Distinct dynamic profiles of microglial activation are associated with progression of Alzheimer's disease

Distinct dynamic profiles of microglial activation are associated with progression of Alzheimer's... Abstract Although brain neuroinflammation may play an instrumental role in the pathophysiology of Alzheimer’s disease, its actual impact on disease progression remains controversial, being reported as either detrimental or protective. This work aimed at investigating the temporal relationship between microglial activation and clinical progression of Alzheimer’s disease. First, in a large cohort of patients with Alzheimer’s disease we analysed the predictive value of microglial activation assessed by 18F-DPA-714 PET imaging on functional, cognitive and MRI biomarkers outcomes after a 2-year follow-up. Second, we analysed the longitudinal progression of 18F-DPA-714 binding in patients with Alzheimer’s disease by comparison with controls, and assessed its influence on clinical progression. At baseline, all participants underwent a clinical assessment, brain MRI, 11C-PiB, 18F-DPA-714 PET imaging and TSPO genotyping. Participants were followed-up annually for 2 years. At the end of the study, subjects were asked to repeat a second 18F-DPA-714-PET imaging. Initial 18F-DPA-714 binding was higher in prodromal (n = 33) and in demented patients with Alzheimer’s disease (n = 19) compared to controls (n = 17). After classifying patients into slow and fast decliners according to functional (Clinical Dementia Rating change) or cognitive (Mini-Mental State Examination score decline) outcomes, we found a higher initial 18F-DPA-714 binding in slow than fast decliners. Negative correlations were observed between initial 18F-DPA-714 binding and the Clinical Dementia Rating Sum of Boxes score increase, the MMSE score loss and the progression of hippocampal atrophy. This suggests that higher initial 18F-DPA-714 binding is associated with better clinical prognosis. Twenty-four patients with Alzheimer’s disease and 15 control subjects performed a second DPA-PET. We observed an increase of 18F-DPA-714 in patients with Alzheimer’s disease as compared with controls (mean 13.2% per year versus 4.2%) both at the prodromal (15.8%) and at the demented stages (8.3%). The positive correlations between change in 18F-DPA-714 binding over time and the three clinical outcome measures (Clinical Dementia Rating, Mini-Mental State Examination, hippocampal atrophy) suggested a detrimental effect on clinical Alzheimer’s disease progression of increased neuroinflammation after the initial PET examination, without correlation with PiB-PET uptake at baseline. High initial 18F-DPA-714 binding was correlated with a low subsequent increase of microglial activation and favourable clinical evolution, whereas the opposite profile was observed when initial 18F-DPA-714 binding was low, independently of disease severity at baseline. Taken together, our results support a pathophysiological model involving two distinct profiles of microglial activation signatures with different dynamics, which differentially impact on disease progression and may vary depending on patients rather than disease stages. Alzheimer’s disease, inflammation, microglia, PET imaging, neuroprotection Introduction Specific protein inclusions and/or aggregates define most neurodegenerative diseases at the pathological level. In Alzheimer’s disease, beyond the abnormal aggregation of amyloid-β peptide and hyperphosphorylated tau proteins, the pathology has a chronic neuroinflammatory component that involves activation of microglia and astrocytes (Heneka et al., 2015). Misfolded and aggregated proteins bind to pattern recognition receptors expressed by microglia and astroglia, and trigger an innate immune response characterized by the release of inflammatory mediators and/or clearance of pathological proteins. Although accumulating evidence indicate that such chronic neuroinflammatory responses play an instrumental role in the disease pathophysiology, it remains unclear whether they are protective, damaging or both depending on disease stages. A marked accumulation and activation of microglia around amyloid-β plaques has been described in post-mortem brains of patients with Alzheimer’s disease and in animal models of Alzheimer’s disease-like pathology (Prokop et al., 2013). One pathophysiological model proposes that microglia are attracted towards amyloid-β deposits, which they internalize and degrade, thus playing a protective role by promoting clearance of amyloid-β from the brain. In later disease stages, microglia may lose this beneficial effect and acquire a ‘toxic’ phenotype as a consequence of chronic activation and continued production of proinflammatory mediators, in a process that may occur independently of amyloid pathology (Prokop et al., 2013; Michaud and Rivest, 2015). In humans, brain neuroinflammation has been investigated using PET imaging of the 18-kDa translocator protein (TSPO) (Kreisl et al., 2013), which is considered as a marker of microglial activation. However, the studies using the first (11C-PK11195) or the second generation of TSPO radiotracers showed somewhat contradictory results (Lagarde et al., 2017). Varying extents and levels of microglial activation have been identified in prodromal Alzheimer’s disease (Okello et al., 2009; Wiley et al., 2009; Kreisl et al., 2013; Fan et al., 2015a; Hamelin et al., 2016). A relationship was reported between microglial activation and cortical amyloid load (Hamelin et al., 2016; Parbo et al., 2017), but a recent publication including patients with typical and atypical Alzheimer’s disease showed that microglial activation is closely associated with markers of neurodegeneration, suggesting a potential interplay with tau pathology (Kreisl et al., 2017). In addition, no difference in 18F-DPA-714 binding was found between early and late onset of Alzheimer’s disease in a larger sample of patients (Hamelin et al., 2016). Finally, only few studies investigated the predictive value of microglial activation in Alzheimer’s disease progression, and our previous work suggested an early protective role in the largest cohort to date (Hamelin et al., 2016). Only three longitudinal studies of microglial activation in Alzheimer’s disease have been reported up to now, the interpretation of which must be taken cautiously due to small sample sizes. Two studies using 11C-PK11195 (Fan et al., 2015b, 2017) showed a surprising reduction of microglial activation in eight patients with mild cognitive impairment with different amyloidosis status, whereas increased 11C-PK11195 binding was reported in the eight patients with Alzheimer’s disease included. Using 11C-PBR28, a mild increase of TSPO binding in 14 patients with Alzheimer’s disease was found associated with clinical worsening (Kreisl et al., 2016). In light of these results, the spectrum of microglial activation patterns and their impact on disease progression in Alzheimer’s disease remain controversial. The present work aimed at first evaluating the predictive value of microglial activation assessed by 18F-DPA-714 PET imaging in a large cohort of patients with Alzheimer’s disease, based on functional, cognitive and MRI biomarkers outcomes after a 2-year follow-up. Second, in the same cohort we aimed at investigating the longitudinal progression of 18F-DPA-714 binding in patients with Alzheimer’s disease by comparison with controls, and assess its influence on clinical progression. Materials and methods Study design and participants All participants were enrolled in the prospective longitudinal IMABio3 study (NCT01775696), which aimed at assessing the brain neuroinflammatory response in Alzheimer’s disease. The study was approved by a French Ethical Committee. All subjects provided written informed consent prior to participating. Patients with Alzheimer’s disease were included according to the following criteria: (i) progressive episodic memory impairment, characterized by a low free recall not normalized with semantic cueing (Sarazin et al., 2007; Dubois et al., 2010); (ii) absence of extrapyramidal signs; (iii) positive pathophysiological markers of Alzheimer’s disease, defined by CSF Alzheimer’s disease profile {score < 0.8, calculated with the formula amyloid-β42/[240+(1.18 T-tau)]} (de Souza et al., 2011) and amyloid Pittsburgh compound B (PiB)-PET imaging [global cortical index (GCI) > 1.45]. These conservative thresholds were used to ensure the pathophysiological diagnosis of Alzheimer’s disease. Patients with Alzheimer’s disease were defined as prodromal Alzheimer’s disease when their Clinical Dementia Rating (CDR) score was 0.5, and as demented-Alzheimer’s disease when their CDR score was ≥1. Controls were recruited according to the following criteria: (i) Mini-Mental State Examination (MMSE) score ≥ 27/30 and normal neuropsychological assessment; (ii) CDR = 0; (iii) no history of neurological or psychiatric disorders; (iv) no memory complaint or cognitive deficit, and (v) negative amyloid PiB-PET imaging (GCI < 1.4). Controls with a positive amyloid PiB-PET imaging (GCI > 1.45) were defined as amyloidosis-positive controls and analysed separately. We did not include subjects with (i) severe cortical or subcortical vascular lesions; (ii) history of autoimmune and inflammatory diseases or chronic migraines; (iii) history of psychiatric disorders; or (iv) suspicion of alcohol or drugs abuse. No subject was treated with corticoid or non-steroidal anti-inflammatory drug (NSAID) or benzodiazepine, known to interfere with TSPO. Blood samples were drawn to characterize APOE and TSPO genotypes (Hamelin et al., 2016). Based on the rs6971 polymorphism within the TSPO gene, we classified all subjects into three groups: high affinity binders, mixed affinity binders or low affinity binders. At baseline, all participants underwent the same procedure including a complete clinical and neuropsychological assessment, 3 T brain MRI, 11C-PiB and 18F-DPA-714 PET imaging (called DPA-1). Participants were then followed up annually with repeated standardized clinical and neuropsychological assessments and 3 T brain MRI, for 2 years. The last year of the study, subjects were asked to undergo a second 18F-DPA-714 PET imaging (called DPA-2), respecting a minimum of 12 months delay after DPA-1 for patients with Alzheimer’s disease and 24 months delay for controls. Among the 73 patients with Alzheimer’s disease who fulfilled all inclusion criteria, initial DPA-1 PET scan could not be carried out for seven patients because of technical reasons. Eight patients were withdrawn during the 2-year follow-up period (one intercurrent disease, two nursing home, three lost to follow-up, two deaths). Finally, 58 patients with Alzheimer’s disease (including six low affinity binders) with DPA-1 completed the 2-year clinical follow-up. Twenty-seven patients with Alzheimer’s disease underwent a second 18F-DPA-714 PET imaging (n = 18 refused to perform it, n = 3 were at a severe dementia stage making it impossible to carry out a PET acquisition; n = 10 were outside the regulatory limits of the study). PET acquisition failed for technical reasons for three of them. Finally, 24 patients with Alzheimer’s disease (18 prodromal Alzheimer’s disease and six demented Alzheimer’s disease), including three low affinity binders had both DPA-1 and DPA-2 PET scans. The mean delay between DPA-1 and DPA-2 was 1.57 ± 0.3 years. Twenty-nine controls with DPA-1 completed the 2-year clinical follow-up. Among them, eight controls were defined as amyloidosis-positive and analysed separately. A second 18F-DPA-714 PET scan was performed for 15/21 amyloidosis negative controls (n = 3 were outside the regulatory time limit; n = 3 refused DPA-2 PET scan) and 6/8 amyloidosis positive controls (n = 2 refused to perform DPA-2 PET scan). The mean delay between DPA-1 and DPA-2 in controls was 2.2 ± 0.2 years. Clinical, functional and cognitive assessment Standardized neurological and neuropsychological examination was performed annually for 2 years, including the MMSE, CDR scale, Montgomery-Asberg Depression Scale (MADRS) and a standardized cognitive battery for assessing verbal and visual episodic memory, executive functions, gesture praxis, visuo-constructive functions and language. Record of medical events and current treatment were also reported. Definitions of slow and fast decliners After 2 years of follow-up, we used two complementary clinical outcomes to define slow (SD) and fast decliners (FD). The functional (fn) progression was assessed on the basis of the CDR change: patients with Alzheimer’s disease were classified as functional slow decliners when CDR was unchanged and as functional fast decliners when CDR increased. The cognitive (co) progression was assessed by the MMSE score decline over time (cognitive slow decliners: ΔMMSE score ≤ 4 points; cognitive fast decliners: ΔMMSE > 4 points). The cut-off value of 4 points was determined according to knowledge based on longitudinal studies, which have shown that the minimal mean annual rate of progression of cognitive impairment was ∼1–2 points when assessed with the MMSE (Kinkingnéhun et al., 2008). Of note, MMSE scores at the last visit were missing for three patients with Alzheimer’s disease because of too severe cognitive decline. MRI acquisition and hippocampal volume measurements Each participant underwent a MRI examination annually during 2 years, using the same scanner at all visits for reducing equipment-related variability. 3D T1-weighted structural MRI acquisitions were obtained on a 3 T MRI scanner (Siemens Trio, 32 channel system, with a 12-channel head coil for signal reception). This sequence provided a high grey/white matter contrast-to-noise ratio and enabled excellent segmentation and accurate co-registration with PET images (Hamelin et al., 2016). Segmentation of the hippocampus was carried out using an automated method (Chupin et al., 2009). For measuring hippocampal atrophy progression and increase the sensitivity to actual change, the two MRI time points were computed together as previously described (Dubois et al., 2015). Hippocampal volumes were adjusted for head size by correcting for total intracranial volume, derived from SPM8 segmentations. We calculated the percentage of left and right hippocampal volume loss as well as the mean of left + right hippocampal volume loss after 2 years of follow-up. 11C-PiB and 18F-DPA-714 PET imaging procedure Data acquisition MRI and PET scans were performed within 4 months of each other. At baseline, 11C-PiB and 18F-DPA-714 PET scans were performed on the same day. All PET scans were performed using the same camera: a High Resolution Research Tomograph (HRRT; CTI/Siemens Molecular Imaging) (de Jong et al., 2007). A 6-min brain transmission scan was performed before injection of each radioligand using a 137Cs point source to correct the emission scan for tissue attenuation. 11C-PiB-PET (362 ± 45 MBq) and 18F-DPA-714 PET (DPA-1: 197 ± 17 MBq and DPA-2: 197 ± 12 MBq) were injected intravenously, and PET dynamic acquisitions in list mode lasted up to 90 min. All corrections (attenuation, normalization, random and scatter coincidences) were incorporated in an iterative OSEM reconstruction. The partial volume effect was corrected by directly incorporating resolution modelling (i.e. Point Spread Function modelling) inside the iterative algorithm (Sureau et al., 2008) so that no further post-correction was needed. Ten iterations using 16 subsets were used. Dynamic data were binned into 27 time frames (6 × 1 min, 7 × 2 min, 14 × 5 min). Reconstructed dynamic data were realigned for motion correction according to the process of frame-to-reference image registration in Pmod (version 3.5; PMOD Technologies Ltd.). Parametric images were created using Brainvisa software (http://brainvisa.info). Standard uptake value (SUV) parametric images were obtained by: (i) averaging late images (intervals of 40–60 min for 11C-PiB and of 60–90 min for 18F-DPA-714) (Lopresti et al., 2005; Lavisse et al., 2015); and (ii) adjusting for body weight and injected radioligand dose. The cerebellar grey matter was used as a pseudo reference region in both 11C-PiB and 18F-DPA-714 PET analyses to obtain a SUVr (de Souza et al., 2011; Hamelin et al., 2016). Volume of interest analysis The same method of anatomical segmentation was used for both 11C-PiB and 18F-DPA-714 PET images, as described previously (Hamelin et al., 2016). Briefly, an automated segmentation of grey matter was performed to each individual 3D T1-weighted MRI scan using the VBM8 package (http://dbm.neuro.uni-jena.de/vbm/) implemented in SPM8 (Institute of Neurology, London, UK; http://www.fil.ion.ucl.ac.uk/spm/). The segmented MRI scans were co-registered with both 11C-PiB and 18F-DPA-714 parametric images of the subject using a standard mutual information algorithm. The automated anatomical labelling (AAL) atlas was normalized to each individual MRI (using the deformation field extracted from VBM8). Each volume of interest was intersected with the T1 MRI grey matter mask to perform a pseudo-atrophy correction. Then, this new labelling volume was registered to the individual PET space of 11C-PiB and 18F-DPA-714 parametric images using their respective transformation extracted from the PET-MRI co-registration. Similarly, cerebellar grey matter was identified for each subject, eroded (4 mm), and used as a pseudo-reference region. Application of the AAL atlas to the PET data allowed the calculation of 11C-PiB and 18F-DPA-714 uptake in 76 anatomical regions. The volumes of interest were defined separately for the left and right hemispheres and were then pooled into greater anatomical volumes of interest, as previously described (de Souza et al., 2011). Briefly, we defined eight volumes of interest: (i) the frontal cortex; (ii) anterior cingulate; (iii) medium cingulate; (iv) posterior cingulate; (v) precuneus; (vi) parietal cortex; (vii) temporal cortex; and (viii) posterior cortex. A mean 11C-PiB and 18F-DPA-714 SUVr were obtained for each region. As a measure of global cortical burden, we calculated a 11C-PiB and 18F-DPA-714 GCI, representing the subject’s mean SUVr of the neocortical regions cited above (de Souza et al., 2011; Hamelin et al., 2016). Voxel-wise analysis We used SPM8, implemented on a MATLAB platform (Mathworks Inc.). All PET images were spatially normalized onto the standard Montreal Neurological Institute (MNI) and were smoothed with an 8 mm full-width at half-maximum Gaussian filter. In the DPA-PET predictive analysis, we used unpaired t-test comparisons with age, TSPO genotype and MMSE score as covariates to compare fast and slow decliners (P < 0.001 uncorrected). In the DPA-PET longitudinal study, we used an unpaired two-sample t-test with TSPO genotype and interval time between the two PET acquisitions as covariates for comparing Alzheimer’s disease and control groups. The threshold was set at P < 0.001 uncorrected. In addition, evolution of DPA-714 binding in subjects was assessed by using the percentage of DPA changes per year, calculated individually at the voxel level in the MNI space, as follows: (DPA−2DPA−1−1) Δt (1) Where DPA-1 and DPA-2 are DPA-714 SUVr at baseline and follow-up, respectively, and Δt the time interval between the two DPA-714 PET scans. Voxels of percentage were then compared between groups using unpaired t-test with age, TSPO genotype and initial MMSE score as covariates. Correlation analyses used multiple regression design, and TSPO genotype, age and initial MMSE score were included as covariates. For all analyses, a minimum-activated voxel threshold of 20 voxels was applied. Statistical analysis In accordance with recent reports, we pooled the mixed affinity and high affinity binders together and used the TSPO genotype as a covariate in all statistical analyses. In the DPA-PET predictive analysis, we excluded 6/58 low affinity binders with Alzheimer’s disease. The analysis of the DPA-PET longitudinal study was similarly carried out without including low affinity binders. However, as the low affinity binders could be considered as their own controls longitudinally, we verified that our results were not changed when including low affinity binders. Data were analysed using SPSS20 (SPSS Inc., Chicago, Illinois), STATISTICA 6 software (Statsoft) and R, a software environment for statistical computing and graphics (http://www.R-project.org/). Normality of distribution was tested using the Shapiro-Wilk test. Differences between groups were assessed using χ2-test, ANOVA, or Kruskall-Wallis tests when appropriate. ANCOVA, adjusted for age, TSPO genotype and initial MMSE score was used for comparing 18F-DPA-714 binding between groups at baseline (DPA-1). For follow-up comparisons and comparisons of percentage of change, the interval time between both 18F-DPA-714 scans was also added as a covariate (see methodological considerations below for more details). We used one-tailed linear partial correlation analyses with age, TSPO genotype and initial MMSE score as covariates. We used one-tailed tests as the correlation analyses followed comparisons between groups, which gave us a hypothesis regarding the direction of the correlation. Results For clarity, the term ‘DPA-PET predictive analysis’ will refer to assessing how initial microglial activation (DPA-1) is associated with the rapidity of Alzheimer’s disease progression after 2 years of clinical follow-up. The term ‘DPA-PET longitudinal analysis’ will refer to analysing the dynamic profile of microglial activation (DPA-2 versus DPA-1) in patients with Alzheimer’s disease as compared to controls, and its relationship with Alzheimer’s disease clinical progression. DPA-PET predictive analysis Population description Fifty-two patients with Alzheimer’s disease (33 at the prodromal stage and 19 at the dementia stage) and 17 controls were studied (after excluding low affinity binders) (Table 1). There was no significant difference in terms of age or educational level between Alzheimer’s disease and control groups. The prevalence of APOE E4 carriers was higher among patients with Alzheimer’s disease. Detailed neuropsychological evaluations of prodromal and dementia Alzheimer’s disease subgroups are summarized in Supplementary Table 1. Treatments were comparable in the prodromal-Alzheimer’s disease and demented-Alzheimer’s disease subgroups, except for the use of memantine, which was more frequent in demented patients (6/19). There was no change in treatment for Alzheimer’s disease during follow-up. Patients with Alzheimer’s disease exhibited a smaller hippocampal volume at baseline and a higher progression of hippocampal atrophy than controls. After 2 years of follow-up, 17 patients with Alzheimer’s disease had unchanged CDR score and were defined as functional slow decliners and 35 had an increased CDR score and were defined as functional fast decliners. Considering the loss of MMSE scores, 23 patients with Alzheimer’s disease were considered as cognitive slow decliners (mean loss of MMSE = 0.9 ± 2.2) and 26 as cognitive fast decliners (mean loss of MMSE = 8.2 ± 3.2). Detailed rates of decline of the neuropsychological variables in the fast and slow decliners groups over 2 years are shown in Supplementary Table 2. At baseline, functional slow and functional fast decliners as well as cognitive slow and cognitive fast decliners did not differ in terms of educational level, disease duration and baseline neuropsychological variables, but differed in age and MMSE score (fast decliners were younger and had a lower MMSE score). Consequently, age and MMSE scores were systematically used as covariates in all subsequent analyses. Table 1 Population characteristics and baseline 18F-DPA-714 (DPA-1) SUVr in anatomical regions between groups Slow and fast decliners defined by functional outcome (CDR change) Slow and fast decliners defined by cognitive outcome (MMSE loss) Controls All AD patients P-value fn-SD fn-FD P-value co-SD co-FD P-value n = 17 n = 52 n = 17 n = 35 n = 23 n = 26 Age (years) 69.4 (6.4) 67 (10.4) 0.3 74.5 (10) 63.4 (8.6) <0.001 72.6 (10.2) 62.7 (7.8) <0.001 Education (years) 12 (4.3) 13.3 (3.6) 0.4 14.2 (3.4) 12.9 (3.6) 0.2 12.8 (3.8) 12.8 (3.2) 0.13 Carrier of APOE e4 (%) 0 (0) 22 (42) <0.001 12 (70) 18 (51) 0.1 14 (61) 15 (58) 0.5 TSPO genotype (MAB:HAB) 8:9 30:22 0.4 13:4 17:18 0.05 15:8 14:12 0.3 Baseline CDR = 0 (%) 18 (100) 0 (0) <0.001 0 (0) 0 (0) 0.06 0 (0) 0 (0) 0.5 Baseline CDR = 0.5 (%) 0 (0) 33 (64) 11 (65) 22 (63) 15 (65) 18 (70) Baseline CDR>0.5 (%) 0 (0) 19 (36) 6 (35) 13 (37) 8 (35) 8 (30) Baseline MMSE score 29.5 (0.6) 21.1 (5.4) <0.001 23.7 (3.5) 19.8 (5.7) 0.01 22.8 (4) 21 (4.7) 0.15 MMSE score loss 0.3 (0.6) 4.8 (4.6) <0.001 1.2 (3.2) 6.7 (4.1) <0.001 0.9 (2.2) 8.2 (3.2) 0.001 MMSE score loss (%)b 0.4 (2.7) 24.3 (24.4) <0.001 5.2 (14) 34.5 (22.7) <0.001 4.4 (10) 42.0 (19) 0.001 Baseline HVa 2.06 (0.2) 1.74 (0.3) <0.001 1.63 (0.2) 1.80 (0.3) 0.03 2.37 (0.5) 2.54 (0.4) 0.26 HV loss (%)c 0.7 (0.1) 9.01 (5) <0.001 8.5 (5) 9.3 (5.3) 0.6 8.2 (5.4) 9.8 (4.9) 0.23 11C-PiB GCI 1.22 (0.1) 2.89 (0.6) <0.001 2.90 (0.6) 2.85 (0.1) 0.5 2.80 (0.6) 3.00 (0.6) 0.13 Baseline DPA-1 SUVr GCI 1.19 (0.1) 1.37 (0.2) <0.001* 1.50 (0.2) 1.30 (0.1) 0.001** 1.43 (0.2) 1.32 (0.2) 0.04** Frontal 1.19 (0.1) 1.34 (0.2) 0.01* 1.53 (0.2) 1.25 (0.2) 0.001** 1.43 (0.2) 1.28 (0.2) 0.07** Anterior cingulate 1.23 (0.2) 1.35 (0.2) 0.10* 1.55 (0.2) 1.25 (0.1) 0.001** 1.47 (0.2) 1.26 (0.2) 0.006** Medium cingulate 1.26 (0.2) 1.42 (0.2) 0.002* 1.60 (0.2) 1.33 (0.2) 0.001** 1.51 (0.2) 1.36 (0.2) 0.05** Posterior cingulate 1.24 (0.2) 1.45 (0.2) 0.001* 1.57 (0.2) 1.39 (0.2) 0.001** 1.50 (0.2) 1.42 (0.2) 0.09** Precuneus 1.14 (0.1) 1.37 (0.2) <0.001* 1.49 (0.2) 1.31 (0.1) 0.001** 1.42 (0.2) 1.33 (0.2) 0.2** Parietal 1.16 (0.1) 1.38 (0.2) <0.001* 1.52 (0.2) 1.32 (0.2) 0.001** 1.44 (0.2) 1.36 (0.2) 0.04** Temporal 1.09 (0.1) 1.22 (0.1) 0.001* 1.28 (0.1) 1.20 (0.1) 0.004** 1.25 (0.2) 1.19 (0.2) 0.04** Occipital 1.16 (0.1) 1.28 (0.1) 0.003* 1.33 (0.2) 1.25 (0.1) 0.03** 1.32 (0.2) 1.25 (0.2) 0.04** Slow and fast decliners defined by functional outcome (CDR change) Slow and fast decliners defined by cognitive outcome (MMSE loss) Controls All AD patients P-value fn-SD fn-FD P-value co-SD co-FD P-value n = 17 n = 52 n = 17 n = 35 n = 23 n = 26 Age (years) 69.4 (6.4) 67 (10.4) 0.3 74.5 (10) 63.4 (8.6) <0.001 72.6 (10.2) 62.7 (7.8) <0.001 Education (years) 12 (4.3) 13.3 (3.6) 0.4 14.2 (3.4) 12.9 (3.6) 0.2 12.8 (3.8) 12.8 (3.2) 0.13 Carrier of APOE e4 (%) 0 (0) 22 (42) <0.001 12 (70) 18 (51) 0.1 14 (61) 15 (58) 0.5 TSPO genotype (MAB:HAB) 8:9 30:22 0.4 13:4 17:18 0.05 15:8 14:12 0.3 Baseline CDR = 0 (%) 18 (100) 0 (0) <0.001 0 (0) 0 (0) 0.06 0 (0) 0 (0) 0.5 Baseline CDR = 0.5 (%) 0 (0) 33 (64) 11 (65) 22 (63) 15 (65) 18 (70) Baseline CDR>0.5 (%) 0 (0) 19 (36) 6 (35) 13 (37) 8 (35) 8 (30) Baseline MMSE score 29.5 (0.6) 21.1 (5.4) <0.001 23.7 (3.5) 19.8 (5.7) 0.01 22.8 (4) 21 (4.7) 0.15 MMSE score loss 0.3 (0.6) 4.8 (4.6) <0.001 1.2 (3.2) 6.7 (4.1) <0.001 0.9 (2.2) 8.2 (3.2) 0.001 MMSE score loss (%)b 0.4 (2.7) 24.3 (24.4) <0.001 5.2 (14) 34.5 (22.7) <0.001 4.4 (10) 42.0 (19) 0.001 Baseline HVa 2.06 (0.2) 1.74 (0.3) <0.001 1.63 (0.2) 1.80 (0.3) 0.03 2.37 (0.5) 2.54 (0.4) 0.26 HV loss (%)c 0.7 (0.1) 9.01 (5) <0.001 8.5 (5) 9.3 (5.3) 0.6 8.2 (5.4) 9.8 (4.9) 0.23 11C-PiB GCI 1.22 (0.1) 2.89 (0.6) <0.001 2.90 (0.6) 2.85 (0.1) 0.5 2.80 (0.6) 3.00 (0.6) 0.13 Baseline DPA-1 SUVr GCI 1.19 (0.1) 1.37 (0.2) <0.001* 1.50 (0.2) 1.30 (0.1) 0.001** 1.43 (0.2) 1.32 (0.2) 0.04** Frontal 1.19 (0.1) 1.34 (0.2) 0.01* 1.53 (0.2) 1.25 (0.2) 0.001** 1.43 (0.2) 1.28 (0.2) 0.07** Anterior cingulate 1.23 (0.2) 1.35 (0.2) 0.10* 1.55 (0.2) 1.25 (0.1) 0.001** 1.47 (0.2) 1.26 (0.2) 0.006** Medium cingulate 1.26 (0.2) 1.42 (0.2) 0.002* 1.60 (0.2) 1.33 (0.2) 0.001** 1.51 (0.2) 1.36 (0.2) 0.05** Posterior cingulate 1.24 (0.2) 1.45 (0.2) 0.001* 1.57 (0.2) 1.39 (0.2) 0.001** 1.50 (0.2) 1.42 (0.2) 0.09** Precuneus 1.14 (0.1) 1.37 (0.2) <0.001* 1.49 (0.2) 1.31 (0.1) 0.001** 1.42 (0.2) 1.33 (0.2) 0.2** Parietal 1.16 (0.1) 1.38 (0.2) <0.001* 1.52 (0.2) 1.32 (0.2) 0.001** 1.44 (0.2) 1.36 (0.2) 0.04** Temporal 1.09 (0.1) 1.22 (0.1) 0.001* 1.28 (0.1) 1.20 (0.1) 0.004** 1.25 (0.2) 1.19 (0.2) 0.04** Occipital 1.16 (0.1) 1.28 (0.1) 0.003* 1.33 (0.2) 1.25 (0.1) 0.03** 1.32 (0.2) 1.25 (0.2) 0.04** Data are mean (SD) or n (%). aMean of left and right hippocampal volume. bFollow-up MMSE score unavailable for three patients with Alzheimer’s disease because of severe cognitive decline. cFollow-up hippocampal volumes available for n = 35 patients with Alzheimer’s disease. *P-value for DPA-1 SUVr comparisons between controls and Alzheimer’s disease patients with TSPO genotype as covariate. **P-value for DPA-1 SUVr comparisons between slow and fast decliners (age, TSPO genotype and baseline MMSE as covariates). AD = Alzheimer’s disease; co-FD/SD = cognitive fast/slow decliner; fn-FD/SD = functional fast/slow decliner; HAB = high affinity binder; HV = hippocampal volume; MAB = mixed affinity binder. Table 1 Population characteristics and baseline 18F-DPA-714 (DPA-1) SUVr in anatomical regions between groups Slow and fast decliners defined by functional outcome (CDR change) Slow and fast decliners defined by cognitive outcome (MMSE loss) Controls All AD patients P-value fn-SD fn-FD P-value co-SD co-FD P-value n = 17 n = 52 n = 17 n = 35 n = 23 n = 26 Age (years) 69.4 (6.4) 67 (10.4) 0.3 74.5 (10) 63.4 (8.6) <0.001 72.6 (10.2) 62.7 (7.8) <0.001 Education (years) 12 (4.3) 13.3 (3.6) 0.4 14.2 (3.4) 12.9 (3.6) 0.2 12.8 (3.8) 12.8 (3.2) 0.13 Carrier of APOE e4 (%) 0 (0) 22 (42) <0.001 12 (70) 18 (51) 0.1 14 (61) 15 (58) 0.5 TSPO genotype (MAB:HAB) 8:9 30:22 0.4 13:4 17:18 0.05 15:8 14:12 0.3 Baseline CDR = 0 (%) 18 (100) 0 (0) <0.001 0 (0) 0 (0) 0.06 0 (0) 0 (0) 0.5 Baseline CDR = 0.5 (%) 0 (0) 33 (64) 11 (65) 22 (63) 15 (65) 18 (70) Baseline CDR>0.5 (%) 0 (0) 19 (36) 6 (35) 13 (37) 8 (35) 8 (30) Baseline MMSE score 29.5 (0.6) 21.1 (5.4) <0.001 23.7 (3.5) 19.8 (5.7) 0.01 22.8 (4) 21 (4.7) 0.15 MMSE score loss 0.3 (0.6) 4.8 (4.6) <0.001 1.2 (3.2) 6.7 (4.1) <0.001 0.9 (2.2) 8.2 (3.2) 0.001 MMSE score loss (%)b 0.4 (2.7) 24.3 (24.4) <0.001 5.2 (14) 34.5 (22.7) <0.001 4.4 (10) 42.0 (19) 0.001 Baseline HVa 2.06 (0.2) 1.74 (0.3) <0.001 1.63 (0.2) 1.80 (0.3) 0.03 2.37 (0.5) 2.54 (0.4) 0.26 HV loss (%)c 0.7 (0.1) 9.01 (5) <0.001 8.5 (5) 9.3 (5.3) 0.6 8.2 (5.4) 9.8 (4.9) 0.23 11C-PiB GCI 1.22 (0.1) 2.89 (0.6) <0.001 2.90 (0.6) 2.85 (0.1) 0.5 2.80 (0.6) 3.00 (0.6) 0.13 Baseline DPA-1 SUVr GCI 1.19 (0.1) 1.37 (0.2) <0.001* 1.50 (0.2) 1.30 (0.1) 0.001** 1.43 (0.2) 1.32 (0.2) 0.04** Frontal 1.19 (0.1) 1.34 (0.2) 0.01* 1.53 (0.2) 1.25 (0.2) 0.001** 1.43 (0.2) 1.28 (0.2) 0.07** Anterior cingulate 1.23 (0.2) 1.35 (0.2) 0.10* 1.55 (0.2) 1.25 (0.1) 0.001** 1.47 (0.2) 1.26 (0.2) 0.006** Medium cingulate 1.26 (0.2) 1.42 (0.2) 0.002* 1.60 (0.2) 1.33 (0.2) 0.001** 1.51 (0.2) 1.36 (0.2) 0.05** Posterior cingulate 1.24 (0.2) 1.45 (0.2) 0.001* 1.57 (0.2) 1.39 (0.2) 0.001** 1.50 (0.2) 1.42 (0.2) 0.09** Precuneus 1.14 (0.1) 1.37 (0.2) <0.001* 1.49 (0.2) 1.31 (0.1) 0.001** 1.42 (0.2) 1.33 (0.2) 0.2** Parietal 1.16 (0.1) 1.38 (0.2) <0.001* 1.52 (0.2) 1.32 (0.2) 0.001** 1.44 (0.2) 1.36 (0.2) 0.04** Temporal 1.09 (0.1) 1.22 (0.1) 0.001* 1.28 (0.1) 1.20 (0.1) 0.004** 1.25 (0.2) 1.19 (0.2) 0.04** Occipital 1.16 (0.1) 1.28 (0.1) 0.003* 1.33 (0.2) 1.25 (0.1) 0.03** 1.32 (0.2) 1.25 (0.2) 0.04** Slow and fast decliners defined by functional outcome (CDR change) Slow and fast decliners defined by cognitive outcome (MMSE loss) Controls All AD patients P-value fn-SD fn-FD P-value co-SD co-FD P-value n = 17 n = 52 n = 17 n = 35 n = 23 n = 26 Age (years) 69.4 (6.4) 67 (10.4) 0.3 74.5 (10) 63.4 (8.6) <0.001 72.6 (10.2) 62.7 (7.8) <0.001 Education (years) 12 (4.3) 13.3 (3.6) 0.4 14.2 (3.4) 12.9 (3.6) 0.2 12.8 (3.8) 12.8 (3.2) 0.13 Carrier of APOE e4 (%) 0 (0) 22 (42) <0.001 12 (70) 18 (51) 0.1 14 (61) 15 (58) 0.5 TSPO genotype (MAB:HAB) 8:9 30:22 0.4 13:4 17:18 0.05 15:8 14:12 0.3 Baseline CDR = 0 (%) 18 (100) 0 (0) <0.001 0 (0) 0 (0) 0.06 0 (0) 0 (0) 0.5 Baseline CDR = 0.5 (%) 0 (0) 33 (64) 11 (65) 22 (63) 15 (65) 18 (70) Baseline CDR>0.5 (%) 0 (0) 19 (36) 6 (35) 13 (37) 8 (35) 8 (30) Baseline MMSE score 29.5 (0.6) 21.1 (5.4) <0.001 23.7 (3.5) 19.8 (5.7) 0.01 22.8 (4) 21 (4.7) 0.15 MMSE score loss 0.3 (0.6) 4.8 (4.6) <0.001 1.2 (3.2) 6.7 (4.1) <0.001 0.9 (2.2) 8.2 (3.2) 0.001 MMSE score loss (%)b 0.4 (2.7) 24.3 (24.4) <0.001 5.2 (14) 34.5 (22.7) <0.001 4.4 (10) 42.0 (19) 0.001 Baseline HVa 2.06 (0.2) 1.74 (0.3) <0.001 1.63 (0.2) 1.80 (0.3) 0.03 2.37 (0.5) 2.54 (0.4) 0.26 HV loss (%)c 0.7 (0.1) 9.01 (5) <0.001 8.5 (5) 9.3 (5.3) 0.6 8.2 (5.4) 9.8 (4.9) 0.23 11C-PiB GCI 1.22 (0.1) 2.89 (0.6) <0.001 2.90 (0.6) 2.85 (0.1) 0.5 2.80 (0.6) 3.00 (0.6) 0.13 Baseline DPA-1 SUVr GCI 1.19 (0.1) 1.37 (0.2) <0.001* 1.50 (0.2) 1.30 (0.1) 0.001** 1.43 (0.2) 1.32 (0.2) 0.04** Frontal 1.19 (0.1) 1.34 (0.2) 0.01* 1.53 (0.2) 1.25 (0.2) 0.001** 1.43 (0.2) 1.28 (0.2) 0.07** Anterior cingulate 1.23 (0.2) 1.35 (0.2) 0.10* 1.55 (0.2) 1.25 (0.1) 0.001** 1.47 (0.2) 1.26 (0.2) 0.006** Medium cingulate 1.26 (0.2) 1.42 (0.2) 0.002* 1.60 (0.2) 1.33 (0.2) 0.001** 1.51 (0.2) 1.36 (0.2) 0.05** Posterior cingulate 1.24 (0.2) 1.45 (0.2) 0.001* 1.57 (0.2) 1.39 (0.2) 0.001** 1.50 (0.2) 1.42 (0.2) 0.09** Precuneus 1.14 (0.1) 1.37 (0.2) <0.001* 1.49 (0.2) 1.31 (0.1) 0.001** 1.42 (0.2) 1.33 (0.2) 0.2** Parietal 1.16 (0.1) 1.38 (0.2) <0.001* 1.52 (0.2) 1.32 (0.2) 0.001** 1.44 (0.2) 1.36 (0.2) 0.04** Temporal 1.09 (0.1) 1.22 (0.1) 0.001* 1.28 (0.1) 1.20 (0.1) 0.004** 1.25 (0.2) 1.19 (0.2) 0.04** Occipital 1.16 (0.1) 1.28 (0.1) 0.003* 1.33 (0.2) 1.25 (0.1) 0.03** 1.32 (0.2) 1.25 (0.2) 0.04** Data are mean (SD) or n (%). aMean of left and right hippocampal volume. bFollow-up MMSE score unavailable for three patients with Alzheimer’s disease because of severe cognitive decline. cFollow-up hippocampal volumes available for n = 35 patients with Alzheimer’s disease. *P-value for DPA-1 SUVr comparisons between controls and Alzheimer’s disease patients with TSPO genotype as covariate. **P-value for DPA-1 SUVr comparisons between slow and fast decliners (age, TSPO genotype and baseline MMSE as covariates). AD = Alzheimer’s disease; co-FD/SD = cognitive fast/slow decliner; fn-FD/SD = functional fast/slow decliner; HAB = high affinity binder; HV = hippocampal volume; MAB = mixed affinity binder. 18F-DPA-714 cortical binding at baseline (DPA-1) Volume of interest analysis We first compared the whole Alzheimer’s disease cohort to controls. DPA-1 binding was significantly higher in patients with Alzheimer’s disease as compared to controls, especially in the temporal and parietal regions (Supplementary Table 3). No significant difference of DPA-1 binding was found between prodromal Alzheimer’s disease and dementia Alzheimer’s disease subjects. There was a positive correlation between the DPA-1 GCI and the baseline MMSE score (r = 0.26, P = 0.037), left hippocampal volume (r = 0.31, P = 0.04) and cortical volume (r = 0.25, P = 0.04), with TSPO genotype, age and CDR as covariates. Because young patients with Alzheimer’s disease had a faster clinical progression over time, we verified that no 18F-DPA-714 uptake difference was observed between patients with early and late onset of Alzheimer’s disease. In addition, we did not find any correlation between age and initial 18F-DPA-714 uptake. Then we compared the DPA-1 binding between slow and fast decliners. DPA-1 binding was significantly higher in functional slow decliners than in functional fast decliners in all cortical regions as well as for the global index (P = 0.001; with age, TSPO genotype and initial MMSE score as covariates) (Table 1 and Fig. 1A). Similar results were observed when comparing DPA-1 between cognitive slow and cognitive fast decliners, although with lower statistical power (GCI: P = 0.04, Table 1). Figure 1 View largeDownload slide DPA-PET predictive study: scatter and box plots showing global and regional 18F-DPA-714 (DPA-1) SUVr in slow and fast Alzheimer’s disease decliners. (A) Comparison between functional slow decliners (n = 17, in red) and functional fast decliners (n = 35, in blue). (B) Comparison between cognitive slow decliners (n = 15, in red) and cognitive fast decliners (n = 18, in blue) in the subgroup of patients with Alzheimer’s disease (AD) at the prodromal stage at inclusion. *P < 0.05 with age, TSPO genotype and initial MMSE as covariates. Figure 1 View largeDownload slide DPA-PET predictive study: scatter and box plots showing global and regional 18F-DPA-714 (DPA-1) SUVr in slow and fast Alzheimer’s disease decliners. (A) Comparison between functional slow decliners (n = 17, in red) and functional fast decliners (n = 35, in blue). (B) Comparison between cognitive slow decliners (n = 15, in red) and cognitive fast decliners (n = 18, in blue) in the subgroup of patients with Alzheimer’s disease (AD) at the prodromal stage at inclusion. *P < 0.05 with age, TSPO genotype and initial MMSE as covariates. Finally, we compared DPA-1 binding in slow and fast decliners within the subgroups of patients at either prodromal stage (CDR = 0.5 at inclusion) or dementia stage (CDR ≥ 1 at baseline) separately. Regarding the prodromal patients with Alzheimer’s disease, DPA-1 binding was significantly higher in functional slow decliners (n = 11) than in functional fast decliners (n = 22), especially in the anterior regions, as well as in cognitive slow decliners (n = 15) than in cognitive fast decliners (n = 18) (Fig. 1). Concerning the demented patients with Alzheimer’s disease, we also found higher DPA-1 binding in the functional slow decliners (n = 6) than in the functional fast decliners (n = 13), but no significant difference was observed between cognitive slow (n = 8) and cognitive fast decliners (n = 8), taking into account that MMSE scores were lacking at the last visit for three patients with Alzheimer’s disease because of too severe cognitive impairment. Voxel-wise comparisons The voxel-wise analysis showed a higher 18F-DPA-714 binding in functional slow decliners when compared to functional fast decliners (P < 0.001, uncorrected; with age, TSPO and MMSE as covariates), which is more pronounced in the anterior regions in prodromal patients with Alzheimer’s disease (Fig. 2A, upper panels). A similar pattern of microglial activation was observed when we compared cognitive slow and cognitive fast decliner patients (P < 0.001, uncorrected; with age, TSPO and MMSE as covariates) (Fig. 2A, lower panels). Figure 2 View largeDownload slide DPA-PET predictive study: statistical parametric mapping analysis of 18F-DPA-714 SUVr at baseline (DPA-1). (A) Voxel-wise comparisons between slow and fast decliners. (i) Comparison of DPA-1 binding at baseline between functional slow decliners (fn-SD) and functional fast decliners (fn-FD) in the whole Alzheimer’s disease population (n = 52, left) and in the prodromal Alzheimer’s disease population (n = 33, right). (ii) Comparison of DPA-1 binding at baseline between cognitive slow decliners (co-SD) and cognitive fast decliners (co-FD) in the whole Alzheimer’s disease population (n = 49, left) and in the prodromal Alzheimer’s disease population (n = 33, right). **P < 0.001 uncorrected with age, TSPO genotype and initial MMSE as covariates. (B) Correlations between DPA-1 binding and functional, cognitive or MRI biomarkers outcomes using voxel-wise analysis. (i) Correlations between DPA-1 binding and the MMSE change (% of progression), in the whole Alzheimer’s disease group (n = 49, left) and in the prodromal Alzheimer’s disease group (n = 33, right). (ii) Correlations between DPA-1 binding and the left hippocampal volume change, in the whole Alzheimer’s disease group (n = 36, left) and in the prodromal Alzheimer’s disease group (n = 26, right). *P < 0.05 with age, TSPO genotype and initial MMSE as covariates. AD = Alzheimer’s disease; HV = hippocampal volume. Figure 2 View largeDownload slide DPA-PET predictive study: statistical parametric mapping analysis of 18F-DPA-714 SUVr at baseline (DPA-1). (A) Voxel-wise comparisons between slow and fast decliners. (i) Comparison of DPA-1 binding at baseline between functional slow decliners (fn-SD) and functional fast decliners (fn-FD) in the whole Alzheimer’s disease population (n = 52, left) and in the prodromal Alzheimer’s disease population (n = 33, right). (ii) Comparison of DPA-1 binding at baseline between cognitive slow decliners (co-SD) and cognitive fast decliners (co-FD) in the whole Alzheimer’s disease population (n = 49, left) and in the prodromal Alzheimer’s disease population (n = 33, right). **P < 0.001 uncorrected with age, TSPO genotype and initial MMSE as covariates. (B) Correlations between DPA-1 binding and functional, cognitive or MRI biomarkers outcomes using voxel-wise analysis. (i) Correlations between DPA-1 binding and the MMSE change (% of progression), in the whole Alzheimer’s disease group (n = 49, left) and in the prodromal Alzheimer’s disease group (n = 33, right). (ii) Correlations between DPA-1 binding and the left hippocampal volume change, in the whole Alzheimer’s disease group (n = 36, left) and in the prodromal Alzheimer’s disease group (n = 26, right). *P < 0.05 with age, TSPO genotype and initial MMSE as covariates. AD = Alzheimer’s disease; HV = hippocampal volume. Correlations between DPA-1 binding and functional, cognitive and MRI biomarkers outcomes In the whole group of patients with Alzheimer’s disease, we found a negative correlation between DPA-1 binding in the frontal regions and the percentage of increase of the CDR-Sum of Boxes (SOB) (r = −0.27, P = 0.03; with age, TSPO and MMSE as covariates). Similarly, we found a negative correlation between global cortical DPA-1 binding and the percentage of decrease of the MMSE score (r = −0.28 P = 0.028 for the GCI; with age, TSPO genotype and initial MMSE as covariates). These results suggest that the higher the DPA-1 binding, the lower the functional and cognitive declines. Considering the prodromal Alzheimer’s disease group separately, a significant negative correlation was observed with the percentage of MMSE decline (P < 0.05), but not with the CDR-SOB change. We also correlated DPA-1 binding with the evolution over time of composite scores reflecting different cognitive domains (long-term memory, short-term memory, instrumental functions, executive functions) in the mild cognitive impairment subjects of the predictive cohort (as patients with mild cognitive impairment were more numerous, and because we had more missing longitudinal neuropsychological data in demented patients, leading to floor effect). We found significant negative correlations between DPA-1 binding in the frontal and anterior cingulate regions and the decline of the composite scores reflecting long-term memory [sum of total recall in the Free and Cued Selective Reminding Test (FCSRT) and temporo-spatial orientation, r = −0.31, P = 0.04 for the frontal region and r = −0.37, P = 0.01 for the anterior cingulate region], and short-term memory (sum of encoding score in the FCSRT and digit spans, r = −0.34, P = 0.029 for the frontal region and r = −0.44, P = 0.006 for the anterior cingulate region). We found no significant correlation with the composite scores reflecting instrumental and executive functions. Voxel-wise analysis confirms these results, showing negative correlations between DPA-1 binding and MMSE score decrease in frontal and parieto-occipital cortical areas (Fig. 2B, upper panels). Progression of hippocampal atrophy was used as a MRI biomarker of Alzheimer’s disease progression. The percentage of decrease of hippocampal volume was negatively correlated with the DPA-1 uptake. Highest correlations were found between the left hippocampus atrophy rate and the frontotemporal DPA-1 uptake in prodromal patients with Alzheimer’s disease. These data suggest that higher DPA-1 uptake was associated with lower loss of hippocampal volume. Voxel-wise analysis confirmed this correlation in the same cortical areas (Fig. 2B, lower panels). There was no significant correlation between DPA-1 uptake and the percentage of decrease of cortical grey matter volume. PiB cortical binding and Alzheimer’s disease progression We did not find any significant results when similar analyses as described above were carried out with using 11C-PiB binding (no difference of PiB binding between slow and fast decliners as defined either by CDR or MMSE scores changes; no correlation between PiB binding and functional, cognitive or MRI outcomes). Thus, 11C-PiB binding was not significantly associated with the rate of clinical progression. As previously published, PiB cortical binding correlated with DPA-1 binding (r = 0.5, P = 0.03). DPA-PET longitudinal analysis Population description All of the following results were obtained without including the low affinity binders. Patients (n = 21) and controls (n = 13) did not differ with regards to age and educational level. The mean loss in MMSE score after 2 years was 5 points in prodromal patients with Alzheimer’s disease, 7.1 points in demented patients with Alzheimer’s disease and 0.03 points in controls. The mean interval between both PET scans was 1.6 years in prodromal patients with Alzheimer’s disease, 1.4 years in demented patients with Alzheimer’s disease and 2.2 years in controls. Detailed neuropsychological data of the population are summarized in Supplementary Table 4. We also verified that DPA-1 binding in the patients who performed DPA-2 was not significantly different from that of those who did not (DPA-1 GCI = 1.37 ± 0.17 in the 21 patients who had both DPA-1 and DPA-2, and DPA-1 GCI = 1.36 ± 0.2 in the 31 patients who did not perform DPA-2). Methodological considerations Cerebellar grey matter as a pseudo-reference region We verified that cerebellar grey matter 18F-DPA-714 uptake (i) did not differ between patients and controls neither in the first (DPA-1) nor in the second (DPA-2) PET scan; (ii) remained stable individually between DPA-1 and DPA-2 in both patients with Alzheimer’s disease (P = 0.1) and controls (P = 0.6), without any difference between groups (Supplementary Fig. 1); and (iii) was not correlated with MMSE score, age, APOE genotype and cortical volume measured by MRI. These results provide additional arguments to those previously published for using the cerebellar grey matter as a pseudo-reference region for quantifying 18F-DPA-714 binding. Quantification of longitudinal changes in 18F-DPA-714 binding Changes in DPA binding were expressed as an annualized percentage. Because this supposes a linear progression over time, we verified for all analyses that results were not modified when using the global percentage of DPA binding variation, while including the delay between both DPA-PET scans as a covariate. Comparisons of DPA-1, DPA-2 and DPA binding progression between controls and patients with Alzheimer’s disease Volume of interest analysis As shown for DPA-1, DPA-2 binding was higher in patients with Alzheimer’s disease than in controls in all volumes of interest, especially at the prodromal stage (P < 0.01) (Supplementary Table 5). In the control group, 18F-DPA-714 binding was stable with time (4.2 ± 4.3% variation per year for the GCI) (Fig. 3A). In contrast, 18F-DPA-714 binding significantly increased over time in the Alzheimer’s disease group as compared to controls (13.2% per year for the GCI) (P = 0.02), especially at the prodromal stage (15.1% per year versus 8.3% per year at the dementia stage). Individual analysis showed heterogeneous 18F-DPA-714 binding progression profiles among patients with Alzheimer’s disease (Fig. 3B and C). Figure 3 View largeDownload slide Longitudinal change of 18F-DPA-714 GCI. (A and B) Individual 18F-DPA-714 GCI at baseline (DPA-1) and follow-up (DPA-2) in controls (n = 13) and in patients with Alzheimer’s disease (n = 21, prodromal Alzheimer’s disease in black, dementia Alzheimer’s disease in grey). (C) Mean 18F-DPA-714 GCI at baseline and follow-up in patients with low DPA-1 binding [in red, SUVr ≤ 1.39, which is the optimal cut-off value for GCI derived from the receiver operating characteristic (ROC) curve analysis for differentiating slow and fast decliners] and with high DPA-1 binding (in blue, SUVr > 1.39) and in controls (in grey). (D) Surface plot representing the percentage of increase of the CDR-SOB score (z-axis) in relation with the DPA-1 cortical binding (x-axis) and the annualized percentage of DPA cortical binding increase (y-axis). Figure 3 View largeDownload slide Longitudinal change of 18F-DPA-714 GCI. (A and B) Individual 18F-DPA-714 GCI at baseline (DPA-1) and follow-up (DPA-2) in controls (n = 13) and in patients with Alzheimer’s disease (n = 21, prodromal Alzheimer’s disease in black, dementia Alzheimer’s disease in grey). (C) Mean 18F-DPA-714 GCI at baseline and follow-up in patients with low DPA-1 binding [in red, SUVr ≤ 1.39, which is the optimal cut-off value for GCI derived from the receiver operating characteristic (ROC) curve analysis for differentiating slow and fast decliners] and with high DPA-1 binding (in blue, SUVr > 1.39) and in controls (in grey). (D) Surface plot representing the percentage of increase of the CDR-SOB score (z-axis) in relation with the DPA-1 cortical binding (x-axis) and the annualized percentage of DPA cortical binding increase (y-axis). Voxel-wise analysis DPA-1 and DPA-2 binding were higher in patients with Alzheimer’s disease than in controls, especially at the prodromal stage in temporo-parietal regions (P < 0.001, uncorrected) (Fig. 4A and B). The annual progression of 18F-DPA-714 binding was higher in patients with Alzheimer’s disease than in controls in fronto-parietal regions (P < 0.01, uncorrected) (Fig. 4C). Figure 4 View largeDownload slide DPA-PET longitudinal study. Statistical parametric mapping comparisons of 18F-DPA-714 SUVr between Alzheimer’s disease group (n = 21) and controls (n = 13) (left) and between prodromal Alzheimer’s disease group (n = 15) and controls (n = 13) (right). (A) DPA-1, **P < 0.001 uncorrected, TSPO genotype as a covariate. (B) DPA-2, **P < 0.001 uncorrected, TSPO genotype and delay time as covariates. (C) Annualized percentage of DPA binding progression, *P < 0.01 uncorrected, age, TSPO genotype and initial CDR as covariates. AD = Alzheimer’s disease. Figure 4 View largeDownload slide DPA-PET longitudinal study. Statistical parametric mapping comparisons of 18F-DPA-714 SUVr between Alzheimer’s disease group (n = 21) and controls (n = 13) (left) and between prodromal Alzheimer’s disease group (n = 15) and controls (n = 13) (right). (A) DPA-1, **P < 0.001 uncorrected, TSPO genotype as a covariate. (B) DPA-2, **P < 0.001 uncorrected, TSPO genotype and delay time as covariates. (C) Annualized percentage of DPA binding progression, *P < 0.01 uncorrected, age, TSPO genotype and initial CDR as covariates. AD = Alzheimer’s disease. Correlation between DPA binding progression and functional, cognitive and MRI biomarker outcome To avoid any confounding bias, age, TSPO binding status and initial MMSE score were defined as covariates for all correlation analyses (either by volume of interest or voxel-wise method). We found positive correlations between the increase in 18F-DPA-714 binding and the annualized percentage of CDR-SOB increase in all volumes of interest, both in the whole group and in prodromal patients with Alzheimer’s disease (Supplementary Table 6). These data suggest that the increase in 18F-DPA-714 binding is associated with functional decline over time even at the prodromal stage. Voxel-wise analysis confirmed these results, showing significant correlations in frontal and parieto-temporal regions (Fig. 5A). Figure 5 View largeDownload slide DPA-PET longitudinal study. Statistical parametric mapping showing correlations between 18F-DPA-714 SUVr [in the Alzheimer’s disease group (left), and in the prodromal Alzheimer’s disease group (right)] and (A) CDR-SOB increase, *P < 0.01 uncorrected; (B) MMSE score loss, *P < 0.01 uncorrected; and (C) left hippocampal volume volume loss, *P < 0.01 uncorrected. Age, TSPO genotype and initial MMSE as covariates. HV = hippocampal volume. Figure 5 View largeDownload slide DPA-PET longitudinal study. Statistical parametric mapping showing correlations between 18F-DPA-714 SUVr [in the Alzheimer’s disease group (left), and in the prodromal Alzheimer’s disease group (right)] and (A) CDR-SOB increase, *P < 0.01 uncorrected; (B) MMSE score loss, *P < 0.01 uncorrected; and (C) left hippocampal volume volume loss, *P < 0.01 uncorrected. Age, TSPO genotype and initial MMSE as covariates. HV = hippocampal volume. We also found a significant positive correlation between the annualized percentage of increase in DPA binding and the decrease of MMSE score over 2 years, especially in the frontal regions (Supplementary Table 6), confirmed by the voxel-wise analysis (Fig. 5B). These results suggest that progression of 18F-DPA-714 binding is associated with cognitive decline over time. Finally, we found significant positive correlations between the annualized percentage of increase in DPA binding and the decrease in left hippocampal volume, in frontal and parieto-temporal cortex over 2 years (Supplementary Table 6). This finding was replicated by the voxel-wise analysis (Fig. 5C). Correlations between PiB-PET at baseline and DPA binding progression or clinical Alzheimer’s disease outcomes We found no correlation between initial 11C-PiB binding and either the percentage of DPA binding progression, CDR-SOB progression, changes in MMSE score or the decrease of hippocampal volume over 2 years. Relationships between DPA-1 binding, DPA binding progression and clinical decline We found a significant negative correlation between the DPA-1 uptake and the DPA binding progression (r = −0.64, P = 0.002 for the GCI with age, TSPO genotype and initial MMSE score as covariates), suggesting that patients with the highest DPA-1 binding are those who tend to have the lowest increase in DPA binding over time (Fig. 3D). Interestingly, unlike the extent of initial DPA-1 binding, its longitudinal increase is associated with higher cognitive and functional decline (Fig. 5) as well as faster hippocampal atrophy. This dynamic pattern is observed in both prodromal and demented stages of Alzheimer’s disease, even if the statistical significance was stronger in prodromal Alzheimer’s disease. Of note, results remained similar when we included age, disease duration, initial MMSE score or initial CDR score as covariates. Amyloidosis positive controls Among six amyloidosis positive controls, who were also analysed, including two low affinity binders, only one developed memory cognitive decline during the 2-year follow-up period. The annualized percentages of increase in 18F-DPA-714 binding in these six amyloidosis-positive controls were +24.66%, −4.27%, −10.35%, −1.22%, −8.07% and +17.58%, the two latter percentages corresponding to the low affinity binders. Of note, the only amyloidosis-positive control who developed memory impairment during the follow-up period was a low affinity binder and had both a relatively low DPA-1 GCI of 1.10, and a relatively high annualized percentage of increase in 18F-DPA-714 binding of +17.58%. However, another amyloidosis positive control (high affinity binder) had a low DPA-1 GCI of 1.12 with an annualized percentage of increase in 18F-DPA-714 binding of 24.66% and did not develop cognitive impairment during follow-up. Discussion This study provides new insights into the temporal relationships between microglial activation as measured by 18F-DPA-714 binding and Alzheimer’s disease clinical progression. First, we found in a large cohort of 52 patients with Alzheimer’s disease that higher initial 18F-DPA-714 binding is associated with better clinical prognosis after a 2-year follow-up. Second, the longitudinal DPA-PET study showed that the subsequent increase in 18F-DPA-714 binding observed in patients with Alzheimer’s disease (as compared with controls) was linked to Alzheimer’s disease worsening. Finally, patients with lowest initial 18F-DPA-714 binding had the greatest subsequent increase of microglial activation and unfavourable clinical evolution, while patients with highest initial 18F-DPA-714 binding had the lowest subsequent increase of microglial activation and more favourable clinical evolution, independently of the initial cortical amyloid load. Altogether, these results support the hypothesis that two distinct dynamic profiles of microglial activation differentially impact on disease progression in patients with Alzheimer’s disease, being either beneficial or detrimental, and may vary depending on patients rather than disease stages. Compared with previous works using TSPO PET imaging in Alzheimer’s disease, one of the strengths of the current study is the larger sample of subjects (Alzheimer’s disease and controls) and the choice of very strict inclusion criteria. In addition to positivity for Alzheimer’s disease pathophysiological markers (PiB-PET and CSF biomarkers), we excluded all subjects with known medical events that could influence TSPO binding, such as history of inflammatory diseases and suspicion of alcohol abuse. No change in Alzheimer’s disease treatment during the follow-up could explain the difference in clinical progression. In addition, we used a high sensitivity PET scanner (HRRT) together with 18F-DPA-714, a second generation TSPO tracer, which provides better sensitivity than 11C-PK11195 for evaluating increased TSPO expression (Chauveau et al., 2009; Yokokura et al., 2017), and for which mixed affinity and high affinity binders can be pooled more easily than when using 11C-PBR28, of known much higher affinity ratio between high affinity and low affinity binders (Owen et al., 2011; Hamelin et al., 2016). Furthermore, TSPO genotype was added as a covariate in all statistical analyses and low affinity binders were excluded from all analyses. The advantage of having a large cohort is counterbalanced by the lack of arterial blood samples, which may be a limitation to fully quantify the DPA-714 binding. The SUVr method assumes no difference in cerebellar binding at baseline and follow-up. While we observed the stability of cerebellar 18F-DPA-714 binding over time in patients with Alzheimer’s disease and controls, in congruence with previous works using 11C-PBR28 radioligand, it has not actually been validated using full quantification of DPA-714 data (Lyoo et al., 2015; Kreisl et al., 2016). Previous transversal studies using TSPO PET imaging reported controversial findings regarding the protective or deleterious impact of microglial activation in Alzheimer’s disease. Some studies showed that 11C-PK11195 binding correlated with clinical severity, cortical volume atrophy and fluorodeoxyglucose (FDG) hypometabolism, suggesting a toxic effect of microglial activation (Cagnin et al., 2001; Edison et al., 2008; Yokokura et al., 2011). Conversely, other reports showed no correlation with cognitive scores regardless of whether patients with mild cognitive impairment progressed to dementia or remained clinically stable (Schuitemaker et al., 2013). Using second generation TSPO tracers, one study suggested that 11C-PBR28 uptake correlated with cortical volume loss and decrease in several cognitive performances (Lyoo et al., 2015). In contrast, we previously showed that 18F-DPA-714 binding correlated positively with MMSE scores and grey matter volume, especially at the prodromal stage, suggesting a protective role of microglial activation (Hamelin et al., 2016). Only three longitudinal studies aimed at understanding TSPO binding progression have been reported in Alzheimer’s disease so far, which used different radioligands and different quantification methods. Two of them used 11C(R)-PK11195 tracer and a cluster-based approach for defining the reference region (Fan et al., 2015b, 2017). After a 16-month follow-up period, voxel-wise analysis suggested that the increase in microglial activation in patients with Alzheimer’s disease (n = 8, MMSE = 21/30) was positively correlated with amyloid deposition and inversely correlated with regional cerebral metabolic rate (Fan et al., 2015b). However, the small number of patients with Alzheimer’s disease, the negativity of amyloid imaging in one of them (calling into question Alzheimer’s disease diagnosis) and the lack of longitudinal follow-up of healthy controls limited the interpretation of these data. The same group reported findings in eight subjects with mild cognitive impairment (MMSE = 27.6/30) with different amyloid status, four of eight being amyloid-negative based on PiB-PET imaging (Fan et al., 2017). The authors described opposite results as compared to those observed in their patients with Alzheimer’s disease, with a longitudinal reduction in microglial activation associated with a marginal increase in amyloid load over 14 months. Fan et al. thus proposed the existence of two peaks of microglial activation in the course of Alzheimer’s disease, with an early protective peak and a late pro-inflammatory deleterious peak. However, the very small number of mild cognitive impairment due to patients with Alzheimer’s disease analysed in this study (only four) raises question about the robustness of such a model. The third longitudinal PET study used 11C-PBR28 and cerebellum as a pseudo-reference region for quantifying TSPO binding in 14 amyloid-positive patients with Alzheimer’s disease at different disease stages (MMSE ranging from 14 to 30; n = 9 patients with CDR = 0.5, n = 5 patients with CDR = 1) and eight controls (Kreisl et al., 2016). 11C-PBR28 binding increased in temporo-parietal regions from 3.9% to 6.3% per year in patients versus 0.5% to 1% per year in controls. The increase in TSPO binding correlated with functional worsening on CDR-SOB and with reduced cortical volume. The small number of slow and fast decliners defined by changes in the CDR score (n = 5 and n = 9, respectively) did not allow investigating the binding progression within the prodromal Alzheimer’s disease subgroups separately. Our results are in accordance with Kreisl et al. (2016), as we found an increase in 18F-DPA-714 binding over 24 months in patients with Alzheimer’s disease as compared with controls (13.2% per year versus 4.2%), both at the prodromal (15.8%) and at the demented stages (8.3%). The positive correlations between the increase in DPA binding and the increase in CDR-SOB scores, MMSE score decrease and rate of hippocampal atrophy suggested a detrimental effect on clinical Alzheimer’s disease progression of the subsequent increase in microglial activation, which was not influenced by the initial fibrillar amyloid load. Determining whether such a profile is similar in prodromal and demented stages of Alzheimer’s disease remains difficult. Indeed, one limitation of our work is inherent to clinical PET longitudinal studies, as the second PET examination could not be performed in patients at severe dementia stage for obvious ethical reasons. Consequently, the number of patients at the prodromal stage (n = 15) was higher than that of patients at the demented stage (n = 6), precluding sufficiently powerful statistical analyses in the demented subgroup. A larger number of subjects would be necessary to draw precise conclusions. This deleterious effect of an increasing microglial activation may appear inconsistent with our findings from the DPA-PET predictive analysis, which evidenced a protective effect on Alzheimer’s disease progression of higher initial 18F-DPA-714 binding, confirming our previous studies (Hamelin et al., 2016). Here we used both functional (CDR score changes) and cognitive (MMSE score decline) outcome measures for defining slow and fast decliners. In addition, we assessed correlations between 18F-DPA-714 binding and clinical and MRI outcome scores by both volume of interest and voxel-wise methods. Importantly, 18F-DPA-714 binding at baseline as well as the subsequent increase in 18F-DPA-714 binding were highly heterogeneous among patients with Alzheimer’s disease. Interestingly, patients with the highest initial 18F-DPA-714 binding display the lowest subsequent increase in 18F-DPA-714 binding over time, and better prognosis. Conversely, patients with the lowest initial 18F-DPA-714 binding show a higher subsequent increase in 18F-DPA-714 binding and a worst prognosis. We must nevertheless acknowledge that in patients with high DPA-1, the course of microglial activation before inclusion in the study remains to be explored. The follow-up of amyloidosis-positive controls was poorly informative, as the only subject who developed memory impairment during the follow-up period was a low affinity binder. The others remained clinically stable over time and exhibited heterogeneous 18F-DPA-714 binding progressions. One of the limitations of this study is the low number of subjects in this subgroup and the too short follow-up duration of these patients. Following-up these subjects for a longer period will be highly informative for deciphering the role of microglial activation in preclinical Alzheimer’s disease. The concept of diverse functional phenotypes of immune cells, ranging from pro-inflammatory to immunosuppressive, has been expended to microglia (Tang and Le, 2016). Importantly, the use of the traditional ‘M1/M2 paradigm’ of activation status is now considered as over-simplified, and functional differentiation patterns of activated microglia extend much beyond the classical proinflammatory M1-like and neuroprotective M2-like phenotypes, rather spanning a full spectrum of activation patterns (Heneka et al., 2014). The balance of pro-inflammatory and neuroprotective microglial activation is highly complex, especially in Alzheimer’s disease, in which microglia may exhibit mixed activation phenotypes. In this line, a recent study on 299 human brains evidenced an alteration of microglial immunophenotypes in association with ageing and the development of Alzheimer’s disease dementia, highlighting the complexity and diversity of microglial responses (Minett et al., 2016). Up to now, no PET tracer has been able to distinguish between different microglial subtypes or to capture the transition between such different activation states (Vivash and O’Brien, 2016). Beyond the current classical model of neuroinflammation, proposing a homogeneous early protective effect that subsequently turns into a later detrimental effect, our data rather suggest a more complex model based on both distinct patterns and dynamics of microglial activation signatures among patients with Alzheimer’s disease, which differently impact on disease progression and contribute to shape diverse clinical evolution profiles. Considering the positive correlation between DPA-1 and PiB binding, the protective microglial signature may reflect an interaction with misfolded deposited proteins and removal of neurotoxic aggregates, especially for fibrillar amyloid deposition (Fan et al., 2015b; Parbo et al., 2017). The detrimental microglial signature could reflect an interplay with neuronal injury processes, possibly independently of amyloid pathology when considering the absence of correlation between PiB uptake and the increase in 18F-DPA-714 binding. Protective or deleterious microglial signatures are both observed at the prodromal stage, and mostly do not seem related to the severity of the disease. Our data rather suggest that the relative extent and dynamics of beneficial and detrimental microglial activation signatures may vary among patients, thus translating into different clinical evolution profiles. Identifying the factors modulating this balance between such different signatures will be of major importance for both prognostic and therapeutic purposes. In this line, our previous studies in a mouse model of Alzheimer’s disease-like pathology suggested that peripheral modulation of a given T cell population could impact on the rate of disease progression at least partially by modulating microglial responses (Dansokho et al., 2016). Hence, different subtypes of patients with Alzheimer’s disease may benefit from different treatment protocols and/or innovative therapeutic approaches targeting inflammatory and immune responses. PET imaging using 18F-DPA-714 appears as a highly valuable tool for identifying target populations and assessing drug efficacy in such studies, even if we cannot exclude that some other biological parameters could also influence Alzheimer’s disease progression. Table 2 Population characteristics for the DPA-PET longitudinal analysis Controls All AD Prodromal AD Dementia AD Amyloidosis controls n = 13 n = 21 n = 15 n = 6 n = 4 TSPO genotype (MAB:HAB) 6:7 14:7 9:6 5:1 2:2 LAB (excluded from further analysis) 2 3 3 0 2 Age (years) 68.9 (6.5) 63.6 (7.4)* 64.6 (7.7) 60.8 (6.3) 73.5 (8.9)†† Education (years) 11.8 (4.2) 13.4 (3.7) 13.6 (4) 12.8 (3.7) 12.6 (3) Carrier of APOE e4 (%) 0 (0) 13 (62)* 10 (66)* 3 (50)* 2 (50)* Baseline CDR = 0 13 0 0 0 4 Baseline CDR = 0.5 0 15 15 0 0 Baseline CDR > 0.5 0 6 0 6 0 CDR change (%) 0 (0%) 16 (76 %)* 11 (73%)* 5 (83%)* 1 (25%)†,†† Baseline MMSE score 29.7 (0.5) 22.2 (4.7)* 24.6 (2.6)*,†† 16.1 (3.3)*,† 29.5 (0.5)†,†† Follow-up MMSE score 29.4 (0.9) 16.6 (7.2)* 19.6 (5.6)*,†† 9 (5)*,† 29 (2)†,†† Mean MMSE score loss (%) 0.03 (2.5) 5 (23.5)* 5 (20)*,†† 7.1 (20)*,† 0.8 (4.8)†,†† Baseline mean HV 2.06 (0.3) 1.8 (0.2)* 1.8 (0.3) 1.8 (0.3) 2.1 (0.3) Mean HV loss, % (SD)a 0.76 (1.5) 8.4 (4.4) 8 (4.5)* 10 (4.5)* 2.2 (3.5)†,†† ††C-PiB GCI 1.2 (0.07) 2.86 (0.6)* 2.7 (0.5)* 3.3 (0.8)* 2.27 (0.6)*,†† Mean delay between two DPA (years) 2.2 (0.2) 1.57 (0.3)* 1.63 (0.3)* 1.42 (0.3)* 2.03 (0.1)†,†† DPA-1 GCI# 1.19 (0.1) 1.37 (0.17)* 1.37 (0.1)* 1.37 (0.1)* 1.31 (0.3) DPA-2 GCI## 1.29 (0.15) 1.66 (0.3)* 1.72 (0.3)* 1.52 (0.2) 1.33 (0.2)† Annual % of DPA GCI increase### 4.2 (4.3) 13.1 (13.6)* 15.0 (13)* 8.3 (15) 2.2 (15) Controls All AD Prodromal AD Dementia AD Amyloidosis controls n = 13 n = 21 n = 15 n = 6 n = 4 TSPO genotype (MAB:HAB) 6:7 14:7 9:6 5:1 2:2 LAB (excluded from further analysis) 2 3 3 0 2 Age (years) 68.9 (6.5) 63.6 (7.4)* 64.6 (7.7) 60.8 (6.3) 73.5 (8.9)†† Education (years) 11.8 (4.2) 13.4 (3.7) 13.6 (4) 12.8 (3.7) 12.6 (3) Carrier of APOE e4 (%) 0 (0) 13 (62)* 10 (66)* 3 (50)* 2 (50)* Baseline CDR = 0 13 0 0 0 4 Baseline CDR = 0.5 0 15 15 0 0 Baseline CDR > 0.5 0 6 0 6 0 CDR change (%) 0 (0%) 16 (76 %)* 11 (73%)* 5 (83%)* 1 (25%)†,†† Baseline MMSE score 29.7 (0.5) 22.2 (4.7)* 24.6 (2.6)*,†† 16.1 (3.3)*,† 29.5 (0.5)†,†† Follow-up MMSE score 29.4 (0.9) 16.6 (7.2)* 19.6 (5.6)*,†† 9 (5)*,† 29 (2)†,†† Mean MMSE score loss (%) 0.03 (2.5) 5 (23.5)* 5 (20)*,†† 7.1 (20)*,† 0.8 (4.8)†,†† Baseline mean HV 2.06 (0.3) 1.8 (0.2)* 1.8 (0.3) 1.8 (0.3) 2.1 (0.3) Mean HV loss, % (SD)a 0.76 (1.5) 8.4 (4.4) 8 (4.5)* 10 (4.5)* 2.2 (3.5)†,†† ††C-PiB GCI 1.2 (0.07) 2.86 (0.6)* 2.7 (0.5)* 3.3 (0.8)* 2.27 (0.6)*,†† Mean delay between two DPA (years) 2.2 (0.2) 1.57 (0.3)* 1.63 (0.3)* 1.42 (0.3)* 2.03 (0.1)†,†† DPA-1 GCI# 1.19 (0.1) 1.37 (0.17)* 1.37 (0.1)* 1.37 (0.1)* 1.31 (0.3) DPA-2 GCI## 1.29 (0.15) 1.66 (0.3)* 1.72 (0.3)* 1.52 (0.2) 1.33 (0.2)† Annual % of DPA GCI increase### 4.2 (4.3) 13.1 (13.6)* 15.0 (13)* 8.3 (15) 2.2 (15) Data are mean (SD) or n (%). aLongitudinal data for hippocampal volumes available for 14 patients with Alzheimer’s disease only. *P-value < 0.05 compared to controls. †P-value < 0.05 compared to prodromal Alzheimer’s disease. ††P-value < 0.05 compared to dementia Alzheimer’s disease. #P-value with TSPO genotype as covariate. ##P-value with TSPO genotype and delay between the DPA exams as covariates. ###P-value with age, TSPO genotype and CDR as covariates. AD = Alzheimer’s disease; HAB/LAB/MAB = high/low/mixed affinity binders; HV = hippocampal volume. Table 2 Population characteristics for the DPA-PET longitudinal analysis Controls All AD Prodromal AD Dementia AD Amyloidosis controls n = 13 n = 21 n = 15 n = 6 n = 4 TSPO genotype (MAB:HAB) 6:7 14:7 9:6 5:1 2:2 LAB (excluded from further analysis) 2 3 3 0 2 Age (years) 68.9 (6.5) 63.6 (7.4)* 64.6 (7.7) 60.8 (6.3) 73.5 (8.9)†† Education (years) 11.8 (4.2) 13.4 (3.7) 13.6 (4) 12.8 (3.7) 12.6 (3) Carrier of APOE e4 (%) 0 (0) 13 (62)* 10 (66)* 3 (50)* 2 (50)* Baseline CDR = 0 13 0 0 0 4 Baseline CDR = 0.5 0 15 15 0 0 Baseline CDR > 0.5 0 6 0 6 0 CDR change (%) 0 (0%) 16 (76 %)* 11 (73%)* 5 (83%)* 1 (25%)†,†† Baseline MMSE score 29.7 (0.5) 22.2 (4.7)* 24.6 (2.6)*,†† 16.1 (3.3)*,† 29.5 (0.5)†,†† Follow-up MMSE score 29.4 (0.9) 16.6 (7.2)* 19.6 (5.6)*,†† 9 (5)*,† 29 (2)†,†† Mean MMSE score loss (%) 0.03 (2.5) 5 (23.5)* 5 (20)*,†† 7.1 (20)*,† 0.8 (4.8)†,†† Baseline mean HV 2.06 (0.3) 1.8 (0.2)* 1.8 (0.3) 1.8 (0.3) 2.1 (0.3) Mean HV loss, % (SD)a 0.76 (1.5) 8.4 (4.4) 8 (4.5)* 10 (4.5)* 2.2 (3.5)†,†† ††C-PiB GCI 1.2 (0.07) 2.86 (0.6)* 2.7 (0.5)* 3.3 (0.8)* 2.27 (0.6)*,†† Mean delay between two DPA (years) 2.2 (0.2) 1.57 (0.3)* 1.63 (0.3)* 1.42 (0.3)* 2.03 (0.1)†,†† DPA-1 GCI# 1.19 (0.1) 1.37 (0.17)* 1.37 (0.1)* 1.37 (0.1)* 1.31 (0.3) DPA-2 GCI## 1.29 (0.15) 1.66 (0.3)* 1.72 (0.3)* 1.52 (0.2) 1.33 (0.2)† Annual % of DPA GCI increase### 4.2 (4.3) 13.1 (13.6)* 15.0 (13)* 8.3 (15) 2.2 (15) Controls All AD Prodromal AD Dementia AD Amyloidosis controls n = 13 n = 21 n = 15 n = 6 n = 4 TSPO genotype (MAB:HAB) 6:7 14:7 9:6 5:1 2:2 LAB (excluded from further analysis) 2 3 3 0 2 Age (years) 68.9 (6.5) 63.6 (7.4)* 64.6 (7.7) 60.8 (6.3) 73.5 (8.9)†† Education (years) 11.8 (4.2) 13.4 (3.7) 13.6 (4) 12.8 (3.7) 12.6 (3) Carrier of APOE e4 (%) 0 (0) 13 (62)* 10 (66)* 3 (50)* 2 (50)* Baseline CDR = 0 13 0 0 0 4 Baseline CDR = 0.5 0 15 15 0 0 Baseline CDR > 0.5 0 6 0 6 0 CDR change (%) 0 (0%) 16 (76 %)* 11 (73%)* 5 (83%)* 1 (25%)†,†† Baseline MMSE score 29.7 (0.5) 22.2 (4.7)* 24.6 (2.6)*,†† 16.1 (3.3)*,† 29.5 (0.5)†,†† Follow-up MMSE score 29.4 (0.9) 16.6 (7.2)* 19.6 (5.6)*,†† 9 (5)*,† 29 (2)†,†† Mean MMSE score loss (%) 0.03 (2.5) 5 (23.5)* 5 (20)*,†† 7.1 (20)*,† 0.8 (4.8)†,†† Baseline mean HV 2.06 (0.3) 1.8 (0.2)* 1.8 (0.3) 1.8 (0.3) 2.1 (0.3) Mean HV loss, % (SD)a 0.76 (1.5) 8.4 (4.4) 8 (4.5)* 10 (4.5)* 2.2 (3.5)†,†† ††C-PiB GCI 1.2 (0.07) 2.86 (0.6)* 2.7 (0.5)* 3.3 (0.8)* 2.27 (0.6)*,†† Mean delay between two DPA (years) 2.2 (0.2) 1.57 (0.3)* 1.63 (0.3)* 1.42 (0.3)* 2.03 (0.1)†,†† DPA-1 GCI# 1.19 (0.1) 1.37 (0.17)* 1.37 (0.1)* 1.37 (0.1)* 1.31 (0.3) DPA-2 GCI## 1.29 (0.15) 1.66 (0.3)* 1.72 (0.3)* 1.52 (0.2) 1.33 (0.2)† Annual % of DPA GCI increase### 4.2 (4.3) 13.1 (13.6)* 15.0 (13)* 8.3 (15) 2.2 (15) Data are mean (SD) or n (%). aLongitudinal data for hippocampal volumes available for 14 patients with Alzheimer’s disease only. *P-value < 0.05 compared to controls. †P-value < 0.05 compared to prodromal Alzheimer’s disease. ††P-value < 0.05 compared to dementia Alzheimer’s disease. #P-value with TSPO genotype as covariate. ##P-value with TSPO genotype and delay between the DPA exams as covariates. ###P-value with age, TSPO genotype and CDR as covariates. AD = Alzheimer’s disease; HAB/LAB/MAB = high/low/mixed affinity binders; HV = hippocampal volume. Acknowledgements The authors would like to thank chemical/radiopharmaceutical and nursing staff of Service Hospitalier Frederic Joliot for the synthesis of 11C-PIB and 18F-DPA-714 and patient management, the team of CENIR (Centre de Neuroimagerie de Recherche) in Salpetriere Hospital for patient management during the MRI acquisition, the staff of the CATI (Centre d’Acquisition et de Traitement automatisé de l’Image) for technical support, the study participants. The authors gratefully acknowledge la Fondation pour la Recherche sur la Maladie d’Alzheimer, and the CEA/I2BM/Neurospin-Paris Descartes University collaboration. Funding French Health Ministry (PHRC) under reference PHRC-0054-N 2010 and Institut Roche de Recherche et Medecine Translationelle, European Union’s Seventh Framework Programme (FP7/2007-2013), grant agreement n° HEALTH-F2-2011-278850 (INMiND). The research leading to these results has received funding from the programme ‘Investissements d'avenir ANR-10-IAIHU-06’. Supplementary material Supplementary material is available at Brain online. Abbreviations Abbreviations CDR Clinical Dementia Rating GCI Global Cortical Index MMSE Mini-Mental State Examination PIB Pittsburgh compound B SUVr standardized uptake value ratio References Cagnin A , Brooks DJ , Kennedy AM , Gunn RN , Myers R , Turkheimer FE , et al. In-vivo measurement of activated microglia in dementia . Lancet 2001 ; 358 : 461 – 7 . Google Scholar CrossRef Search ADS PubMed Chauveau F , Van Camp N , Dollé F , Kuhnast B , Hinnen F , Damont A , et al. Comparative evaluation of the translocator protein radioligands 11C-DPA-713, 18F-DPA-714, and 11C-PK11195 in a rat model of acute neuroinflammation . J Nucl Med 2009 ; 50 : 468 – 76 . Google Scholar CrossRef Search ADS PubMed Chupin M , Hammers A , Liu RSN , Colliot O , Burdett J , Bardinet E , et al. Automatic segmentation of the hippocampus and the amygdala driven by hybrid constraints: method and validation . Neuroimage 2009 ; 46 : 749 – 61 . Google Scholar CrossRef Search ADS PubMed Dansokho C , Ait Ahmed D , Aid S , Toly-Ndour C , Chaigneau T , Calle V , et al. Regulatory T cells delay disease progression in Alzheimer-like pathology . Brain 2016 ; 139 : 1237 – 51 . Google Scholar CrossRef Search ADS PubMed de Souza LC , Corlier F , Habert M-O , Uspenskaya O , Maroy R , Lamari F , et al. Similar amyloid-β burden in posterior cortical atrophy and Alzheimer’s disease . Brain 2011 ; 134 : 2036 – 43 . Google Scholar CrossRef Search ADS PubMed Dubois B , Chupin M , Hampel H , Lista S , Cavedo E , Croisile B , et al. Donepezil decreases annual rate of hippocampal atrophy in suspected prodromal Alzheimer’s disease . Alzheimers Dement 2015 ; 11 : 1041 – 9 . Google Scholar CrossRef Search ADS PubMed Dubois B , Feldman HH , Jacova C , Cummings JL , Dekosky ST , Barberger-Gateau P , et al. Revising the definition of Alzheimer’s disease: a new lexicon . Lancet Neurol 2010 ; 9 : 1118 – 27 . Google Scholar CrossRef Search ADS PubMed Edison P , Archer HA , Gerhard A , Hinz R , Pavese N , Turkheimer FE , et al. Microglia, amyloid, and cognition in Alzheimer’s disease: an [11C](R)PK11195-PET and [11C]PIB-PET study . Neurobiol Dis 2008 ; 32 : 412 – 19 . Google Scholar CrossRef Search ADS PubMed Fan Z , Aman Y , Ahmed I , Chetelat G , Landeau B , Ray Chaudhuri K , et al. Influence of microglial activation on neuronal function in Alzheimer’s and Parkinson’s disease dementia . Alzheimers Dement 2015a ; 11 : 608 – 21.e7 . Google Scholar CrossRef Search ADS Fan Z , Brooks DJ , Okello A , Edison P . An early and late peak in microglial activation in Alzheimer’s disease trajectory . Brain 2017 ; 140 : 792 – 803 . Google Scholar PubMed Fan Z , Okello AA , Brooks DJ , Edison P . Longitudinal influence of microglial activation and amyloid on neuronal function in Alzheimer’s disease . Brain 2015b ; 138 : 3685 – 98 . Google Scholar CrossRef Search ADS Hamelin L , Lagarde J , Dorothée G , Leroy C , Labit M , Comley RA , et al. Early and protective microglial activation in Alzheimer’s disease: a prospective study using 18F-DPA-714 PET imaging . Brain 2016 ; 139 : 1252 – 64 . Google Scholar CrossRef Search ADS PubMed Heneka MT , Carson MJ , El Khoury J , Landreth GE , Brosseron F , Feinstein DL , et al. Neuroinflammation in Alzheimer’s disease . Lancet Neurol 2015 ; 14 : 388 – 405 . Google Scholar CrossRef Search ADS PubMed Heneka MT , Kummer MP , Latz E . Innate immune activation in neurodegenerative disease . Nat Rev Immunol 2014 ; 14 : 463 – 77 . Google Scholar CrossRef Search ADS PubMed de Jong HWAM , van Velden FHP , Kloet RW , Buijs FL , Boellaard R , Lammertsma AA . Performance evaluation of the ECAT HRRT: an LSO-LYSO double layer high resolution, high sensitivity scanner . Phys Med Biol 2007 ; 52 : 1505 – 26 . Google Scholar CrossRef Search ADS PubMed Kinkingnéhun S , Sarazin M , Lehéricy S , Guichart-Gomez E , Hergueta T , Dubois B . VBM anticipates the rate of progression of Alzheimer disease: a 3-year longitudinal study . Neurology 2008 ; 70 : 2201 – 11 . Google Scholar CrossRef Search ADS PubMed Kreisl WC , Lyoo CH , Liow J-S , Snow J , Page E , Jenko KJ , et al. Distinct patterns of increased translocator protein in posterior cortical atrophy and amnestic Alzheimer’s disease . Neurobiol Aging 2017 ; 51 : 132 – 40 . Google Scholar CrossRef Search ADS PubMed Kreisl WC , Lyoo CH , Liow J-S , Wei M , Snow J , Page E , et al. ( 11)C-PBR28 binding to translocator protein increases with progression of Alzheimer’s disease . Neurobiol Aging 2016 ; 44 : 53 – 61 . Google Scholar CrossRef Search ADS PubMed Kreisl WC , Lyoo CH , McGwier M , Snow J , Jenko KJ , Kimura N , et al. In vivo radioligand binding to translocator protein correlates with severity of Alzheimer’s disease . Brain 2013 ; 136 : 2228 – 38 . Google Scholar CrossRef Search ADS PubMed Lagarde J , Sarazin M , Bottlaender M . In vivo PET imaging of neuroinflammation in Alzheimer’s disease . J Neural Transm (Vienna) 2017 , in press . doi: 10.1007/s00702-017-1731-x. Lavisse S , García-Lorenzo D , Peyronneau M-A , Bodini B , Thiriez C , Kuhnast B , et al. Optimized quantification of translocator protein radioligand 18F-DPA-714 uptake in the brain of genotyped healthy volunteers . J Nucl Med 2015 ; 56 : 1048 – 54 . Google Scholar CrossRef Search ADS PubMed Lopresti BJ , Klunk WE , Mathis CA , Hoge JA , Ziolko SK , Lu X , et al. Simplified quantification of Pittsburgh Compound B amyloid imaging PET studies: a comparative analysis . J Nucl Med 2005 ; 46 : 1959 – 72 . Google Scholar PubMed Lyoo CH , Ikawa M , Liow J-S , Zoghbi SS , Morse CL , Pike VW , et al. Cerebellum can serve as a pseudo-reference region in Alzheimer disease to detect neuroinflammation measured with pet radioligand binding to translocator protein . J Nucl Med 2015 ; 56 : 701 – 6 . Google Scholar CrossRef Search ADS PubMed Michaud J-P , Rivest S . Anti-inflammatory signaling in microglia exacerbates Alzheimer's disease-related pathology . Neuron 2015 ; 85 : 450 – 2 . Google Scholar CrossRef Search ADS PubMed Minett T , Classey J , Matthews FE , Fahrenhold M , Taga M , Brayne C , et al. Microglial immunophenotype in dementia with Alzheimer’s pathology . J Neuroinflammation 2016 ; 13 : 135 . Google Scholar CrossRef Search ADS PubMed Okello A , Edison P , Archer HA , Turkheimer FE , Kennedy J , Bullock R , et al. Microglial activation and amyloid deposition in mild cognitive impairment: a PET study . Neurology 2009 ; 72 : 56 – 62 . Google Scholar CrossRef Search ADS PubMed Owen DRJ , Gunn RN , Rabiner EA , Bennacef I , Fujita M , Kreisl WC , et al. Mixed-affinity binding in humans with 18-kDa translocator protein ligands . J Nucl Med 2011 ; 52 : 24 – 32 . Google Scholar CrossRef Search ADS PubMed Parbo P , Ismail R , Hansen KV , Amidi A , Mårup FH , Gottrup H , et al. Brain inflammation accompanies amyloid in the majority of mild cognitive impairment cases due to Alzheimer’s disease . Brain 2017 ; 140 : 2002 – 11 . Google Scholar CrossRef Search ADS PubMed Prokop S , Miller KR , Heppner FL . Microglia actions in Alzheimer’s disease . Acta Neuropathol 2013 ; 126 : 461 – 77 . Google Scholar CrossRef Search ADS PubMed Sarazin M , Berr C , De Rotrou J , Fabrigoule C , Pasquier F , Legrain S , et al. Amnestic syndrome of the medial temporal type identifies prodromal AD: a longitudinal study . Neurology 2007 ; 69 : 1859 – 67 . Google Scholar CrossRef Search ADS PubMed Schuitemaker A , Kropholler MA , Boellaard R , van der Flier WM , Kloet RW , van der Doef TF , et al. Microglial activation in Alzheimer’s disease: an (R)-[11C]PK11195 positron emission tomography study . Neurobiol Aging 2013 ; 34 : 128 – 36 . Google Scholar CrossRef Search ADS PubMed Sureau FC , Reader AJ , Comtat C , Leroy C , Ribeiro M-J , Buvat I , et al. Impact of image-space resolution modeling for studies with the high-resolution research tomograph . J Nucl Med 2008 ; 49 : 1000 – 8 . Google Scholar CrossRef Search ADS PubMed Tang Y , Le W . Differential roles of M1 and M2 microglia in neurodegenerative diseases . Mol Neurobiol 2016 ; 53 : 1181 – 94 . Google Scholar CrossRef Search ADS PubMed Vivash L , O’Brien TJ . Imaging microglial activation with TSPO PET: lighting up neurologic diseases? J Nucl Med 2016 ; 57 : 165 – 8 . Google Scholar CrossRef Search ADS PubMed Wiley CA , Lopresti BJ , Venneti S , Price J , Klunk WE , DeKosky ST , et al. Carbon 11-labeled Pittsburgh Compound B and carbon 11-labeled (R)-PK11195 positron emission tomographic imaging in Alzheimer disease . Arch Neurol 2009 ; 66 : 60 – 7 . Google Scholar CrossRef Search ADS PubMed Yokokura M , Mori N , Yagi S , Yoshikawa E , Kikuchi M , Yoshihara Y , et al. In vivo changes in microglial activation and amyloid deposits in brain regions with hypometabolism in Alzheimer’s disease . Eur J Nucl Med Mol Imaging 2011 ; 38 : 343 – 51 . Google Scholar CrossRef Search ADS PubMed Yokokura M , Terada T , Bunai T , Nakaizumi K , Takebayashi K , Iwata Y , et al. Depiction of microglial activation in aging and dementia: positron emission tomography with [(11)C]DPA713 versus [(11)C](R)PK11195 . J Cereb Blood Flow Metab 2017 ; 37 : 877 – 889 . Google Scholar CrossRef Search ADS PubMed © The Author(s) (2018). Published by Oxford University Press on behalf of the Guarantors of Brain. All rights reserved. For permissions, please email: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Brain Oxford University Press

Distinct dynamic profiles of microglial activation are associated with progression of Alzheimer's disease

Loading next page...
 
/lp/ou_press/distinct-dynamic-profiles-of-microglial-activation-are-associated-with-nA8c8uFt8J
Publisher
Oxford University Press
Copyright
© The Author(s) (2018). Published by Oxford University Press on behalf of the Guarantors of Brain. All rights reserved. For permissions, please email: journals.permissions@oup.com
ISSN
0006-8950
eISSN
1460-2156
D.O.I.
10.1093/brain/awy079
Publisher site
See Article on Publisher Site

Abstract

Abstract Although brain neuroinflammation may play an instrumental role in the pathophysiology of Alzheimer’s disease, its actual impact on disease progression remains controversial, being reported as either detrimental or protective. This work aimed at investigating the temporal relationship between microglial activation and clinical progression of Alzheimer’s disease. First, in a large cohort of patients with Alzheimer’s disease we analysed the predictive value of microglial activation assessed by 18F-DPA-714 PET imaging on functional, cognitive and MRI biomarkers outcomes after a 2-year follow-up. Second, we analysed the longitudinal progression of 18F-DPA-714 binding in patients with Alzheimer’s disease by comparison with controls, and assessed its influence on clinical progression. At baseline, all participants underwent a clinical assessment, brain MRI, 11C-PiB, 18F-DPA-714 PET imaging and TSPO genotyping. Participants were followed-up annually for 2 years. At the end of the study, subjects were asked to repeat a second 18F-DPA-714-PET imaging. Initial 18F-DPA-714 binding was higher in prodromal (n = 33) and in demented patients with Alzheimer’s disease (n = 19) compared to controls (n = 17). After classifying patients into slow and fast decliners according to functional (Clinical Dementia Rating change) or cognitive (Mini-Mental State Examination score decline) outcomes, we found a higher initial 18F-DPA-714 binding in slow than fast decliners. Negative correlations were observed between initial 18F-DPA-714 binding and the Clinical Dementia Rating Sum of Boxes score increase, the MMSE score loss and the progression of hippocampal atrophy. This suggests that higher initial 18F-DPA-714 binding is associated with better clinical prognosis. Twenty-four patients with Alzheimer’s disease and 15 control subjects performed a second DPA-PET. We observed an increase of 18F-DPA-714 in patients with Alzheimer’s disease as compared with controls (mean 13.2% per year versus 4.2%) both at the prodromal (15.8%) and at the demented stages (8.3%). The positive correlations between change in 18F-DPA-714 binding over time and the three clinical outcome measures (Clinical Dementia Rating, Mini-Mental State Examination, hippocampal atrophy) suggested a detrimental effect on clinical Alzheimer’s disease progression of increased neuroinflammation after the initial PET examination, without correlation with PiB-PET uptake at baseline. High initial 18F-DPA-714 binding was correlated with a low subsequent increase of microglial activation and favourable clinical evolution, whereas the opposite profile was observed when initial 18F-DPA-714 binding was low, independently of disease severity at baseline. Taken together, our results support a pathophysiological model involving two distinct profiles of microglial activation signatures with different dynamics, which differentially impact on disease progression and may vary depending on patients rather than disease stages. Alzheimer’s disease, inflammation, microglia, PET imaging, neuroprotection Introduction Specific protein inclusions and/or aggregates define most neurodegenerative diseases at the pathological level. In Alzheimer’s disease, beyond the abnormal aggregation of amyloid-β peptide and hyperphosphorylated tau proteins, the pathology has a chronic neuroinflammatory component that involves activation of microglia and astrocytes (Heneka et al., 2015). Misfolded and aggregated proteins bind to pattern recognition receptors expressed by microglia and astroglia, and trigger an innate immune response characterized by the release of inflammatory mediators and/or clearance of pathological proteins. Although accumulating evidence indicate that such chronic neuroinflammatory responses play an instrumental role in the disease pathophysiology, it remains unclear whether they are protective, damaging or both depending on disease stages. A marked accumulation and activation of microglia around amyloid-β plaques has been described in post-mortem brains of patients with Alzheimer’s disease and in animal models of Alzheimer’s disease-like pathology (Prokop et al., 2013). One pathophysiological model proposes that microglia are attracted towards amyloid-β deposits, which they internalize and degrade, thus playing a protective role by promoting clearance of amyloid-β from the brain. In later disease stages, microglia may lose this beneficial effect and acquire a ‘toxic’ phenotype as a consequence of chronic activation and continued production of proinflammatory mediators, in a process that may occur independently of amyloid pathology (Prokop et al., 2013; Michaud and Rivest, 2015). In humans, brain neuroinflammation has been investigated using PET imaging of the 18-kDa translocator protein (TSPO) (Kreisl et al., 2013), which is considered as a marker of microglial activation. However, the studies using the first (11C-PK11195) or the second generation of TSPO radiotracers showed somewhat contradictory results (Lagarde et al., 2017). Varying extents and levels of microglial activation have been identified in prodromal Alzheimer’s disease (Okello et al., 2009; Wiley et al., 2009; Kreisl et al., 2013; Fan et al., 2015a; Hamelin et al., 2016). A relationship was reported between microglial activation and cortical amyloid load (Hamelin et al., 2016; Parbo et al., 2017), but a recent publication including patients with typical and atypical Alzheimer’s disease showed that microglial activation is closely associated with markers of neurodegeneration, suggesting a potential interplay with tau pathology (Kreisl et al., 2017). In addition, no difference in 18F-DPA-714 binding was found between early and late onset of Alzheimer’s disease in a larger sample of patients (Hamelin et al., 2016). Finally, only few studies investigated the predictive value of microglial activation in Alzheimer’s disease progression, and our previous work suggested an early protective role in the largest cohort to date (Hamelin et al., 2016). Only three longitudinal studies of microglial activation in Alzheimer’s disease have been reported up to now, the interpretation of which must be taken cautiously due to small sample sizes. Two studies using 11C-PK11195 (Fan et al., 2015b, 2017) showed a surprising reduction of microglial activation in eight patients with mild cognitive impairment with different amyloidosis status, whereas increased 11C-PK11195 binding was reported in the eight patients with Alzheimer’s disease included. Using 11C-PBR28, a mild increase of TSPO binding in 14 patients with Alzheimer’s disease was found associated with clinical worsening (Kreisl et al., 2016). In light of these results, the spectrum of microglial activation patterns and their impact on disease progression in Alzheimer’s disease remain controversial. The present work aimed at first evaluating the predictive value of microglial activation assessed by 18F-DPA-714 PET imaging in a large cohort of patients with Alzheimer’s disease, based on functional, cognitive and MRI biomarkers outcomes after a 2-year follow-up. Second, in the same cohort we aimed at investigating the longitudinal progression of 18F-DPA-714 binding in patients with Alzheimer’s disease by comparison with controls, and assess its influence on clinical progression. Materials and methods Study design and participants All participants were enrolled in the prospective longitudinal IMABio3 study (NCT01775696), which aimed at assessing the brain neuroinflammatory response in Alzheimer’s disease. The study was approved by a French Ethical Committee. All subjects provided written informed consent prior to participating. Patients with Alzheimer’s disease were included according to the following criteria: (i) progressive episodic memory impairment, characterized by a low free recall not normalized with semantic cueing (Sarazin et al., 2007; Dubois et al., 2010); (ii) absence of extrapyramidal signs; (iii) positive pathophysiological markers of Alzheimer’s disease, defined by CSF Alzheimer’s disease profile {score < 0.8, calculated with the formula amyloid-β42/[240+(1.18 T-tau)]} (de Souza et al., 2011) and amyloid Pittsburgh compound B (PiB)-PET imaging [global cortical index (GCI) > 1.45]. These conservative thresholds were used to ensure the pathophysiological diagnosis of Alzheimer’s disease. Patients with Alzheimer’s disease were defined as prodromal Alzheimer’s disease when their Clinical Dementia Rating (CDR) score was 0.5, and as demented-Alzheimer’s disease when their CDR score was ≥1. Controls were recruited according to the following criteria: (i) Mini-Mental State Examination (MMSE) score ≥ 27/30 and normal neuropsychological assessment; (ii) CDR = 0; (iii) no history of neurological or psychiatric disorders; (iv) no memory complaint or cognitive deficit, and (v) negative amyloid PiB-PET imaging (GCI < 1.4). Controls with a positive amyloid PiB-PET imaging (GCI > 1.45) were defined as amyloidosis-positive controls and analysed separately. We did not include subjects with (i) severe cortical or subcortical vascular lesions; (ii) history of autoimmune and inflammatory diseases or chronic migraines; (iii) history of psychiatric disorders; or (iv) suspicion of alcohol or drugs abuse. No subject was treated with corticoid or non-steroidal anti-inflammatory drug (NSAID) or benzodiazepine, known to interfere with TSPO. Blood samples were drawn to characterize APOE and TSPO genotypes (Hamelin et al., 2016). Based on the rs6971 polymorphism within the TSPO gene, we classified all subjects into three groups: high affinity binders, mixed affinity binders or low affinity binders. At baseline, all participants underwent the same procedure including a complete clinical and neuropsychological assessment, 3 T brain MRI, 11C-PiB and 18F-DPA-714 PET imaging (called DPA-1). Participants were then followed up annually with repeated standardized clinical and neuropsychological assessments and 3 T brain MRI, for 2 years. The last year of the study, subjects were asked to undergo a second 18F-DPA-714 PET imaging (called DPA-2), respecting a minimum of 12 months delay after DPA-1 for patients with Alzheimer’s disease and 24 months delay for controls. Among the 73 patients with Alzheimer’s disease who fulfilled all inclusion criteria, initial DPA-1 PET scan could not be carried out for seven patients because of technical reasons. Eight patients were withdrawn during the 2-year follow-up period (one intercurrent disease, two nursing home, three lost to follow-up, two deaths). Finally, 58 patients with Alzheimer’s disease (including six low affinity binders) with DPA-1 completed the 2-year clinical follow-up. Twenty-seven patients with Alzheimer’s disease underwent a second 18F-DPA-714 PET imaging (n = 18 refused to perform it, n = 3 were at a severe dementia stage making it impossible to carry out a PET acquisition; n = 10 were outside the regulatory limits of the study). PET acquisition failed for technical reasons for three of them. Finally, 24 patients with Alzheimer’s disease (18 prodromal Alzheimer’s disease and six demented Alzheimer’s disease), including three low affinity binders had both DPA-1 and DPA-2 PET scans. The mean delay between DPA-1 and DPA-2 was 1.57 ± 0.3 years. Twenty-nine controls with DPA-1 completed the 2-year clinical follow-up. Among them, eight controls were defined as amyloidosis-positive and analysed separately. A second 18F-DPA-714 PET scan was performed for 15/21 amyloidosis negative controls (n = 3 were outside the regulatory time limit; n = 3 refused DPA-2 PET scan) and 6/8 amyloidosis positive controls (n = 2 refused to perform DPA-2 PET scan). The mean delay between DPA-1 and DPA-2 in controls was 2.2 ± 0.2 years. Clinical, functional and cognitive assessment Standardized neurological and neuropsychological examination was performed annually for 2 years, including the MMSE, CDR scale, Montgomery-Asberg Depression Scale (MADRS) and a standardized cognitive battery for assessing verbal and visual episodic memory, executive functions, gesture praxis, visuo-constructive functions and language. Record of medical events and current treatment were also reported. Definitions of slow and fast decliners After 2 years of follow-up, we used two complementary clinical outcomes to define slow (SD) and fast decliners (FD). The functional (fn) progression was assessed on the basis of the CDR change: patients with Alzheimer’s disease were classified as functional slow decliners when CDR was unchanged and as functional fast decliners when CDR increased. The cognitive (co) progression was assessed by the MMSE score decline over time (cognitive slow decliners: ΔMMSE score ≤ 4 points; cognitive fast decliners: ΔMMSE > 4 points). The cut-off value of 4 points was determined according to knowledge based on longitudinal studies, which have shown that the minimal mean annual rate of progression of cognitive impairment was ∼1–2 points when assessed with the MMSE (Kinkingnéhun et al., 2008). Of note, MMSE scores at the last visit were missing for three patients with Alzheimer’s disease because of too severe cognitive decline. MRI acquisition and hippocampal volume measurements Each participant underwent a MRI examination annually during 2 years, using the same scanner at all visits for reducing equipment-related variability. 3D T1-weighted structural MRI acquisitions were obtained on a 3 T MRI scanner (Siemens Trio, 32 channel system, with a 12-channel head coil for signal reception). This sequence provided a high grey/white matter contrast-to-noise ratio and enabled excellent segmentation and accurate co-registration with PET images (Hamelin et al., 2016). Segmentation of the hippocampus was carried out using an automated method (Chupin et al., 2009). For measuring hippocampal atrophy progression and increase the sensitivity to actual change, the two MRI time points were computed together as previously described (Dubois et al., 2015). Hippocampal volumes were adjusted for head size by correcting for total intracranial volume, derived from SPM8 segmentations. We calculated the percentage of left and right hippocampal volume loss as well as the mean of left + right hippocampal volume loss after 2 years of follow-up. 11C-PiB and 18F-DPA-714 PET imaging procedure Data acquisition MRI and PET scans were performed within 4 months of each other. At baseline, 11C-PiB and 18F-DPA-714 PET scans were performed on the same day. All PET scans were performed using the same camera: a High Resolution Research Tomograph (HRRT; CTI/Siemens Molecular Imaging) (de Jong et al., 2007). A 6-min brain transmission scan was performed before injection of each radioligand using a 137Cs point source to correct the emission scan for tissue attenuation. 11C-PiB-PET (362 ± 45 MBq) and 18F-DPA-714 PET (DPA-1: 197 ± 17 MBq and DPA-2: 197 ± 12 MBq) were injected intravenously, and PET dynamic acquisitions in list mode lasted up to 90 min. All corrections (attenuation, normalization, random and scatter coincidences) were incorporated in an iterative OSEM reconstruction. The partial volume effect was corrected by directly incorporating resolution modelling (i.e. Point Spread Function modelling) inside the iterative algorithm (Sureau et al., 2008) so that no further post-correction was needed. Ten iterations using 16 subsets were used. Dynamic data were binned into 27 time frames (6 × 1 min, 7 × 2 min, 14 × 5 min). Reconstructed dynamic data were realigned for motion correction according to the process of frame-to-reference image registration in Pmod (version 3.5; PMOD Technologies Ltd.). Parametric images were created using Brainvisa software (http://brainvisa.info). Standard uptake value (SUV) parametric images were obtained by: (i) averaging late images (intervals of 40–60 min for 11C-PiB and of 60–90 min for 18F-DPA-714) (Lopresti et al., 2005; Lavisse et al., 2015); and (ii) adjusting for body weight and injected radioligand dose. The cerebellar grey matter was used as a pseudo reference region in both 11C-PiB and 18F-DPA-714 PET analyses to obtain a SUVr (de Souza et al., 2011; Hamelin et al., 2016). Volume of interest analysis The same method of anatomical segmentation was used for both 11C-PiB and 18F-DPA-714 PET images, as described previously (Hamelin et al., 2016). Briefly, an automated segmentation of grey matter was performed to each individual 3D T1-weighted MRI scan using the VBM8 package (http://dbm.neuro.uni-jena.de/vbm/) implemented in SPM8 (Institute of Neurology, London, UK; http://www.fil.ion.ucl.ac.uk/spm/). The segmented MRI scans were co-registered with both 11C-PiB and 18F-DPA-714 parametric images of the subject using a standard mutual information algorithm. The automated anatomical labelling (AAL) atlas was normalized to each individual MRI (using the deformation field extracted from VBM8). Each volume of interest was intersected with the T1 MRI grey matter mask to perform a pseudo-atrophy correction. Then, this new labelling volume was registered to the individual PET space of 11C-PiB and 18F-DPA-714 parametric images using their respective transformation extracted from the PET-MRI co-registration. Similarly, cerebellar grey matter was identified for each subject, eroded (4 mm), and used as a pseudo-reference region. Application of the AAL atlas to the PET data allowed the calculation of 11C-PiB and 18F-DPA-714 uptake in 76 anatomical regions. The volumes of interest were defined separately for the left and right hemispheres and were then pooled into greater anatomical volumes of interest, as previously described (de Souza et al., 2011). Briefly, we defined eight volumes of interest: (i) the frontal cortex; (ii) anterior cingulate; (iii) medium cingulate; (iv) posterior cingulate; (v) precuneus; (vi) parietal cortex; (vii) temporal cortex; and (viii) posterior cortex. A mean 11C-PiB and 18F-DPA-714 SUVr were obtained for each region. As a measure of global cortical burden, we calculated a 11C-PiB and 18F-DPA-714 GCI, representing the subject’s mean SUVr of the neocortical regions cited above (de Souza et al., 2011; Hamelin et al., 2016). Voxel-wise analysis We used SPM8, implemented on a MATLAB platform (Mathworks Inc.). All PET images were spatially normalized onto the standard Montreal Neurological Institute (MNI) and were smoothed with an 8 mm full-width at half-maximum Gaussian filter. In the DPA-PET predictive analysis, we used unpaired t-test comparisons with age, TSPO genotype and MMSE score as covariates to compare fast and slow decliners (P < 0.001 uncorrected). In the DPA-PET longitudinal study, we used an unpaired two-sample t-test with TSPO genotype and interval time between the two PET acquisitions as covariates for comparing Alzheimer’s disease and control groups. The threshold was set at P < 0.001 uncorrected. In addition, evolution of DPA-714 binding in subjects was assessed by using the percentage of DPA changes per year, calculated individually at the voxel level in the MNI space, as follows: (DPA−2DPA−1−1) Δt (1) Where DPA-1 and DPA-2 are DPA-714 SUVr at baseline and follow-up, respectively, and Δt the time interval between the two DPA-714 PET scans. Voxels of percentage were then compared between groups using unpaired t-test with age, TSPO genotype and initial MMSE score as covariates. Correlation analyses used multiple regression design, and TSPO genotype, age and initial MMSE score were included as covariates. For all analyses, a minimum-activated voxel threshold of 20 voxels was applied. Statistical analysis In accordance with recent reports, we pooled the mixed affinity and high affinity binders together and used the TSPO genotype as a covariate in all statistical analyses. In the DPA-PET predictive analysis, we excluded 6/58 low affinity binders with Alzheimer’s disease. The analysis of the DPA-PET longitudinal study was similarly carried out without including low affinity binders. However, as the low affinity binders could be considered as their own controls longitudinally, we verified that our results were not changed when including low affinity binders. Data were analysed using SPSS20 (SPSS Inc., Chicago, Illinois), STATISTICA 6 software (Statsoft) and R, a software environment for statistical computing and graphics (http://www.R-project.org/). Normality of distribution was tested using the Shapiro-Wilk test. Differences between groups were assessed using χ2-test, ANOVA, or Kruskall-Wallis tests when appropriate. ANCOVA, adjusted for age, TSPO genotype and initial MMSE score was used for comparing 18F-DPA-714 binding between groups at baseline (DPA-1). For follow-up comparisons and comparisons of percentage of change, the interval time between both 18F-DPA-714 scans was also added as a covariate (see methodological considerations below for more details). We used one-tailed linear partial correlation analyses with age, TSPO genotype and initial MMSE score as covariates. We used one-tailed tests as the correlation analyses followed comparisons between groups, which gave us a hypothesis regarding the direction of the correlation. Results For clarity, the term ‘DPA-PET predictive analysis’ will refer to assessing how initial microglial activation (DPA-1) is associated with the rapidity of Alzheimer’s disease progression after 2 years of clinical follow-up. The term ‘DPA-PET longitudinal analysis’ will refer to analysing the dynamic profile of microglial activation (DPA-2 versus DPA-1) in patients with Alzheimer’s disease as compared to controls, and its relationship with Alzheimer’s disease clinical progression. DPA-PET predictive analysis Population description Fifty-two patients with Alzheimer’s disease (33 at the prodromal stage and 19 at the dementia stage) and 17 controls were studied (after excluding low affinity binders) (Table 1). There was no significant difference in terms of age or educational level between Alzheimer’s disease and control groups. The prevalence of APOE E4 carriers was higher among patients with Alzheimer’s disease. Detailed neuropsychological evaluations of prodromal and dementia Alzheimer’s disease subgroups are summarized in Supplementary Table 1. Treatments were comparable in the prodromal-Alzheimer’s disease and demented-Alzheimer’s disease subgroups, except for the use of memantine, which was more frequent in demented patients (6/19). There was no change in treatment for Alzheimer’s disease during follow-up. Patients with Alzheimer’s disease exhibited a smaller hippocampal volume at baseline and a higher progression of hippocampal atrophy than controls. After 2 years of follow-up, 17 patients with Alzheimer’s disease had unchanged CDR score and were defined as functional slow decliners and 35 had an increased CDR score and were defined as functional fast decliners. Considering the loss of MMSE scores, 23 patients with Alzheimer’s disease were considered as cognitive slow decliners (mean loss of MMSE = 0.9 ± 2.2) and 26 as cognitive fast decliners (mean loss of MMSE = 8.2 ± 3.2). Detailed rates of decline of the neuropsychological variables in the fast and slow decliners groups over 2 years are shown in Supplementary Table 2. At baseline, functional slow and functional fast decliners as well as cognitive slow and cognitive fast decliners did not differ in terms of educational level, disease duration and baseline neuropsychological variables, but differed in age and MMSE score (fast decliners were younger and had a lower MMSE score). Consequently, age and MMSE scores were systematically used as covariates in all subsequent analyses. Table 1 Population characteristics and baseline 18F-DPA-714 (DPA-1) SUVr in anatomical regions between groups Slow and fast decliners defined by functional outcome (CDR change) Slow and fast decliners defined by cognitive outcome (MMSE loss) Controls All AD patients P-value fn-SD fn-FD P-value co-SD co-FD P-value n = 17 n = 52 n = 17 n = 35 n = 23 n = 26 Age (years) 69.4 (6.4) 67 (10.4) 0.3 74.5 (10) 63.4 (8.6) <0.001 72.6 (10.2) 62.7 (7.8) <0.001 Education (years) 12 (4.3) 13.3 (3.6) 0.4 14.2 (3.4) 12.9 (3.6) 0.2 12.8 (3.8) 12.8 (3.2) 0.13 Carrier of APOE e4 (%) 0 (0) 22 (42) <0.001 12 (70) 18 (51) 0.1 14 (61) 15 (58) 0.5 TSPO genotype (MAB:HAB) 8:9 30:22 0.4 13:4 17:18 0.05 15:8 14:12 0.3 Baseline CDR = 0 (%) 18 (100) 0 (0) <0.001 0 (0) 0 (0) 0.06 0 (0) 0 (0) 0.5 Baseline CDR = 0.5 (%) 0 (0) 33 (64) 11 (65) 22 (63) 15 (65) 18 (70) Baseline CDR>0.5 (%) 0 (0) 19 (36) 6 (35) 13 (37) 8 (35) 8 (30) Baseline MMSE score 29.5 (0.6) 21.1 (5.4) <0.001 23.7 (3.5) 19.8 (5.7) 0.01 22.8 (4) 21 (4.7) 0.15 MMSE score loss 0.3 (0.6) 4.8 (4.6) <0.001 1.2 (3.2) 6.7 (4.1) <0.001 0.9 (2.2) 8.2 (3.2) 0.001 MMSE score loss (%)b 0.4 (2.7) 24.3 (24.4) <0.001 5.2 (14) 34.5 (22.7) <0.001 4.4 (10) 42.0 (19) 0.001 Baseline HVa 2.06 (0.2) 1.74 (0.3) <0.001 1.63 (0.2) 1.80 (0.3) 0.03 2.37 (0.5) 2.54 (0.4) 0.26 HV loss (%)c 0.7 (0.1) 9.01 (5) <0.001 8.5 (5) 9.3 (5.3) 0.6 8.2 (5.4) 9.8 (4.9) 0.23 11C-PiB GCI 1.22 (0.1) 2.89 (0.6) <0.001 2.90 (0.6) 2.85 (0.1) 0.5 2.80 (0.6) 3.00 (0.6) 0.13 Baseline DPA-1 SUVr GCI 1.19 (0.1) 1.37 (0.2) <0.001* 1.50 (0.2) 1.30 (0.1) 0.001** 1.43 (0.2) 1.32 (0.2) 0.04** Frontal 1.19 (0.1) 1.34 (0.2) 0.01* 1.53 (0.2) 1.25 (0.2) 0.001** 1.43 (0.2) 1.28 (0.2) 0.07** Anterior cingulate 1.23 (0.2) 1.35 (0.2) 0.10* 1.55 (0.2) 1.25 (0.1) 0.001** 1.47 (0.2) 1.26 (0.2) 0.006** Medium cingulate 1.26 (0.2) 1.42 (0.2) 0.002* 1.60 (0.2) 1.33 (0.2) 0.001** 1.51 (0.2) 1.36 (0.2) 0.05** Posterior cingulate 1.24 (0.2) 1.45 (0.2) 0.001* 1.57 (0.2) 1.39 (0.2) 0.001** 1.50 (0.2) 1.42 (0.2) 0.09** Precuneus 1.14 (0.1) 1.37 (0.2) <0.001* 1.49 (0.2) 1.31 (0.1) 0.001** 1.42 (0.2) 1.33 (0.2) 0.2** Parietal 1.16 (0.1) 1.38 (0.2) <0.001* 1.52 (0.2) 1.32 (0.2) 0.001** 1.44 (0.2) 1.36 (0.2) 0.04** Temporal 1.09 (0.1) 1.22 (0.1) 0.001* 1.28 (0.1) 1.20 (0.1) 0.004** 1.25 (0.2) 1.19 (0.2) 0.04** Occipital 1.16 (0.1) 1.28 (0.1) 0.003* 1.33 (0.2) 1.25 (0.1) 0.03** 1.32 (0.2) 1.25 (0.2) 0.04** Slow and fast decliners defined by functional outcome (CDR change) Slow and fast decliners defined by cognitive outcome (MMSE loss) Controls All AD patients P-value fn-SD fn-FD P-value co-SD co-FD P-value n = 17 n = 52 n = 17 n = 35 n = 23 n = 26 Age (years) 69.4 (6.4) 67 (10.4) 0.3 74.5 (10) 63.4 (8.6) <0.001 72.6 (10.2) 62.7 (7.8) <0.001 Education (years) 12 (4.3) 13.3 (3.6) 0.4 14.2 (3.4) 12.9 (3.6) 0.2 12.8 (3.8) 12.8 (3.2) 0.13 Carrier of APOE e4 (%) 0 (0) 22 (42) <0.001 12 (70) 18 (51) 0.1 14 (61) 15 (58) 0.5 TSPO genotype (MAB:HAB) 8:9 30:22 0.4 13:4 17:18 0.05 15:8 14:12 0.3 Baseline CDR = 0 (%) 18 (100) 0 (0) <0.001 0 (0) 0 (0) 0.06 0 (0) 0 (0) 0.5 Baseline CDR = 0.5 (%) 0 (0) 33 (64) 11 (65) 22 (63) 15 (65) 18 (70) Baseline CDR>0.5 (%) 0 (0) 19 (36) 6 (35) 13 (37) 8 (35) 8 (30) Baseline MMSE score 29.5 (0.6) 21.1 (5.4) <0.001 23.7 (3.5) 19.8 (5.7) 0.01 22.8 (4) 21 (4.7) 0.15 MMSE score loss 0.3 (0.6) 4.8 (4.6) <0.001 1.2 (3.2) 6.7 (4.1) <0.001 0.9 (2.2) 8.2 (3.2) 0.001 MMSE score loss (%)b 0.4 (2.7) 24.3 (24.4) <0.001 5.2 (14) 34.5 (22.7) <0.001 4.4 (10) 42.0 (19) 0.001 Baseline HVa 2.06 (0.2) 1.74 (0.3) <0.001 1.63 (0.2) 1.80 (0.3) 0.03 2.37 (0.5) 2.54 (0.4) 0.26 HV loss (%)c 0.7 (0.1) 9.01 (5) <0.001 8.5 (5) 9.3 (5.3) 0.6 8.2 (5.4) 9.8 (4.9) 0.23 11C-PiB GCI 1.22 (0.1) 2.89 (0.6) <0.001 2.90 (0.6) 2.85 (0.1) 0.5 2.80 (0.6) 3.00 (0.6) 0.13 Baseline DPA-1 SUVr GCI 1.19 (0.1) 1.37 (0.2) <0.001* 1.50 (0.2) 1.30 (0.1) 0.001** 1.43 (0.2) 1.32 (0.2) 0.04** Frontal 1.19 (0.1) 1.34 (0.2) 0.01* 1.53 (0.2) 1.25 (0.2) 0.001** 1.43 (0.2) 1.28 (0.2) 0.07** Anterior cingulate 1.23 (0.2) 1.35 (0.2) 0.10* 1.55 (0.2) 1.25 (0.1) 0.001** 1.47 (0.2) 1.26 (0.2) 0.006** Medium cingulate 1.26 (0.2) 1.42 (0.2) 0.002* 1.60 (0.2) 1.33 (0.2) 0.001** 1.51 (0.2) 1.36 (0.2) 0.05** Posterior cingulate 1.24 (0.2) 1.45 (0.2) 0.001* 1.57 (0.2) 1.39 (0.2) 0.001** 1.50 (0.2) 1.42 (0.2) 0.09** Precuneus 1.14 (0.1) 1.37 (0.2) <0.001* 1.49 (0.2) 1.31 (0.1) 0.001** 1.42 (0.2) 1.33 (0.2) 0.2** Parietal 1.16 (0.1) 1.38 (0.2) <0.001* 1.52 (0.2) 1.32 (0.2) 0.001** 1.44 (0.2) 1.36 (0.2) 0.04** Temporal 1.09 (0.1) 1.22 (0.1) 0.001* 1.28 (0.1) 1.20 (0.1) 0.004** 1.25 (0.2) 1.19 (0.2) 0.04** Occipital 1.16 (0.1) 1.28 (0.1) 0.003* 1.33 (0.2) 1.25 (0.1) 0.03** 1.32 (0.2) 1.25 (0.2) 0.04** Data are mean (SD) or n (%). aMean of left and right hippocampal volume. bFollow-up MMSE score unavailable for three patients with Alzheimer’s disease because of severe cognitive decline. cFollow-up hippocampal volumes available for n = 35 patients with Alzheimer’s disease. *P-value for DPA-1 SUVr comparisons between controls and Alzheimer’s disease patients with TSPO genotype as covariate. **P-value for DPA-1 SUVr comparisons between slow and fast decliners (age, TSPO genotype and baseline MMSE as covariates). AD = Alzheimer’s disease; co-FD/SD = cognitive fast/slow decliner; fn-FD/SD = functional fast/slow decliner; HAB = high affinity binder; HV = hippocampal volume; MAB = mixed affinity binder. Table 1 Population characteristics and baseline 18F-DPA-714 (DPA-1) SUVr in anatomical regions between groups Slow and fast decliners defined by functional outcome (CDR change) Slow and fast decliners defined by cognitive outcome (MMSE loss) Controls All AD patients P-value fn-SD fn-FD P-value co-SD co-FD P-value n = 17 n = 52 n = 17 n = 35 n = 23 n = 26 Age (years) 69.4 (6.4) 67 (10.4) 0.3 74.5 (10) 63.4 (8.6) <0.001 72.6 (10.2) 62.7 (7.8) <0.001 Education (years) 12 (4.3) 13.3 (3.6) 0.4 14.2 (3.4) 12.9 (3.6) 0.2 12.8 (3.8) 12.8 (3.2) 0.13 Carrier of APOE e4 (%) 0 (0) 22 (42) <0.001 12 (70) 18 (51) 0.1 14 (61) 15 (58) 0.5 TSPO genotype (MAB:HAB) 8:9 30:22 0.4 13:4 17:18 0.05 15:8 14:12 0.3 Baseline CDR = 0 (%) 18 (100) 0 (0) <0.001 0 (0) 0 (0) 0.06 0 (0) 0 (0) 0.5 Baseline CDR = 0.5 (%) 0 (0) 33 (64) 11 (65) 22 (63) 15 (65) 18 (70) Baseline CDR>0.5 (%) 0 (0) 19 (36) 6 (35) 13 (37) 8 (35) 8 (30) Baseline MMSE score 29.5 (0.6) 21.1 (5.4) <0.001 23.7 (3.5) 19.8 (5.7) 0.01 22.8 (4) 21 (4.7) 0.15 MMSE score loss 0.3 (0.6) 4.8 (4.6) <0.001 1.2 (3.2) 6.7 (4.1) <0.001 0.9 (2.2) 8.2 (3.2) 0.001 MMSE score loss (%)b 0.4 (2.7) 24.3 (24.4) <0.001 5.2 (14) 34.5 (22.7) <0.001 4.4 (10) 42.0 (19) 0.001 Baseline HVa 2.06 (0.2) 1.74 (0.3) <0.001 1.63 (0.2) 1.80 (0.3) 0.03 2.37 (0.5) 2.54 (0.4) 0.26 HV loss (%)c 0.7 (0.1) 9.01 (5) <0.001 8.5 (5) 9.3 (5.3) 0.6 8.2 (5.4) 9.8 (4.9) 0.23 11C-PiB GCI 1.22 (0.1) 2.89 (0.6) <0.001 2.90 (0.6) 2.85 (0.1) 0.5 2.80 (0.6) 3.00 (0.6) 0.13 Baseline DPA-1 SUVr GCI 1.19 (0.1) 1.37 (0.2) <0.001* 1.50 (0.2) 1.30 (0.1) 0.001** 1.43 (0.2) 1.32 (0.2) 0.04** Frontal 1.19 (0.1) 1.34 (0.2) 0.01* 1.53 (0.2) 1.25 (0.2) 0.001** 1.43 (0.2) 1.28 (0.2) 0.07** Anterior cingulate 1.23 (0.2) 1.35 (0.2) 0.10* 1.55 (0.2) 1.25 (0.1) 0.001** 1.47 (0.2) 1.26 (0.2) 0.006** Medium cingulate 1.26 (0.2) 1.42 (0.2) 0.002* 1.60 (0.2) 1.33 (0.2) 0.001** 1.51 (0.2) 1.36 (0.2) 0.05** Posterior cingulate 1.24 (0.2) 1.45 (0.2) 0.001* 1.57 (0.2) 1.39 (0.2) 0.001** 1.50 (0.2) 1.42 (0.2) 0.09** Precuneus 1.14 (0.1) 1.37 (0.2) <0.001* 1.49 (0.2) 1.31 (0.1) 0.001** 1.42 (0.2) 1.33 (0.2) 0.2** Parietal 1.16 (0.1) 1.38 (0.2) <0.001* 1.52 (0.2) 1.32 (0.2) 0.001** 1.44 (0.2) 1.36 (0.2) 0.04** Temporal 1.09 (0.1) 1.22 (0.1) 0.001* 1.28 (0.1) 1.20 (0.1) 0.004** 1.25 (0.2) 1.19 (0.2) 0.04** Occipital 1.16 (0.1) 1.28 (0.1) 0.003* 1.33 (0.2) 1.25 (0.1) 0.03** 1.32 (0.2) 1.25 (0.2) 0.04** Slow and fast decliners defined by functional outcome (CDR change) Slow and fast decliners defined by cognitive outcome (MMSE loss) Controls All AD patients P-value fn-SD fn-FD P-value co-SD co-FD P-value n = 17 n = 52 n = 17 n = 35 n = 23 n = 26 Age (years) 69.4 (6.4) 67 (10.4) 0.3 74.5 (10) 63.4 (8.6) <0.001 72.6 (10.2) 62.7 (7.8) <0.001 Education (years) 12 (4.3) 13.3 (3.6) 0.4 14.2 (3.4) 12.9 (3.6) 0.2 12.8 (3.8) 12.8 (3.2) 0.13 Carrier of APOE e4 (%) 0 (0) 22 (42) <0.001 12 (70) 18 (51) 0.1 14 (61) 15 (58) 0.5 TSPO genotype (MAB:HAB) 8:9 30:22 0.4 13:4 17:18 0.05 15:8 14:12 0.3 Baseline CDR = 0 (%) 18 (100) 0 (0) <0.001 0 (0) 0 (0) 0.06 0 (0) 0 (0) 0.5 Baseline CDR = 0.5 (%) 0 (0) 33 (64) 11 (65) 22 (63) 15 (65) 18 (70) Baseline CDR>0.5 (%) 0 (0) 19 (36) 6 (35) 13 (37) 8 (35) 8 (30) Baseline MMSE score 29.5 (0.6) 21.1 (5.4) <0.001 23.7 (3.5) 19.8 (5.7) 0.01 22.8 (4) 21 (4.7) 0.15 MMSE score loss 0.3 (0.6) 4.8 (4.6) <0.001 1.2 (3.2) 6.7 (4.1) <0.001 0.9 (2.2) 8.2 (3.2) 0.001 MMSE score loss (%)b 0.4 (2.7) 24.3 (24.4) <0.001 5.2 (14) 34.5 (22.7) <0.001 4.4 (10) 42.0 (19) 0.001 Baseline HVa 2.06 (0.2) 1.74 (0.3) <0.001 1.63 (0.2) 1.80 (0.3) 0.03 2.37 (0.5) 2.54 (0.4) 0.26 HV loss (%)c 0.7 (0.1) 9.01 (5) <0.001 8.5 (5) 9.3 (5.3) 0.6 8.2 (5.4) 9.8 (4.9) 0.23 11C-PiB GCI 1.22 (0.1) 2.89 (0.6) <0.001 2.90 (0.6) 2.85 (0.1) 0.5 2.80 (0.6) 3.00 (0.6) 0.13 Baseline DPA-1 SUVr GCI 1.19 (0.1) 1.37 (0.2) <0.001* 1.50 (0.2) 1.30 (0.1) 0.001** 1.43 (0.2) 1.32 (0.2) 0.04** Frontal 1.19 (0.1) 1.34 (0.2) 0.01* 1.53 (0.2) 1.25 (0.2) 0.001** 1.43 (0.2) 1.28 (0.2) 0.07** Anterior cingulate 1.23 (0.2) 1.35 (0.2) 0.10* 1.55 (0.2) 1.25 (0.1) 0.001** 1.47 (0.2) 1.26 (0.2) 0.006** Medium cingulate 1.26 (0.2) 1.42 (0.2) 0.002* 1.60 (0.2) 1.33 (0.2) 0.001** 1.51 (0.2) 1.36 (0.2) 0.05** Posterior cingulate 1.24 (0.2) 1.45 (0.2) 0.001* 1.57 (0.2) 1.39 (0.2) 0.001** 1.50 (0.2) 1.42 (0.2) 0.09** Precuneus 1.14 (0.1) 1.37 (0.2) <0.001* 1.49 (0.2) 1.31 (0.1) 0.001** 1.42 (0.2) 1.33 (0.2) 0.2** Parietal 1.16 (0.1) 1.38 (0.2) <0.001* 1.52 (0.2) 1.32 (0.2) 0.001** 1.44 (0.2) 1.36 (0.2) 0.04** Temporal 1.09 (0.1) 1.22 (0.1) 0.001* 1.28 (0.1) 1.20 (0.1) 0.004** 1.25 (0.2) 1.19 (0.2) 0.04** Occipital 1.16 (0.1) 1.28 (0.1) 0.003* 1.33 (0.2) 1.25 (0.1) 0.03** 1.32 (0.2) 1.25 (0.2) 0.04** Data are mean (SD) or n (%). aMean of left and right hippocampal volume. bFollow-up MMSE score unavailable for three patients with Alzheimer’s disease because of severe cognitive decline. cFollow-up hippocampal volumes available for n = 35 patients with Alzheimer’s disease. *P-value for DPA-1 SUVr comparisons between controls and Alzheimer’s disease patients with TSPO genotype as covariate. **P-value for DPA-1 SUVr comparisons between slow and fast decliners (age, TSPO genotype and baseline MMSE as covariates). AD = Alzheimer’s disease; co-FD/SD = cognitive fast/slow decliner; fn-FD/SD = functional fast/slow decliner; HAB = high affinity binder; HV = hippocampal volume; MAB = mixed affinity binder. 18F-DPA-714 cortical binding at baseline (DPA-1) Volume of interest analysis We first compared the whole Alzheimer’s disease cohort to controls. DPA-1 binding was significantly higher in patients with Alzheimer’s disease as compared to controls, especially in the temporal and parietal regions (Supplementary Table 3). No significant difference of DPA-1 binding was found between prodromal Alzheimer’s disease and dementia Alzheimer’s disease subjects. There was a positive correlation between the DPA-1 GCI and the baseline MMSE score (r = 0.26, P = 0.037), left hippocampal volume (r = 0.31, P = 0.04) and cortical volume (r = 0.25, P = 0.04), with TSPO genotype, age and CDR as covariates. Because young patients with Alzheimer’s disease had a faster clinical progression over time, we verified that no 18F-DPA-714 uptake difference was observed between patients with early and late onset of Alzheimer’s disease. In addition, we did not find any correlation between age and initial 18F-DPA-714 uptake. Then we compared the DPA-1 binding between slow and fast decliners. DPA-1 binding was significantly higher in functional slow decliners than in functional fast decliners in all cortical regions as well as for the global index (P = 0.001; with age, TSPO genotype and initial MMSE score as covariates) (Table 1 and Fig. 1A). Similar results were observed when comparing DPA-1 between cognitive slow and cognitive fast decliners, although with lower statistical power (GCI: P = 0.04, Table 1). Figure 1 View largeDownload slide DPA-PET predictive study: scatter and box plots showing global and regional 18F-DPA-714 (DPA-1) SUVr in slow and fast Alzheimer’s disease decliners. (A) Comparison between functional slow decliners (n = 17, in red) and functional fast decliners (n = 35, in blue). (B) Comparison between cognitive slow decliners (n = 15, in red) and cognitive fast decliners (n = 18, in blue) in the subgroup of patients with Alzheimer’s disease (AD) at the prodromal stage at inclusion. *P < 0.05 with age, TSPO genotype and initial MMSE as covariates. Figure 1 View largeDownload slide DPA-PET predictive study: scatter and box plots showing global and regional 18F-DPA-714 (DPA-1) SUVr in slow and fast Alzheimer’s disease decliners. (A) Comparison between functional slow decliners (n = 17, in red) and functional fast decliners (n = 35, in blue). (B) Comparison between cognitive slow decliners (n = 15, in red) and cognitive fast decliners (n = 18, in blue) in the subgroup of patients with Alzheimer’s disease (AD) at the prodromal stage at inclusion. *P < 0.05 with age, TSPO genotype and initial MMSE as covariates. Finally, we compared DPA-1 binding in slow and fast decliners within the subgroups of patients at either prodromal stage (CDR = 0.5 at inclusion) or dementia stage (CDR ≥ 1 at baseline) separately. Regarding the prodromal patients with Alzheimer’s disease, DPA-1 binding was significantly higher in functional slow decliners (n = 11) than in functional fast decliners (n = 22), especially in the anterior regions, as well as in cognitive slow decliners (n = 15) than in cognitive fast decliners (n = 18) (Fig. 1). Concerning the demented patients with Alzheimer’s disease, we also found higher DPA-1 binding in the functional slow decliners (n = 6) than in the functional fast decliners (n = 13), but no significant difference was observed between cognitive slow (n = 8) and cognitive fast decliners (n = 8), taking into account that MMSE scores were lacking at the last visit for three patients with Alzheimer’s disease because of too severe cognitive impairment. Voxel-wise comparisons The voxel-wise analysis showed a higher 18F-DPA-714 binding in functional slow decliners when compared to functional fast decliners (P < 0.001, uncorrected; with age, TSPO and MMSE as covariates), which is more pronounced in the anterior regions in prodromal patients with Alzheimer’s disease (Fig. 2A, upper panels). A similar pattern of microglial activation was observed when we compared cognitive slow and cognitive fast decliner patients (P < 0.001, uncorrected; with age, TSPO and MMSE as covariates) (Fig. 2A, lower panels). Figure 2 View largeDownload slide DPA-PET predictive study: statistical parametric mapping analysis of 18F-DPA-714 SUVr at baseline (DPA-1). (A) Voxel-wise comparisons between slow and fast decliners. (i) Comparison of DPA-1 binding at baseline between functional slow decliners (fn-SD) and functional fast decliners (fn-FD) in the whole Alzheimer’s disease population (n = 52, left) and in the prodromal Alzheimer’s disease population (n = 33, right). (ii) Comparison of DPA-1 binding at baseline between cognitive slow decliners (co-SD) and cognitive fast decliners (co-FD) in the whole Alzheimer’s disease population (n = 49, left) and in the prodromal Alzheimer’s disease population (n = 33, right). **P < 0.001 uncorrected with age, TSPO genotype and initial MMSE as covariates. (B) Correlations between DPA-1 binding and functional, cognitive or MRI biomarkers outcomes using voxel-wise analysis. (i) Correlations between DPA-1 binding and the MMSE change (% of progression), in the whole Alzheimer’s disease group (n = 49, left) and in the prodromal Alzheimer’s disease group (n = 33, right). (ii) Correlations between DPA-1 binding and the left hippocampal volume change, in the whole Alzheimer’s disease group (n = 36, left) and in the prodromal Alzheimer’s disease group (n = 26, right). *P < 0.05 with age, TSPO genotype and initial MMSE as covariates. AD = Alzheimer’s disease; HV = hippocampal volume. Figure 2 View largeDownload slide DPA-PET predictive study: statistical parametric mapping analysis of 18F-DPA-714 SUVr at baseline (DPA-1). (A) Voxel-wise comparisons between slow and fast decliners. (i) Comparison of DPA-1 binding at baseline between functional slow decliners (fn-SD) and functional fast decliners (fn-FD) in the whole Alzheimer’s disease population (n = 52, left) and in the prodromal Alzheimer’s disease population (n = 33, right). (ii) Comparison of DPA-1 binding at baseline between cognitive slow decliners (co-SD) and cognitive fast decliners (co-FD) in the whole Alzheimer’s disease population (n = 49, left) and in the prodromal Alzheimer’s disease population (n = 33, right). **P < 0.001 uncorrected with age, TSPO genotype and initial MMSE as covariates. (B) Correlations between DPA-1 binding and functional, cognitive or MRI biomarkers outcomes using voxel-wise analysis. (i) Correlations between DPA-1 binding and the MMSE change (% of progression), in the whole Alzheimer’s disease group (n = 49, left) and in the prodromal Alzheimer’s disease group (n = 33, right). (ii) Correlations between DPA-1 binding and the left hippocampal volume change, in the whole Alzheimer’s disease group (n = 36, left) and in the prodromal Alzheimer’s disease group (n = 26, right). *P < 0.05 with age, TSPO genotype and initial MMSE as covariates. AD = Alzheimer’s disease; HV = hippocampal volume. Correlations between DPA-1 binding and functional, cognitive and MRI biomarkers outcomes In the whole group of patients with Alzheimer’s disease, we found a negative correlation between DPA-1 binding in the frontal regions and the percentage of increase of the CDR-Sum of Boxes (SOB) (r = −0.27, P = 0.03; with age, TSPO and MMSE as covariates). Similarly, we found a negative correlation between global cortical DPA-1 binding and the percentage of decrease of the MMSE score (r = −0.28 P = 0.028 for the GCI; with age, TSPO genotype and initial MMSE as covariates). These results suggest that the higher the DPA-1 binding, the lower the functional and cognitive declines. Considering the prodromal Alzheimer’s disease group separately, a significant negative correlation was observed with the percentage of MMSE decline (P < 0.05), but not with the CDR-SOB change. We also correlated DPA-1 binding with the evolution over time of composite scores reflecting different cognitive domains (long-term memory, short-term memory, instrumental functions, executive functions) in the mild cognitive impairment subjects of the predictive cohort (as patients with mild cognitive impairment were more numerous, and because we had more missing longitudinal neuropsychological data in demented patients, leading to floor effect). We found significant negative correlations between DPA-1 binding in the frontal and anterior cingulate regions and the decline of the composite scores reflecting long-term memory [sum of total recall in the Free and Cued Selective Reminding Test (FCSRT) and temporo-spatial orientation, r = −0.31, P = 0.04 for the frontal region and r = −0.37, P = 0.01 for the anterior cingulate region], and short-term memory (sum of encoding score in the FCSRT and digit spans, r = −0.34, P = 0.029 for the frontal region and r = −0.44, P = 0.006 for the anterior cingulate region). We found no significant correlation with the composite scores reflecting instrumental and executive functions. Voxel-wise analysis confirms these results, showing negative correlations between DPA-1 binding and MMSE score decrease in frontal and parieto-occipital cortical areas (Fig. 2B, upper panels). Progression of hippocampal atrophy was used as a MRI biomarker of Alzheimer’s disease progression. The percentage of decrease of hippocampal volume was negatively correlated with the DPA-1 uptake. Highest correlations were found between the left hippocampus atrophy rate and the frontotemporal DPA-1 uptake in prodromal patients with Alzheimer’s disease. These data suggest that higher DPA-1 uptake was associated with lower loss of hippocampal volume. Voxel-wise analysis confirmed this correlation in the same cortical areas (Fig. 2B, lower panels). There was no significant correlation between DPA-1 uptake and the percentage of decrease of cortical grey matter volume. PiB cortical binding and Alzheimer’s disease progression We did not find any significant results when similar analyses as described above were carried out with using 11C-PiB binding (no difference of PiB binding between slow and fast decliners as defined either by CDR or MMSE scores changes; no correlation between PiB binding and functional, cognitive or MRI outcomes). Thus, 11C-PiB binding was not significantly associated with the rate of clinical progression. As previously published, PiB cortical binding correlated with DPA-1 binding (r = 0.5, P = 0.03). DPA-PET longitudinal analysis Population description All of the following results were obtained without including the low affinity binders. Patients (n = 21) and controls (n = 13) did not differ with regards to age and educational level. The mean loss in MMSE score after 2 years was 5 points in prodromal patients with Alzheimer’s disease, 7.1 points in demented patients with Alzheimer’s disease and 0.03 points in controls. The mean interval between both PET scans was 1.6 years in prodromal patients with Alzheimer’s disease, 1.4 years in demented patients with Alzheimer’s disease and 2.2 years in controls. Detailed neuropsychological data of the population are summarized in Supplementary Table 4. We also verified that DPA-1 binding in the patients who performed DPA-2 was not significantly different from that of those who did not (DPA-1 GCI = 1.37 ± 0.17 in the 21 patients who had both DPA-1 and DPA-2, and DPA-1 GCI = 1.36 ± 0.2 in the 31 patients who did not perform DPA-2). Methodological considerations Cerebellar grey matter as a pseudo-reference region We verified that cerebellar grey matter 18F-DPA-714 uptake (i) did not differ between patients and controls neither in the first (DPA-1) nor in the second (DPA-2) PET scan; (ii) remained stable individually between DPA-1 and DPA-2 in both patients with Alzheimer’s disease (P = 0.1) and controls (P = 0.6), without any difference between groups (Supplementary Fig. 1); and (iii) was not correlated with MMSE score, age, APOE genotype and cortical volume measured by MRI. These results provide additional arguments to those previously published for using the cerebellar grey matter as a pseudo-reference region for quantifying 18F-DPA-714 binding. Quantification of longitudinal changes in 18F-DPA-714 binding Changes in DPA binding were expressed as an annualized percentage. Because this supposes a linear progression over time, we verified for all analyses that results were not modified when using the global percentage of DPA binding variation, while including the delay between both DPA-PET scans as a covariate. Comparisons of DPA-1, DPA-2 and DPA binding progression between controls and patients with Alzheimer’s disease Volume of interest analysis As shown for DPA-1, DPA-2 binding was higher in patients with Alzheimer’s disease than in controls in all volumes of interest, especially at the prodromal stage (P < 0.01) (Supplementary Table 5). In the control group, 18F-DPA-714 binding was stable with time (4.2 ± 4.3% variation per year for the GCI) (Fig. 3A). In contrast, 18F-DPA-714 binding significantly increased over time in the Alzheimer’s disease group as compared to controls (13.2% per year for the GCI) (P = 0.02), especially at the prodromal stage (15.1% per year versus 8.3% per year at the dementia stage). Individual analysis showed heterogeneous 18F-DPA-714 binding progression profiles among patients with Alzheimer’s disease (Fig. 3B and C). Figure 3 View largeDownload slide Longitudinal change of 18F-DPA-714 GCI. (A and B) Individual 18F-DPA-714 GCI at baseline (DPA-1) and follow-up (DPA-2) in controls (n = 13) and in patients with Alzheimer’s disease (n = 21, prodromal Alzheimer’s disease in black, dementia Alzheimer’s disease in grey). (C) Mean 18F-DPA-714 GCI at baseline and follow-up in patients with low DPA-1 binding [in red, SUVr ≤ 1.39, which is the optimal cut-off value for GCI derived from the receiver operating characteristic (ROC) curve analysis for differentiating slow and fast decliners] and with high DPA-1 binding (in blue, SUVr > 1.39) and in controls (in grey). (D) Surface plot representing the percentage of increase of the CDR-SOB score (z-axis) in relation with the DPA-1 cortical binding (x-axis) and the annualized percentage of DPA cortical binding increase (y-axis). Figure 3 View largeDownload slide Longitudinal change of 18F-DPA-714 GCI. (A and B) Individual 18F-DPA-714 GCI at baseline (DPA-1) and follow-up (DPA-2) in controls (n = 13) and in patients with Alzheimer’s disease (n = 21, prodromal Alzheimer’s disease in black, dementia Alzheimer’s disease in grey). (C) Mean 18F-DPA-714 GCI at baseline and follow-up in patients with low DPA-1 binding [in red, SUVr ≤ 1.39, which is the optimal cut-off value for GCI derived from the receiver operating characteristic (ROC) curve analysis for differentiating slow and fast decliners] and with high DPA-1 binding (in blue, SUVr > 1.39) and in controls (in grey). (D) Surface plot representing the percentage of increase of the CDR-SOB score (z-axis) in relation with the DPA-1 cortical binding (x-axis) and the annualized percentage of DPA cortical binding increase (y-axis). Voxel-wise analysis DPA-1 and DPA-2 binding were higher in patients with Alzheimer’s disease than in controls, especially at the prodromal stage in temporo-parietal regions (P < 0.001, uncorrected) (Fig. 4A and B). The annual progression of 18F-DPA-714 binding was higher in patients with Alzheimer’s disease than in controls in fronto-parietal regions (P < 0.01, uncorrected) (Fig. 4C). Figure 4 View largeDownload slide DPA-PET longitudinal study. Statistical parametric mapping comparisons of 18F-DPA-714 SUVr between Alzheimer’s disease group (n = 21) and controls (n = 13) (left) and between prodromal Alzheimer’s disease group (n = 15) and controls (n = 13) (right). (A) DPA-1, **P < 0.001 uncorrected, TSPO genotype as a covariate. (B) DPA-2, **P < 0.001 uncorrected, TSPO genotype and delay time as covariates. (C) Annualized percentage of DPA binding progression, *P < 0.01 uncorrected, age, TSPO genotype and initial CDR as covariates. AD = Alzheimer’s disease. Figure 4 View largeDownload slide DPA-PET longitudinal study. Statistical parametric mapping comparisons of 18F-DPA-714 SUVr between Alzheimer’s disease group (n = 21) and controls (n = 13) (left) and between prodromal Alzheimer’s disease group (n = 15) and controls (n = 13) (right). (A) DPA-1, **P < 0.001 uncorrected, TSPO genotype as a covariate. (B) DPA-2, **P < 0.001 uncorrected, TSPO genotype and delay time as covariates. (C) Annualized percentage of DPA binding progression, *P < 0.01 uncorrected, age, TSPO genotype and initial CDR as covariates. AD = Alzheimer’s disease. Correlation between DPA binding progression and functional, cognitive and MRI biomarker outcome To avoid any confounding bias, age, TSPO binding status and initial MMSE score were defined as covariates for all correlation analyses (either by volume of interest or voxel-wise method). We found positive correlations between the increase in 18F-DPA-714 binding and the annualized percentage of CDR-SOB increase in all volumes of interest, both in the whole group and in prodromal patients with Alzheimer’s disease (Supplementary Table 6). These data suggest that the increase in 18F-DPA-714 binding is associated with functional decline over time even at the prodromal stage. Voxel-wise analysis confirmed these results, showing significant correlations in frontal and parieto-temporal regions (Fig. 5A). Figure 5 View largeDownload slide DPA-PET longitudinal study. Statistical parametric mapping showing correlations between 18F-DPA-714 SUVr [in the Alzheimer’s disease group (left), and in the prodromal Alzheimer’s disease group (right)] and (A) CDR-SOB increase, *P < 0.01 uncorrected; (B) MMSE score loss, *P < 0.01 uncorrected; and (C) left hippocampal volume volume loss, *P < 0.01 uncorrected. Age, TSPO genotype and initial MMSE as covariates. HV = hippocampal volume. Figure 5 View largeDownload slide DPA-PET longitudinal study. Statistical parametric mapping showing correlations between 18F-DPA-714 SUVr [in the Alzheimer’s disease group (left), and in the prodromal Alzheimer’s disease group (right)] and (A) CDR-SOB increase, *P < 0.01 uncorrected; (B) MMSE score loss, *P < 0.01 uncorrected; and (C) left hippocampal volume volume loss, *P < 0.01 uncorrected. Age, TSPO genotype and initial MMSE as covariates. HV = hippocampal volume. We also found a significant positive correlation between the annualized percentage of increase in DPA binding and the decrease of MMSE score over 2 years, especially in the frontal regions (Supplementary Table 6), confirmed by the voxel-wise analysis (Fig. 5B). These results suggest that progression of 18F-DPA-714 binding is associated with cognitive decline over time. Finally, we found significant positive correlations between the annualized percentage of increase in DPA binding and the decrease in left hippocampal volume, in frontal and parieto-temporal cortex over 2 years (Supplementary Table 6). This finding was replicated by the voxel-wise analysis (Fig. 5C). Correlations between PiB-PET at baseline and DPA binding progression or clinical Alzheimer’s disease outcomes We found no correlation between initial 11C-PiB binding and either the percentage of DPA binding progression, CDR-SOB progression, changes in MMSE score or the decrease of hippocampal volume over 2 years. Relationships between DPA-1 binding, DPA binding progression and clinical decline We found a significant negative correlation between the DPA-1 uptake and the DPA binding progression (r = −0.64, P = 0.002 for the GCI with age, TSPO genotype and initial MMSE score as covariates), suggesting that patients with the highest DPA-1 binding are those who tend to have the lowest increase in DPA binding over time (Fig. 3D). Interestingly, unlike the extent of initial DPA-1 binding, its longitudinal increase is associated with higher cognitive and functional decline (Fig. 5) as well as faster hippocampal atrophy. This dynamic pattern is observed in both prodromal and demented stages of Alzheimer’s disease, even if the statistical significance was stronger in prodromal Alzheimer’s disease. Of note, results remained similar when we included age, disease duration, initial MMSE score or initial CDR score as covariates. Amyloidosis positive controls Among six amyloidosis positive controls, who were also analysed, including two low affinity binders, only one developed memory cognitive decline during the 2-year follow-up period. The annualized percentages of increase in 18F-DPA-714 binding in these six amyloidosis-positive controls were +24.66%, −4.27%, −10.35%, −1.22%, −8.07% and +17.58%, the two latter percentages corresponding to the low affinity binders. Of note, the only amyloidosis-positive control who developed memory impairment during the follow-up period was a low affinity binder and had both a relatively low DPA-1 GCI of 1.10, and a relatively high annualized percentage of increase in 18F-DPA-714 binding of +17.58%. However, another amyloidosis positive control (high affinity binder) had a low DPA-1 GCI of 1.12 with an annualized percentage of increase in 18F-DPA-714 binding of 24.66% and did not develop cognitive impairment during follow-up. Discussion This study provides new insights into the temporal relationships between microglial activation as measured by 18F-DPA-714 binding and Alzheimer’s disease clinical progression. First, we found in a large cohort of 52 patients with Alzheimer’s disease that higher initial 18F-DPA-714 binding is associated with better clinical prognosis after a 2-year follow-up. Second, the longitudinal DPA-PET study showed that the subsequent increase in 18F-DPA-714 binding observed in patients with Alzheimer’s disease (as compared with controls) was linked to Alzheimer’s disease worsening. Finally, patients with lowest initial 18F-DPA-714 binding had the greatest subsequent increase of microglial activation and unfavourable clinical evolution, while patients with highest initial 18F-DPA-714 binding had the lowest subsequent increase of microglial activation and more favourable clinical evolution, independently of the initial cortical amyloid load. Altogether, these results support the hypothesis that two distinct dynamic profiles of microglial activation differentially impact on disease progression in patients with Alzheimer’s disease, being either beneficial or detrimental, and may vary depending on patients rather than disease stages. Compared with previous works using TSPO PET imaging in Alzheimer’s disease, one of the strengths of the current study is the larger sample of subjects (Alzheimer’s disease and controls) and the choice of very strict inclusion criteria. In addition to positivity for Alzheimer’s disease pathophysiological markers (PiB-PET and CSF biomarkers), we excluded all subjects with known medical events that could influence TSPO binding, such as history of inflammatory diseases and suspicion of alcohol abuse. No change in Alzheimer’s disease treatment during the follow-up could explain the difference in clinical progression. In addition, we used a high sensitivity PET scanner (HRRT) together with 18F-DPA-714, a second generation TSPO tracer, which provides better sensitivity than 11C-PK11195 for evaluating increased TSPO expression (Chauveau et al., 2009; Yokokura et al., 2017), and for which mixed affinity and high affinity binders can be pooled more easily than when using 11C-PBR28, of known much higher affinity ratio between high affinity and low affinity binders (Owen et al., 2011; Hamelin et al., 2016). Furthermore, TSPO genotype was added as a covariate in all statistical analyses and low affinity binders were excluded from all analyses. The advantage of having a large cohort is counterbalanced by the lack of arterial blood samples, which may be a limitation to fully quantify the DPA-714 binding. The SUVr method assumes no difference in cerebellar binding at baseline and follow-up. While we observed the stability of cerebellar 18F-DPA-714 binding over time in patients with Alzheimer’s disease and controls, in congruence with previous works using 11C-PBR28 radioligand, it has not actually been validated using full quantification of DPA-714 data (Lyoo et al., 2015; Kreisl et al., 2016). Previous transversal studies using TSPO PET imaging reported controversial findings regarding the protective or deleterious impact of microglial activation in Alzheimer’s disease. Some studies showed that 11C-PK11195 binding correlated with clinical severity, cortical volume atrophy and fluorodeoxyglucose (FDG) hypometabolism, suggesting a toxic effect of microglial activation (Cagnin et al., 2001; Edison et al., 2008; Yokokura et al., 2011). Conversely, other reports showed no correlation with cognitive scores regardless of whether patients with mild cognitive impairment progressed to dementia or remained clinically stable (Schuitemaker et al., 2013). Using second generation TSPO tracers, one study suggested that 11C-PBR28 uptake correlated with cortical volume loss and decrease in several cognitive performances (Lyoo et al., 2015). In contrast, we previously showed that 18F-DPA-714 binding correlated positively with MMSE scores and grey matter volume, especially at the prodromal stage, suggesting a protective role of microglial activation (Hamelin et al., 2016). Only three longitudinal studies aimed at understanding TSPO binding progression have been reported in Alzheimer’s disease so far, which used different radioligands and different quantification methods. Two of them used 11C(R)-PK11195 tracer and a cluster-based approach for defining the reference region (Fan et al., 2015b, 2017). After a 16-month follow-up period, voxel-wise analysis suggested that the increase in microglial activation in patients with Alzheimer’s disease (n = 8, MMSE = 21/30) was positively correlated with amyloid deposition and inversely correlated with regional cerebral metabolic rate (Fan et al., 2015b). However, the small number of patients with Alzheimer’s disease, the negativity of amyloid imaging in one of them (calling into question Alzheimer’s disease diagnosis) and the lack of longitudinal follow-up of healthy controls limited the interpretation of these data. The same group reported findings in eight subjects with mild cognitive impairment (MMSE = 27.6/30) with different amyloid status, four of eight being amyloid-negative based on PiB-PET imaging (Fan et al., 2017). The authors described opposite results as compared to those observed in their patients with Alzheimer’s disease, with a longitudinal reduction in microglial activation associated with a marginal increase in amyloid load over 14 months. Fan et al. thus proposed the existence of two peaks of microglial activation in the course of Alzheimer’s disease, with an early protective peak and a late pro-inflammatory deleterious peak. However, the very small number of mild cognitive impairment due to patients with Alzheimer’s disease analysed in this study (only four) raises question about the robustness of such a model. The third longitudinal PET study used 11C-PBR28 and cerebellum as a pseudo-reference region for quantifying TSPO binding in 14 amyloid-positive patients with Alzheimer’s disease at different disease stages (MMSE ranging from 14 to 30; n = 9 patients with CDR = 0.5, n = 5 patients with CDR = 1) and eight controls (Kreisl et al., 2016). 11C-PBR28 binding increased in temporo-parietal regions from 3.9% to 6.3% per year in patients versus 0.5% to 1% per year in controls. The increase in TSPO binding correlated with functional worsening on CDR-SOB and with reduced cortical volume. The small number of slow and fast decliners defined by changes in the CDR score (n = 5 and n = 9, respectively) did not allow investigating the binding progression within the prodromal Alzheimer’s disease subgroups separately. Our results are in accordance with Kreisl et al. (2016), as we found an increase in 18F-DPA-714 binding over 24 months in patients with Alzheimer’s disease as compared with controls (13.2% per year versus 4.2%), both at the prodromal (15.8%) and at the demented stages (8.3%). The positive correlations between the increase in DPA binding and the increase in CDR-SOB scores, MMSE score decrease and rate of hippocampal atrophy suggested a detrimental effect on clinical Alzheimer’s disease progression of the subsequent increase in microglial activation, which was not influenced by the initial fibrillar amyloid load. Determining whether such a profile is similar in prodromal and demented stages of Alzheimer’s disease remains difficult. Indeed, one limitation of our work is inherent to clinical PET longitudinal studies, as the second PET examination could not be performed in patients at severe dementia stage for obvious ethical reasons. Consequently, the number of patients at the prodromal stage (n = 15) was higher than that of patients at the demented stage (n = 6), precluding sufficiently powerful statistical analyses in the demented subgroup. A larger number of subjects would be necessary to draw precise conclusions. This deleterious effect of an increasing microglial activation may appear inconsistent with our findings from the DPA-PET predictive analysis, which evidenced a protective effect on Alzheimer’s disease progression of higher initial 18F-DPA-714 binding, confirming our previous studies (Hamelin et al., 2016). Here we used both functional (CDR score changes) and cognitive (MMSE score decline) outcome measures for defining slow and fast decliners. In addition, we assessed correlations between 18F-DPA-714 binding and clinical and MRI outcome scores by both volume of interest and voxel-wise methods. Importantly, 18F-DPA-714 binding at baseline as well as the subsequent increase in 18F-DPA-714 binding were highly heterogeneous among patients with Alzheimer’s disease. Interestingly, patients with the highest initial 18F-DPA-714 binding display the lowest subsequent increase in 18F-DPA-714 binding over time, and better prognosis. Conversely, patients with the lowest initial 18F-DPA-714 binding show a higher subsequent increase in 18F-DPA-714 binding and a worst prognosis. We must nevertheless acknowledge that in patients with high DPA-1, the course of microglial activation before inclusion in the study remains to be explored. The follow-up of amyloidosis-positive controls was poorly informative, as the only subject who developed memory impairment during the follow-up period was a low affinity binder. The others remained clinically stable over time and exhibited heterogeneous 18F-DPA-714 binding progressions. One of the limitations of this study is the low number of subjects in this subgroup and the too short follow-up duration of these patients. Following-up these subjects for a longer period will be highly informative for deciphering the role of microglial activation in preclinical Alzheimer’s disease. The concept of diverse functional phenotypes of immune cells, ranging from pro-inflammatory to immunosuppressive, has been expended to microglia (Tang and Le, 2016). Importantly, the use of the traditional ‘M1/M2 paradigm’ of activation status is now considered as over-simplified, and functional differentiation patterns of activated microglia extend much beyond the classical proinflammatory M1-like and neuroprotective M2-like phenotypes, rather spanning a full spectrum of activation patterns (Heneka et al., 2014). The balance of pro-inflammatory and neuroprotective microglial activation is highly complex, especially in Alzheimer’s disease, in which microglia may exhibit mixed activation phenotypes. In this line, a recent study on 299 human brains evidenced an alteration of microglial immunophenotypes in association with ageing and the development of Alzheimer’s disease dementia, highlighting the complexity and diversity of microglial responses (Minett et al., 2016). Up to now, no PET tracer has been able to distinguish between different microglial subtypes or to capture the transition between such different activation states (Vivash and O’Brien, 2016). Beyond the current classical model of neuroinflammation, proposing a homogeneous early protective effect that subsequently turns into a later detrimental effect, our data rather suggest a more complex model based on both distinct patterns and dynamics of microglial activation signatures among patients with Alzheimer’s disease, which differently impact on disease progression and contribute to shape diverse clinical evolution profiles. Considering the positive correlation between DPA-1 and PiB binding, the protective microglial signature may reflect an interaction with misfolded deposited proteins and removal of neurotoxic aggregates, especially for fibrillar amyloid deposition (Fan et al., 2015b; Parbo et al., 2017). The detrimental microglial signature could reflect an interplay with neuronal injury processes, possibly independently of amyloid pathology when considering the absence of correlation between PiB uptake and the increase in 18F-DPA-714 binding. Protective or deleterious microglial signatures are both observed at the prodromal stage, and mostly do not seem related to the severity of the disease. Our data rather suggest that the relative extent and dynamics of beneficial and detrimental microglial activation signatures may vary among patients, thus translating into different clinical evolution profiles. Identifying the factors modulating this balance between such different signatures will be of major importance for both prognostic and therapeutic purposes. In this line, our previous studies in a mouse model of Alzheimer’s disease-like pathology suggested that peripheral modulation of a given T cell population could impact on the rate of disease progression at least partially by modulating microglial responses (Dansokho et al., 2016). Hence, different subtypes of patients with Alzheimer’s disease may benefit from different treatment protocols and/or innovative therapeutic approaches targeting inflammatory and immune responses. PET imaging using 18F-DPA-714 appears as a highly valuable tool for identifying target populations and assessing drug efficacy in such studies, even if we cannot exclude that some other biological parameters could also influence Alzheimer’s disease progression. Table 2 Population characteristics for the DPA-PET longitudinal analysis Controls All AD Prodromal AD Dementia AD Amyloidosis controls n = 13 n = 21 n = 15 n = 6 n = 4 TSPO genotype (MAB:HAB) 6:7 14:7 9:6 5:1 2:2 LAB (excluded from further analysis) 2 3 3 0 2 Age (years) 68.9 (6.5) 63.6 (7.4)* 64.6 (7.7) 60.8 (6.3) 73.5 (8.9)†† Education (years) 11.8 (4.2) 13.4 (3.7) 13.6 (4) 12.8 (3.7) 12.6 (3) Carrier of APOE e4 (%) 0 (0) 13 (62)* 10 (66)* 3 (50)* 2 (50)* Baseline CDR = 0 13 0 0 0 4 Baseline CDR = 0.5 0 15 15 0 0 Baseline CDR > 0.5 0 6 0 6 0 CDR change (%) 0 (0%) 16 (76 %)* 11 (73%)* 5 (83%)* 1 (25%)†,†† Baseline MMSE score 29.7 (0.5) 22.2 (4.7)* 24.6 (2.6)*,†† 16.1 (3.3)*,† 29.5 (0.5)†,†† Follow-up MMSE score 29.4 (0.9) 16.6 (7.2)* 19.6 (5.6)*,†† 9 (5)*,† 29 (2)†,†† Mean MMSE score loss (%) 0.03 (2.5) 5 (23.5)* 5 (20)*,†† 7.1 (20)*,† 0.8 (4.8)†,†† Baseline mean HV 2.06 (0.3) 1.8 (0.2)* 1.8 (0.3) 1.8 (0.3) 2.1 (0.3) Mean HV loss, % (SD)a 0.76 (1.5) 8.4 (4.4) 8 (4.5)* 10 (4.5)* 2.2 (3.5)†,†† ††C-PiB GCI 1.2 (0.07) 2.86 (0.6)* 2.7 (0.5)* 3.3 (0.8)* 2.27 (0.6)*,†† Mean delay between two DPA (years) 2.2 (0.2) 1.57 (0.3)* 1.63 (0.3)* 1.42 (0.3)* 2.03 (0.1)†,†† DPA-1 GCI# 1.19 (0.1) 1.37 (0.17)* 1.37 (0.1)* 1.37 (0.1)* 1.31 (0.3) DPA-2 GCI## 1.29 (0.15) 1.66 (0.3)* 1.72 (0.3)* 1.52 (0.2) 1.33 (0.2)† Annual % of DPA GCI increase### 4.2 (4.3) 13.1 (13.6)* 15.0 (13)* 8.3 (15) 2.2 (15) Controls All AD Prodromal AD Dementia AD Amyloidosis controls n = 13 n = 21 n = 15 n = 6 n = 4 TSPO genotype (MAB:HAB) 6:7 14:7 9:6 5:1 2:2 LAB (excluded from further analysis) 2 3 3 0 2 Age (years) 68.9 (6.5) 63.6 (7.4)* 64.6 (7.7) 60.8 (6.3) 73.5 (8.9)†† Education (years) 11.8 (4.2) 13.4 (3.7) 13.6 (4) 12.8 (3.7) 12.6 (3) Carrier of APOE e4 (%) 0 (0) 13 (62)* 10 (66)* 3 (50)* 2 (50)* Baseline CDR = 0 13 0 0 0 4 Baseline CDR = 0.5 0 15 15 0 0 Baseline CDR > 0.5 0 6 0 6 0 CDR change (%) 0 (0%) 16 (76 %)* 11 (73%)* 5 (83%)* 1 (25%)†,†† Baseline MMSE score 29.7 (0.5) 22.2 (4.7)* 24.6 (2.6)*,†† 16.1 (3.3)*,† 29.5 (0.5)†,†† Follow-up MMSE score 29.4 (0.9) 16.6 (7.2)* 19.6 (5.6)*,†† 9 (5)*,† 29 (2)†,†† Mean MMSE score loss (%) 0.03 (2.5) 5 (23.5)* 5 (20)*,†† 7.1 (20)*,† 0.8 (4.8)†,†† Baseline mean HV 2.06 (0.3) 1.8 (0.2)* 1.8 (0.3) 1.8 (0.3) 2.1 (0.3) Mean HV loss, % (SD)a 0.76 (1.5) 8.4 (4.4) 8 (4.5)* 10 (4.5)* 2.2 (3.5)†,†† ††C-PiB GCI 1.2 (0.07) 2.86 (0.6)* 2.7 (0.5)* 3.3 (0.8)* 2.27 (0.6)*,†† Mean delay between two DPA (years) 2.2 (0.2) 1.57 (0.3)* 1.63 (0.3)* 1.42 (0.3)* 2.03 (0.1)†,†† DPA-1 GCI# 1.19 (0.1) 1.37 (0.17)* 1.37 (0.1)* 1.37 (0.1)* 1.31 (0.3) DPA-2 GCI## 1.29 (0.15) 1.66 (0.3)* 1.72 (0.3)* 1.52 (0.2) 1.33 (0.2)† Annual % of DPA GCI increase### 4.2 (4.3) 13.1 (13.6)* 15.0 (13)* 8.3 (15) 2.2 (15) Data are mean (SD) or n (%). aLongitudinal data for hippocampal volumes available for 14 patients with Alzheimer’s disease only. *P-value < 0.05 compared to controls. †P-value < 0.05 compared to prodromal Alzheimer’s disease. ††P-value < 0.05 compared to dementia Alzheimer’s disease. #P-value with TSPO genotype as covariate. ##P-value with TSPO genotype and delay between the DPA exams as covariates. ###P-value with age, TSPO genotype and CDR as covariates. AD = Alzheimer’s disease; HAB/LAB/MAB = high/low/mixed affinity binders; HV = hippocampal volume. Table 2 Population characteristics for the DPA-PET longitudinal analysis Controls All AD Prodromal AD Dementia AD Amyloidosis controls n = 13 n = 21 n = 15 n = 6 n = 4 TSPO genotype (MAB:HAB) 6:7 14:7 9:6 5:1 2:2 LAB (excluded from further analysis) 2 3 3 0 2 Age (years) 68.9 (6.5) 63.6 (7.4)* 64.6 (7.7) 60.8 (6.3) 73.5 (8.9)†† Education (years) 11.8 (4.2) 13.4 (3.7) 13.6 (4) 12.8 (3.7) 12.6 (3) Carrier of APOE e4 (%) 0 (0) 13 (62)* 10 (66)* 3 (50)* 2 (50)* Baseline CDR = 0 13 0 0 0 4 Baseline CDR = 0.5 0 15 15 0 0 Baseline CDR > 0.5 0 6 0 6 0 CDR change (%) 0 (0%) 16 (76 %)* 11 (73%)* 5 (83%)* 1 (25%)†,†† Baseline MMSE score 29.7 (0.5) 22.2 (4.7)* 24.6 (2.6)*,†† 16.1 (3.3)*,† 29.5 (0.5)†,†† Follow-up MMSE score 29.4 (0.9) 16.6 (7.2)* 19.6 (5.6)*,†† 9 (5)*,† 29 (2)†,†† Mean MMSE score loss (%) 0.03 (2.5) 5 (23.5)* 5 (20)*,†† 7.1 (20)*,† 0.8 (4.8)†,†† Baseline mean HV 2.06 (0.3) 1.8 (0.2)* 1.8 (0.3) 1.8 (0.3) 2.1 (0.3) Mean HV loss, % (SD)a 0.76 (1.5) 8.4 (4.4) 8 (4.5)* 10 (4.5)* 2.2 (3.5)†,†† ††C-PiB GCI 1.2 (0.07) 2.86 (0.6)* 2.7 (0.5)* 3.3 (0.8)* 2.27 (0.6)*,†† Mean delay between two DPA (years) 2.2 (0.2) 1.57 (0.3)* 1.63 (0.3)* 1.42 (0.3)* 2.03 (0.1)†,†† DPA-1 GCI# 1.19 (0.1) 1.37 (0.17)* 1.37 (0.1)* 1.37 (0.1)* 1.31 (0.3) DPA-2 GCI## 1.29 (0.15) 1.66 (0.3)* 1.72 (0.3)* 1.52 (0.2) 1.33 (0.2)† Annual % of DPA GCI increase### 4.2 (4.3) 13.1 (13.6)* 15.0 (13)* 8.3 (15) 2.2 (15) Controls All AD Prodromal AD Dementia AD Amyloidosis controls n = 13 n = 21 n = 15 n = 6 n = 4 TSPO genotype (MAB:HAB) 6:7 14:7 9:6 5:1 2:2 LAB (excluded from further analysis) 2 3 3 0 2 Age (years) 68.9 (6.5) 63.6 (7.4)* 64.6 (7.7) 60.8 (6.3) 73.5 (8.9)†† Education (years) 11.8 (4.2) 13.4 (3.7) 13.6 (4) 12.8 (3.7) 12.6 (3) Carrier of APOE e4 (%) 0 (0) 13 (62)* 10 (66)* 3 (50)* 2 (50)* Baseline CDR = 0 13 0 0 0 4 Baseline CDR = 0.5 0 15 15 0 0 Baseline CDR > 0.5 0 6 0 6 0 CDR change (%) 0 (0%) 16 (76 %)* 11 (73%)* 5 (83%)* 1 (25%)†,†† Baseline MMSE score 29.7 (0.5) 22.2 (4.7)* 24.6 (2.6)*,†† 16.1 (3.3)*,† 29.5 (0.5)†,†† Follow-up MMSE score 29.4 (0.9) 16.6 (7.2)* 19.6 (5.6)*,†† 9 (5)*,† 29 (2)†,†† Mean MMSE score loss (%) 0.03 (2.5) 5 (23.5)* 5 (20)*,†† 7.1 (20)*,† 0.8 (4.8)†,†† Baseline mean HV 2.06 (0.3) 1.8 (0.2)* 1.8 (0.3) 1.8 (0.3) 2.1 (0.3) Mean HV loss, % (SD)a 0.76 (1.5) 8.4 (4.4) 8 (4.5)* 10 (4.5)* 2.2 (3.5)†,†† ††C-PiB GCI 1.2 (0.07) 2.86 (0.6)* 2.7 (0.5)* 3.3 (0.8)* 2.27 (0.6)*,†† Mean delay between two DPA (years) 2.2 (0.2) 1.57 (0.3)* 1.63 (0.3)* 1.42 (0.3)* 2.03 (0.1)†,†† DPA-1 GCI# 1.19 (0.1) 1.37 (0.17)* 1.37 (0.1)* 1.37 (0.1)* 1.31 (0.3) DPA-2 GCI## 1.29 (0.15) 1.66 (0.3)* 1.72 (0.3)* 1.52 (0.2) 1.33 (0.2)† Annual % of DPA GCI increase### 4.2 (4.3) 13.1 (13.6)* 15.0 (13)* 8.3 (15) 2.2 (15) Data are mean (SD) or n (%). aLongitudinal data for hippocampal volumes available for 14 patients with Alzheimer’s disease only. *P-value < 0.05 compared to controls. †P-value < 0.05 compared to prodromal Alzheimer’s disease. ††P-value < 0.05 compared to dementia Alzheimer’s disease. #P-value with TSPO genotype as covariate. ##P-value with TSPO genotype and delay between the DPA exams as covariates. ###P-value with age, TSPO genotype and CDR as covariates. AD = Alzheimer’s disease; HAB/LAB/MAB = high/low/mixed affinity binders; HV = hippocampal volume. Acknowledgements The authors would like to thank chemical/radiopharmaceutical and nursing staff of Service Hospitalier Frederic Joliot for the synthesis of 11C-PIB and 18F-DPA-714 and patient management, the team of CENIR (Centre de Neuroimagerie de Recherche) in Salpetriere Hospital for patient management during the MRI acquisition, the staff of the CATI (Centre d’Acquisition et de Traitement automatisé de l’Image) for technical support, the study participants. The authors gratefully acknowledge la Fondation pour la Recherche sur la Maladie d’Alzheimer, and the CEA/I2BM/Neurospin-Paris Descartes University collaboration. Funding French Health Ministry (PHRC) under reference PHRC-0054-N 2010 and Institut Roche de Recherche et Medecine Translationelle, European Union’s Seventh Framework Programme (FP7/2007-2013), grant agreement n° HEALTH-F2-2011-278850 (INMiND). The research leading to these results has received funding from the programme ‘Investissements d'avenir ANR-10-IAIHU-06’. Supplementary material Supplementary material is available at Brain online. Abbreviations Abbreviations CDR Clinical Dementia Rating GCI Global Cortical Index MMSE Mini-Mental State Examination PIB Pittsburgh compound B SUVr standardized uptake value ratio References Cagnin A , Brooks DJ , Kennedy AM , Gunn RN , Myers R , Turkheimer FE , et al. In-vivo measurement of activated microglia in dementia . Lancet 2001 ; 358 : 461 – 7 . Google Scholar CrossRef Search ADS PubMed Chauveau F , Van Camp N , Dollé F , Kuhnast B , Hinnen F , Damont A , et al. Comparative evaluation of the translocator protein radioligands 11C-DPA-713, 18F-DPA-714, and 11C-PK11195 in a rat model of acute neuroinflammation . J Nucl Med 2009 ; 50 : 468 – 76 . Google Scholar CrossRef Search ADS PubMed Chupin M , Hammers A , Liu RSN , Colliot O , Burdett J , Bardinet E , et al. Automatic segmentation of the hippocampus and the amygdala driven by hybrid constraints: method and validation . Neuroimage 2009 ; 46 : 749 – 61 . Google Scholar CrossRef Search ADS PubMed Dansokho C , Ait Ahmed D , Aid S , Toly-Ndour C , Chaigneau T , Calle V , et al. Regulatory T cells delay disease progression in Alzheimer-like pathology . Brain 2016 ; 139 : 1237 – 51 . Google Scholar CrossRef Search ADS PubMed de Souza LC , Corlier F , Habert M-O , Uspenskaya O , Maroy R , Lamari F , et al. Similar amyloid-β burden in posterior cortical atrophy and Alzheimer’s disease . Brain 2011 ; 134 : 2036 – 43 . Google Scholar CrossRef Search ADS PubMed Dubois B , Chupin M , Hampel H , Lista S , Cavedo E , Croisile B , et al. Donepezil decreases annual rate of hippocampal atrophy in suspected prodromal Alzheimer’s disease . Alzheimers Dement 2015 ; 11 : 1041 – 9 . Google Scholar CrossRef Search ADS PubMed Dubois B , Feldman HH , Jacova C , Cummings JL , Dekosky ST , Barberger-Gateau P , et al. Revising the definition of Alzheimer’s disease: a new lexicon . Lancet Neurol 2010 ; 9 : 1118 – 27 . Google Scholar CrossRef Search ADS PubMed Edison P , Archer HA , Gerhard A , Hinz R , Pavese N , Turkheimer FE , et al. Microglia, amyloid, and cognition in Alzheimer’s disease: an [11C](R)PK11195-PET and [11C]PIB-PET study . Neurobiol Dis 2008 ; 32 : 412 – 19 . Google Scholar CrossRef Search ADS PubMed Fan Z , Aman Y , Ahmed I , Chetelat G , Landeau B , Ray Chaudhuri K , et al. Influence of microglial activation on neuronal function in Alzheimer’s and Parkinson’s disease dementia . Alzheimers Dement 2015a ; 11 : 608 – 21.e7 . Google Scholar CrossRef Search ADS Fan Z , Brooks DJ , Okello A , Edison P . An early and late peak in microglial activation in Alzheimer’s disease trajectory . Brain 2017 ; 140 : 792 – 803 . Google Scholar PubMed Fan Z , Okello AA , Brooks DJ , Edison P . Longitudinal influence of microglial activation and amyloid on neuronal function in Alzheimer’s disease . Brain 2015b ; 138 : 3685 – 98 . Google Scholar CrossRef Search ADS Hamelin L , Lagarde J , Dorothée G , Leroy C , Labit M , Comley RA , et al. Early and protective microglial activation in Alzheimer’s disease: a prospective study using 18F-DPA-714 PET imaging . Brain 2016 ; 139 : 1252 – 64 . Google Scholar CrossRef Search ADS PubMed Heneka MT , Carson MJ , El Khoury J , Landreth GE , Brosseron F , Feinstein DL , et al. Neuroinflammation in Alzheimer’s disease . Lancet Neurol 2015 ; 14 : 388 – 405 . Google Scholar CrossRef Search ADS PubMed Heneka MT , Kummer MP , Latz E . Innate immune activation in neurodegenerative disease . Nat Rev Immunol 2014 ; 14 : 463 – 77 . Google Scholar CrossRef Search ADS PubMed de Jong HWAM , van Velden FHP , Kloet RW , Buijs FL , Boellaard R , Lammertsma AA . Performance evaluation of the ECAT HRRT: an LSO-LYSO double layer high resolution, high sensitivity scanner . Phys Med Biol 2007 ; 52 : 1505 – 26 . Google Scholar CrossRef Search ADS PubMed Kinkingnéhun S , Sarazin M , Lehéricy S , Guichart-Gomez E , Hergueta T , Dubois B . VBM anticipates the rate of progression of Alzheimer disease: a 3-year longitudinal study . Neurology 2008 ; 70 : 2201 – 11 . Google Scholar CrossRef Search ADS PubMed Kreisl WC , Lyoo CH , Liow J-S , Snow J , Page E , Jenko KJ , et al. Distinct patterns of increased translocator protein in posterior cortical atrophy and amnestic Alzheimer’s disease . Neurobiol Aging 2017 ; 51 : 132 – 40 . Google Scholar CrossRef Search ADS PubMed Kreisl WC , Lyoo CH , Liow J-S , Wei M , Snow J , Page E , et al. ( 11)C-PBR28 binding to translocator protein increases with progression of Alzheimer’s disease . Neurobiol Aging 2016 ; 44 : 53 – 61 . Google Scholar CrossRef Search ADS PubMed Kreisl WC , Lyoo CH , McGwier M , Snow J , Jenko KJ , Kimura N , et al. In vivo radioligand binding to translocator protein correlates with severity of Alzheimer’s disease . Brain 2013 ; 136 : 2228 – 38 . Google Scholar CrossRef Search ADS PubMed Lagarde J , Sarazin M , Bottlaender M . In vivo PET imaging of neuroinflammation in Alzheimer’s disease . J Neural Transm (Vienna) 2017 , in press . doi: 10.1007/s00702-017-1731-x. Lavisse S , García-Lorenzo D , Peyronneau M-A , Bodini B , Thiriez C , Kuhnast B , et al. Optimized quantification of translocator protein radioligand 18F-DPA-714 uptake in the brain of genotyped healthy volunteers . J Nucl Med 2015 ; 56 : 1048 – 54 . Google Scholar CrossRef Search ADS PubMed Lopresti BJ , Klunk WE , Mathis CA , Hoge JA , Ziolko SK , Lu X , et al. Simplified quantification of Pittsburgh Compound B amyloid imaging PET studies: a comparative analysis . J Nucl Med 2005 ; 46 : 1959 – 72 . Google Scholar PubMed Lyoo CH , Ikawa M , Liow J-S , Zoghbi SS , Morse CL , Pike VW , et al. Cerebellum can serve as a pseudo-reference region in Alzheimer disease to detect neuroinflammation measured with pet radioligand binding to translocator protein . J Nucl Med 2015 ; 56 : 701 – 6 . Google Scholar CrossRef Search ADS PubMed Michaud J-P , Rivest S . Anti-inflammatory signaling in microglia exacerbates Alzheimer's disease-related pathology . Neuron 2015 ; 85 : 450 – 2 . Google Scholar CrossRef Search ADS PubMed Minett T , Classey J , Matthews FE , Fahrenhold M , Taga M , Brayne C , et al. Microglial immunophenotype in dementia with Alzheimer’s pathology . J Neuroinflammation 2016 ; 13 : 135 . Google Scholar CrossRef Search ADS PubMed Okello A , Edison P , Archer HA , Turkheimer FE , Kennedy J , Bullock R , et al. Microglial activation and amyloid deposition in mild cognitive impairment: a PET study . Neurology 2009 ; 72 : 56 – 62 . Google Scholar CrossRef Search ADS PubMed Owen DRJ , Gunn RN , Rabiner EA , Bennacef I , Fujita M , Kreisl WC , et al. Mixed-affinity binding in humans with 18-kDa translocator protein ligands . J Nucl Med 2011 ; 52 : 24 – 32 . Google Scholar CrossRef Search ADS PubMed Parbo P , Ismail R , Hansen KV , Amidi A , Mårup FH , Gottrup H , et al. Brain inflammation accompanies amyloid in the majority of mild cognitive impairment cases due to Alzheimer’s disease . Brain 2017 ; 140 : 2002 – 11 . Google Scholar CrossRef Search ADS PubMed Prokop S , Miller KR , Heppner FL . Microglia actions in Alzheimer’s disease . Acta Neuropathol 2013 ; 126 : 461 – 77 . Google Scholar CrossRef Search ADS PubMed Sarazin M , Berr C , De Rotrou J , Fabrigoule C , Pasquier F , Legrain S , et al. Amnestic syndrome of the medial temporal type identifies prodromal AD: a longitudinal study . Neurology 2007 ; 69 : 1859 – 67 . Google Scholar CrossRef Search ADS PubMed Schuitemaker A , Kropholler MA , Boellaard R , van der Flier WM , Kloet RW , van der Doef TF , et al. Microglial activation in Alzheimer’s disease: an (R)-[11C]PK11195 positron emission tomography study . Neurobiol Aging 2013 ; 34 : 128 – 36 . Google Scholar CrossRef Search ADS PubMed Sureau FC , Reader AJ , Comtat C , Leroy C , Ribeiro M-J , Buvat I , et al. Impact of image-space resolution modeling for studies with the high-resolution research tomograph . J Nucl Med 2008 ; 49 : 1000 – 8 . Google Scholar CrossRef Search ADS PubMed Tang Y , Le W . Differential roles of M1 and M2 microglia in neurodegenerative diseases . Mol Neurobiol 2016 ; 53 : 1181 – 94 . Google Scholar CrossRef Search ADS PubMed Vivash L , O’Brien TJ . Imaging microglial activation with TSPO PET: lighting up neurologic diseases? J Nucl Med 2016 ; 57 : 165 – 8 . Google Scholar CrossRef Search ADS PubMed Wiley CA , Lopresti BJ , Venneti S , Price J , Klunk WE , DeKosky ST , et al. Carbon 11-labeled Pittsburgh Compound B and carbon 11-labeled (R)-PK11195 positron emission tomographic imaging in Alzheimer disease . Arch Neurol 2009 ; 66 : 60 – 7 . Google Scholar CrossRef Search ADS PubMed Yokokura M , Mori N , Yagi S , Yoshikawa E , Kikuchi M , Yoshihara Y , et al. In vivo changes in microglial activation and amyloid deposits in brain regions with hypometabolism in Alzheimer’s disease . Eur J Nucl Med Mol Imaging 2011 ; 38 : 343 – 51 . Google Scholar CrossRef Search ADS PubMed Yokokura M , Terada T , Bunai T , Nakaizumi K , Takebayashi K , Iwata Y , et al. Depiction of microglial activation in aging and dementia: positron emission tomography with [(11)C]DPA713 versus [(11)C](R)PK11195 . J Cereb Blood Flow Metab 2017 ; 37 : 877 – 889 . Google Scholar CrossRef Search ADS PubMed © The Author(s) (2018). Published by Oxford University Press on behalf of the Guarantors of Brain. All rights reserved. For permissions, please email: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

BrainOxford University Press

Published: Mar 28, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off