Daniel Ferreira

OBJECTIVE: To test the hypothesis that distinct subtypes of Alzheimer disease (AD) exist and underlie the heterogeneity within AD, we conducted a systematic review and meta-analysis on AD subtype studies based on postmortem and neuroimaging data. METHODS: EMBASE, PubMed, and Web of Science databases were consulted until July 2019. RESULTS: Neuropathology and neuroimaging studies have consistently identified 3 subtypes of AD based on the distribution of tau-related pathology and regional brain atrophy: typical, limbic-predominant, and hippocampal-sparing AD. A fourth subtype, minimal atrophy AD, has been identified in several neuroimaging studies. Typical AD displays tau-related pathology and atrophy both in hippocampus and association cortex and has a pooled frequency of 55%. Limbic-predominant, hippocampal-sparing, and minimal atrophy AD had a pooled frequency of 21%, 17%, and 15%, respectively. Between-subtype differences were found in age at onset, age at assessment, sex distribution, years of education, global cognitive status, disease duration, APOE ε4 genotype, and CSF biomarker levels. CONCLUSION: We identified 2 core dimensions of heterogeneity: typicality and severity. We propose that these 2 dimensions determine individuals' belonging to one of the AD subtypes based on the combination of protective factors, risk factors, and concomitant non-AD brain pathologies. This model is envisioned to aid with framing hypotheses, study design, interpretation of results, and understanding mechanisms in future subtype studies. Our model can be used along the A/T/N classification scheme for AD biomarkers. Unraveling the heterogeneity within AD is critical for implementing precision medicine approaches and for ultimately developing successful disease-modifying drugs for AD.

Atrophy patterns on MRI can reliably predict three neuropathological subtypes of Alzheimer's disease (AD): typical, limbic-predominant, or hippocampal-sparing. A method to enable their investigation in the clinical routine is still lacking. We aimed to (1) validate the combined use of visual rating scales for identification of AD subtypes; (2) characterise these subtypes at baseline and over two years; and (3) investigate how atrophy patterns and non-memory cognitive domains contribute to memory impairment. AD patients were classified as either typical AD (n = 100), limbic-predominant (n = 33), or hippocampal-sparing (n = 35) by using the Scheltens' scale for medial temporal lobe atrophy (MTA), the Koedam's scale for posterior atrophy (PA), and the Pasquier's global cortical atrophy scale for frontal atrophy (GCA-F). A fourth group with no atrophy was also identified (n = 30). 230 healthy controls were also included. There was great overlap among subtypes in demographic, clinical, and cognitive variables. Memory performance was more dependent on non-memory cognitive functions in hippocampal-sparing and the no atrophy group. Hippocampal-sparing and the no atrophy group showed less aggressive disease progression. Visual rating scales can be used to identify distinct AD subtypes. Recognizing AD heterogeneity is important and visual rating scales may facilitate investigation of AD heterogeneity in clinical routine.

OBJECTIVE: To investigate the particular pathology of the Arctic APP (APParc) early-onset familial Alzheimer disease (eoFAD) mutation for the first time in vivo with PET in comparison with other eoFAD mutations and sporadic Alzheimer disease (sAD). METHODS: We examined 2 APParc mutation carriers together with 5 noncarrier siblings cross-sectionally with (11)C-labeled Pittsburgh compound B (PiB) and (18)F-fluorodeoxyglucose (FDG) PET, as well as MRI, CSF biomarkers, and neuropsychological tests. Likewise, we examined 7 patients with sAD, 1 carrier of a presenilin 1 (PSEN1) mutation, 1 carrier of the Swedish APP (APPswe) mutation, and 7 healthy controls (HCs). RESULTS: Cortical PiB retention was very low in the APParc mutation carriers while cerebral glucose metabolism and CSF levels of Aβ(1-42), total and phosphorylated tau were clearly pathologic. This was in contrast to the PSEN1 and APPswe mutation carriers revealing high PiB retention in the cortex and the striatum in combination with abnormal glucose metabolism and CSF biomarkers, and the patients with sAD who showed typically high cortical PiB retention and pathologic CSF levels as well as decreased glucose metabolism when compared with HCs. CONCLUSIONS: The lack of fibrillar β-amyloid (Aβ) as visualized by PiB PET in APParc mutation carriers suggests, given the reduced glucose metabolism and levels of Aβ(1-42) in CSF, that other forms of Aβ such as oligomers and protofibrils are important for the pathologic processes leading to clinical Alzheimer disease.

Deep learning (DL) methods have in recent years yielded impressive results in medical imaging, with the potential to function as clinical aid to radiologists. However, DL models in medical imaging are often trained on public research cohorts with images acquired with a single scanner or with strict protocol harmonization, which is not representative of a clinical setting. The aim of this study was to investigate how well a DL model performs in unseen clinical datasets-collected with different scanners, protocols and disease populations-and whether more heterogeneous training data improves generalization. In total, 3117 MRI scans of brains from multiple dementia research cohorts and memory clinics, that had been visually rated by a neuroradiologist according to Scheltens' scale of medial temporal atrophy (MTA), were included in this study. By training multiple versions of a convolutional neural network on different subsets of this data to predict MTA ratings, we assessed the impact of including images from a wider distribution during training had on performance in external memory clinic data. Our results showed that our model generalized well to datasets acquired with similar protocols as the training data, but substantially worse in clinical cohorts with visibly different tissue contrasts in the images. This implies that future DL studies investigating performance in out-of-distribution (OOD) MRI data need to assess multiple external cohorts for reliable results. Further, by including data from a wider range of scanners and protocols the performance improved in OOD data, which suggests that more heterogeneous training data makes the model generalize better. To conclude, this is the most comprehensive study to date investigating the domain shift in deep learning on MRI data, and we advocate rigorous evaluation of DL models on clinical data prior to being certified for deployment.

BACKGROUND: Targeted delivery of nerve growth factor (NGF) has emerged as a potential therapy for Alzheimer's disease (AD) due to its regenerative effects on basal forebrain cholinergic neurons. This hypothesis has been tested in patients with AD using encapsulated cell biodelivery of NGF (NGF-ECB) in a first-in-human study. We report our results from a third-dose cohort of patients receiving second-generation NGF-ECB implants with improved NGF secretion. METHODS: Four patients with mild to moderate AD were recruited to participate in an open-label, phase Ib dose escalation study with a 6-month duration. Each patient underwent stereotactic implant surgery with four NGF-ECB implants targeted at the cholinergic basal forebrain. The NGF secretion of the second-generation implants was improved by using the Sleeping Beauty transposon gene expression technology and an improved three-dimensional internal scaffolding, resulting in production of about 10 ng NGF/device/day. RESULTS: All patients underwent successful implant procedures without complications, and all patients completed the study, including implant removal after 6 months. Upon removal, 13 of 16 implants released NGF, 8 implants released NGF at the same rate or higher than before the implant procedure, and 3 implants failed to release detectable amounts of NGF. Of 16 adverse events, none was NGF-, or implant-related. Changes from baseline values of cholinergic markers in cerebrospinal fluid (CSF) correlated with cortical nicotinic receptor expression and Mini Mental State Examination score. Levels of neurofilament light chain (NFL) protein increased in CSF after NGF-ECB implant, while glial fibrillary acidic protein (GFAP) remained stable. CONCLUSIONS: The data derived from this patient cohort demonstrate the safety and tolerability of sustained NGF release by a second-generation NGF-ECB implant to the basal forebrain, with uneventful surgical implant and removal of NGF-ECB implants in a new dosing cohort of four patients with AD. TRIAL REGISTRATION: ClinicalTrials.gov identifier: NCT01163825 . Registered on 14 Jul 2010.