ARTICLE OPEN ACCESS Cortical microstructural correlates of astrocytosis in autosomal-dominant Alzheimer disease Eduard Vilaplana, MSc, PhD,* Elena Rodriguez-Vieitez, MSc, PhD,* Daniel Ferreira, PhD, Victor Montal, MSc, Correspondence Ove Almkvist, PhD, Anders Wall, PhD, Alberto Lleó, MD, PhD, Eric Westman, PhD, Caroline Graff, MD, PhD, Dr. Nordberg Juan Fortea, MD, PhD, and Agneta Nordberg, MD, PhD Agneta.K.Nordberg@ki.se ® Neurology 2020;94:e2026-e2036. doi:10.1212/WNL.0000000000009405 Abstract Objective To study the macrostructural and microstructural MRI correlates of brain astrocytosis, mea- sured with 11C-deuterium-L-deprenyl (11C-DED)–PET, in familial autosomal-dominant Alz- heimer disease (ADAD). Methods The total sample (n = 31) comprised ADAD mutation carriers (n = 10 presymptomatic, 39.2 ± 10.6 years old; n = 3 symptomatic, 55.5 ± 2.0 years old) and noncarriers (n = 18, 44.0 ± 13.7 years old) belonging to families with mutations in either the presenilin-1 or amyloid precursor protein genes. All participants underwent structural and diffusion MRI and neuropsychological assessment, and 20 participants (6 presymptomatic and 3 symptomatic mutation carriers and 11 noncarriers) also underwent 11C-DED-PET. Results Vertex-wise interaction analyses revealed a differential relationship between carriers and non- carriers in the association between 11C-DED binding and estimated years to onset (EYO) and between cortical mean diffusivity (MD) and EYO. These differences were due to higher 11C-DED binding in presymptomatic carriers, with lower binding in symptomatic carriers compared to noncarriers, and to lower cortical MD in presymptomatic carriers, with higher MD in symp- tomatic carriers compared to noncarriers. Using a vertex-wise local correlation approach, 11 C-DED binding was negatively correlated with cortical MD and positively correlated with cortical thickness. Conclusions Our proof-of-concept study is the first to show that microstructural and macrostructural changes can reflect underlying neuroinflammatory mechanisms in early stages of Alzheimer disease (AD). The findings support a role for neuroinflammation in AD pathogenesis, with potential implications for the correct interpretation of neuroimaging biomarkers as surrogate endpoints in clinical trials. *These authors contributed equally to this work. From the Memory Unit, Department of Neurology (E.V., V.M., A.L., J.F.), Hospital de la Santa Creu i Sant Pau, Biomedical Research Institute Sant Pau, Universitat Autònoma de Barcelona; Centro de Investigación Biomédica en Red de Enfermedades Neurodegenerativas, CIBERNED (E.V., V.M., A.L., J.F.), Madrid, Spain; Department of Neurobiology (E.R.-V., D.F., O.A., E.W., A.N.), Care Sciences and Society, Center for Alzheimer Research, Division of Clinical Geriatrics, and Division of Neurogeriatrics (C.G.), Karolinska Institutet, Stockholm Department of Psychology (O.A.), Stockholm University; The Aging Brain Unit (O.A., A.N.) and Unit for Hereditary Dementias (C.G.), Theme Aging, Karolinska University Hospital, Stockholm; Department of Surgical Sciences, Section of Nuclear Medicine & PET (A.W.), Uppsala University, Sweden; and Department of Neuroimaging (E.W.), Centre for Neuroimaging Sciences, Institute of Psychiatry, Psychology and Neuroscience, King’s College London, United Kingdom. Go to Neurology.org/N for full disclosures. Funding information and disclosures deemed relevant by the authors, if any, are provided at the end of the article. The Article Processing Charge was funded by Swedish Research Council. This is an open access article distributed under the terms of the Creative Commons Attribution License 4.0 (CC BY), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. e2026 Copyright © 2020 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology. Glossary AD = Alzheimer disease; ADAD = autosomal-dominant Alzheimer disease; 11C-DED = 11C-deuterium-L-deprenyl; CTh = cortical thickness; DTI = diffusion tensor imaging; DWI = diffusion-weighted imaging; EYO = estimated number of years to symptom onset; FWE = family-wise error; GM = gray matter; MAO-B = monoamine oxidase-B; MD = mean diffusivity; TE = echo time; TR = repetition time. Alzheimer disease (AD) is a complex disorder in which mul- mutation status. Therefore, the study includes mutation car- tiple pathophysiologic features coexist.1 In a small proportion riers and noncarriers, all recruited and examined following of patients, AD is hereditary due to autosomal-dominant identical procedures, without known selection bias. mutations2 with an early and rather predictable mutation- specific age at onset, allowing the investigation of pre- Symptom onset in mutation carriers is defined as the time at symptomatic brain changes.2,3 which the first clinically relevant cognitive symptoms appeared, as either experienced by the patient or noticed by near relatives. Neuroinflammation is postulated as a key player in AD In this cohort, the average age at onset was earliest (36 ± 2 pathogenesis.4,5 While most PET imaging studies of neuro- years) in PSEN1 Ile143Thr mutation carriers, while it was inflammation have studied microgliosis,6,7 few PET tracers similar in carriers of the other 3 mutations: PSEN1 His163Tyr exist for astrocytosis. The most common is 11C-deuterium-L- (52 ± 7 years), APPswe KM670/671NL (54 ± 5 years), and deprenyl (11C-DED), which targets monoamine oxidase-B APParc Glu693Gly (56 ± 3 years).19 The average age at onset (MAO-B).8–11 Using 11C-DED-PET in a longitudinal autosomal- in each family was calculated from medical records for disease dominant AD (ADAD) cohort, we previously reported pre- onset in individuals from that family (5, 9, 24, and 12 indi- symptomatic astrocyte activation followed by decline along viduals for the 4 mutation types, respectively). The estimated disease progression.12 number of years to symptom onset (EYO) was calculated for each carrier or noncarrier participant by subtracting the indi- The complexity of AD requires multimodal approaches. Using vidual’s age from the average age at onset for the respective structural and diffusion MRI, we recently proposed a model of family. The concept of EYO in noncarriers is artificial. We only gray matter (GM) changes in sporadic AD,13–16 in which an use it in the interaction analyses to have the participants in the early presymptomatic phase of decreased cortical mean diffu- same temporal dimension. sivity (MD) and increased cortical thickness (CTh) is followed by increased cortical MD and decreased CTh at symptomatic In our study, symptomatic carriers had been clinically di- stages. Although the origin of these structural changes is unclear, agnosed with either mild cognitive impairment20 or AD de- previous biological evidence suggests a role for neuronal or glial mentia.21 Presymptomatic carriers had no cognitive complaints remodeling and hypertrophy.15,17,18 Whether astrocytosis has and did not fulfill the criteria for mild cognitive impairment or a measurable structural correlate in AD is unknown. Identifying AD dementia. Clinicians and researchers in contact with or structural correlates of brain inflammation is of potential utility examining the ADAD research participants were blind to the for clinical trial design, specifically when interpreting the mutation status. Diagnoses were made during a consensus structural changes observed in immunization or in trials tar- meeting where a geriatrician/neurologist, a neuropsychologist, geting neuroinflammation. In this proof-of-concept study, we and a nurse discussed the outcome of the participant aimed to (1) assess astrocytosis in ADAD using 11C-DED-PET; assessment. (2) investigate microstructural and macrostructural measures in ADAD; and (3) investigate the microstructural and macro- The study included 31 participants (table 1). All available structural correlates of astrocytosis in ADAD. ADAD mutation carriers who had both MRI and diffusion- weighted imaging (DWI) data were selected. An age- and sex- matched (to both presymptomatic and symptomatic carriers) Methods group of noncarriers who also had MRI and DWI data were Study design and participants used as a control group for the mutation carriers. All partic- Individuals from families with known ADAD mutations were ipants underwent a comprehensive clinical and imaging ex- recruited through the Unit for Hereditary Dementias, which amination that included a medical history, neurologic and provides genetic counseling at Theme Aging, Karolinska Uni- psychiatric examination, EEG, MRI, APOE genotyping, and versity Hospital (Stockholm, Sweden). The participants with neuropsychological assessment. Moreover, a subset of partic- ADAD in this study are part of an ongoing prospective research ipants (n = 20, table 2) underwent a 11C-DED PET scan. This study at Karolinska Institutet that started in 1993 involving was acquired within 3.8 ± 3.7 months of the MRI scan, except families that carry 1 of 4 mutation types. All family members for 2 participants: 1 presymptomatic carrier had the 11C-DED were invited to participate, and those who accepted were in- PET scan 1.7 years before the MRI and 1 symptomatic carrier cluded. Recruitment was performed blind to participants’ had the scan 3.3 years after the MRI. When analyses involving Neurology.org/N Neurology | Volume 94, Number 19 | May 12, 2020 e2027 Table 1 Demographic and clinical information for the whole sample Noncarriers pMC sMC Total N 18 10 3a 31 b b Sex, M/F 9/9 2/1 Age, y 44.0 (13.7) 39.2 (10.6) 55.5 (2.0) 43.6 (12.7) Estimated years to symptom onset −7.4 (10.6) −12.6 (8.1) 1.47 (4.0) −8.2 (10.0) Education, y 11.6 (2.0) 13.3 (1.6) 10.3 (1.2) 12.0 (2.0) APOE «4 carriers, % 27 50 67 39 Global cognition, z score 0.20 (0.49) 0.44 (0.40) −1.06 (2.43) 0.15 (0.87) Episodic memory, z score −0.14 (0.64) 0.04 (0.72) −1.47 (1.38) −0.21 (0.83) Abbreviations: pMC = presymptomatic mutation carriers; sMC = symptomatic mutation carriers. Continuous data are presented as means (SD). a Two sMC were diagnosed with mild cognitive impairment and 1 with Alzheimer disease dementia. b Sex distribution for pMC and for the whole sample is not revealed to preserve confidentiality. comparisons between MRI and PET imaging data were re- sequence and a diffusion tensor imaging (DTI) MRI sequence peated excluding these 2 participants, the results did not change in a 3T Siemens (Munich, Germany) Trio scanner. The ac- significantly (not shown). quisition measures of T1 MRI included the following: repeti- tion time/echo time (TR/TE) 1,780/3.42 ms, inversion time Standard protocol approvals, registrations, 900 ms, 192 sagittal slices, voxel size 1 × 1 × 1 mm3, and flip and patient consents angle = 9°. DTI was performed using a spin echoplanar imaging All participants provided written informed consent to par- sequence (TR/TE 8,000/97 ms, 60 axial slices, voxel size 2 ×2 ticipate in the study, which was conducted according to the ×2.4 mm3) with 30 orientations for the diffusion-sensitizing Declaration of Helsinki and subsequent revisions. Ethical gradients (b-value of 1,000 s/mm2). Further details of the approval was obtained from the regional Human Ethics procedure can be found elsewhere.12,22 Committee of Stockholm and the Faculty of Medicine and Radiation Hazard Ethics Committee of Uppsala University Structural MRI was preprocessed using FreeSurfer 6.0 (surfer. Hospital, Sweden. nmr.mgh.harvard.edu).23 All cortical segmentations were inspected visually to detect processing errors, which were MRI acquisition and processing corrected if necessary as is customary in MRI surface-based All participants (n = 31) underwent a structural 3D T1 analyses.14,15 Of the initial 33 participants, 2 (1 mutation carrier magnetization-prepared rapid acquisition gradient echo and 1 noncarrier) had been excluded from the analysis because Table 2 Demographic and clinical information in the PET subset Noncarriers pMC sMC Total a N 11 6 3 20 b b Sex, M/F 6/5 2/1 Age, y 47.1 (13.6) 43.4 (7.8) 55.5 (2.0) 47.2 (11.5) Estimated years to symptom onset −5.6 (11.4) −9.6 (8.2) 1.4 (3.9) −5.8 (10.0) Education, y 11.7 (2.0) 12.8 (1.9) 10.3 (1.2) 11.9 (2.0) APOE «4 carriers, % 27 50 67 40 Global cognition, z score 0.29 (0.44) 0.53 (0.42) −1.06 (2.43) 0.16 (1.03) Episodic memory, z score −0.18 (0.46) −0.27 (0.45) −1.47 (1.38) −0.40 (0.76) Abbreviations: pMC = presymptomatic mutation carriers; sMC = symptomatic mutation carriers. Continuous data are presented as mean (SD). a Two sMC were diagnosed with mild cognitive impairment and 1 with Alzheimer disease dementia. b Sex distribution for pMC and for the whole sample is not revealed to preserve confidentiality. e2028 Neurology | Volume 94, Number 19 | May 12, 2020 Neurology.org/N of severe segmentation errors (6%), thus 31 were finally in- midpoint between pial and white matter FreeSurfer surfaces as cluded (table 1). in the diffusion analyses. Diffusion imaging data were processed with an in-house Neuropsychological assessment surface-based DTI approach,15 which uses tools from the Participants were assessed using a comprehensive battery of FSL (FMRIB Software Library) (fsl.fmrib.ox.ac.uk/fsl/ neuropsychological tests, including memory, attention, lan- fslwiki, version 5.0.9) and FreeSurfer 6.0 packages. This guage, executive, and visuospatial functions.29 Raw scores were surface-based approach takes advantage of recent meth- converted to z scores using a reference group from the Kar- odologic advances24–26 to overcome the limitations of tra- olinska University Hospital, and combined into 2 composite ditional voxel-based approaches when used in analyses of the scores for global cognition (9 subtests) and episodic memory cortical mantle. First, it reduces the contribution from CSF (3 subtests).29 Cronbach α was used to assess the internal and white matter signal in GM voxels that can confound the consistency of each composite score. The episodic memory cortical MD measures. Second, it applies a surface-based composite (Cronbach α = 0.73) was useful to capture early smoothing procedure, as it has been shown that volume- impairment,29 while the global cognitive composite (Cronbach based analysis techniques may be sensitive to the across- α = 0.67) represented an aggregate of various nonmemory voxel smoothing kernel size.27 In the surface-based DTI domains.29 The neuropsychological assessment was performed approach, images were motion-corrected, skull-stripped, and within 1.0 ± 3.3 months from the date of the MRI scan. diffusion tensor–fitted. The diffusion images were then coregistered to each participant’s T1 native space using the Statistical analysis bbregister tool in FreeSurfer 6.0. The cortical MD maps Before any statistical data analysis of the PET and MRI, a 2D resulting from the DTI fitting were then sampled in the full-width half-maximum Gaussian kernel of 20 mm across the midpoint between white and pial surfaces generated by cortical mantle was applied to 11C-DED PET, MD, and CTh FreeSurfer, projected onto the participant’s cortical surface surfaces. space, and registered to the FreeSurfer standard space for subsequent analysis. To address the main objective of this study, investigating the microstructural and macrostructural MRI correlates of PET image acquisition and processing astrocytosis using 11C-DED PET at different stages of ADAD, A subset of participants (n = 20) underwent 11C-DED PET the statistical analysis was conducted in 3 steps. imaging at the Uppsala PET Centre, Uppsala University, Sweden. Briefly, 60-minute dynamic 11C-DED images were In step 1, we assessed whether mutation carriers and non- acquired on ECAT EXACT HR+ (Siemens/CTI) and GE carriers had different associations with EYO for 3 imaging (Chicago, IL) Discovery ST PET/CT scanners (mean injected modalities: 11C-DED PET, cortical MD, and CTh. To this end, dose, 221 ± 65 MBq), reconstructed, and motion vertex-wise linear regression models were assessed in Free- corrected.12,22 All PET emission data were reconstructed with Surfer, with each imaging measure as a dependent variable and filtered backprojection using a 4-mm Hanning filter, resulting EYO as the independent predictor. We then statistically tested in a transaxial spatial resolution of 5 mm in the field of view. for a differential relationship of each imaging modality with The matrix included 128 × 128 pixels, and a zoom factor of 2.5 EYO between mutation carriers (n = 9 for 11C-DED PET, was used. All 19 reconstructed frames (4 × 30 s, 8 × 60 s, 4 ×300 n = 13 for cortical MD and CTh) and noncarriers (n = 11 for 11 s, and 3 ×600 s) were realigned for motion correction using the C-DED PET, n = 18 for cortical MD and CTh), using sex as second frame as reference, with subsequent time frames being covariate. successively realigned to the previous one. For 11C-DED PET quantification, a modified-reference Patlak model10,28 was ap- In step 2, to further investigate 11C-DED PET, cortical MD, plied to the 20–60 minutes dynamic 11C-DED PET images and CTh differences across disease stages, the mutation carriers using the cerebellar GM as modified reference region to gen- were stratified into presymptomatic or symptomatic, and these erate individual parametric Patlak slope images (units: min−1), 2 subgroups were each compared to the noncarrier group. assuming a cerebellar GM slope of 0.01 min−1. 11C-DED Vertex-wise general linear models were assessed in FreeSurfer binding was then expressed as the ratio of 11C-DED slope in with each imaging measure as a dependent variable, and age each brain voxel to that in the cerebellar GM. and sex as covariates. Group comparisons were carried out between each of the mutation carrier groups (n = 6 pre- Once 11C-DED PET binding had been calculated, 10–60 symptomatic and n = 3 symptomatic for 11C-DED PET, n = 10 minutes averaged PET images were used to coregister the PET presymptomatic and n = 3 symptomatic for cortical MD and volume to each participant’s native T1 using the mri_coreg tool CTh) and the noncarrier group (n = 11 for 11C-DED PET, n = in FreeSurfer 6.0. Although the parametric 11C-DED images 18 for cortical MD and CTh). were originally generated in native 11C-DED PET space, the images were subsequently projected onto the cortical surface In step 3, we investigated the local association between space for direct comparison between the 11C-DED PET and 11 C-DED binding and cortical microstructural and macro- MRI data. The cortical 11C-DED binding was sampled in the structural MRI measures using vertex-wise correlation Neurology.org/N Neurology | Volume 94, Number 19 | May 12, 2020 e2029 analyses between 11C-DED binding and cortical MD, and memory from the noncarriers. Symptomatic carriers tended to between 11C-DED binding and CTh, in FreeSurfer. These have lower episodic memory scores than noncarriers (Mann- associations were tested for the mutation carrier group alone Whitney Z = −1.95, p = 0.052), while the respective comparison (n = 9). for global cognition did not reach statistical significance. 11 All the vertex-wise analyses described above were corrected C-DED binding in mutation carriers for multiple comparisons within FreeSurfer by using a cluster Figure 1 shows the results of the linear regression models used extension criterion in a Monte Carlo simulation with 10,000 to investigate the association between 11C-DED binding and repeats, with the family-wise error (FWE) correction settled EYO in mutation carriers and noncarriers. Figure 1A illustrates at p < 0.05. Only clusters that survived the multiple- the clusters representing a significant differential relationship comparisons correction are shown. For each analysis, all sig- between 11C-DED binding and EYO depending on the mu- nificant clusters were isolated, averaged, and plotted in scat- tation status (carrier/noncarrier) in the PET subset (n = 20). terplots or box-and-whisker plots for illustrative purposes. Significant clusters (FWE-corrected, p < 0.05) emerged mostly within bilateral temporal and frontal regions (figure 1A). For Group analyses for continuous nonimaging variables (de- illustrative purposes, the average of all significant clusters was mographic, clinical, and neuropsychological variables) were plotted separately for mutation carriers and noncarriers (figure performed using analysis of variance with Tukey post hoc 1B); coordinates of all significant clusters are available in table corrections or Kruskal-Wallis tests, as appropriate. The χ 2 test e-1 (doi.org/10.5061/dryad.585581j). 11C-DED binding was was applied for categorical variables. Statistical analyses were negatively associated with EYO in carriers while no significant performed using R statistical software (r-project.org). association was observed in noncarriers. The vertex-wise whole brain group comparison of 11C-DED binding between pre- Data availability symptomatic carriers and noncarriers showed a pattern of in- Anonymized data will be shared by request from any qualified creased 11C-DED binding in the presymptomatic carriers over investigator for the sole purpose of replicating procedures and clusters involving precentral, parietal, and precuneus regions results presented in the report provided that data transfer is in (figure 2A; FWE-corrected, p < 0.05). For illustrative purposes, agreement with EU legislation on the general data protection the comparison between presymptomatic carriers and non- regulation. carriers within the average of all significant clusters is shown by a box-and-whisker plot (figure 2B); coordinates of all signifi- cant clusters are available in table e-2 (doi.org/10.5061/dryad. Results 585581j). In contrast, 11C-DED binding appeared to be similar Demographic and clinical data are summarized in table 1 for the or even decreased for the symptomatic carriers vs the non- whole cohort (n = 31) and in table 2 for the subset of partic- carriers (figure e-1 and table e-3, doi.org/10.5061/dryad. ipants with 11C-DED PET image data (n = 20). There were no 585581j). significant differences in age, proportion of mutation carriers, or proportion of symptomatic carriers between the whole co- Cortical microstructural and macrostructural hort and the PET subset. There were no significant differences measures in mutation carriers in age between noncarriers and presymptomatic or symp- Figure 3 shows the results of the linear regression models used tomatic carriers in the whole sample (table 1) or in the PET to investigate the association between cortical MD and EYO in subset (table 2). There were no significant differences in the mutation carriers and noncarriers. Figure 3A illustrates the proportion of APOE e4 allele carriers between groups in the clusters representing a significant differential relationship be- whole cohort and in the PET subset. Sex information for the tween MD and EYO depending on the mutation status presymptomatic carriers and the whole cohort, as well as in- (carrier/noncarrier) in the whole cohort (n = 31). The sig- dividual family membership of the data points in the figures, are nificant clusters (FWE-corrected, p < 0.05) in figure 3A show not revealed for confidentiality reasons. a widespread bilateral cortical pattern, including temporo- parietal, precuneus, posterior and anterior cingulate, and frontal Neuropsychological profiles regions. For illustrative purposes, the average of all significant In the whole cohort (table 1), the presymptomatic carrier group clusters was plotted against EYO for mutation carriers and did not differ significantly in global cognition or episodic memory noncarriers separately (figure 3B). In mutation carriers, a pat- from the noncarrier group. Symptomatic carriers had the lowest tern of increasing MD with disease progression (as measured scores for global cognition (z = −1.06 ± 2.43) and episodic by EYO) was observed, while no change in relation to EYO was memory (z = −1.47 ± 1.38) of all groups. They had significantly seen in noncarriers. No result survived multiple comparisons poorer episodic memory scores than noncarriers (Mann- when comparing the relationship of CTh with EYO between Whitney Z = −2.11, p = 0.035); however, the difference in carriers and noncarriers (not shown). global cognition between symptomatic carriers and noncarriers was not statistically significant. Similarly, in the subset of partic- Cortical MD was reduced in presymptomatic mutation ipants with PET imaging data (table 2), presymptomatic carriers carriers compared to noncarriers in clusters including pari- did not significantly differ in either global cognition or episodic etal, frontal, temporal, and occipital regions (figure 4A). e2030 Neurology | Volume 94, Number 19 | May 12, 2020 Neurology.org/N 11 Figure 1 Differential relationship between estimated number of years to symptom onset (EYO) and C-deuterium-L- deprenyl (11C-DED) PET binding in mutation carriers (MC) vs noncarriers (NC) (A) Surface map represents clusters of significant differential relationship between 11C-DED PET binding and EYO depending on the mutation status (MC/NC). Only clusters surviving multiple-comparison correction are depicted (family-wise error [FWE]–corrected, p < 0.05); coordinates of all significant clusters are available in table e-1 (doi.org/10.5061/dryad.585581j). (B) Linear regression trajectories of 11C-DED PET binding averaged over the left hemisphere (LH) vs EYO in MC (red) and NC (blue), including 95% confidence bands around the model predictions (illustrative purposes). RH = right hemisphere. Figure 4B shows the average MD within all significant These analyses were repeated including APOE e4 allele status clusters in a box-and-whisker plot between presymptomatic (carrier vs noncarrier) as a covariate and the results did not carriers and noncarriers. On the other hand, cortical MD was change. increased over widespread clusters including bilateral temporo-parietal, cingulate, and frontal regions in symp- Microstructural and macrostructural MRI tomatic carriers vs noncarriers (figure 4C). Figure 4D shows correlates of 11C-DED binding the average MD within all significant clusters in a box-and- A vertex-wise map-to-map correlation analysis was carried out whisker plot between symptomatic carriers and noncarriers. to assess the local association between 11C-DED binding and Finally, there were subtle increases in CTh in pre- brain microstructure and macrostructure within the mutation symptomatic carriers at the uncorrected level, and decreases carriers who had data on all imaging biomarkers (n = 9) (figure in CTh in symptomatic carriers (figure e-2, doi.org/10. 5). Higher levels of 11C-DED binding were related to lower 5061/dryad.585581j). cortical MD in temporo-parietal regions (figure 5A, FWE- Figure 2 Group comparison of 11C-deuterium-L-deprenyl (11C-DED) PET binding in presymptomatic mutation carriers (pMC) vs noncarriers (NC) (A) Surface map represents clusters of increased (in red–yellow) 11C-DED PET binding in pMC compared to NC; only clusters surviving multiple- comparison correction are depicted (family-wise error [FWE]–corrected, p < 0.05); coordinates of all significant clusters are available in table e-2 (doi.org/10.5061/dryad.585581j). (B) Box-and- whisker plot compares the average 11C-DED binding over the left hemisphere (LH) between pMC (red) and NC (blue) (illustrative purposes). RH = right hemisphere. Neurology.org/N Neurology | Volume 94, Number 19 | May 12, 2020 e2031 Figure 3 Differential relationship between estimated number of years to symptom onset (EYO) and cortical mean diffu- sivity (MD) in mutation carriers (MC) and noncarriers (NC) (A) Surface map represents clusters of significant differential relationship between cortical MD and EYO depending on the mutation status (MC/NC). Only clusters surviving multiple-comparison correction are depicted (family-wise error [FWE]–corrected, p < 0.05); coordinates of all significant clusters are available in table e-4 (doi.org/10.5061/dryad.585581j). (B) Linear regression trajectories of cortical MD averaged over the left hemisphere (LH) vs EYO in MC (red) and in NC (blue), including 95% confidence bands around the model predictions (illustrative purposes). RH = right hemisphere. corrected, p < 0.05) and to increased CTh (figure 5C). The the astrocyte phenotype indicating loss of function. Importantly, average of all significant clusters illustrating these significant similarly to previous works,12 we observed a diffuse pattern of associations between 11C-DED binding and cortical MD and increased 11C-DED binding involving parietal and precuneus CTh is plotted in figure 5, B and D, respectively. areas. These areas are susceptible to amyloid deposition but are relatively spared of tau deposition at early disease stages. Our results thus suggest that increased CTh (and reduced MD) are related to inflammation in early preclinical stages whereas re- Discussion duced CTh (and increased MD) are more closely related to tau In this study, we investigated the structural MRI correlates of aggregation, which drives local neurodegeneration in later pre- brain astrocytosis as measured by 11C-DED PET in patients in clinical and prodromal AD stages.33 early presymptomatic stages of ADAD. We report that 11C- DED binding has a measurable structural correlate in the form Only 2 previous studies have reported cortical MD decreases of decreased cortical MD and increased CTh in ADAD. in presymptomatic ADAD. Fortea and collaborators34 found decreased cortical MD in a group of presymptomatic PSEN1 We first assessed 11C-DED binding in ADAD. Our results, using carriers. These findings were later replicated in a separate a surface-based approach, confirmed previously published cohort reported by Ryan and collaborators,35 who found results in the same cohort using regional and voxel-wise decreased MD in asymptomatic carriers in subcortical struc- approaches12 in which 11C-DED binding peaked in early pre- tures, and increased MD in symptomatic carriers. The CTh symptomatic ADAD and then decreased with EYO toward the results in our presymptomatic group, although only at trend symptomatic stage. This result suggests that 11C-DED PET can level, also agree with previous studies reporting increases in track the inflammatory processes that occur in the early phases CTh or volume in presymptomatic ADAD.34,36,37 Overall, our of the disease and highlights the role of neuroinflammation in results are in agreement with previous studies that have AD pathogenesis. Preclinical studies have shown that astrocytes assessed brain microstructure and macrostructure in contribute to the clearance of β-amyloid species.30,31 The ex- ADAD.34,35 In preclinical sporadic AD, nonlinear trajectories posure of astrocytes to soluble β-amyloid species was reported have been described for MD and CTh.13,15 This nonlinear to promote astrocytic MAO-B upregulation,32 which may ex- trajectory would be the result of the transition from an plain the observed high 11C-DED binding in presymptomatic amyloid-negative healthy stage to an amyloid-positive pre- ADAD. In contrast, at late disease stages there is preclinical clinical and then clinical disease stage. However, data sup- evidence for astrocytic dysfunction and atrophy observed in porting nonlinear trajectories in ADAD are less clear,34 aged transgenic mouse brains.17 In our study, the observed suggesting that participants harboring a mutation present decline in 11C-DED binding toward the symptomatic stage may abnormal increases in CTh from very early ages.37 Given thus be a sign of astrocyte cell loss or of progressive changes in these previous reports in ADAD and the small sample size of e2032 Neurology | Volume 94, Number 19 | May 12, 2020 Neurology.org/N a common underlying neuroinflammatory process as measured Figure 4 Group comparisons of cortical mean diffusivity by 11C-DED binding. (MD) in presymptomatic mutation carriers (pMC) and symptomatic mutation carriers (sMC) vs Our results are biologically plausible. Neuroinflammation has noncarriers (NC) been postulated as central to AD pathogenesis,4 and the ele- vated 11C-DED binding in presymptomatic carriers was expected, as previously published.12 In this presymptomatic stage, the inflammatory process would produce changes in cell phenotype including increased cell volume (neuronal or glial swelling) and cell number (glial recruitment and activation) that could explain the decrease in cortical MD.15 Indeed, it has been reported that changes in cell volume or glial activation can alter the microstructural properties of brain tissue.39 Studies conducted by us and others that have found decreases in dif- fusivity in presymptomatic ADAD34,35 have interpreted the results using the same biological rationale. Recent advances in PET technology have allowed this hypothesis, which was driven by animal studies and pathologic data, to be now tested in vivo. Consonant with our main hypothesis, our 11C-DED binding results favor the interpretation that astrocyte reactivity may underlie the observed local structural changes in pre- symptomatic carriers. In symptomatic carriers, the observed increase in cortical MD was consistent with our previous study in the symptomatic stage of sporadic AD.15 In this stage, loss of tissue integrity and breakdown of cell membranes and intracellular organelles would result in water molecules moving more easily, and thus in increased diffusivity.40 (A) Surface map represents clusters of reduced (in blue) cortical MD in pMC compared to NC; only clusters surviving multiple-comparison correction are Interestingly, not all brain areas that showed astrocytosis pre- depicted (family-wise error [FWE]–corrected, p < 0.05); coordinates of all significant clusters are available in table e-5 (doi.org/10.5061/dryad. sented decreased cortical MD, and vice versa. Several factors 585581j). (B) Box-and-whisker plot compares the average cortical MD within might account for this mismatch. First, only clusters surviving the left hemisphere between pMC (red) and NC (blue) (illustrative purposes). (C) Surface map represents clusters of increased (in red) cortical MD in sMC multiple comparisons are shown. The uncorrected map-to-map compared to NC; only clusters surviving multiple-comparison correction are depicted (FWE-corrected, p < 0.05); coordinates of all significant clusters are correlations between structural measures and 11C-DED binding available in table e-5. (D) Box-and-whisker plot compares the average MD showed increased spatial concordance between those measures within the left hemisphere (LH) between sMC (red) and NC (blue) (illustrative purposes). RH = right hemisphere. (not shown). Second, although our study shows that both phenomena co-occur, other pathophysiologic factors apart from astrocytosis, such as microgliosis,7,41 neuronal hypertrophy, and our study, we were restricted to testing linear models, but changes in cell membrane permeability, could account for future studies with larger cohorts are warranted to test more cortical MD changes.18,42 In addition, in this study we only complex nonlinear trajectories. measured one marker of astrocytosis (11C-DED binding), and therefore we cannot exclude that other local processes measured The main finding of our study was the significant map-to-map using alternative markers of astrocytosis or microgliosis7,41 may local association between 11C-DED binding and cortical mi- contribute to MD changes in any given region. crostructure and macrostructure in ADAD. This is the first time that evidence of the structural impact of brain astrocytosis as The results of this study have several implications. First, it is measured by 11C-DED binding has been reported in ADAD. evident from our diffusivity results and previous studies by us Interestingly, 11C-DED binding had a negative association with and others in patients with sporadic AD that cortical MD is MD, but a positive association with CTh. This finding is con- decreased in early preclinical stages.15,43 This finding is fol- sistent with the divergent behavior of MD and CTh measures lowed by widespread increases in cortical MD once the disease as previously reported in sporadic AD and in frontotemporal advances to the symptomatic stage.44,45 A neuroinflammatory dementia,15,38 where increased CTh was related to decreased mechanism common to both sporadic and familial AD may MD, while atrophy co-occurs with increased MD. The novelty underlie these similar patterns of cortical microstructural and of the present study is that, although MD and CTh have di- macrostructural changes, reinforcing the role of neuro- vergent directions, they are both topographic biomarkers inflammation in disease pathogenesis. Second, we have shown whereby decreased MD and increased CTh both reflect that brain inflammation has a structural correlate and, more Neurology.org/N Neurology | Volume 94, Number 19 | May 12, 2020 e2033 Figure 5 Local associations between 11C-deuterium-L-deprenyl (11C-DED) PET binding and cortical microstructure and macrostructure in mutation carriers (A) Surface map represents clusters of significant vertex-wise negative correlations (in blue) between astrocytosis as measured by 11C-DED PET binding and cortical mean diffusivity (MD). Only clusters surviving multiple-comparison correction are depicted (family-wise error [FWE]–corrected, p < 0.05); coordinates of all significant clusters are available in table e-6 (doi.org/10.5061/dryad.585581j). (B) Linear regression of cortical MD (vertical axis) vs 11C-DED PET binding (horizontal axis) averaged within the left hemisphere (LH), including a 95% confidence band around the model prediction (illustrative purposes). (C) Surface map represents clusters of significant vertex-wise correlations between astrocytosis as measured by 11C-DED PET binding and cortical thickness (CTh). Only clusters surviving multiple-comparison correction are depicted (FWE-corrected, p < 0.05); coordinates of all significant clusters are available in table e-6. (D) Linear regression of CTh (vertical axis) vs 11C-DED PET binding (horizontal axis) evaluated within the LH, including a 95% confidence band around the model prediction (illustrative purposes). RH = right hemisphere. importantly, that these brain changes can be measured when cortical microstructure biomarkers are more sensitive than analyzing both the diffusion properties of brain tissue and its macrostructure. In this regard, we recently reported that cor- CTh. The fact that cortical MD is related to brain astrocytosis tical MD is more sensitive than CTh to detect neurodegener- reinforces the value of assessing the microstructural properties ative processes in frontotemporal dementia38 and sporadic AD of the tissue, as suggested previously,15,40 and further research is (in preparation). In addition, this study could affect the in- warranted to investigate the potential utility of cortical MD as terpretation of MRI biomarkers as surrogate markers in clinical an early biomarker. A third important implication involves trials. What kind of biomarker changes would be expected in clinical trial design. Inflammation biomarkers are of interest in trials aimed at reducing glial activation and neuroinflammation? clinical trials, for patient stratification and to track biological If our interpretation is correct and brain inflammation produces effects of drugs.4 In this respect, our results using 11C-DED cortical thickening and decreased cortical diffusivity, a drug that PET and recent studies emphasizing the role of astrocyte effectively decreases brain inflammation could produce the biomarkers in AD46,47 motivate further research on the use of opposite effect: it would contribute to cortical thinning and astrocyte PET biomarkers in clinical trials. Our results also increased diffusivity with respect to the pretreatment state. This showed that both MD and CTh are topographic biomarkers counterintuitive effect would support the notion that secondary to neuroinflammatory processes as measured by brain shrinkage after immunotherapy in active (AN1792)48 11 C-DED PET, supporting the potential added value of MD and passive (solanezumab49 and bapineuzumab50) immuniza- and CTh as AD biomarkers. The fact that the CTh results were tion trials was caused directly or indirectly by reducing less prominent than the cortical MD results may indicate that inflammation. e2034 Neurology | Volume 94, Number 19 | May 12, 2020 Neurology.org/N The strengths of this study include its multimodal imaging Foundation, the Åke Wiberg Foundation, and the Departament approach and the accurate surface-based method, which de Salut de la Generalitat de Catalunya, Pla estratègic de recerca i were chosen to overcome processing limitations high- innovació en salut (PERIS) 2016–2020 (SLT006/17/95). lighted in the literature.15 More fundamentally, the unique cohort and multimodal MRI-PET data in this study provide valuable insights into the structural correlates of neuro- Disclosure inflammation in ADAD. The study has some limitations. None of the authors has a conflict of interest related to this First, the number of participants included is small and the study. Go to Neurology.org/N for full disclosures. results should be interpreted with caution and as a pre- liminary, proof-of-concept study. Nevertheless, we em- Publication history phasize the rarity of the condition, the uniqueness of this Received by Neurology May 2, 2019. Accepted in final form sample with 11C-DED PET, the unicentric nature of this November 18, 2019. cohort, and that only results that survived multiple com- parisons are presented. Second, diffusion MRI is particu- larly susceptible to field artifacts. The lack of gradient field Appendix Authors maps did not allow us to perform a physics-based correc- Name Location Contribution tion of echoplanar imaging distortion. Finally, although Eduard Universitat Conception and design of the cortical MD and astrocytosis are related, given the cross- Vilaplana, Autònoma de study, acquisition and analysis of sectional design of the study, we could not infer causality MSc, PhD Barcelona, Spain data, drafting and revision of the manuscript critically for with the presented data. Only longitudinal studies with intellectual content long follow-up can establish the sequence of events that Elena Karolinska Conception and design of the take place in AD. Rodriguez- Institutet, study, acquisition and analysis of Vieitez, MSc, Stockholm, data, drafting and revision of the PhD Sweden manuscript critically for This study shows that astrocytosis is an early event in familial intellectual content AD that peaks during the presymptomatic stage. Importantly, Daniel Karolinska Conception and design of the this presymptomatic astrocytosis has a structural brain correlate Ferreira, PhD Institutet, study, acquisition and analysis of that is measurable as decreased cortical MD and increased Stockholm, data, drafting and revision of the cortical thickness. Changes in brain astrocytosis, microstruc- Sweden manuscript critically for intellectual content ture, and macrostructure occur simultaneously as the disease progresses, leading to decreased astrocyte activation in the Victor Montal, Universitat Conception and design of the MSc Autònoma de study, drafting and revision of symptomatic phase of the disease as diffusivity increases and Barcelona, Spain the manuscript critically for the cortex thins. These results should be considered in clinical intellectual content trials so that neuroimaging biomarkers can be interpreted Ove Almkvist, Karolinska Acquisition and analysis of data, correctly when used as outcome measures, and also when PhD Institutet, revision of the manuscript Stockholm, critically for intellectual content modeling predicted outcomes in response to treatment. Sweden Anders Wall, Uppsala Acquisition and analysis of data, Acknowledgment PhD University, revision of the manuscript The authors thank the participants in the PET studies and Sweden critically for intellectual content their relatives; the staff at the Uppsala PET Center and the Alberto Lleó, Universitat Conception and design of the Memory Clinic at Karolinska University Hospital, Hud- MD, PhD Autònoma de study, drafting and revision of dinge; and Dr. Anne Kinhult Ståhlbom for clinical pro- Barcelona, Spain the manuscript critically for intellectual content fessional help. Eric Westman, Karolinska Conception and design of the PhD Institutet, study, acquisition and analysis of Study funding Stockholm, data, revision of the manuscript This work was financially supported by grants from the Swedish Sweden critically for intellectual content Research Council (projects 2017-02965, 2017-06086, K2014- Caroline Graff, Karolinska Acquisition and analysis of data, 61×-05817), the Swedish Foundation for Strategic Research MD, PhD Institutet, revision of the manuscript Stockholm, critically for intellectual content (SSF, project RB13-0192), the Knut and Alice Wallenberg Sweden Foundation, the Stockholm County Council-Karolinska Insti- Juan Fortea, Universitat Conception and design of the tutet regional agreement on medical training and clinical re- MD, PhD Autònoma de study, acquisition and analysis of search (ALF grant), GE Healthcare (unrestricted grant), the Barcelona, Spain data, drafting and revision of the manuscript critically for Swedish Brain Foundation, the Alzheimer Foundation in Swe- intellectual content den, the Dementia Association, Swedish Brain Power, the EU FP7 large-scale integrating project INMiND (uni-muenster.de/ Agneta Karolinska Conception and design of the Nordberg, MD, Institutet, study, acquisition and analysis of INMiND), the Foundation for Old Servants, Karolinska Insti- PhD Stockholm, data, drafting and revision of the tutet’s Foundation for Aging Research, Gun and Bertil Stohne’s Sweden manuscript critically for intellectual content Foundation, Loo and Hans Osterman’s Foundation, the Åhlén Neurology.org/N Neurology | Volume 94, Number 19 | May 12, 2020 e2035 References 26. Coalson TS, Van Essen DC, Glasser MF. The impact of traditional neuroimaging methods on the spatial localization of cortical areas. Proc Natl Acad Sci USA 2018; 1. Scheltens P, Blennow K, Breteler MMB, et al. Alzheimer’s disease. Lancet 2016;388: 115:E6356–E6365. 505–517. 27. Jones DK, Symms MR, Cercignani M, Howard RJ. The effect of filter size on VBM 2. Bateman RJ, Aisen PS, De Strooper B, et al. Autosomal-dominant Alzheimer’s disease: analyses of DT-MRI data. Neuroimage 2005;26:546–554. a review and proposal for the prevention of Alzheimer’s disease. Alzheimers Res Ther 28. Patlak CS, Blasberg RG. Graphical evaluation of blood-to-brain transfer constants from 2011;3:1. 3. Gordon BA, Blazey TM, Su Y, et al. Spatial patterns of neuroimaging biomarker multiple-time uptake data. Generalizations. J Cereb Blood Flow Metab 1985;5:584–590. change in individuals from families with autosomal dominant Alzheimer’s disease: 29. Almkvist O, Rodriguez-Vieitez E, Thordardottir S, et al. Predicting cognitive decline a longitudinal study. Lancet Neurol 2018;17:241–250. across four Decades in mutation carriers and non-carriers in autosomal-dominant 4. Heneka MT, Carson MJ, El Khoury J, et al. Neuroinflammation in Alzheimer’s Alzheimer’s disease. J Int Neuropsychol Soc 2017;23:195–203. disease. Lancet Neurol 2015;14:388–405. 30. Thal DR. The role of astrocytes in amyloid beta-protein toxicity and clearance. Exp 5. Lambert JC, Ibrahim-Verbaas CA, Harold D, et al. Meta-analysis of 74,046 individuals Neurol 2012;236:1–5. identifies 11 new susceptibility loci for Alzheimer’s disease. Nat Genet 2013;45: 31. Wyss-Coray T, Loike JD, Brionne TC, et al. Adult mouse astrocytes degrade amyloid- 1452–1458. beta in vitro and in situ. Nat Med 2003;9:453–457. 6. Varrone A, Nordberg A. Molecular imaging of neuroinflammation in Alzheimer’s 32. Song W, Zhou LJ, Zheng SX, Zhu XZ. Amyloid-beta 25-35 peptide induces expression of disease. Clin Transl Imaging 2015;3:437–447. monoamine oxidase B in cultured rat astrocytes. Acta Pharmacol Sin 2000;21:557–563. 7. Stefaniak J, O’Brien J. Imaging of neuroinflammation in dementia: a review. J Neurol 33. Bischof GN, Jessen F, Fliessbach K, et al. Impact of tau and amyloid burden on glucose Neurosurg Psychiatry 2016;87:21–28. metabolism in Alzheimer’s disease. Ann Clin Transl Neurol 2016;3:934–939. 8. Ekblom J, Jossan SS, Bergstrom M, Oreland L, Walum E, Aquilonius SM. Monoamine 34. Fortea J, Sala-Llonch R, Bartres-Faz D, et al. Increased cortical thickness and caudate oxidase-B in astrocytes. Glia 1993;8:122–132. volume precede atrophy in PSEN1 mutation carriers. J Alzheimers Dis 2010;22: 9. Fowler JS, Logan J, Volkow ND, Wang GJ. Translational neuroimaging: positron 909–922. emission tomography studies of monoamine oxidase. Mol Imaging Biol 2005;7: 35. Ryan NS, Keihaninejad S, Shakespeare TJ, et al. Magnetic resonance imaging evidence 377–387. for presymptomatic change in thalamus and caudate in familial Alzheimer’s disease. 10. Johansson A, Engler H, Blomquist G, et al. Evidence for astrocytosis in ALS dem- onstrated by [11C](L)-deprenyl-D2 PET. J Neurol Sci 2007;255:17–22. Brain 2013;136:1399–1414. 11. Tong J, Rathitharan G, Meyer JH, et al. Brain monoamine oxidase B and A in human 36. Lee GJ, Lu PH, Medina LD, et al. Regional brain volume differences in symptomatic parkinsonian dopamine deficiency disorders. Brain 2017;140:2460–2474. and presymptomatic carriers of familial Alzheimer’s disease mutations. J Neurol 12. Rodriguez-Vieitez E, Saint-Aubert L, Carter SF, et al. Diverging longitudinal changes Neurosurg Psychiatry 2013;84:154–162. in astrocytosis and amyloid PET in autosomal dominant Alzheimer’s disease. Brain 37. Quiroz YT, Schultz AP, Chen K, et al. Brain imaging and blood biomarker abnor- 2016;139:922–936. malities in children with autosomal dominant Alzheimer disease: a cross-sectional 13. Fortea J, Sala-Llonch R, Bartres-Faz D, et al. Cognitively preserved subjects with study. JAMA Neurol 2015;72:912–919. transitional cerebrospinal fluid ss-amyloid 1-42 values have thicker cortex in Alz- 38. Illan-Gala I, Montal V, Borrego-Ecija S, et al. Cortical microstructure in the behavioral heimer’s disease vulnerable areas. Biol Psychiatry 2011;70:183–190. variant of frontotemporal dementia: looking beyond atrophy. Brain 2019;142: 14. Fortea J, Vilaplana E, Alcolea D, et al. Cerebrospinal fluid beta-amyloid and phospho- 1121–1133. tau biomarker interactions affecting brain structure in preclinical Alzheimer disease. 39. Roitbak T, Sykova E. Diffusion barriers evoked in the rat cortex by reactive astro- Ann Neurol 2014;76:223–230. gliosis. Glia 1999;28:40–48. 15. Montal V, Vilaplana E, Alcolea D, et al. Cortical microstructural changes along the 40. Weston PS, Simpson IJ, Ryan NS, Ourselin S, Fox NC. Diffusion imaging changes in Alzheimer’s disease continuum. Alzheimers Dement 2018;14:340–351. grey matter in Alzheimer’s disease: a potential marker of early neurodegeneration. 16. Pegueroles J, Vilaplana E, Montal V, et al. Longitudinal brain structural changes in preclinical Alzheimer’s disease. Alzheimers Dement 2017;13:499–509. Alzheimers Res Ther 2015;7:47. 17. De Strooper B, Karran E. The cellular phase of Alzheimer’s disease. Cell 2016;164: 41. Fan Z, Brooks DJ, Okello A, Edison P. An early and late peak in microglial activation in 603–615. Alzheimer’s disease trajectory. Brain 2017;140:792–803. 18. Iacono D, O’Brien R, Resnick SM, et al. Neuronal hypertrophy in asymptomatic 42. Le Bihan D. Looking into the functional architecture of the brain with diffusion MRI. Alzheimer disease. J Neuropathol Exp Neurol 2008;67:578–589. Nat Rev Neurosci 2003;4:469–480. 19. Thordardottir S, Stahlbom AK, Ferreira D, et al. Preclinical cerebrospinal fluid and 43. Racine AM, Adluru N, Alexander AL, et al. Associations between white matter mi- volumetric magnetic resonance imaging biomarkers in Swedish familial Alzheimer’s crostructure and amyloid burden in preclinical Alzheimer’s disease: a multimodal disease. J Alzheimers Dis 2015;43:1393–1402. imaging investigation. Neuroimage Clin 2014;4:604–614. 20. Petersen RC. Mild cognitive impairment as a diagnostic entity. J Intern Med 2004; 44. Douaud G, Menke RA, Gass A, et al. Brain microstructure reveals early abnormalities 256:183–194. more than two years prior to clinical progression from mild cognitive impairment to 21. McKhann G, Drachman D, Folstein M, Katzman R, Price D, Stadlan EM. Clinical Alzheimer’s disease. J Neurosci 2013;33:2147–2155. diagnosis of Alzheimer’s disease: report of the NINCDS-ADRDA work group under 45. Scola E, Bozzali M, Agosta F, et al. A diffusion tensor MRI study of patients with MCI the auspices of Department of Health and Human Services Task Force on Alzheimer’s and AD with a 2-year clinical follow-up. J Neurol Neurosurg Psychiatry 2010;81: disease. Neurology 1984;34:939–944. 798–805. 22. Carter SF, Scholl M, Almkvist O, et al. Evidence for astrocytosis in prodromal Alz- 46. Arranz AM, De Stroper B. The role of astroglia in Alzheimer’s disease: pathophysi- heimer disease provided by 11C-deuterium-L-deprenyl: a multitracer PET paradigm ology and clinical implications. Lancet Neurol 2019;18:406–414. combining 11C-Pittsburgh compound B and 18F-FDG. J Nucl Med 2012;53:37–46. 47. Carter SF, Herholz K, Rosa-Neto P, Pellerin L, Nordberg A, Zimmer ER. Astrocyte 23. Fischl B, Dale AM. Measuring the thickness of the human cerebral cortex from magnetic resonance images. Proc Natl Acad Sci USA 2000;97:11050–11055. biomarkers in Alzheimer’s disease. Trends Mol Med 2019;25:77–95. 24. Beer AL, Plank T, Meyer G, Greenlee MW. Combined diffusion-weighted and 48. Fox NC, Black RS, Gilman S, et al. Effects of A beta immunization (AN1792) on MRI functional magnetic resonance imaging reveals a temporal-occipital network involved measures of cerebral volume in Alzheimer disease. Neurology 2005;64:1563–1572. in auditory-visual object processing. Front Integr Neurosci 2013;7:5. 49. Doody RS, Thomas RG, Farlow M, et al. Phase 3 trials of solanezumab for mild-to- 25. Wu M, Lu LH, Lowes A, et al. Development of superficial white matter and its moderate Alzheimer’s disease. N Engl J Med 2014;370:311–321. structural interplay with cortical gray matter in children and adolescents. Hum Brain 50. Salloway S, Sperling R, Fox NC, et al. Two phase 3 trials of bapineuzumab in mild-to- Mapp 2014;35:2806–2816. moderate Alzheimer’s disease. N Engl J Med 2014;370:322–333. e2036 Neurology | Volume 94, Number 19 | May 12, 2020 Neurology.org/N