11C-Acetate PET Imaging in Patients with Multiple Sclerosis

Background Activation of glial cells is a cardinal feature in multiple sclerosis (MS) pathology, and acetate has been reported to be selectively uptaken by astrocytes in the CNS. The aim of this study was to investigate the efficacy of PET with 11C-acetate for MS diagnosis. Materials and Methods Six patients with relapsing-remitting MS and 6 healthy volunteers (HV) were enrolled. The 11C-acetate brain uptake on PET was measured in patients with MS and HV. Volume-of-interest analysis of cerebral gray and white matter based on the segmentation technique for co-registered MRI and voxel-based statistical parametric analysis were performed. Correlation between 11C-acetate uptake and the lesion number in T1- and T2- weighted MR images were also assessed. Results The standardized uptake value (SUV) of 11C-acetate was increased in both white and gray matter in MS patients compared to HV. Voxel-based statistical analysis revealed a significantly increased SUV relative to that in the bilateral thalami (SUVt) in a broad area of white matter, particularly in the subcortical white matter of MS patients. The numbers of T2 lesions and T1 black holes were significantly correlated with SUV of 11C-acetate in white and gray matter. Conclusions The 11C-acetate uptake significantly increased in MS patients and correlated to the number of MRI lesions. These preliminary data suggest that 11C-acetate PET can be a useful clinical examination for MS patients.


Introduction
Multiple sclerosis (MS) is an inflammatory demyelinating autoimmune disease of the CNS [1]. Although MRI is recognized as the most informative surrogate marker [2], the diagnostic value of MRI in MS remains insufficient [3]. Glial activation is a key feature in the neuroinflammatory MS pathology, and glial activation from the early phase of MS is suggested by a MRS study [4]. Microglial activation has also been shown in PET studies [5,6]. However, astrocyte activation in MS has not been evaluated in vivo due to the lack of an appropriate radioligand, despite the astrocytosis observed from the early phase of disease and the important role potentially played by astrocytes [4,7,8].
Acetate is converted into fatty acids by the key enzyme acetyl-CoA synthase and metabolized in the citric acid cycle. 11 C-acetate has been used as a tracer to evaluate cardiac oxidative metabolism [9] and later used as a PET biomarker in patients with renal cell carcinoma, hepatocellular carcinoma, prostate cancer, and multiple myeloma [10,11,12,13]. In the CNS, 11 C-acetate PET has proven useful for the diagnosis of astrocytoma [14] because acetate is preferentially absorbed into astrocytes by the monocarboxylate transporter (MCT) [15,16]. Notably, the expression of MCT is increased in MS brains [17]. Therefore, we surmised that 11 Cacetate PET could be a useful diagnostic tool in combination with MRI, and we investigated the utility of 11 C-acetate PET for the diagnosis of MS and evaluated the astrocyte activity in the MS brain.

Subjects and clinical evaluation
Six patients with relapsing-remitting MS were evaluated. All patients were in the remission phase. Disability was assessed based on the Expanded Disability Status Scale (EDSS) [18]. Six healthy volunteers (HV) served as normal controls (Table 1). This study was approved by the Ethics Committee of Osaka University Hospital, and written informed consent was obtained from each participant.

PET
PET was performed using a SET-3000 GCT/X scanner (Shimadzu Corp., Kyoto, Japan). 11 C-acetate was synthesized by carbonation of Grignard reagent followed by acid hydrolysis. 11 Ccarbon dioxide reacted with methylmagnesium bromide followed by hydrolysis with hydrochloric acid to yield 11 C acetic acid [19]. The radio chemical purity was greater than 98%. A total of 370 MBq of the tracer was administered intravenously, and a 20min emission acquisition was initiated 20 min later. PET images were obtained in a 3-D mode. The images were reconstructed using a filtered-back projection method after 3D Gaussian smoothing with a 6-mm full width at half maximum (FWHM). Scatter correction was performed using a hybrid dual-energy window method combined with a convolution-subtraction method, and the true scatter-free component of the standard photopeak window was estimated sonographically. All PET images were reconstructed in 2566256699 anisotropic voxels, with each voxel measuring 16162.6 mm.

Data analysis
Whole brain VOI analysis. All procedures were performed using a personal computer (DELL Precision T7400; DELL Inc., Round Rock, TX, USA) running on Microsoft Windows 7 (Microsoft Corp., Redmond, WA, USA). The 3D T1-weighted MRI scan was re-sliced in the native space of each subject using a 1.061.061.0 mm voxel size. The results were first categorized as GM, WM, and CSF, then spatially normalized using the unified model [20] of Statistical Parametric Mapping (SPM) 8 (Wellcome Department of Imaging Neuroscience: http://www.fil.ion.ucl.ac. uk/spm/) according to the optimized voxel-based morphometry (VBM) protocol [21]. This generated both spatial normalization matrices and inverse spatial normalization matrices. The resulting normalized GM map was transformed into native space using an inverse spatial normalization matrix. To generate VOI for GM and/or WM, binary mask images for the GM and/or WM were created using the segmented images in the native space of each subject. The binary mask image boundary was set at 35% of the maximum GM or WM concentration as described in previous studies [22,23].
The 11 C-acetate PET images were co-registered with the resliced 3D T1-weighted MRI using the SPM8 registration function based on the mutual information. The co-registration precision was inspected with the ''Check Registration'' tool in SPM8. Then, the co-registered PET images were spatially transformed using normalization and/or inverse normalization matrices identical to those generated in the previously described Table 1. Patient data and demographics. automatic segmentation process. The 11 C -acetate uptake in the GM and WM VOI was analyzed using the binary masks within the native space. To minimize contamination from the spill-in effect of adjacent brain segments, the spill-in-free VOIs of GM and WM were generated by the VOI erosion process. First, the binary masks were blurred by convolution using the point spread function of the PET scanner (presumably a simple isotropic Gaussian kernel with a FWHM of 8 mm). The spill-in-free gray matter maskĜ G is expressed as follows:Ĝ where x is a voxel, G is the gray matter binary mask, andW W (x) is the blurred image of the white matter binary mask (i.e. spill-in fraction from the white matter to the voxel x). The spill-in-free white matter mask was also constructed as described above. Spillin from CSF was assumed as zero. VOI analysis for 11 C-acetate uptake using spill-in-free GM and WM masks was also performed ( Figure S1). The relative standardized uptake value (SUVt) served as the uptake indicator for analysis; the regional standardized uptake value (SUV) was divided by the mean SUV within the bilateral thalami of each subject. The Mann-Whitney U test was performed to determine significance of 11 C -acetate SUVt differences between MS and HV. The significance level was designated at p,0.05.
Voxel-based statistical analysis. Voxel-based whole brain SUVt in the MS and HV groups was compared using Statistical Parametric Mapping (SPM) 8 (Wellcome Department of Imaging Neuroscience). The spatially normalized PET images were smoothed using a 12-mm FWHM isotropic Gaussian kernel, which conditions the residuals to conform more closely to the Gaussian random field model underlying the statistical adjustment of the p values. The SPM statistical model used voxel-by-voxel ''two-sample T-test with covariates,'' which designated age as a nuisance variable in order to detect voxels showing a significant age-adjusted SUVt difference between the MS and HV groups.
11C-acetate uptake and the MR images correlation assessment. The T2 and T1 black hole lesions were independently recorded visually by three observers. T1 black holes were defined as visible hypointense regions on the T1-weighted images coincident with a high signal intense region on the T2-weighted images. Each MRI mask image was divided into its hemispheres to create the hemispheric VOIs. Pearson product moment correlation analyses were performed to assess the association between the number of MRI lesions, and the SUV of 11 C-acetate was accessed from the hemispheric VOIs of the GM and WM. Statistical significance was designated at p,0.05.

Statistical analysis
The data in Fig 1 and Table S1 were analyzed using the Mann-Whitney U test. ANCOVA was used to assess the differences between age-adjusted groups illustrated in Fig 2B-G, and Pearson product moment correlation analyses were performed for data in Fig 3, SPSS 14.J was used for statistical analysis.

Results
VOI analysis of the 11 C-acetate SUV revealed that the mean SUV was higher in the MS patients than in the HV in all regions assessed (Fig 1A). To evaluate the regional distribution of 11 Cacetate uptake independent of physiologic variation in the whole brain, we calculated the relative uptake value (SUVt), which is the regional SUV divided by the mean SUV within the bilateral thalami of each participant (Fig 1B). The thalamus served as the reference region because it is rarely involved in MS pathology [24], and the SUV difference between the thalamus of HV and MS patients was the least among brain regions, as shown in Fig 1A. Each regional SUVt in the MS patients were increased particularly in the parietal, occipital, and insula regions.
Spatially normalized group mean images of 11 C-acetate SUVt automatically segmented based on MRI showed increased uptake in both WM and GM in MS patients (Fig 2A). The SUVt of MS patients was significantly higher than that of HV in both WM (p = 0.002) and GM (p = 0.001). In addition, all six MS patients had a significantly higher WM/GM SUV ratio than the six HV (p = 0.009) (Fig 2B-D). This trend was consistently observed even after accommodating spill-in effect from adjacent brain segments (Fig 2E-G). Collectively, the 11 C-acetate uptake significantly increased in both the WM and GM of MS patients, and this increase was more predominant in WM. The whole brain SPM analysis revealed a significant increase in SUVt of voxel cluster in MS patients compared to HV, primarily in the subcortical frontal, parietal, and occipital regions; no voxels showed a significantly lower SUVt in MS patients compared to HV (Fig 2H).
The voxel-based t-statistic for the WM tracts showed a significantly increased mean T-score, predominantly in the superior longitudinal fasciculus, posterior thalamic radiation, and sagittal stratum, with the highest local maximum T-score in the corpus callosum (Table S2).
We then assessed potential correlation between 11 C-acetate SUV and MRI brain lesions. The mean SUV in WM was significantly correlated to the number of T1 black holes (R 2 = 0.5059, p = 0.009) and T2 lesions (R 2 = 0.4594, p = 0.015) (Fig 3A, B). The mean SUV in GM also correlated to the number of T1 black holes (R 2 = 0.4088, p = 0.025) and T2 lesions (R 2 = 0.3952, p = 0.029) (Fig 3C, D). The correlation to the EDSS score and disease duration did not reach statistical significance.

Discussion
There have been few studies imaging astrocytes in vivo using 11 C-acetate PET. In MS, astrocyte proliferation [25] and formation of scars composing a dense network of hypertrophic cells are characteristics of the MS histopathology [8]. An increased MCT expression in astrocytes within MS lesions was recently shown by immunohistochemical analysis [17], which suggest an increase in astrocyte metabolism. However, latent autoantibodymediated astrocyte damage [26] supposedly decreases the metabolic activity, and therefore, the metabolic activity of astrocytes in MS brains remains undetermined. In this study, we observed a significantly increased brain uptake of the radioligand 11 C-acetate in MS patients. Our study revealed for the first time that astrocytes are generally activated in MS brains based on the acetate metabolism.
Representative studies showed that a higher value in the kinetic parameter, which indicates the washout level of 11 C-acetate, reflects the astrocyte reactivity in normal rats and healthy humans [27]. In MS, however, compared to HV, the pathologic changes in the severity of 11 C-acetate accumulation may be much more prominent than the changes related to physiologic activation in healthy humans. Therefore, a slight increase in the washout speed may be inapparent in the PET SUV in MS. Furthermore, because the perfusion in the normal appearing white matter decreased in MS [28], the increase in 11 C-acetate uptake by static PET may be underestimated due to a reduced CBF in MS.
11C-Acetate PET for Multiple Sclerosis  The increased uptake was more pronounced in the WM, although a significant increase was observed in both the WM and GM. A significantly increased uptake was observed primarily within the subcortical WM on the voxel-based statistical analysis (Fig. 2H). On the voxel-based statistical analysis of the WM tracts, the distribution of the increased acetate uptake was similar to that in regions of axonal damage in DTI studies (Table S2). Recent voxel-and tract-based analyses in DTI studies revealed widespread damage to the subcortical WM, particularly in the sagittal stratum, corpus callosum, posterior thalamic radiation, and corona radiata [29]. These data suggested that the region-dependent increased acetate uptake was induced by the reactive astrocyte coexisting with heterogeneously dispersed MS lesions detected in DTI studies (Fig. 2H and Table S2). Although inflammatory WM demyelination detected by conventional MRI is a cardinal feature of MS, pathologic changes exist even in normal appearing WM and GM [30]. Astrocyte pathology precedes demyelination in an animal model [31]; astrocyte hypertrophy occurs at the leading edge of acute MS lesions, followed later by astrocytic scarring [8]. Thus, the altered astrocyte activation is presumably involved in MS pathophysiology [4,7,32]. Correlation between the radial diffusivity quantified by DTI and T1 black hole formation are recognized markers of axonal loss and tissue destruction [33,34]. In the present study, the strongest correlation was detected between the mean SUV in WM and the T1 black hole number, suggesting that the mean SUV may correlate with axonal damage. The mean SUV in GM also increased and correlated with the number of MRI lesions, suggesting cortical astrocyte involvement in MS pathology. Cortical involvement and subsequent cognitive decline occur in approximately half of MS patients [35]. However, little information exists on the pathophysiologic involvement of cortical astrocytes [36]. Normally, astrocytes supply lactate to neurons for oxidation [37], and metabolic dysfunction of neurons and glial cell activation likely occurs in the MS brain [25]. Moreover, astrocytes are associated with preclinical axonal damage in an animal model of MS [38]. These results suggest that the increased 11 C-acetate uptake within GM may reflect astrocyte-associated cortical damage in MS.
The present study has a few limitations. First, 11 C-acetate uptake in MS plaques was not assessed separately because most plaques were so small that a partial volume effect caused by the relatively low resolution of PET was inevitable. Second, the analysis was performed on static PET data instead of kinetic parameters. In the present study, the data acquired between 20 to 40 min after tracer administration were summed to build static uptake images because the time activity curve stabilized after 20 min (data not shown). Regional uptake distribution may be contaminated by the dispersion of radioactive metabolites. However, in our study, 1-11 C-acetate was used, and its dispersion of labeled metabolites was the smallest among the various types of acetate tracers [39,40]. In addition, because almost all the tracer was first absorbed through MCT-1 expressed within astrocytes according to their reactivity, the summed radioactivity is thought to reflect the first uptake of 11 C-acetate and its subsequent 11C-Acetate PET for Multiple Sclerosis PLOS ONE | www.plosone.org metabolism by reactive astrocytes. Finally, because the mean age was higher in the control group than in the MS group, we used the ANCOVA to assess the differences among the age-adjusted SUVt. Although the mean age of MS patients was generally lower than that of the healthy volunteers, age did not significantly affect the increased uptake of 11 C-acetate in MS patients.

Conclusions
The present study suggests that the pathologic white matter changes in patients with MS can be detected by non-invasive static 11 C-acetate PET, which may be an effective MS diagnostic tool. Development of clinically applicable monocarbonic acid tracers labeled with longer half-life radioactive nuclides are needed, as are further studies enrolling more participants, including those in the early and relapse phases. Figure S1 Binary mask imaging parameters for VOI analysis. The scheme of VOI analysis is described. A: 11   S1 Relative 11 C-acetate biodistribution in the CNS. The mean SUVt of each lesion in the CNS was analyzed and for group comparison between HV and MS patients, the Mann-Whitney U test was performed. (DOC) Table S2 Regional T-scores from voxel-based statistical comparison in WM. Voxel-based statistical comparison in white matter tracts was performed. The positive T-scores indicate an increased 11 C-acetate uptake in the MS patients compared to the HV. (DOC)

Supporting Information
Text S1 Supplementary methods. Methods for ''Voxelbased statistical analysis for WM tracts'' are described with references. (DOC)