European Journal of Nuclear Medicine and Molecular Imaging (2022) 49:4589–4600 https://doi.org/10.1007/s00259-022-05915-5 ORIGINAL ARTICLE Neuroimaging analyses from a randomized, controlled study to evaluate plasma exchange with albumin replacement in mild‑to‑moderate Alzheimer’s disease: additional results from the AMBAR study Gemma Cuberas‑Borrós1,2 · Isabel Roca2 · Joan Castell‑Conesa2 · Laura Núñez3 · Mercè Boada4,5 · Oscar L. López6 · Carlota Grifols3 · Miquel Barceló3 · Deborah Pareto7 · Antonio Páez3 Received: 21 February 2022 / Accepted: 14 July 2022 / Published online: 22 July 2022 © The Author(s) 2022 Abstract Purpose This study was designed to detect structural and functional brain changes in Alzheimer’s disease (AD) patients treated with therapeutic plasma exchange (PE) with albumin replacement, as part of the recent AMBAR phase 2b/3 clinical trial. Methods Mild-to-moderate AD patients were randomized into four arms: three arms receiving PE with albumin (one with low-dose albumin, and two with low/high doses of albumin alternated with IVIG), and a placebo (sham PE) arm. All arms underwent 6 weeks of weekly conventional PE followed by 12 months of monthly low-volume PE. Magnetic resonance imag- ing (MRI) volumetric analyses and regional and statistical parametric mapping (SPM) analysis on 18F-fluorodeoxyglucose positron emission tomography (18FDG-PET) were performed. Results MRI analyses (n = 198 patients) of selected subcortical structures showed fewer volume changes from baseline to final visit in the high albumin + IVIG treatment group (p < 0.05 in 3 structures vs. 4 to 9 in other groups). The high albu- min + IVIG group showed no statistically significant reduction of right hippocampus. SPM 18FDG-PET analyses (n = 213 patients) showed a worsening of metabolic activity in the specific areas affected in AD (posterior cingulate, precuneus, and parieto-temporal regions). The high-albumin + IVIG treatment group showed the greatest metabolic stability over the course of the study, i.e., the smallest percent decline in metabolism (MaskAD), and least progression of defect compared to placebo. Conclusions PE with albumin replacement was associated with fewer deleterious changes in subcortical structures and less metabolic decline compared to the typical of the progression of AD. This effect was more marked in the group treated with high albumin + IVIG. Trial registration (AMBAR trial registration: EudraCT#: 2011–001,598-25; ClinicalTrials.gov ID: NCT01561053). Keywords Alzheimer’s disease · Plasmapheresis · Plasma exchange · Albumin · Intravenous immunoglobulin This article is part of the Topical Collection on Neurology – Dementia 4 * Gemma Cuberas‑Borrós Ace Alzheimer Center Barcelona - Universitat Internacional gemma.cuberas.borros@gmail.com de Catalunya, Barcelona, Spain 5 1 Centro de Investigación Biomédica en Red de Enfermedades Research & Innovation Unit, Althaia Xarxa Assistencial Neurodegenerativas (CIBERNED), Instituto de Salud Carlos Universitària de Manresa, Carrer Dr. Joan Soler 1‑3, III, Madrid, Spain 08242 Manresa, Spain 6 2 Departments of Neurology and Psychiatry, University Department of Nuclear Medicine, Hospital Universitari Vall of Pittsburgh School of Medicine, Pittsburgh, PA, USA d’Hebrón, Universitat Autònoma de Barcelona, Barcelona, 7 Spain Radiology Department (IDI), Hospital Universitari Vall 3 d’Hebron, Barcelona, Spain Alzheimer’s Research Group, Grifols, Barcelona, Spain 13 Vol.:(0123456789) 4590 European Journal of Nuclear Medicine and Molecular Imaging (2022) 49:4589–4600 Introduction Material and methods Dementia is a growing public health concern worldwide. It Objective and study design is estimated that the number of people living with dementia will be 131.5 million by 2050 [1]. Alzheimer’s disease (AD) The neuroimaging objective of the AMBAR study was to accounts for 60–80% of these cases of dementia [2]. The detect functional and structural changes in the brains of behavioral symptoms of AD include cognitive impairment, mild-to-moderate AD patients treated with PE with thera- memory loss, depression, and deficits in communication and peutic albumin replacement in comparison with placebo spatial perception [3]. Microscopically, AD is characterized (sham-PE) patients. These outcomes were assessed using by neurofibrillary tangles (intracellular) and amyloid plaques magnetic resonance imaging (MRI) and 18F-fluorodeoxyglu- (extracellular) made up of phosphorylated tau proteins and cose positron emission tomography (18FDG-PET). amyloid β peptides (Aβ), respectively [4, 5]. Both of these The AMBAR trial enrolled patients at 41 sites: 19 in defects are thought to contribute to neuronal cell death. Spain and 22 in the USA. This study was conducted in com- In the efforts to attack the underlying disease process in pliance all applicable regulatory guidelines and the Interna- AD, one of the primary targets has been Aβ. Accumulation tional Council for Harmonization of Technical Requirements of Aβ in brain tissue and vasculature is thought to be an for Pharmaceuticals for Human Use (ICH) Good Clinical important part of the AD process [6]. Although the thera- Practice (GCP). Each patient and caregiver, family member, peutic strategy of targeting Aβ to improve function in AD or legal representative provided informed consent prior to patients has been until recently largely unsuccessful [7–9], enrollment. The Institutional Review Boards (IRBs) or Eth- a potential clinical benefit associated with reduction of Aβ ics Committees from the sites and the Health Authorities plaques using anti-Aβ antibodies is currently contribut- from the countries where the study was performed approved ing to the understanding of Aβ pathophysiology [10–13]. the protocol, the informed consent form, and the patient Furthermore, additional interest in Aβ has been piqued by information sheets. the observation that sequestration of Aβ in the circulation resulted in a decrease in Aβ in the cerebrospinal fluid (CSF) Patient population [14]. This led to the “sink hypothesis” that removal of Aβ from the circulation could lead to a decrease in Aβ in the In the AMBAR trial, 322 patients were evaluable and 232 CSF and the brain. patients (72%) completed the study. Reasons for discontinu- Given that Aβ in blood is bound to albumin (1:1) [15], ation are summarized in Supplemental Table S1 for placebo it was proposed that plasma exchange (PE) with albumin and active treatment groups. Further details have been previ- replacement could be used to remove Aβ from the peripheral ously reported elsewhere [23]. circulation. Removal of the patient’s plasma would result in Inclusion criteria were men and women between 55 and removal of albumin-bound Aβ. PE would be followed by 85 years of age with a diagnosis of probable AD dementia replacement of the lost volume along with exogenous albu- according to the National Institute of Neurological and Com- min to avoid hypovolemia and maintain osmotic pressure munication Disorders and Stroke and the Alzheimer’s Dis- [16]. This could result in further sequestration of Aβ through ease and Related Disorders Association (NINCDS-ADRDA) binding to the fresh Aβ-free [17] albumin. Additional PE criteria [24], with baseline Mini-Mental State Exam could produce further reductions in CSF/brain Aβ through (MMSE) score from 18 to 26 [25], and on a stable dose equilibration with plasma Aβ [18]. of AChEIs and/or memantine. Exclusion criteria included The neuroimaging studies described in this paper were cerebrovascular disease, IgA deficiency, low hemoglobin, part of a multicenter, randomized, controlled, phase 2b/3 high plasma creatinine, uncontrolled hypertension, liver dis- study to evaluate the efficacy and safety of PE with albumin ease, heart diseases, any condition in which the treatment replacement with or without alternated intravenous immuno- was contraindicated or not feasible, and fewer than 6 years globulin (IVIG) in AD patients (AMBAR Trial: EudraCT#: of education. Full details of eligibility criteria are available 2011–001,598-25; ClinicalTrials.gov ID: NCT01561053). elsewhere [22, 23]. In a previous phase 2 study [19], positive clinical changes The population was 54% female with an average age of associated with PE treatment were found to parallel neuro- 69.0 ± 7.7 (mean ± SD) years. The patients were equally imaging data [20]. Similarly, the previously published clini- divided in terms of mild (n = 161, baseline Mini-Mental cal results from the AMBAR study showed that PE treatment State Examination [MMSE] score: 22–26) and moderate slowed the decline in cognitive, functional, and neuropsy- AD (n = 161, baseline MMSE score: 18–21). Patients were chological outcomes in AD patients, while improving their free of cerebrovascular disease and had been diagnosed quality of life [21–23]. with AD since 2.4 ± 2.4 years based on criteria from the 13 European Journal of Nuclear Medicine and Molecular Imaging (2022) 49:4589–4600 4591 NINCDS-ADRDA criteria [24]. These patients had a base- magnetization-prepared rapid gradient-echo (MP-RAGE) line MMSE score of 21.3 ± 2.6 [25] and were being treated pulse sequences included in a vendor-supplied package, with acetylcholinesterase inhibitors and/or memantine at and using the same acquisition protocol (TR = 2300 ms, a stable dose. Further details of demographic and clinical TE = 2.98 ms, field of view 256 mm, and 192 sagittal parti- characteristics of patients have been previously published tions, voxel size = 1 × 1 × 1 mm). [23] and summarized for placebo and active treatment groups in Supplemental Table S1. 18 FDG‑PET acquisition Treatment groups Patients who were claustrophobic or non-cooperative, those without suitable venous access to administer the tracer, and Eligible patients were randomly assigned one of the four those taking corticosteroids at the time of the test were groups (1:1:1:1), three PE-treated groups and one placebo excluded. Patients fasted for at least 4 h prior to the pro- group. During intensive treatment phase, the three treat- cedure to provide optimal brain 18FDG uptake that was not ment groups received weekly conventional therapeutic PE influenced by increased blood glucose levels. (TPE) with albumin (Albutein® 5%, Grifols) replacement, Before the injection of 18FDG, blood glucose levels were through peripheral or central venous access, for 6 weeks. checked to assure that they were lower than 160 mg/dL. This was followed by a 12-month maintenance period with Patients were placed in a quiet, dimly lit room several min- monthly low-volume PE (LVPE) during which three dif- utes before 18FDG administration, and during the 18FDG ferent treatment modalities were administered: one group uptake phase (at least 20 min), talking, reading, or listening received LVPE with low-dose (20 g; 100 mL) 20% albumin were avoided. Sedation of patients was not allowed. (Albutein® 20%, Grifols); one group received low-dose A dose of 185 ± 18 MBq (5 ± 0.5 mCi) 18FDG was admin- albumin alternated with low-dose (10 g; 200 mL) IVIG (Fle- istered, and 30 min after injection, the images were acquired. bogamma® 5% DIF, Grifols) every 4 months; and one group Acquisition of CT images was carried out first. Next, 18FDG- received high-dose (40 g; 200 mL) 20% albumin alternated PET images were acquired in 3D mode, with an acquisition with high-dose (20 g; 400 mL) IVIG every 4 months. The duration of 15 min, matrix size 128 × 128, slice thickness placebo group received a simulated non-invasive PE (sham) (Z) of 2.0 mm, and pixel size (X, Y) of 2.0 mm × 2.0 mm. procedure that did not involve any fluid transfer [22, 23]. The images were reconstructed with iterative algorithm (15 Placebo patients were subjected to the same procedures and subsets and 4 iterations). evaluations as PE-treated patients. The patients, caregivers, The images were corrected for attenuation with the CT and neuroimaging evaluators were blinded as to the therapy scan, and were reviewed to assess possible artifacts and received. movement. Eventually the reconstructed images were reori- ented on the orbitomeatal axis. Image acquisition Individual image analysis MRI and 18FDG-PET measures were taken at baseline (month 0; M0), the intermediate visit after the intensive Longitudinal analysis of the 18FDG-PET and MRI scans was treatment phase (month 2; M2), month 9 (M9), and the final limited to patients who completed all study visits and had visit (month 14; M14). MRI scans were also performed at analyzable scans at all visits. month 6 (M6). Images were acquired at different imaging sites, MRI MRI quantification with a 1.5 T magnet scanner, and 18F-FDG-PET with PET/ CT scanners. Each participating imaging site acquired and FIRST software from the Oxford Centre for Functional Mag- reconstructed the data according to a standardized protocol netic Resonance Imaging of the Brain (FMRIB) Software following the Alzheimer’s Disease Neuroimaging Initiative Library (FSL) (Oxford, UK) [27] and SPM8 software (Wel- (ADNI) recommendations [26]. come Trust Centre for Neuroimaging, London, UK) running on MATLAB (MATLAB R2012b, MathWorks, Natick, MA, MRI acquisition USA) were used to compute brain volumetry. FIRST, a fully automated method, was applied to the Patients were asked to lie in supine position, the center of MPRAGE images to derive subcortical volumes. MRI the coil was aligned with the line of the eyes, they were images were reoriented according to the anatomic FSL instructed not to move, and the head was gently stabilized atlas, using the fslorient2std tool. Next, the neck region was with cushions. Whole-brain T1-weighted images of 1-mm extracted with the robustfov tool and ran FIRST to determine isotropic resolution were obtained at isocenter using the the volume of the right and left thalamus, putamen, caudate, 13 4592 European Journal of Nuclear Medicine and Molecular Imaging (2022) 49:4589–4600 pallidum, hippocampus, amygdala, accumbens area, and templates created from the normal-reference template, each brainstem plus the fourth ventricle. patient voxel was assigned a value based on a preset range In addition, the following cortical volumetries were ana- (see Supplemental Table S3). These values were used to lyzed: total intracranial volume, overall gray matter (GM), create an individual defect map. Each of these values was white matter (WM), and cerebrospinal fluid (CSF) volumes, assigned a color indicating the severity of the lesion. For an together with fractions of all these measures with respect to individual subject, these parametric analyses could be com- the total intracranial volume (TIV). MRI scans were nor- pared over the course of the study. A particular voxel could malized and segmented by using the default unified seg- be evaluated over time and evaluated as to whether it had mentation methods of SPM8 to obtain GM, WM, and CSF improved (voxels that showed a defect at M0 and were not a segmented masks. Gray matter and white matter volumes defect at M14, or voxels that had an increase of Z-score of were obtained and segmented, as well as cerebrospinal fluid more than 0.5), maintained (defect or normal at M0 and the (CSF) for each patient, and the TIV was calculated as the same at M14), or worsened (a voxel that was not a defect at sum of gray, white, and CSF volumes, allowing generation M0, but was a defect at M14, or a voxel with a Z-score that of the corresponding gray and white matter relative to the worsened by 0.5) relative to the control database. total intracranial volume. Finally, to minimize the effects of cerebral atrophy on the creation of the individual metabolic defect pattern, a correc- 18 FDG‑PET quantification tion mask was created from the gray matter probability maps of the same patient and applied to the result to remove those In the first part of this study, a quantitative approach based voxels which may correspond to areas of atrophy or increase/ on voxel-by-voxel analysis was performed to assess the dilation of the ventricular cavities. global metabolic activity of the entire brain (by creating In order to allow comparisons between treatment groups parametric images of the metabolic defect pattern and its and comparisons over time within treatment group, a follow-up), as well as individual brain region analysis to regional analysis was performed by extracting the mean identify individual cerebral metabolic abnormalities. uptake values from the regions of the Automatic Anatomical Image pre-processing was carried out by using SPM8. Labeling (AAL) atlas [31], using an in-house algorithm, and To start, 18FDG-PET images were spatially normalized onto statistical analyses were conducted. AAL atlas is a parcel- the Montreal Neurological Institute (MNI) space, intensity lation of a standardized MRI scan from the MNI and mean normalized to the cerebellum mean (as this is a large and counts were extracted from its 116 mapped brain regions stable region and it is known to be minimally affected by (see Supplemental Table S4). age-related changes) [28–30] and spatially smoothed using When the 18FDG-PET scans were analyzed for the 116 an isotropic Gaussian kernel with 8 mm FWHM to increase brain areas (VOIs) using the AAL atlas, three parameters the signal-to-noise ratio and to account for the subtle varia- were measured: (1) metabolism intensity (mean counts per tions in anatomical structures. each region inside the cerebral atrophy correction mask The individual parametric images of the metabolic defect computed from the normalized and smoothed image sets); pattern were computed using the intensity normalized and (2) defect extension (% of abnormal voxels per each region smoothed image sets comparing them to an in-house and inside the cerebral atrophy correction mask computed from tracer-, and age-matched normal-reference brain 18FDG-PET the individual defect map), and (3) defect intensity (degree template (in MNI space) based on 48 healthy elderly con- of defect per each region inside the cerebral atrophy correc- trols recruited for other unicentric study (see Supplemental tion mask computed from the individual defect map). Table S2 for details). All subjects in the control population Finally, the baseline defect pattern was computed for all were scanned using same scanner, with the same method- groups by computing the mean image from all individual ology and protocols as in the present study. This approach defect maps. compares voxel-by-voxel each patient’s 18FDG-PET scan with those of the healthy controls. This comparison allowed the detection of metabolic defects based on the Z-score for Voxel‑based analysis with statistical parametric an individual voxel and comparison to the mean and stand- mapping (SPM) ard deviation of the reference template. A voxel that, when compared to a control voxel at the same anatomic position Voxel-based analysis with SPM was used to establish the in MNI space in the normal-reference brain 18FDG-PET pattern of tissue abnormalities from the disease and to assess template, had a value that is lower than the average by at the global group effects of treatment over time. least two standard deviations was considered a defect. Based The paired t-test implemented in SPM8 was applied con- on the Z-score for an individual voxel and comparison to sidering a statistical threshold of p < 0.001 uncorrected and the mean and the standard deviation of the distribution in relation to the unspecific hypometabolic voxels. Those 13 European Journal of Nuclear Medicine and Molecular Imaging (2022) 49:4589–4600 4593 detected voxels that survive a cluster thresholding using a 96.3% of MRI), and n = 183 at M14 (representing 81.7% of 18 minimum cluster size of k = 50 were considered. FDG-PET and 92.4% of MRI). SPM maps showing contrasts were superimposed onto a T1 template using MRIcron on 12 axial views covering the Brain volumetry analysis by MRI entire brain. To identify the location of the significant voxels obtained in the SPM analysis, the Wake Forest University No changes were reported between treatment groups at each (WFU) PickAtlas tool [32] was used. cross-sectional time point and in all subcortical and cortical For each treatment group, the evolution of k-extent areas and globally there was a loss in volume throughout the (metabolism loss over time (change from baseline)/longi- 14-month period. tudinal analysis: M2 vs M0; M9 vs M0, and M14 vs M0) There was a significant volume loss of GM (in absolute was shown in maximal intensity projection (MIP) view. The values and fraction) in all treatment groups and placebo from measurement of defect extent and defect intensity took into M0 to M14. WM in absolute values were not significantly consideration all the voxels within the VOI. decreased in any treatment group and placebo group, and The SPM maps showing metabolic decline from M0 WM fraction was only reduced in the placebo group, but to M14 in the placebo group allowed identification of the not in any active treatment groups. On the contrary, signifi- regions of the brain involved in the natural progression of cant volume loss of CSF (in absolute values and fraction) AD. These data were used to create the Alzheimer’s Disease was observed in all active treatment groups, but not in the mask (MaskAD). The MaskAD when applied to extract the placebo group from baseline to final visit (M14). mean uptake from the scans of each individual in the treat- When volumetry for subcortical structures was analyzed ment groups and allowed the determination of the metabolic longitudinally (from M0 to M14), the placebo group showed defect intensity in the treatment groups compared to placebo. a significant volumetry loss in 4 areas (hippocampus [L, R] and accumbens [L, R]). For the low albumin group, significant decreases in volume were shown in 9 areas Statistics (caudate [L, R], hippocampus [L, R], accumbens [L, R], amygdala [L], putamen [R], and pallidum [R]). The low Descriptive statistics were calculated for all measured varia- albumin + IVIG group showed a significant volumetry loss bles (mean, median, standard deviation, 25th percentile, 75th in 8 areas (thalamus-proper [L, R], caudate [L, R], accum- percentile, minimum, and maximum). Comparisons between bens [L, R], putamen [L], and hippocampus [R]). For the treatment groups and longitudinal comparisons (M0, M2, high albumin + IVIG, there was a significant volumetry loss M6, M9, and M14) were made using analysis of variance in 3 areas: amygdala [L], caudate [R], and accumbens [R]. (ANOVA). Pairwise comparisons of different time points Therefore, nucleus accumbens [R] was significantly reduced within a treatment group were made using Student’s t-test in all treatment groups (active and sham), whereas caudate with a significance level set at p < 0.05. Statistical analyses [R] was significantly reduced in all active treatment groups, were performed using SAS software (version 9.2, Cary, NC, but not in placebo group. Pallidum [L] and brain stem 4th USA). ventricle areas were not significantly decreased for all active treatment and placebo groups. All groups except high albu- min + IVIG showed a statistically significant reduction of Results hippocampus [R]. Globally, fewer statistically significant changes in vol- Samples for image processing ume were observed over time in the high albumin + IVIG treatment group (p < 0.05 in 3 structures vs. 4 to 9 in other Of the total study population of 322 evaluable patients from groups) (Table 1). AMBAR study, 18FDG-PET scans were missing or unable to be analyzed for 13 patients at baseline (M0). At final visit Brain metabolism analysis by 18FDG‑PET (M14), 224 images were available. For longitudinal SPM 18 analysis, only patients that completed all study visits and FDG-PET analyses showed that the baseline defect pattern without images unable to be processed could be included was consistent with AD (posterior cingulate, both parietal, leaving 213 patients for 18FDG-PET longitudinal analysis. temporal [slightly more extensive in the left hemisphere], For MRI, 55 patients had missing or unreadable scans at M0. and slight extension to the frontal) and was similar between At M14, 198 images were available. This was the database all treatment groups. Similar defect patterns were obtained available for analysis of VOI. for mild and moderate AD patients. No differences were Most patients had both 18FDG-PET and MRI scans avail- observed in the defect pattern between mild and moderate able: n = 257 at M0 (representing 83.2% of 18FDG-PET and AD patients or between treatment groups. However, defect 13 4594 European Journal of Nuclear Medicine and Molecular Imaging (2022) 49:4589–4600 pattern in moderate AD patients was more intense (evi- changed between M0 and M14 are shown in Table 2. The denced by the color of the defect, greener, which means high albumin + IVIG treatment group showed the low- more standard deviations in relation to the normality data- est percentage of areas with significant changes in these base) than in mild AD patients, and more extensive (Fig. 1). parameters. Most of the changes in metabolic intensity Changes in metabolism intensity, defect extension, and were seen in the limbic structures, temporal and pari- defect intensity in all brain areas that had significantly etal cortex, and the cerebellum. The primary brain areas Table 1  Subcortical structures Placebo PE-treatment with significant (p < 0.05) structural (volumetric) changes Low Albumin Low Albumin + IVIG High Albumin + IVIG measured by MRI between baseline and month 14 in Accumbens [L, R] Accumbens [L, R] Accumbens [L, R] Accumbens [R] Alzheimer’s disease patients Hippocampus [L, R] Amygdala [L] Caudate [L, R] Amygdala [L] treated with plasma exchange Caudate [L, R] Hippocampus [R] Caudate [R] (PE) with low- or high-dose albumin with or without Hippocampus [L, R] Putamen [L] intravenous immunoglobulin Pallidum [R] Thalamus-proper [L, R] (IVIG) Putamen [R] Total: 4/15 (26.7%) Total: 9/15 (60%) Total: 8/15 (53.3%) Total: 3/15 (20%) Fig. 1  Defect pattern in 18 F-flurodeoxyglucose positron emission tomography (18FDG- PET) analysis at baseline in Alzheimer’s disease patients treated with plasma exchange with low- or high-dose albumin with or without intravenous immunoglobulin (IVIG). Axial view Table 2  Change in metabolism intensity, defect extension, and defect line and month 14 in Alzheimer’s disease patients treated with plasma intensity in all brain areas (N = 116) expressed as a number and per- exchange (PE) with low- or high-dose albumin with or without intra- centage of areas that significantly changed (p < 0.05) between base- venous immunoglobulin (IVIG) Placebo PE-treatment Low albumin Low albumin + IVIG High albumin + IVIG Metabolism intensity 21 (18.1%) 58 (50.0%) 44 (37.9%) 29 (25.0%) Defect extension 42 (36.2%) 63 (54.3%) 65 (56.0%) 36 (31.0%) Defect intensity 56 (48.3%) 52 (44.8%) 58 (50.0%) 47 (40.5%) 13 European Journal of Nuclear Medicine and Molecular Imaging (2022) 49:4589–4600 4595 affected were the same as those seen when measuring posterior cingulum, precuneus, and parietal cortex extending metabolic intensity. into the temporal region (Fig. 2A). Analysis of the areas that showed worsening over the study period indicated that the Parametric analysis of 18FDG‑PET areas affected were the posterior cingulum, precuneus, and parietal and temporal regions of the cortex (Fig. 2B). The The metabolic activity pattern is shown in Fig. 2. The areas high albumin + IVIG treatment group showed the greatest that showed maintenance of metabolic activity were the stability over the course of the study. Fig. 2  Triangulation view of the metabolic maintenance pattern (A: upper panel) and worsening pattern (B: lower panel) (para- metric 18FDG-PET analysis) over the complete study period (baseline [M0] to month 14 [M14]) in Alzheimer’s disease patients treated with plasma exchange with low- or high- dose albumin with or without intravenous immunoglobulin (IVIG). All patients 13 4596 European Journal of Nuclear Medicine and Molecular Imaging (2022) 49:4589–4600 The percentage decline in metabolism (MaskAD) from decline during the study period was less in the mild AD M0 to M14 is shown in Table 3. When all the patient data group than in the moderate AD group. High albumin + IVIG was considered together, the least amount of decline was had the greatest effect in the mild AD group. In the moder- seen in the high albumin + IVIG group. By AD severity, ate AD group, the least decline was seen in the low albu- min group. Table 3 shows that area of significant metabolic decline was less in the high albumin + IVIG group than in Table 3  Percentage of metabolism loss in MaskAD between baseline any of the other groups. This was true for the pooled data and month 14 in Alzheimer’s disease patients treated with plasma and when the patients were divided by AD severity. Greater exchange (PE) with low or high dose albumin with or without intrave- loss of metabolism (increased MaskAD size) was seen in the nous immunoglobulin (IVIG), stratified by disease severity placebo, low albumin, and low albumin + IVIG groups than Placebo PE-treatment in the high albumin + IVIG group. The high albumin + IVIG Low albu- Low albu- High albu- group showed the smallest MaskAD size in all three brain min min + IVIG min + IVIG regions (temporal, parietal, and frontal). Figure 3 shows the SPM analysis, where there was very All patients − 4.3 − 3.3 − 3.6 − 2.9 little change in metabolism in mild AD patients treated Mild AD − 1.6 − 2.1 − 2.9 − 0.9 with high albumin + IVIG. Some deterioration was seen in Moderate AD − 6.6 − 3.6 − 4.3 − 4.2 the other three treatment groups. As in the whole patient Fig. 3  Maximum intensity projection (MIP) view of changes of SPM between baseline and month 14 in Alzheimer’s disease patients treated with plasma exchange with low- or high-dose albumin with or without intravenous immunoglobulin (IVIG), stratified by disease severity 13 European Journal of Nuclear Medicine and Molecular Imaging (2022) 49:4589–4600 4597 population, the effects of high albumin + IVIG were seen types of dementia [35, 36]. However, 18FDG-PET imag- consistently across brain regions. There was essentially no ing is a more valuable tool in the assessment of patients change in metabolism in the parietal and frontal lobes over with AD [37]. More recently, MRI has demonstrated utility the course of the study in the high albumin + IVIG treat- due to a greater capability to distinguish anatomical struc- ment group while metabolism declined in the other treat- tures in the brain and provide more precise images [38]. In ment groups. this study 18FDG-PET and MRI scans performed on AD Table 4 shows that area of significant metabolic decline patients throughout the course of the study were compared (size of mask k-extent) in the SPM analysis was smallest in with scans performed at baseline (M0) and scans from a the high albumin + IVIG group. This was true for the pooled population without AD. Moreover, SPM is considered the data and when the patients were divided by AD severity. gold standard for dementia diagnosis [39]. With 213 patients Much greater loss of metabolism is reflected in the MIP view analyzed (49 to 60 per treatment group), this study is one of from the other treatment groups (Fig. 3). the largest performed using this technique in a clinical trial. The effects of treatment were more noticeable in the mild MRI has been shown to detect structural changes in the AD patient group. As shown in Supplemental Fig. S1A, at brain associated with AD including cerebral atrophy and M14 the metabolic intensity decreased while the extension enlargement of the ventricles and sulci [40, 41]. Our brain and intensity of defect slightly increased. As with the repre- volumetry studies showed that the high albumin + IVIG sentative example of moderate AD patients (Supplemental group had fewer structural changes from M0 to M14, includ- Fig. S1B), the reduction of metabolic intensity at M14 was ing the stabilization of the hippocampus volume. Impor- seen across all cerebral lobes, and more evident the exten- tantly, the hippocampus plays a key role in developing sion and intensity of defect, which doubled. new memories [42] and is one of the first areas in the brain affected by AD [43]. 18 FDG-PET analyses of the effects of PE on AD using Discussion these scans showed that baseline pattern of defect was con- sistent with AD [44–47] and similar between all treatment Diagnosis and progression of AD are currently based on groups. After treatment, metabolic deficits were still seen, clinical observation, scoring on various assessments of cog- but in fewer brain areas and were less extensive in the high nitive function, and surrogate biochemical markers [33]. albumin + IVIG treatment group than in the other treatment Most patients in the AMBAR study, who were enrolled groups or the placebo group. This was particularly evidenced based on the presence of the AD clinical syndrome, had in defect extension and defect intensity, the parameters baseline CSF Aβ42 levels ≤ 600 pg/mL [23]. This is in agree- directly related to neuronal damage rather than metabolic ment with patient populations and methods of determination intensity. These effects were observed primarily in the brain in other studies [23]. However, over the years, there has been regions most affected by AD (posterior cingulate, precuneus increasing interest in the diagnostic utility of neuroimaging. and parieto-temporal lobes) [48–50]. This study explored 18FDG-PET and MRI results from the When looking at the longitudinal metabolic decline in all AMBAR study patients. brain areas (MaskAD or number of voxels showing a meta- Among neuroimaging techniques, computed axial tomog- bolic decline at the end of the study compared to baseline), raphy was found to be useful in diagnosing some types of mild AD patients showed a pattern of metabolic decline dementia but is of limited usefulness in discriminating over time slower than moderate AD patients, as expected. AD [34]. Single positron emission computed tomography However, the increase in intensity and extension of defect (SPECT) has also shown some utility in distinguishing some were not significantly different between any of the treatment groups and placebo. This could be explained by the fact that minimal changes in the high albumin + IVIG group were Table 4  Metabolism loss in SPM 18FDG-PET in size of mask k-extent (in voxels) between baseline and month 14 in Alzheimer’s balanced by the extensive pattern of defect in the two low disease patients treated with plasma exchange (PE) with low- or high- albumin groups (with or without IVIG), which had change dose albumin with or without intravenous immunoglobulin (IVIG), patterns similar to placebo. This suggests that the mild AD stratified by disease severity patients’ group was more stable over time, regardless of the Placebo PE-treatment treatment arm. Low albumin Low albu- High albu- Interestingly, the clinical results showed that PE-treated min + IVIG min + IVIG mild AD patients scored better in verbal memory and lan- All patients 83,998 118,074 143,269 30,822 guage fluency tests (particularly in Rey auditory verbal Mild AD 2560 8050 9494 293 learning test [RAVLT], semantic verbal fluency [SVF], Moderate AD 101,251 91,978 104,177 49,706 and phonetic verbal fluency PVF]) [21]. This paralleled the reduced metabolic deficit observed in the posterior cingulate 13 4598 European Journal of Nuclear Medicine and Molecular Imaging (2022) 49:4589–4600 gyrus and parietal lobe, where damage is suggestive of AD marked in the group treated with high albumin + IVIG pathology in patients with language deficits [44, 48]. Moreo- compared to the low albumin groups (with or without ver, 18FDG uptake in the posterior cingulate-precuneus or in IVIG) and in moderate AD patients. Mild AD patients language-related Brodmann areas has been shown to corre- seemed metabolically more stable over time, regardless late the main indices of RAVLT in patients with mild cogni- of the treatment arm. tive impairment (MCI) [51] and in AD [52]. These results are important because PE could stabilize hypometabolism or Supplementary Information The online version contains supplemen- tary material available at https://d​ oi.o​ rg/1​ 0.1​ 007/s​ 00259-0​ 22-0​ 5915-5. hypoperfusion in mild AD. These defects are predictors of a rapid progression to more severe AD [53]. More specifically, Acknowledgements Jordi Bozzo PhD, CMPP and Michael K James, the angular region, along with the posterior cingulate, are PhD (Grifols) are acknowledged for medical writing and editorial sup- regions that present greater progression of metabolic loss. port in the preparation of this manuscript. This finding is in agreement with the findings of Rosselli et al. [54] in which they demonstrated with structural con- Author contribution GC-B was the lead investigator of the imaging branch of the trial. AP was the medical director of the trial and con- nectivity techniques the integrative role of the angular region ducted the study oversight as well as provided guidance and expertise in linguistic functions. on plasma-derived medicines. MBo and OLL were the coordinating In moderate AD patients, there was a significant differ- investigators of the trial in Spain and the USA, respectively. GC-B, AP, ence in the extension of defect pattern between high albu- LN, MBo, and OLL were responsible for the design and methods of the trial. MBo participated in patient recruitment and data collection. min + IVIG and the rest of treatment arms. Low albumin LN, CG, and MBa were responsible of the trial clinical operations. (regardless of IVIG) presented extension defect patterns GC-B, IR, JC-C, and DP contributed to neuroimage data collection similar to placebo, whereas high albumin + IVIG group had and/or interpretation. MBo, OLL, LN, CG, MBa, and AP contributed less loss of metabolism by intensity and extension than the to the discussion on Alzheimer’s disease. GC-B, LN, CG, MBa and AP participated in the manuscript writing up. All authors revised rest. This suggests that the higher dose of albumin was nec- and approved the final version of the manuscript and agreed to the essary to achieve the neuroprotective effects. submission. Previously published data from pilot and phase 2 stud- ies showed that the positive clinical effects of PE in AD Funding The AMBAR study is sponsored by Grifols, a manufacturer patients [19, 55] were correlated with positive neuroimag- of therapeutic human serum albumin and intravenous immune globu- lin. GC-B, IR, JC-C, DP, MBo, and OLL received direct or indirect ing data [20, 55]. Taking these findings to a further step, funding from Grifols to carry out the study and the preparation of the dose separation and separate mild-to-moderate AD patients manuscript. were assessed in this phase 2b/3 study. Clinical results showed that some of the positive cognitive and behavioral Data availability Data reported in this manuscript are available within effects of PE appeared to be more pronounced in moderate the article and/or its supplementary materials. Additional data from the AMBAR study (EudraCT#: 2011–001,598-25; ClinicalTrials.gov AD patients [23], while memory, language, and processing ID: NCT01561053) are available from the corresponding author upon speed abilities appeared to be more pronounced in mild AD reasonable request. patients [21]. In the present neuroimaging analyses, some positive effects of PE treatment were seen in the mild AD Declarations group, while others were seen in the moderate AD group, particularly with high albumin + IVIG. Ethics approval This study was conducted in compliance all applicable A potential limitation of this study is that some of the regulatory guidelines and the International Council for Harmonization of Technical Requirements for Pharmaceuticals for Human Use (ICH) analyses such as metabolism intensity, pattern of defect, and Good Clinical Practice (GCP). Institutional Review Boards (IRBs) or defect intensity are semi-quantitative, which could conceal Ethics Committees from the sites and the Health Authorities from the some of the complexity of AD etiopathology. Additionally, countries where the study was performed approved the protocol, the a study period longer than 14 months may be necessary to informed consent form, and the patient information sheets. fully assess whether the treatment has a meaningful impact Consent to participate Each patient and caregiver, family member, or on the course of AD in these patients. Determining the exact legal representative provided informed consent prior to enrollment. relationship between the effects of PE on brain structure and function (as measured by 18FDG-PET and MRI) and Consent to publish Not applicable. cognitive and behavioral effects will require additional Competing interests MBo has been a consultant for Araclon, Avid, investigation. Bayer, Elan, Grifols, Janssen/Pfizer, Lilly, Neuroptix, Nutricia, Roche, In conclusion, PE treatment in AD patients was associ- Sanofi, Biogen, and Servier; and received fees for lectures and funds ated with fewer deleterious changes in selected subcortical for research from Araclon, Esteve, Grifols, Janssen, Novartis, Nutricia, Piramal, Pfizer-Wyett, Roche, and Servier. OLL has been a consult- structures and less metabolic decline in the specific areas ant for Grifols and Lundbeck. LN, MBa, CG, and AP are full-time affected in AD than the typical progression of the disease, employees of Grifols. as observed in the placebo group. These effects were more 13 European Journal of Nuclear Medicine and Molecular Imaging (2022) 49:4589–4600 4599 Open Access This article is licensed under a Creative Commons Attri- proteins and erythrocytes: implications for their quantitation in bution 4.0 International License, which permits use, sharing, adapta- plasma. Biochem Biophys Res Commun. 2000;268:750–6. tion, distribution and reproduction in any medium or format, as long 16. Meca-Lallana JE, Rodriguez-Hilario H, Martinez-Vidal S, as you give appropriate credit to the original author(s) and the source, Saura-Lujan I, Carreton-Ballester A, Escribano-Soriano JB, provide a link to the Creative Commons licence, and indicate if changes et al. Plasmapheresis: its use in multiple sclerosis and other were made. The images or other third party material in this article are demyelinating processes of the central nervous system. An included in the article's Creative Commons licence, unless indicated observation study. Rev Neurol. 2003;37:917–26. otherwise in a credit line to the material. If material is not included in 17. Costa M, Ortiz AM, Jorquera JI. Therapeutic albumin binding the article's Creative Commons licence and your intended use is not to remove amyloid-beta. J Alzheimers Dis. 2012;29:159–70. permitted by statutory regulation or exceeds the permitted use, you will 18. Huang Y, Potter R, Sigurdson W, Kasten T, Connors R, Morris need to obtain permission directly from the copyright holder. To view a JC, et al. beta-amyloid dynamics in human plasma. Arch Neurol. copy of this licence, visit http://​creat​iveco​mmons.​org/​licen​ses/​by/4.​0/. 2012;69:1591–7. 19. Boada M, Anaya F, Ortiz P, Olazarán J, Shua-Haim JR, Obis- esan TO, et al. Efficacy and safety of plasma exchange with 5% albumin to modify cerebrospinal fluid and plasma amyloid-β References concentrations and cognition outcomes in Alzheimer’s disease patients: a multicenter, randomized, controlled clinical trial. J 1. Prince M, Wimo A, Guerchet M, Ali G-C, Wu Y-T, Prina M. Alzheimers Dis. 2017;56:129–43. World alzheimer report 2015 - The global impact of dementia: 20. Cuberas-Borros G, Roca I, Boada M, Tarraga L, Hernan- An analysis of prevalence, incidence, cost and trends. London: dez I, Buendia M, et al. Longitudinal neuroimaging analysis Alzheimer's Dis Int. 2015;p84. in mild-moderate Alzheimer’s disease patients treated with 2. Alzheimer's disease facts and figures. Alzheimers dement. 2020. plasma exchange with 5% human albumin. J Alzheimers Dis. https://​doi.​org/​10.​1002/​alz.​12068. 2018;61:321–32. 3. Feldman HH, Jacova C, Robillard A, Garcia A, Chow T, Borrie 21. Boada M, Lopez O, Olazaran J, Nunez L, Pfeffer M, G. P-R, M, et al. Diagnosis and treatment of dementia: 2. Diagnosis et al. Neuropsychological, neuropsychiatric and quality of life Cmaj. 2008;178:825–36. assessments in Alzheimer’s disease patients treated with plasma 4. Blennow K, de Leon MJ, Zetterberg H. Alzheimer’s disease. exchange with albumin replacement from the AMBAR Study. Lancet. 2006;368:387–403. Alzheimers Dement. 2021. https://​doi.​org/​10.​1002/​alz.​12477. 5. Ittner LM, Gotz J. Amyloid-beta and tau–a toxic pas de deux in 22. Boada M, Lopez O, Nunez L, Szczepiorkowski ZM, Torres M, Alzheimer’s disease. Nat Rev Neurosci. 2011;12:65–72. Grifols C, et al. Plasma exchange for Alzheimer’s disease Manage- 6. Selkoe DJ, Hardy J. The amyloid hypothesis of Alzheimer’s ment by Albumin Replacement (AMBAR) trial: study design and disease at 25 years. EMBO Mol Med. 2016;8:595–608. progress. Alzheimers Dement (N Y). 2019;5:61–9. 7. Brody M, Liu E, Di J, Lu M, Margolin RA, Werth JL, et al. A 23. Boada M, Lopez O, Olazaran J, Nunez L, Pfeffer M, Paricio M, phase II, randomized, double-blind, placebo-controlled study of et al. A randomized, controlled clinical trial of plasma exchange safety, pharmacokinetics, and biomarker results of subcutaneous with albumin replacement for Alzheimer’s disease: primary results bapineuzumab in patients with mild to moderate Alzheimer’s of the AMBAR Study. Alzheimers Dement. 2020;16:1412–25. disease. J Alzheimers Dis. 2016;54:1509–19. 24. McKhann G, Drachman D, Folstein M, Katzman R, Price D, Stad- 8. Cummings JL, Morstorf T, Zhong K. Alzheimer’s disease drug- lan EM. Clinical diagnosis of Alzheimer’s disease: report of the development pipeline: few candidates, frequent failures. Alzhei- NINCDS-ADRDA Work Group under the auspices of Depart- mers Res Ther. 2014;6:37. ment of Health and Human Services Task Force on Alzheimer’s 9. Vandenberghe R, Rinne JO, Boada M, Katayama S, Scheltens P, Disease. Neurology. 1984;34:939–44. Vellas B, et al. Bapineuzumab for mild to moderate Alzheimer’s 25 Folstein MF, Folstein SE, McHugh PR. “Mini-mental state”. A disease in two global, randomized, phase 3 trials. Alzheimers practical method for grading the cognitive state of patients for the Res Ther. 2016;8:18. clinician. J Psychiatr Res. 1975;12:189–98. 10. Cummings J, Aisen P, Lemere C, Atri A, Sabbagh M, Sallo- 26. Alzheimer’s disease neuroimaging initiative 2 (ADNI2). Proce- way S. Aducanumab produced a clinically meaningful benefit dures manual. 2010. Available at: https://​adni.​loni.​usc.​edu/​wp-​ in association with amyloid lowering. Alzheimers Res Ther. ​ anual.p​ df. Accessed conten​ t/u​ pload​ s/2​ 008/0​ 7/a​ dni2-p​ roced​ ures-m 2021;13:98. 22 May 2022 11. Klein G, Delmar P, Voyle N, Rehal S, Hofmann C, Abi-Saab D, 27. Patenaude B, Smith SM, Kennedy DN, Jenkinson M. A Bayesian et al. Gantenerumab reduces amyloid-β plaques in patients with model of shape and appearance for subcortical brain segmenta- prodromal to moderate Alzheimer’s disease: a PET substudy tion. Neuroimage. 2011;56:907–22. interim analysis. Alzheimers Res Ther. 2019;11:101. 28. Borghammer P. Perfusion and metabolism imaging studies in Par- 12. Mintun MA, Lo AC, Duggan Evans C, Wessels AM, Ardayfio kinson’s disease. Dan Med J. 2012;59:B4466. PA, Andersen SW, et al. Donanemab in early Alzheimer’s dis- 29. López-González FJ, Silva-Rodríguez J, Paredes-Pacheco J, ease. N Engl J Med. 2021;384:1691–704. Niñerola-Baizán A, Efthimiou N, Martín-Martín C, et al. Intensity 13. Swanson CJ, Zhang Y, Dhadda S, Wang J, Kaplow J, Lai RYK, normalization methods in brain FDG-PET quantification. Neuro- et al. A randomized, double-blind, phase 2b proof-of-concept image. 2020;222:117229. clinical trial in early Alzheimer’s disease with lecanemab, an 30. Brugnolo A, De Carli F, Pagani M, Morbelli S, Jonsson C, Chin- anti-Aβ protofibril antibody. Alzheimers Res Ther. 2021;13:80. carini A, et al. Head-to-head comparison among semi-quantifi- 14. Matsuoka Y, Saito M, LaFrancois J, Saito M, Gaynor K, Olm cation tools of brain FDG-PET to aid the diagnosis of prodromal V, et al. Novel therapeutic approach for the treatment of Alz- Alzheimer’s disease. J Alzheimers Dis. 2019;68:383–94. heimer’s disease by peripheral administration of agents with an 31. Tzourio-Mazoyer N, Landeau B, Papathanassiou D, Crivello F, affinity to beta-amyloid. J Neurosci. 2003;23:29–33. Etard O, Delcroix N, et al. Automated anatomical labeling of acti- 15. Kuo YM, Kokjohn TA, Kalback W, Luehrs D, Galasko DR, vations in SPM using a macroscopic anatomical parcellation of Chevallier N, et al. Amyloid-beta peptides interact with plasma the MNI MRI single-subject brain. Neuroimage. 2002;15:273–89. 13 4600 European Journal of Nuclear Medicine and Molecular Imaging (2022) 49:4589–4600 32. Maldjian JA, Laurienti PJ, Kraft RA, Burdette JH. An automated 46 Mosconi L. Brain glucose metabolism in the early and specific method for neuroanatomic and cytoarchitectonic atlas-based inter- diagnosis of Alzheimer’s disease. FDG-PET studies in MCI and rogation of fMRI data sets. Neuroimage. 2003;19:1233–9. AD. Euro J Nucl Med Mol Imaging. 2005;32:486–510. 33. Dubois B, Feldman HH, Jacova C, Hampel H, Molinuevo 47. Pagani M, Dessi B, Morbelli S, Brugnolo A, Salmaso D, Piccini JL, Blennow K, et al. Advancing research diagnostic criteria A, et al. MCI patients declining and not-declining at mid-term fol- for Alzheimer’s disease: the IWG-2 criteria. Lancet Neurol. low-up: FDG-PET findings. Curr Alzheimer Res. 2010;7:287–94. 2014;13:614–29. 48. Lehmann M, Rohrer JD, Clarkson MJ, Ridgway GR, Scahill RI, 34. Naugle RI, Cullum CM, Bigler ED, Massman PJ. Neuropsycho- Modat M, et al. Reduced cortical thickness in the posterior cin- logical and computerized axial tomography volume characteris- gulate gyrus is characteristic of both typical and atypical Alzhei- tics of empirically derived dementia subgroups. J Nerv Ment Dis. mer’s disease. J Alzheimers Dis. 2010;20:587–98. 1985;173:596–604. 49. Yokoi T, Watanabe H, Yamaguchi H, Bagarinao E, Masuda M, 35. Sakurai T, Kawashima S, Satake S, Miura H, Tokuda H, Toba K. Imai K, et al. Involvement of the precuneus/posterior cingulate Differential subtypes of diabetic older adults diagnosed with Alz- cortex is significant for the development of Alzheimer’s disease: heimer’s disease. Geriatr Gerontol Int. 2014;14(Suppl 2):62–70. a PET (THK5351, PiB) and resting fMRI study. Front Aging Neu- 36. Song IU, Chung YA, Chung SW, Jeong J. Early diagnosis of Alz- rosci. 2018;10:304. heimer’s disease and Parkinson’s disease associated with dementia 50. Bailly M, Destrieux C, Hommet C, Mondon K, Cottier J-P, Beau- using cerebral perfusion SPECT. Dement Geriatr Cogn Disord. fils E, et al. Precuneus and cingulate cortex atrophy and hypome- 2014;37:276–85. tabolism in patients with Alzheimer’s disease and mild cognitive 37. Marcus C, Mena E, Subramaniam RM. Brain PET in the diagnosis impairment: MRI and 18F-FDG PET quantitative analysis using of Alzheimer’s disease. Clin Nucl Med. 2014;39:e413–26. FreeSurfer. Biomed Res Int. 2015;2015:583931. 38. Boelaarts L, Scheltens P, de Jonghe J. Does MRI increase the 51. Brugnolo A, Morbelli S, Arnaldi D, De Carli F, Accardo J, Bossert diagnostic confidence of physicians in an outpatient memory I, et al. Metabolic correlates of Rey auditory verbal learning test clinic. Dement Geriatr Cogn Dis Extra. 2016;6:242–51. in elderly subjects with memory complaints. J Alzheimers Dis. 39. Perani D, Della Rosa PA, Cerami C, Gallivanone F, Fallanca F, 2014;39:103–13. Vanoli EG, et al. Validation of an optimized SPM procedure for 52. Chiaravalloti A, Ricci M, Di Biagio D, Filippi L, Martorana A, FDG-PET in dementia diagnosis in a clinical setting. NeuroImage Schillaci O. The brain metabolic correlates of the main indices of Clinical. 2014;6:445–54. https://​doi.​org/​10.​1016/j.​nicl.​2014.​10.​ neuropsychological assessment in Alzheimer’s disease. J Person 009. Med. 2020;10:25. 40. Apostolova LG, Green AE, Babakchanian S, Hwang KS, 53. Drzezga A, Lautenschlager N, Siebner H, Riemenschneider Chou Y-Y, Toga AW, et al. Hippocampal atrophy and ventricu- M, Willoch F, Minoshima S, et al. Cerebral metabolic changes lar enlargement in normal aging, mild cognitive impairment accompanying conversion of mild cognitive impairment into Alz- (MCI), and Alzheimer Disease. Alzheimer Dis Assoc Disord. heimer’s disease: a PET follow-up study. Eur J Nucl Med Mol 2012;26:17–27. Imaging. 2003;30:1104–13. 41. Cai K, Xu H, Guan H, Zhu W, Jiang J, Cui Y, et al. Identifi- 54. Rosselli M, Ardila A, Bernal B. Angular gyrus connectivity model cation of early-stage Alzheimer’s disease using sulcal mor- for language: a functional neuroimaging meta-analysis. Rev Neu- phology and other common neuroimaging indices. PloS one. rol. 2015;60:495–503. 2017;12:e0170875. 55. Boada M, Ortiz P, Anaya F, Hernandez I, Munoz J, Nunez L, 42. Bisaz R, Travaglia A, Alberini CM. The neurobiological bases of et al. Amyloid-targeted therapeutics in Alzheimer’s disease: use memory formation: from physiological conditions to psychopa- of human albumin in plasma exchange as a novel approach for thology. Psychopathology. 2014;47:347–56. Abeta mobilization. Drug News Perspect. 2009;22:325–39. 43. Jahn H. Memory loss in Alzheimer’s disease. Dialogues Clin Neu- rosci. 2013;15:445–54. Publisher's note Springer Nature remains neutral with regard to 44 Brown RK, Bohnen NI, Wong KK, Minoshima S, Frey KA. Brain jurisdictional claims in published maps and institutional affiliations. PET in suspected dementia: patterns of altered FDG metabolism. Radiographics. 2014;34:684–701. 45. Jack CR Jr, Knopman DS, Jagust WJ, Shaw LM, Aisen PS, Weiner MW, et al. Hypothetical model of dynamic biomarkers of the Alz- heimer’s pathological cascade. Lancet Neurol. 2010;9:119–28. 13