]-FEPPA kinetic rate constants (Rusjan et al., 2011) were used to generate the noise-free TACs: K1=0.09 mL cm−3 min−1, K1/k2=0.57 mL cm−3, k3=0.11 min−1, and k4=0.01 min−1 with 2 h scanning length. The noise level for each frame was modeled with a Gaussian distribution with standard deviation (SD) according to the formula described previously (Logan et al., 2001):

SDi=SFeλtiend+tistart2ci(tiend−tistart) (1)

where i is the frame number, ci is the noise-free simulated radioactivity, tiend is the end time of the ith frame, tistart is the start time of ith frame, λ is the radioisotope decay constant (which is 109.8 min for F-18 radioligands), and SF is a scaling factor to adjust the level of noise. The noise was generated with random number based on Gaussian distribution and added to the radiotracer activity in each frame. One thousand noisy TACs were generated at different noise levels of 10%, 13%, and 20% by setting SF values in Eq. 1 as 8.5, 10, and 15, respectively. These noise levels cover a range of TAC noise typically contained within the WM-ROIs studied. Percent TAC noise was calculated as the average ratio of mean SD to mean of radioactivity of all 33 time frames. The 2TCM was applied to simulated TAC data to estimate [18F]-FEPPA VT. Mean [18F]-FEPPA VT for a given noise level was calculated as the average VT values of simulated TACs, excluding outliers, defined as COV value ≥100%. Noise-induced bias was calculated as percent deviation of sample mean VT from the true noise-free value. Noise-induced variability of simulated VT at each noise level was calculated as sample SD. The noise-induced variability for VT was also expressed as percent coefficient of variation (CV), calculated as ratio of SD/mean × 100% (excluding outliers).

Simulation study to assess the variability and identifiability of VT estimates when k3 increases

Monte Carlo simulations were performed to assess the identifiability of VT when k3 increases to simulate (pathological) situations of increased TSPO density in WM-ROI. The simulation study was performed under an ideal, noise-free condition as well as under conditions of the same three noise levels: 10%, 13%, and 20% (i.e., SF value was set to 8.5, 10, and 15) as implemented previously.

Statistical analysis

Goodness of fit was evaluated using the Akaike Information Criterion (AIC) (Akaike, 1974) and the model selection criterion (MSC) (MicroMath, 1995) as calculated in PMOD (PMOD Technologies). Lower AIC and higher MSC values were indicative of a better fit. VT estimates from the 2TCM and 1TCM were compared with paired t-test. Regression analyses were performed to examine the effects of age and rs6971 polymorphism on [18F]-FEPPA VT in each WM-ROI.

Statistical analyses were performed using SPSS Statistics 17.0. The threshold for significance was set at P< 0.05, two-tailed, for all analyses.

RESULTS

Subjects

Thirty-two healthy individuals were included in the study (12 males, 20 females; age 48.1 ± 17.9, age range: 19–78 years old). All subjects were free of any current medical and psychiatric illness. Twenty-nine subjects were not taking any prescription or over-the-counter medication, while three of 32 individuals were taking antihypertensive or cholesterol-lowering therapy. Visual inspection of PD-weighted MRI revealed no evidence of significant vascular lesion except in one subject where evidence of WM hyperintensities was observed surrounding the anterior and posterior horn of the lateral ventricles.

Genetic analysis revealed 21 HABs, 11 MABs, and no LABs. There was no difference in age (F(1, 30)=0.39; P=0.537) or MMSE scores (F(1,30)=1.996; P=0.171) between HAB and MABs. Additionally, there was no difference in the tracer parameters between the genetic groups (amount injected (F(1, 30)=0.006; P=0.937), mass injected (F(1,30) =0.040; P=0.843), specific activity at injection (F(1,30)=0.720; P=0.403)).

Time-activity curves

The appearance of TACs in all studied WM tracts was different from that previously observed in the grey matter (Rusjan et al., 2011). WM TACs showed a lower peak and slower wash-out compared to the grey matter, with peak SUV reached within 15 min of tracer administration. The TAC for the CB shows the highest peak (SUV 0.92) followed by PLIC (SUV 0.81), SLF (SUV 0.60), and CC (SUV 0.59). After 2 h, the activity decreased by 14% of the peak value for CC, 33% for CB, 15% for SLF, and 27% for PLIC. The average SUV showed almost a complete overlap between the HAB and MAB subjects in all ROIs (Fig. 1).

Fig. 1.

Fig. 1.

Open in a new tab

The average concentration of radioactivity after injection of [18F]-FEPPA, given as SUV, separated by the genetic groups: HAB (n=21) and MAB (n=11) in the CC and CB (A) and in the SLF and PLIC (B). Overall, the SUV curves showed relatively low uptake and slow wash-out in all the WM-ROI examined. The SUV curves also showed almost a complete overlap between HAB and MAB.

Kinetic analysis

For each of the WM tracts studied, the 2TCM provided a better fit than the 1TCM (Fig. 2), with lower AIC and higher MSC values (paired t-test AIC and MSC for all WM-ROI < 0.001). Depending on the WM-ROI, approximately 10% of our samples (the number varies between two and four subjects,) present VT values with very low identifiability (COV>30%). However, including or excluding these subjects did not affect the average results. Table I showing that the mean [18F]-FEPPA VT values across all four WM-ROI were around 7 mL cm23, with moderate identifiability, indicated as mean COVs ranging between 15% and 19%. A summary of kinetic parameters separated by HAB and MAB is also presented in Table I, showing no difference in mean (F(1,30)=1.44; P=0.24) and identifiability (F(1, 30)=1.09; P=0.31) of [18F]-FEPPA VT values across all four WM-ROIs.

Fig. 2.

Fig. 2.

Open in a new tab

The time-activity data and curve fitting for CC for a typical subject. 2TC model provided significantly better fitting that did 1TCM for all subjects. SUV, standardized uptake value.

TABLE I.

Summary of kinetic rate constants and [18F]-FEPPA VT estimated with 2TCM; n=32

K1 (mL cm−3 min−1) K1/k2 (mL cm−3) k3 (1 min−1) k4 (1 min−1) VT(mL cm−3)
All subjects (n = 32)
  CC 0.09 ± 0.03 (6.8) 0.58 ± 0.22 (12.8) 0.11 ± 0.06 (17.5) 0.01 ± 0.00 (28.4) 6.75 ± 2.59 (15.2)
  CB 0.22 ± 0.20 (13.0) 0.64 ± 0.42 (16.1) 0.19 ± 0.13 (25.6) 0.01 ± 15.26 (32.7) 7.54 ± 2.15 (16.7)
  SLF 0.09 ± 0.03 (7.5) 0.66 ± 0.23 (14.4) 0.10 ± 0.04 (19.6) 0.01 ± 0.00 (34.1) 7.18 ± 2.65 (18.8)
  PLIC 0.13 ± 0.05 (10.4) 0.77 ± 0.37 (16.1) 0.12 ± 0.08 (24.0) 0.01 ± 0.00 (37.4) 7.50 ± 2.49 (17.2)
  Total WM 0.08 ± 0.03 (5.0) 0.64 ± 0.15 (8.7) 0.08 ± 0.02 (12.8) 0.01 ± 0.00 (25.4) 7.21 ± 2.33 (10.4)
HAB; n = 21
  CC 0.09 ± 0.03 (6.1) 0.62 ± 0.16 (12.1) 0.10 ± 0.03 (18.4) 0.01 ± 0.00 (33.9) 6.92 ± 2.82 (15.2)
  CB 0.24 ± 0.23 (13.9) 0.62 ± 0.36 (14.9) 0.20 ± 0.14 (24.6) 0.01 ± 0.01 (33.9) 7.89 ± 2.18 (17.5)
  SLF 0.09 ± 0.03 (7.8) 0.70 ± 0.22 (13.1) 0.09 ± 0.02 (18.4) 0.01 ± 0.00 (35.2) 7.60 ± 2.89 (19.8)
  PLIC 0.12 ± 0.04 (11.1) 0.77 ± 0.37 (16.1) 0.12 ± 0.06 (23.5) 0.01 ± 0.00 (41.4) 7.84 ± 2.53 (19.6)
  Total WM 0.08 ± 0.02 (4.8) 0.66 ± 0.14 (8.3) 0.08 ± 0.02 (12.0) 0.01 ± 0.00 (25.9) 7.46 ± 2.54 (14.8)
MAB; n = 11
  CC 0.09 ± 0.04 (8.2) 0.50 ± 0.28 (14.1) 0.14 ± 0.09 (15.9) 0.01 ± 0.00 (26.1) 6.43 ± 2.16 (15.2)
  CB 0.16 ± 0.11 (11.4) 0.69 ± 0.54 (18.5) 0.17 ± 0.13 (27.5) 0.01 ± 0.01 (30.5) 6.85 ± 0.01 (15.1)
  SLF 0.09 ± 0.04 (6.9) 0.60 ± 0.25 (16.9) 0.11 ± 0.06 (22.0) 0.01 ± 0.00 (31.9) 6.38 ± 2.00 (16.7)
  PLIC 0.13 ± 0.06 (9.0) 0.78 ± 0.38 (15.9) 0.13 ± 0.11 (25.0) 0.01 ± 0.00 (29.8) 6.87 ± 2.39 (12.6)
  Total WM 0.07 ± 0.03 (5.2) 0.60 ± 0.16 (9.7) 0.09 ± 0.02 (14.5) 0.01 ± 0.00 (24.4) 6.83 ± 1.92 (13.5)
Open in a new tab

Data are mean ± SD, with mean COV in parentheses.

A time stability analysis shows that VT increases with the scan duration for all WM-ROIs (Fig. 3). In comparison to the VT estimated with 120 min scan length, there was an average difference of 17%, 14%, 10%, 8%, 6%, 6%, and 3% for the scanning length duration of 85, 90, 95, 100, 105, 110, and 115 min, respectively.

Fig. 3.

Fig. 3.

Open in a new tab

The average time convergence of VT to 120 min value across the four WM-ROIs.

Simulation studies

The effect of noise on identifiability in 2TCM estimation of [18F]-FEPPA VT

Table II shows the percent bias, variability, and identifiability for simulated data at different noise levels. The true VT value of a noise-free TAC was 6.95 mL cm3. The identifiability of VT becomes lower, and the variability increases as the noise level increases. However, despite the decrease in identifiability, the noise-induced bias in VT was relatively small. At 10% noise level, the mean identifiability of VT was increased to 15%, and the variability (SD and CV) of the simulated VT was 0.93 (CV=13%). Despite this relatively large decrease in identifiability, there was only an approximately 2% change in the mean VT value relative to the true VT value (6.95 increased to 7.10). Similarly, at 20% noise level, the identifiability and variability of VT was around 24%, but the estimated VT value remains relatively stable, showing a modest 6% increase from 6.95 to 7.34.

TABLE II.

Summary of simulation data examining the effect of increase in noise levels and TSPO density on the bias, variability, and identifiability of [18F]-FEPPA VT

VT CV Identifiability VT %COV
SF Noise level (%) Increase k3 Mean SD SD/mean Mean SD N* Bias (%)
TRUE 0 1 6.95 n/a n/a 0.5 n/a n/a n/a
1.3 8.87 n/a n/a 0.4 n/a n/a n/a
1.6 10.77 n/a n/a 0.4 n/a n/a n/a
1.9 12.70 n/a n/a 0.4 n/a n/a n/a
8.5 10 1 7.10 0.93 13% 14.79 5.97 9 2%
1.3 9.15 1.35 15% 14.91 6.02 3 3%
1.6 11.10 1.50 14% 15.10 5.94 0 3%
1.9 13.02 1.88 14% 15.16 4.76 2 3%
10 13 1 7.22 1.29 18% 19.49 9.55 15 4%
1.3 9.28 1.71 18% 19.73 9.62 9 5%
1.6 11.35 2.32 20% 20.13 10.30 9 5%
1.9 13.29 2.64 20% 20.06 8.85 18 5%
15 20 1 7.34 1.74 24% 29.74 15.49 61 6%
1.3 9.52 2.34 25% 30.45 16.32 48 7%
1.6 11.83 3.08 26% 32.09 16.56 71 10%
1.9 13.77 3.58 26% 32.43 16.00 52 8%
Open in a new tab

Data were analyzed using the 2TCM. Mean and SD were reported for various SF values (i.e., % noise levels) and k3 values. N* indicates the number of outliers out of 1000 experiments (not in percentage). Outliers were arbitrarily defined as VT estimates with unacceptable identifiability (COV > 100%).

The effect of increasing k3 on the variability and identifiability of [18F]-FEPPA VT under conditions of several noise levels

As we increase the noise level, the identifiability of VT decreases. Our results showed that increasing k3 did not have a large effect on the identifiability and variability of VT. For example, at the highest noise level, increasing k3 by 90% resulted in the increases in COV from approximately 30% to 32%, while the variability of VT was increased from 24% to 26%. Table II summarizes the results of the simulations performed with k3 increased by 30%, 60%, and 90% under a noise-free condition and with 10%, 13%, and 20% noise levels.

[18F]-FEPPA VT in the WM tracts across healthy adult life span

The kinetic parameters and VT for both HAB and MAB are summarized in Table I. In all WM-ROIs examined, the measured mean [18F]-FEPPA VT values were higher in HAB than in MAB, showing a range of 8–19% difference between the groups. Regression analyses with [18F]-FEPPA VT as dependent variable, and age and genetic group as predictors, revealed no significant effect of age or rs6978 polymorphism in any of the WM-ROIs examined (Table III). Figure 4 shows scatter plots illustrating the non-significant relationship between age –and [18F]-FEPPA VT in all four WM-ROIs. In addition, there was no relationship between age and [18F]-FEPPA VT in the total WM.

TABLE III.

Regression analyses to explore the relationship between age and [18F]-FEPPA VT in the white matte regions, adjusted for the rs6971 polymorphism

Age effect
 Genetic effect
Regions F (1,29) P F (1,29) P
CC 0.140 0.711 0.265 0.610
CB 0.298 0.589 1.758 0.195
SLF 0.592 0.448 1.652 0.209
PLIC 0.118 0.734 1.028 0.319
Total WM 1.217 0.279 0.528 0.473
Open in a new tab
Fig. 4.

Fig. 4.

Open in a new tab

The relationship between age and [18F]-FEPPA VT in four white matter tracts, showing no significant association controlling for genetic status (HAB and MAB, all P> 0.05)

DISCUSSION

This work describes the evaluation of [18F]-FEPPA binding quantification in four WM regions. Consistent with our initial report of [18F]-FEPPA binding in the whole brain WM (Rusjan et al., 2011), the binding in the four WM-ROIs showed slower kinetics characterized by lower uptake and slower wash-out in comparison to the grey matter regions. The 2TCM describes the kinetics of [18F]-FEPPA better than 1TCM, as indicated by the lower AIC and higher MSC values. The slow wash-out of [18F]-FEPPA in the WM regions makes the fitting of time-activity data more difficult. We fitted 2TCM to our simulated TAC to examine the effect of noise on bias, variability and identifiability of VT. The identifiability of VT is a measure of how well the 2TCM fits the data. The fitting of TAC data with higher noise level tends to have larger bias and lower identifiability (higher COV). However, the bias resulted from the noise was small, as indicated by relatively small differences in VT values between those calculated under noise levels and those calculated under an ideal noise-free condition. Although the noise-induced bias was small, the variability was high. For a relatively high noise level (20%), the identifiability and variability of VT were increased to about 30% and 24%, respectively, while the noise-induced bias was only around 6%.

As shown in Figure 3, the VT value increases with a longer scan length, although the change in VT becomes smaller as it is approaching the 120-min time point. These results are consistent with our previous report for 90, 120, 180 min (Rusjan et al., 2011) and similar to previous observations for the other second-generation TSPO radioligands, such as [11C]-PBR28, [11C]-DPA713, and [18F]-PBR06 in humans (Endres et al., 2009; Fujimura et al., 2006; Fujita et al., 2008).

Given the prevalence of activated microglia in WM lesions (Hayes et al., 1987; Simpson et al., 2007), our hope was to extend the utility of [18F]-FEPPA PET to measure increased TSPO expression in clinical populations with WM disease. We performed a simulation study by increasing k3 to examine the effect of increased TSPO density on the identifiability and variability of VT. Increasing the k3 value did not change the identifiability or variability of VT. In a condition with 20% noise level and almost twofold increase in TSPO density, the identifiability and variability of VT were only minimally increased by 9% and 8%, respectively (i.e., the identifiability and variability of VT were increased to 32% and 26%, respectively). These results indicate that the identifiability and variability of VT estimates were similar for both low and high TSPO density conditions, giving support to the potential utility of [18F]-FEPPA to capture a moderate to large increase of TSPO density in WM regions.

The results of the noise simulation study revealed a caveat of TSPO quantification in WM regions. As the noise level increases to 20%, the estimated VT value remains stable, but the number of outliers (i.e., fitting with COV≥100%) increases from 0% to 6%. The noise level in the time-activity data will likely be greater in smaller size WM-ROIs. Previous PET studies have detected elevated TSPO expression in WM lesions occurring in stroke and MS patients (Oh et al., 2011; Thiel et al., 2010), which are arguably small WM regions. In our study, the volume of WM-ROIs studied was between 2 and 19 cm3. If the WM lesion of interest is smaller than approximately 2 cm3, as might be true for WM plaques, then the noise level in the time-activity data would likely be greater than the 20% noise level observed in our real and simulation data. In this case, there will be more variability in the estimation of VT. Given the potential impact of noise on the identifiability of VT, caution should be exercised when interpreting results in WM-ROIs that are smaller than the ones we studied here.

In a clinical setting, the sample size required to detect a difference in [18F]-FEPPA VT will vary depending on the size of WM lesions studied. Based on the mean and variability of VT in our data (as shown in Table II), using the F-test ANOVA to calculate the required sample size, 21 subjects will be needed per group to detect 30% difference between them, assuming effect size d=0.6, alpha=0.05, and power=0.8.

Surprisingly, we did not find a significant difference in [18F]-FEPPA binding between the TSPO polymorphism genetic groups in the WM. Our previous work indicated that [18F]-FEPPA VT is approximately 30% higher in HAB relative to MAB in grey matter ROIs (Mizrahi et al., 2012). In the WM-ROIs, the measured [18F]-FEPPA VT values were observed to be on average about 15% higher in HAB than in MAB, although this difference did not reach statistical significance. In our study sample, we noted that a single MAB subject presented VT values that were almost two times higher than the group mean in all WM-ROIs. We conducted an exploratory post hoc onetailed T test analysis (hypothesising increase VT in HABs), excluding this MAB subject. We found that the differences between HAB and MAB were significant in SLF (t=1.80, P=0.04), PLIC (t=1.72; P= 0.05), and CB (t=1.65; P=0.05), but not in CC (t=1.03, P=0.16). Although the MAB outlier presented higher VT, the 2TCM fitting of the time-activity data of this subject is comparable to the rest of the MAB subjects in the study (the COV ranges between 9% to 14% depending on the WM regions), suggesting that the increases in VT values were not due to poor fitting of the TACs. In addition, consistent with the VT values observed in the WM regions, the VT values in several grey matter regions for this subject were also almost twofold higher as compared to the group mean (data not shown). The higher VT values of the MAB outlier could also be explained by a lower plasma input function, which is consistent with our observation in this dataset.

We propose several possible reasons to explain the lack of differences in [18F]-FEPPA VT between the HAB and MAB. First, assuming that the distribution volume of the free and nonspecific compartment (VND) is the same for both WM and grey matter regions, the lower VT values in the WM regions imply that the fraction of VT attributed to the specific binding in the WM regions is lower. If this were the case, the difference in VT between HAB and MAB would be expected to be smaller in the WM regions. Therefore, assuming that the noise level in the TAC is the only source of within-subject variability, a larger sample size (more than 21 subjects in each group) would be needed to detect a statistically significant difference in VT between HAB and MAB in the WM regions. Alternatively, it is also possible that the VND is not the same for HAB and MAB. However, in the absence of a blocking study, this cannot be verified experimentally. Finally, there is a possibility that the higher myelin lipid content around the axon would contribute to a greater [18F]-FEPPA nonspecific binding in the WM regions. However, in our experiment we cannot determine with certainty how much of [18F]-FEPPA binding is specific and how much of it is free or nonspecific. If we assume that the VND is greater in the WM regions than the grey matter regions, then we would expect the specific binding in the WM regions to be even smaller, and therefore even a greater sample size would be required to detect a significant difference in VT values between HAB and MAB. It remains to be demonstrated whether FEPPA VT between HAB and MAB can be differentiated in subjects with WM inflammatory conditions, where higher signal to noise ratio might be present due to elevated TSPO expression.

We hypothesize that the lack of statistically significant differences between HAB and MAB might be due to the inherently low TSPO expression in normal WM. This view is supported with results from both in vitro and in vivo TSPO studies, which have also demonstrated a relatively lower TSPO expression in healthy WM tissue (Kumar et al., 2012; Owen et al., 2010). Although [3H]-FEPPA has a relatively higher in vitro affinity for TSPO than [3H]-PK11195 or [3H]PBR28 (Wilson et al., 2008), the relatively low TSPO expression in WM might not be sufficient for [18F]-FEPPA to differentiate HAB and MAB in low TSPO density areas like healthy WM. This idea is analogous to findings from previous human whole body PET imaging with [11C]-PK11195 (Kreisl et al., 2010). The existence of different affinity groups has been shown for most second generation TSPO radioligands (Owen et al., 2011); however, [11C]-PK11195 was only able to differentiate affinity groups in organs with high TSPO density (such as in lung and heart) but not in organs with relatively lower TSPO density (such as the brain) (Kreisl et al., 2010). Also, it is possible to attribute the lack of differences in binding between genetic groups to factors other than TSPO density. For example, we noted that the tracer delivery ratio from plasma to brain (K1) for the WM tract was lower (K15 0.09 mL cm−3 min−1 in the CC) compared to those previously observed in grey matter ROIs (e.g., K1=0.19 mL cm−3 min−1 in the temporal cortex (Rusjan et al., 2011)).

In this study, we explored the relationship between age and [18F]-FEPPA binding in WM regions across the adult life span. A previous immunohistochemical examination has reported a relatively low TSPO expression in the WM of healthy human brains as compared to that of diseased brains (Cosenza et al., 2009). In line with this ex vivo evaluation, previous in vivo PET studies have also detected increases in TSPO expression in WM lesions, such as those occurring in MS and stoke patients (Oh et al., 2011; Thiel et al., 2010). In our healthy control sample, there were no associations between age and [18F]-FEPPA binding in any of the four WM-ROIs studied. This observation is in agreement with results from a recent in vivo PET study in healthy volunteers, which also found no association between age and [11C]-PK11195 binding in WM throughout the brain (Kumar et al., 2012). Similarly, in vivo examinations of TSPO expression in grey matter regions using several other TSPO radioligands, including [18F]-FEPPA, have also detected no significant association between age and TSPO binding in the grey matter regions of cognitively intact healthy subjects (Debruyne et al., 2003; Suridjan et al., 2013; Yasuno et al., 2008). However, two studies using other second-generation TSPO radioligands, such as [11C]vinpocentine (Gulyas et al., 2011) and [11C]-PBR111 (Guo et al., 2013) have reported significant positive associations between age and TSPO expression throughout the grey matter regions. The discrepancy in the results among these studies might be attributed to differences in the sample demographics (i.e., age range, sample size), methodology (i.e. standardized update values versus VT as outcome measure), and pharmacokinetics of radioligands used. Several DTI studies performed in healthy human subjects, however, have found significant associations between age and declines in microstructural integrity of these WM tracts (Kerchner et al., 2012; Salat et al., 2005; Sullivan et al., 2006, 2010). These studies showed that the decline in the WM microstructural integrity was significantly related to cognitive impairment commonly observed in older individuals (Lu et al., 2013; Peters, 2002). It is possible to attribute the lack of association between age and TSPO binding in our study to the demographic characteristic of our study participants. First, we only included older subjects that did not have evidence of cognitive impairment as indicated by their normal MMSE scores. Second, vascular risk factors are associated with the occurrence of microstructural lesions in WM tissue and may contribute to appearance of neuroinflammation in the affected area. However, all of our study participants would be considered to be “super healthy,” as none of them had any history of past cardiovascular events or even risk factors associated with cardiovascular and cerebrovascular disease. Only three of the 32 participants were taking antihypertensive or cholesterol-lowering medications. Although WM hyperintensities are often detected in older individuals, there were no detectable vascular lesions in most participants upon visual inspection of their PD-weighted MRI in our study. We detected the appearance of periventricular WM hyperintensities only in one person, which was visible surrounding the anterior and posterior horn of the lateral ventricles.

As shown in the time-activity data, the wash-out of [18F]-FEPPA in the WM-ROIs is quite slow. However, we have previously shown that the 2 h acquisition time is sufficient to provide a reliable estimation of VT in both grey and WM regions, as indicated by the small bias and relatively small change in identifiability of VT from 3 to 2 h scan time (Rusjan et al., 2011). Although a longer scanning time might be slightly more beneficial from the point of view of identifiability, it has some practical limitations in terms of the feasibility of human PET studies. More importantly, our previous study has shown that the correlation of VT between the 2 and 3 h scan time is excellent in the WM as a whole (Rusjan et al., 2011).

The organization of WM fibre tracts in the human brain is heterogeneous and can be quite complex. As a result, partial volume error (PVE) might differentially affect the accuracy of VT quantification among WM-ROIs studied. The partial volume effects depend on the proximity of the different tissues types and the ROI size. The WM regions that are closer to the grey matter areas would be more affected by the spill-in partial volume effects from the grey matter regions. Moreover, the quantification of radioligand binding in smaller ROIs is generally more compromised by the partial volume effects. Therefore, given that the PLIC and CB are smaller in size as compared to the SLF or CC, it is possible that there might be greater partial volume errors in these regions. The lack of correction for PVE presents a limitation of the present work. It should be noted that the spill-in radioactivity contribution to the measured signal would come from the grey matter. Although we expect the mathematical bases of the PVE correction (PVEC) to remain, to our knowledge the algorithms of PVEC in the field have been validated to correct grey matter radioactivity concentration from the spill-in of radiotracer from the WM. To minimize the effect of PVE in our quantification, our automatic ROI delineation method was set up in a conservative manner, ensuring that the edges of the ROIs were excluded, as previously described in (Rusjan et al., 2006). Importantly, post hoc analyses revealed no associations between [18F]-FEPPA VT and ROI volume in any of the WM regions examined (Supporting Information Figs. 2A–2D).

Our results provide evidence for the potential utility and limitations of [18F]-FEPPA PET as an imaging probe to index TSPO expression in WM. Given the remarkable differences in kinetics of [18F]-FEPPA between the grey and WM areas, the results of this study represent a step forward toward our ability to quantify neuroinflammation/microglia activation in a variety of disease conditions involving WM. Future studies may combine DTI and [18F]-FEPPA PET to evaluate neuroinflammation in patient populations in regions of the brain that have reduced WM integrity.

Supplementary Material

Supplementary Figure Legends

ACKNOWLEDGMENTS

The authors thank technicians Alvina Ng and Laura Nguyen, chemistry staff Jun Parkes, Armando Garcia, Winston Stableford, and Ming Wong, and engineers Terry Bell and Ted Harris-Brandts for their assistance with this project.

Contract grant sponsors: Scottish Rite Charitable Foundation and the Alzheimer’s Society of Canada.

Footnotes

Additional Supporting Information may be found in the online version of this article.

REFERENCES

  1. Akaike H. 1974. A new look at the statistical model identification. IEEE Trans Automat Contr 19:716–723. [Google Scholar]
  2. Ashburner J, Friston KJ. 2005. Unified segmentation. Neuroimage 26:839–851. [DOI] [PubMed] [Google Scholar]
  3. Banati RB, Newcombe J, Gunn RN, Cagnin A, Turkheimer F, Heppner F, Price G, Wegner F, Giovannoni G, Miller DH, Perkin GD, Smith T, Hewson AK, Bydder G, Kreutzberg GW, Jones T, Cuzner ML, Myers R. 2000. The peripheral benzodiazepine binding site in the brain in multiple sclerosis: quantitative in vivo imaging of microglia as a measure of disease activity. Brain 123 (Pt 11):2321–2337. [DOI] [PubMed] [Google Scholar]
  4. Cagnin A, Brooks DJ, Kennedy AM, Gunn RN, Myers R, Turkheimer FE, Jones T, Banati RB. 2001. In-vivo measurement of activated microglia in dementia. Lancet 358:461–467. [DOI] [PubMed] [Google Scholar]
  5. Chen MK, Guilarte TR. 2008. Translocator protein 18 kDa (TSPO): Molecular sensor of brain injury and repair. Pharmacol Ther 118: 1–17. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Cosenza-Nashat M, Zhao ML, Suh HS, Morgan J, Natividad R, Morgello S, Lee SC. 2009. Expression of the translocator protein of 18 kDa by microglia, macrophages and astrocytes based on immunohistochemical localization in abnormal human brain. Neuropathol Appl Neurobiol 35:306–328. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Debruyne JC, Versijpt J, Van Laere KJ, De Vos F, Keppens J, Strijckmans K, Achten E, Slegers G, Dierckx RA, Korf J, De Reuck JL. 2003. PET visualization of microglia in multiple sclerosis patients using [11C]PK11195. Eur J Neurol 10:257–264. [DOI] [PubMed] [Google Scholar]
  8. Denes A, Vidyasagar R, Feng J, Narvainen J, McColl BW, Kauppinen RA, Allan SM. 2007. Proliferating resident microglia after focal cerebral ischaemia in mice. J Cereb Blood Flow Metab 27:1941–1953. [DOI] [PubMed] [Google Scholar]
  9. DiPatre PL, Gelman BB. 1997. Microglial cell activation in aging and Alzheimer disease: Partial linkage with neurofibrillary tangle burden in the hippocampus. J Neuropathol Exp Neurol 56:143–149. [DOI] [PubMed] [Google Scholar]
  10. Endres CJ, Pomper MG, James M, Uzuner O, Hammoud DA, Watkins CC, Reynolds A, Hilton J, Dannals RF, Kassiou M. 2009. Initial evaluation of 11C-DPA-713, a novel TSPO PET ligand, in humans. J Nucl Med 50:1276–1282. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Folstein MF, Folstein SE, McHugh PR. 1975. “Mini-mental state”. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 12:189–198. [DOI] [PubMed] [Google Scholar]
  12. Fujimura Y, Ikoma Y, Yasuno F, Suhara T, Ota M, Matsumoto R, Nozaki S, Takano A, Kosaka J, Zhang MR, Nakao R, Suzuki K, Kato N, Ito H. 2006. Quantitative analyses of 18F-FEDAA1106 binding to peripheral benzodiazepine receptors in living human brain. J Nucl Med 47:43–50. [PubMed] [Google Scholar]
  13. Fujita M, Imaizumi M, Zoghbi SS, Fujimura Y, Farris AG, Suhara T, Hong J, Pike VW, Innis RB. 2008. Kinetic analysis in healthy humans of a novel positron emission tomography radioligand to image the peripheral benzodiazepine receptor, a potential biomarker for inflammation. Neuroimage 40:43–52. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Guilarte TR, Kuhlmann AC, O’Callaghan JP, Miceli RC. 1995. Enhanced expression of peripheral benzodiazepine receptors in trimethyltin-exposed rat brain: A biomarker of neurotoxicity. Neurotoxicology 16:441–450. [PubMed] [Google Scholar]
  15. Gulyas B, Vas A, Toth M, Takano A, Varrone A, Cselenyi Z, Schain M, Mattsson P, Halldin C. 2011. Age and disease related changes in the translocator protein (TSPO) system in the human brain: Positron emission tomography measurements with [11C]vinpocetine. Neuroimage 56:1111–1121. [DOI] [PubMed] [Google Scholar]
  16. Guo Q, Colasanti A, Owen DR, Onega M, Kamalakaran A, Bennacef I, Matthews PM, Rabiner EA, Turkheimer FE, Gunn RN. 2013. Quantification of the specific translocator protein signal of 18F-PBR111 in healthy humans: A genetic polymorphism effect on in vivo binding. J Nucl Med 54:1915–1923. [DOI] [PubMed] [Google Scholar]
  17. Hayes GM, Woodroofe MN, Cuzner ML. 1987. Microglia are the major cell type expressing MHC class II in human white matter. J Neurol Sci 80:25–37. [DOI] [PubMed] [Google Scholar]
  18. Huang H, Fan X, Weiner M, Martin-Cook K, Xiao G, Davis J, Devous M, Rosenberg R, Diaz-Arrastia R. 2012. Distinctive disruption patterns of white matter tracts in Alzheimer’s disease with full diffusion tensor characterization. Neurobiol Aging 33: 2029–2045. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Innis RB, Cunningham VJ, Delforge J, Fujita M, Gjedde A, Gunn RN, Holden J, Houle S, Huang SC, Ichise M, Iida H, Ito H, Kimura Y, Koeppe RA, Knudsen GM, Knuuti J, Lammertsma AA, Laruelle M, Logan J, Maguire RP, Mintun MA, Morris ED, Parsey R, Price JC, Slifstein M, Sossi V, Suhara T, Votaw JR, Wong DF, Carson RE. 2007. Consensus nomenclature for in vivo imaging of reversibly binding radioligands. J Cereb Blood Flow Metab 27:1533–1539. [DOI] [PubMed] [Google Scholar]
  20. Kerchner GA, Racine CA, Hale S, Wilheim R, Laluz V, Miller BL, Kramer JH. 2012. Cognitive processing speed in older adults: Relationship with white matter integrity. PLoS One 7:e50425. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Kreisl WC, Fujita M, Fujimura Y, Kimura N, Jenko KJ, Kannan P, Hong J, Morse CL, Zoghbi SS, Gladding RL, Jacobson S, Oh U, Pike VW, Innis RB. 2010. Comparison of [(11)C]-(R)-PK 11195 and [(11)C]PBR28, two radioligands for translocator protein (18 kDa) in human and monkey: Implications for positron emission tomographic imaging of this inflammation biomarker. Neuroimage 49:2924–2932. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Kreisl WC, Jenko KJ, Hines CS, Hyoung Lyoo C, Corona W, Morse CL, Zoghbi SS, Hyde T, Kleinman JE, Pike VW, McMahon FJ, Innis RB. 2013. A genetic polymorphism for translocator protein 18 kDa affects both in vitro and in vivo radioligand binding in human brain to this putative biomarker of neuroinflammation. J Cereb Blood Flow Metab 33:53–58. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Kumar A, Muzik O, Shandal V, Chugani D, Chakraborty P, Chugani HT. 2012. Evaluation of age-related changes in translocator protein (TSPO) in human brain using (11)C-[R]-PK11195 PET. J Neuroinflammation 9:232. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Leenders KL, Perani D, Lammertsma AA, Heather JD, Buckingham P, Healy MJ, Gibbs JM, Wise RJ, Hatazawa J, Herold S, et al. 1990. Cerebral blood flow, blood volume and oxygen utilization. Normal values and effect of age. Brain 113 (Pt 1): 27–47. [DOI] [PubMed] [Google Scholar]
  25. Logan J, Fowler JS, Volkow ND, Ding YS, Wang GJ, Alexoff DL. 2001. A strategy for removing the bias in the graphical analysis method. J Cereb Blood Flow Metab 21:307–320. [DOI] [PubMed] [Google Scholar]
  26. Lu PH, Lee GJ, Tishler TA, Meghpara M, Thompson PM, Bartzokis G. 2013. Myelin breakdown mediates age-related slowing in cognitive processing speed in healthy elderly men. Brain Cogn 81: 131–138. [DOI] [PubMed] [Google Scholar]
  27. Marazziti D, Pancioli-Guadagnucci ML, Rotondo A, Giannaccini G, Martini C, Lucacchini A, Cassano GB. 1994. Age-related changes in peripheral benzodiazepine receptors of human platelets. J Psychiatry Neurosci 19:136–139. [PMC free article] [PubMed] [Google Scholar]
  28. MicroMath. 1995. MicroMath Scientist Handbook Rev 7EEF. MicroMath: Salt Lake City, pp 467 [Google Scholar]
  29. Miller KR, Streit WJ. 2007. The effects of aging, injury and disease on microglial function: A case for cellular senescence. Neuron Glia Biol 3:245–253. [DOI] [PubMed] [Google Scholar]
  30. Mizrahi R, Rusjan PM, Kennedy J, Pollock B, Mulsant B, Suridjan I, De Luca V, Wilson AA, Houle S. 2012. Translocator protein (18 kDa) polymorphism (rs6971) explains in-vivo brain binding affinity of the PET radioligand [(18)F]-FEPPA. J Cereb Blood Flow Metab 32:968–972. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Mori S, Oishi K, Jiang H, Jiang L, Li X, Akhter K, Hua K, Faria AV, Mahmood A, Woods R, Toga AW, Pike GB, Neto PR, Evans A, Zhang J, Huang H, Miller MI, van Zijl P, Mazziotta J. 2008. Stereotaxic white matter atlas based on diffusion tensor imaging in an ICBM template. Neuroimage 40:570–582. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Oh U, Fujita M, Ikonomidou VN, Evangelou IE, Matsuura E, Harberts E, Fujimura Y, Richert ND, Ohayon J, Pike VW, Zhang Y, Zoghbi SS, Innis RB, Jacobson S. 2011. Translocator protein PET imaging for glial activation in multiple sclerosis. J Neuroimmune Pharmacol 6:354–361. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Ouchi Y, Yoshikawa E, Sekine Y, Futatsubashi M, Kanno T, Ogusu T, Torizuka T. 2005. Microglial activation and dopamine terminal loss in early Parkinson’s disease. Ann Neurol 57:168–175. [DOI] [PubMed] [Google Scholar]
  34. Owen DR, Gunn RN, Rabiner EA, Bennacef I, Fujita M, Kreisl WC, Innis RB, Pike VW, Reynolds R, Matthews PM, Parker CA. 2011. Mixed-affinity binding in humans with 18-kDa translocator protein ligands. J Nucl Med 52:24–32. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Owen DR, Howell OW, Tang SP, Wells LA, Bennacef I, Bergstrom M, Gunn RN, Rabiner EA, Wilkins MR, Reynolds R, Matthews PM, Parker CA. 2010. Two binding sites for [3H]PBR28 in human brain: implications for TSPO PET imaging of neuroinflammation. J Cereb Blood Flow Metab 30:1608–1618. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Owen DR, Yeo AJ, Gunn RN, Song K, Wadsworth G, Lewis A, Rhodes C, Pulford DJ, Bennacef I, Parker CA, Stjean PL, Cardon LR, Mooser VE, Matthews PM, Rabiner EA, Rubio JP. 2012. An 18-kDa translocator protein (TSPO) polymorphism explains differences in binding affinity of the PET radioligand PBR28. J Cereb Blood Flow Metab 32:1–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Peters A. 2002. The effects of normal aging on myelin and nerve fibers: A review. J Neurocytol 31:581–593. [DOI] [PubMed] [Google Scholar]
  38. Rusjan P, Mamo D, Ginovart N, Hussey D, Vitcu I, Yasuno F, Tetsuya S, Houle S, Kapur S. 2006. An automated method for the extraction of regional data from PET images. Psychiatry Res 147: 79–89. [DOI] [PubMed] [Google Scholar]
  39. Rusjan PM, Wilson AA, Bloomfield PM, Vitcu I, Meyer JH, Houle S, Mizrahi R. 2011. Quantitation of translocator protein binding in human brain with the novel radioligand [(18)F]-FEPPA and positron emission tomography. J Cereb Blood Flow Metab 31:1807–1816. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Salat DH, Tuch DS, Greve DN, van der Kouwe AJ, Hevelone ND, Zaleta AK, Rosen BR, Fischl B, Corkin S, Rosas HD, Dale AM. 2005. Age-related alterations in white matter microstructure measured by diffusion tensor imaging. Neurobiol Aging 26:1215–1227. [DOI] [PubMed] [Google Scholar]
  41. Schuitemaker A, van der Doef TF, Boellaard R, van der Flier WM, Yaqub M, Windhorst AD, Barkhof F, Jonker C, Kloet RW, Lammertsma AA, Scheltens P, van Berckel BN. 2010. Microglial activation in healthy aging. Neurobiol Aging 33:1067–1072. [DOI] [PubMed] [Google Scholar]
  42. Sexton CE, Kalu UG, Filippini N, Mackay CE, Ebmeier KP. 2011. A meta-analysis of diffusion tensor imaging in mild cognitive impairment and Alzheimer’s disease. Neurobiol Aging 32: 2322.e15–18. [DOI] [PubMed] [Google Scholar]
  43. Sheng JG, Mrak RE, Griffin WS. 1998. Enlarged and phagocytic, but not primed, interleukin-1 alpha-immunoreactive microglia increase with age in normal human brain. Acta Neuropathol 95: 229–234. [DOI] [PubMed] [Google Scholar]
  44. Simpson JE, Ince PG, Higham CE, Gelsthorpe CH, Fernando MS, Matthews F, Forster G, O’Brien JT, Barber R, Kalaria RN, Brayne C, Shaw PJ, Stoeber K, Williams GH, Lewis CE, Wharton SB. 2007. Microglial activation in white matter lesions and nonlesional white matter of ageing brains. Neuropathol Appl Neurobiol 33:670–683. [DOI] [PubMed] [Google Scholar]
  45. Sloane JA, Hollander W, Moss MB, Rosene DL, Abraham CR. 1999. Increased microglial activation and protein nitration in white matter of the aging monkey. Neurobiol Aging 20:395–405. [DOI] [PubMed] [Google Scholar]
  46. Streit WJ, Sammons NW, Kuhns AJ, Sparks DL. 2004. Dystrophic microglia in the aging human brain. Glia 45:208–212. [DOI] [PubMed] [Google Scholar]
  47. Studholme C, Hill DL, Hawkes DJ. 1997. Automated threedimensional registration of magnetic resonance and positron emission tomography brain images by multiresolution optimization of voxel similarity measures. Med Phys 24:25–35. [DOI] [PubMed] [Google Scholar]
  48. Sullivan EV, Adalsteinsson E, Pfefferbaum A. 2006. Selective age-related degradation of anterior callosal fiber bundles quantified in vivo with fiber tracking. Cereb Cortex 16:1030–1039. [DOI] [PubMed] [Google Scholar]
  49. Sullivan EV, Rohlfing T, Pfefferbaum A. 2010. Quantitative fiber tracking of lateral and interhemispheric white matter systems in normal aging: relations to timed performance. Neurobiol Aging 31:464–481. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Suridjan I, Rusjan PM, Voineskos AN, Selvanathan T, Setiawan E, Strafella AP, Wilson AA, Meyer JH, Houle S, Mizrahi R. 2014. Neuroinflammation in healthy aging: A PET study using a novel Translocator Protein 18kDa (TSPO) radioligand, [(18)F]-FEPPA. Neuroimage 84:868–875. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Thiel A, Radlinska BA, Paquette C, Sidel M, Soucy JP, Schirrmacher R, Minuk J. 2010. The temporal dynamics of poststroke neuroinflammation: A longitudinal diffusion tensor imaging-guided PET study with 11C-PK11195 in acute subcortical stroke. J Nucl Med 51:1404–1412. [DOI] [PubMed] [Google Scholar]
  52. Venneti S, Lopresti BJ, Wang G, Hamilton RL, Mathis CA, Klunk WE, Apte UM, Wiley CA. 2009. PK11195 labels activated microglia in Alzheimer’s disease and in vivo in a mouse model using PET. Neurobiol Aging 30:1217–1226. [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Venneti S, Wang G, Nguyen J, Wiley CA. 2008. The positron emission tomography ligand DAA1106 binds with high affinity to activated microglia in human neurological disorders. J Neuropathol Exp Neurol 67:1001–1010. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Voineskos AN, Rajji TK, Lobaugh NJ, Miranda D, Shenton ME, Kennedy JL, Pollock BG, Mulsant BH. 2012. Age-related decline in white matter tract integrity and cognitive performance: A DTI tractography and structural equation modeling study. Neurobiol Aging 33:21–34. [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Wilson AA, Garcia A, Parkes J, McCormick P, Stephenson KA, Houle S, Vasdev N. 2008. Radiosynthesis and initial evaluation of [18F]-FEPPA for PET imaging of peripheral benzodiazepine receptors. Nucl Med Biol 35:305–314. [DOI] [PubMed] [Google Scholar]
  56. Yasuno F, Ota M, Kosaka J, Ito H, Higuchi M, Doronbekov TK, Nozaki S, Fujimura Y, Koeda M, Asada T, Suhara T. 2008. Increased binding of peripheral benzodiazepine receptor in Alzheimer’s disease measured by positron emission tomography with [11C]DAA1106. Biol Psychiatry 64:835–841. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Figure Legends
NIHMS926793-supplement-Supplementary_Figure_Legends.doc (1.2MB, doc)

RESOURCES