The Dame Barbara Windsor Dementia Goals Programme was launched by the UK Government to accelerate the development and delivery of new treatments for dementia. We present the recommendations from the Scientific Advisory Board, to enable timely access to therapies for the wider population, reducing health system burden while improving patient outcomes. The recommendations focus on three areas: (i) establishing a new dynamic national patient registry for clinical trial recruitment; (ii) the use of biomarkers to improve early and accurate diagnosis; and (iii) a framework for end-to-end implementation across the landscape of healthcare, research and regulators. A Brain Aging Registry for Biomarkers, Access to trials, Research and Adoption would support recruitment, monitoring, and personalized care. Embedding digital and biomarker innovations into routine care would improve personalized and equitable dementia services, with earlier diagnosis and more effective prevention. Robust patient and public involvement is required, to ensure transparency, trustworthiness, and meaningful participation.

Synaptic proteins are critical for maintaining healthy neuronal transmission but can also drive prion-like trans-synaptic spreading of pathological aggregated proteins found in many neurodegenerative diseases, including Alzheimer's disease. Recognizing technological limitations for in situ identification of synaptic proteins, we built a modular synaptic chimeric antigen receptor (synCAR) for capture of synaptic proteins by fusing a single-chain antibody fragment with the post-synaptic protein neurolignin-1 (NLGN1). Using our synCAR platform, we show that anchoring the tau antibody PHF1 to the synapse effectively captures pathogenic synaptic tau species. Expressing PHF1 synCAR at synapses in mouse primary and human neuronal tau seeding models results in increased tau aggregation, likely due to concentrating pathological tau seeds at the synapse. These findings provide the first published method for the isolation and modulation of synaptic protein function within relevant biological contexts, highlighting synCAR as a relevant instrumental platform for synaptic protein research and neurodegenerative disease drug development.

Therapeutic interventions to block extracellular tau seeding to prevent endogenous tau aggregation and progression of Alzheimer's disease pathology are currently being investigated in clinical trials. However, the translation of promising preclinical findings to benefit clinical outcomes remains problematic due to the lack of pathophysiological models that recapitulate key features of sporadic Alzheimer's disease-related tauopathies. We developed a primary neuronal tau (hTau) seeding and propagation model. Neurons expressing wild-type human tau protein at a physiological level, seeded with a sub-nanomolar tau derived from Alzheimer's disease brain tissues, rapidly and robustly form tau aggregates and develop impaired mitochondrial function. Resulting aggregates are quantitatively measured using automated high-content algorithms. The considerable pathophysiological relevance, coupled with a highly sensitive dynamic range, makes this assay a valuable model system for studying tau pathobiology and an efficient screening tool for modulators of tau aggregation. Using this model, we demonstrate that by targeting a phosphorylation-specific epitope of tau, an antibody effectively stops tau aggregation.

Alzheimer disease (AD) is the most common form of dementia affecting more than 6 million people in the United States. Currently, 3 monospecific antibodies targeting different Amyloid β (Aβ) species have been approved by the US FDA as disease modifying therapeutics for treatment in early AD patients with amyloid pathology. ABBV-916 is a clinical stage human IgG1 monoclonal antibody which binds to N-terminal truncated, pyroglutamate-modified at amino acid position 3, Aβ (AβpE3). The current study characterized ABBV-916 using human tissue samples and amyloid precursor protein (APP) transgenic mice. ABBV-916 selectively bound to recombinant AβpE3-42 fibrils and native amyloid plaques in unfixed AD brain tissue but did not bind targets in human CSF. ABBV-916 significantly reduced dense plaques from brain tissue that were co-cultured with hiPSC-derived phagocytes. In APPPS1-21 mice, ABBV‑916 bound plaques in a dose-dependent manner after a single intravenous injection. In addition, three months of weekly administration of ABBV-916 murine surrogate antibody significantly decreased amyloid plaques in APPPS1-21 mice. In vivo two-photon imaging revealed that the murine version of ABBV-916 inhibited the growth of the plaques in APPPS1-21 mice. ABBV-916 murine surrogate antibody recruited microglia to plaques within 24-48 hours after a single intraperitoneal injection in Cx3cr1-tdTomato/APPPS1-21 mice. Importantly, in contrast to a positive control antibody, ABBV‑916 murine precursor antibody did not cause microhemorrhage in aged APPPS1-21 mice. Taken together, our results suggest that ABBV-916 is a promising drug candidate. Clinical testing is on-going to evaluate the plaque removal and safety profiles of ABBV-916 in AD patients.

Aggregation of misfolded α-synuclein (α-syn) is a key characteristic feature of Parkinson’s disease (PD) and related synucleinopathies. The nature of these aggregates and their contribution to cellular dysfunction is still not clearly elucidated. We employed mass spectrometry-based total and phospho-proteomics to characterize the underlying molecular and biological changes due to α-syn aggregation using the M83 mouse primary neuronal model of PD. We identified gross changes in the proteome that coincided with the formation of large Lewy body-like α-syn aggregates in these neurons. We used protein-protein interaction (PPI)-based network analysis to identify key protein clusters modulating specific biological pathways that may be dysregulated and identified several mechanisms that regulate protein homeostasis (proteostasis). The observed changes in the proteome may include both homeostatic compensation and dysregulation due to α-syn aggregation and a greater understanding of both processes and their role in α-syn-related proteostasis may lead to improved therapeutic options for patients with PD and related disorders.

Human microglia are critically involved in Alzheimer’s disease (AD) progression, as shown by genetic and molecular studies. However, their role in tau pathology progression in human brain has not been well described. Here, we characterized 32 human donors along progression of AD pathology, both in time—from early to late pathology—and in space—from entorhinal cortex (EC), inferior temporal gyrus (ITG), prefrontal cortex (PFC) to visual cortex (V2 and V1)—with biochemistry, immunohistochemistry, and single nuclei-RNA-sequencing, profiling a total of 337,512 brain myeloid cells, including microglia. While the majority of microglia are similar across brain regions, we identified a specific subset unique to EC which may contribute to the early tau pathology present in this region. We calculated conversion of microglia subtypes to diseased states and compared conversion patterns to those from AD animal models. Targeting genes implicated in this conversion, or their upstream/downstream pathways, could halt gene programs initiated by early tau progression. We used expression patterns of early tau progression to identify genes whose expression is reversed along spreading of spatial tau pathology (EC > ITG > PFC > V2 > V1) and identified their potential involvement in microglia subtype conversion to a diseased state. This study provides a data resource that builds on our knowledge of myeloid cell contribution to AD by defining the heterogeneity of microglia and brain macrophages during both temporal and regional pathology aspects of AD progression at an unprecedented resolution.

Astrocytes play a critical role in brain homeostasis and normal functions but their changes along the spatiotemporal progression of Alzheimer’s disease (AD) neuropathology remain largely unknown. Here we performed single-nucleus RNA-sequencing on brain regions along the stereotypical progression of AD pathology from donors ranging the entire normal aging-AD continuum comprising 628,943 astrocyte nuclei from 32 donors across 5 brain regions. We discovered temporal gene-expression-trajectories with gene sets differentially activated at various disease stages. Surprisingly, a gene set enriched in proteostasis and energy metabolism, was upregulated in late-stage but unexpectedly returned to baseline levels in end-stage, suggesting exhaustion of response in “burnt-out” astrocytes. The spatial gene-expression-trajectories revealed that astrocytic genes of tripartite synapses are dysregulated in parallel to the stereotypical progression of tangle pathology across regions. We identified astrocyte heterogeneity across brain regions with a continuum from homeostatic to reactive cells through “intermediate” transitional states. These findings suggest complex astrocytic dysfunction in AD neurodegeneration.

Two phase-III clinical trials with anti-amyloid peptide antibodies have met their primary goal, i.e. slowing of Alzheimer's disease (AD) progression. However, antibody therapy may not be the optimal therapeutic modality for AD prevention, as we will discuss in the context of the earlier small molecules described as "γ-secretase modulators" (GSM). We review here the structure, function, and pathobiology of γ-secretases, with a focus on how mutations in presenilin genes result in early-onset AD. Significant progress has been made in generating compounds that act in a manner opposite to pathogenic presenilin mutations: they stabilize the proteinase-substrate complex, thereby increasing the processivity of substrate cleavage and altering the size spectrum of Aβ peptides produced. We propose the term "γ-secretase allosteric stabilizers" (GSAS) to distinguish these compounds from the rather heterogenous class of GSM. The GSAS represent, in theory, a precision medicine approach to the prevention of amyloid deposition, as they specifically target a discrete aspect in a complex cell biological signalling mechanism that initiates the pathological processes leading to Alzheimer's disease.

In Alzheimer’s disease (AD) progression, amyloid beta load uniformly increases across the brain cortex, while neurofibrillary tangles progressively spread in a stereotypical pattern from entorhinal to visual cortex. AD is known to have a strong genetic link to microglia and recently, expression profiling at single cell resolution has identified disease‐associated microglia populations with differential amyloid‐beta vs. tau pathology association in human AD brain. However, prior human studies focusing on microglia included a limited number of donors and/or cells per donor, and how microglial transcriptomes change with progression of tau pathology remains largely unknown.

Microglia, astrocytes, and neurons undergo profound functional alterations in response to amyloid-β (Aβ) plaques and neurofibrillary tangles (NFTs) in Alzheimer’s disease (AD). Recent studies using human induced pluripotent stem cell-derived glial cells and neurons, APOE knock-in mice, and human brain bulk RNA-seq have implicated the APOE genotype in these changes, but lack spatial information with respect to Aβ plaques and NFTs. Here we tested the hypotheses that (1) transcriptomic differences between AD and control subjects are maximum within and near Aβ plaques and NFT-bearing neurons; (2) Aβ plaques and NFTs are associated with distinct transcriptomic changes; and (3) the APOE genotype differentially impacts the transcriptomic changes associated with Aβ plaques and NFTs in the AD brain. Laser capture microdissection (LCM) in cryostat sections was performed from superior temporal gyrus (BA22) cortex of human AD (n = 10, including n = 5 APOE ε4/ε4, n = 4 APOE ε3/ε3, and n = 1 APOE ε3/ε4) and age- and sex-matched controls (n = 8). Thioflavin-S-positive Aβ plaques, the 50 µm halo around them, NFTs with the 50 µm halo around them, and areas far (>50 µm) from plaques and NFTs were laser-capture microdissected and subjected to RNA-sequencing to identify differentially expressed genes (DEGs). Aβ plaques had a greater impact on the transcriptome than NFTs (i.e., higher number of DEGs, both upregulated [logFC>0, unadj. P -value<0.05: 2,623 vs. 1,230 genes; adj. P -value<0.05: 1,152 vs. 0 genes] and downregulated [logFC>0, unadj. P -value<0.05: 2,563 vs. 1,056 genes; adj. P -value<0.05: 827 vs. 0 genes]) relative to control cortex. Aβ plaques were characterized by upregulated microglial and downregulated neuronal genes, whereas NFTs had primarily downregulated neuronal genes. A gradient of Aβ plaques > plaque halo > NFTs > far areas was evident, with upregulation of neuroinflammation and downregulation of synaptic neurotransmission and energy metabolism-related gene sets. Comparing APOE ε4 and APOE ε3 age- and sex-matched homozygotes revealed greater changes in APOE ε4 homozygotes across locations. Aβ plaques and, to a lesser extent, NFTs, concentrate the bulk of transcriptomic changes in the AD cortex. The APOE ε4 allele is associated with greater microglial and neuronal transcriptomic responses to Aβ plaques and NFTs, compared to APOE ε3 carriers. These findings will inform future spatial transcriptomics studies.

INTRODUCTION:Omics studies have revealed that various brain cell types undergo profound molecular changes in Alzheimer's disease (AD) but the spatial relationships with plaques and tangles and APOE -linked differences remain unclear. METHODS:We performed laser capture microdissection of Aβ plaques, the 50μm halo around them, tangles with the 50μm halo around them, and areas distant (>50μm) from plaques and tangles in the temporal cortex of AD and control donors, followed by RNA-sequencing. RESULTS:Aβ plaques exhibited upregulated microglial (neuroinflammation/phagocytosis) and downregulated neuronal (neurotransmission/energy metabolism) genes, whereas tangles had mostly downregulated neuronal genes. Aβ plaques had more differentially expressed genes than tangles. We identified a gradient Aβ plaque>peri-plaque>tangle>distant for these changes. AD APOE ε4 homozygotes had greater changes than APOE ε3 across locations, especially within Aβ plaques. DISCUSSION:Transcriptomic changes in AD consist primarily of neuroinflammation and neuronal dysfunction, are spatially associated mainly with Aβ plaques, and are exacerbated by the APOE ε4 allele.

Vascular endothelial cells play an important role in maintaining brain health, but their contribution to Alzheimer's disease (AD) is obscured by limited understanding of the cellular heterogeneity in normal aged brain and in disease. To address this, we performed single nucleus RNAseq on tissue from 32 human AD and non-AD donors (19 female, 13 male) each with five cortical regions: entorhinal cortex, inferior temporal gyrus, prefrontal cortex, visual association cortex, and primary visual cortex. Analysis of 51,586 endothelial cells revealed unique gene expression patterns across the five regions in non-AD donors. Alzheimer's brain endothelial cells were characterized by upregulated protein folding genes and distinct transcriptomic differences in response to amyloid β plaques and cerebral amyloid angiopathy. This dataset demonstrates previously unrecognized regional heterogeneity in the endothelial cell transcriptome in both aged non-AD and AD brain. SIGNIFICANCE STATEMENT In this work, we show that vascular endothelial cells collected from five different brain regions display surprising variability in gene expression. In the presence of Alzheimer's disease pathology, endothelial cell gene expression is dramatically altered with clear differences in regional and temporal changes. These findings help explain why certain brain regions appear to differ in susceptibility to disease-related vascular remodeling events that may impact blood flow.

Amyloid-β plaques and neurofibrillary tangles in the Alzheimer’s disease (AD) brain are accompanied by prominent morphological and functional changes in astrocytes, collectively termed reactive astrogliosis. Single-nuclei RNA-sequencing (snRNA-seq) has begun to unveil the molecular underpinnings of AD reactive astrocytes but technical challenges, including low numbers of nuclei and/or sequencing depth, and the lack of comparisons across brain regions have limited the full picture. Here we present the largest snRNA-seq study of astrocytes to date across five brain regions. Nuclei were isolated from five brain areas of n = 32 donors with increasing AD neuropathology (total n = 160 samples). The five brain areas were chosen based on their hierarchical accumulation of tau pathology: entorhinal cortex (EC)> inferior temporal gyrus (BA20) > dorsolateral prefrontal cortex (BA46) > secondary visual cortex (V2 or BA18/19) > primary visual cortex (V1 or BA17). To enrich in astrocytic nuclei, NEUN+ and OLIG2+ nuclei were separated via FACS, whereas NEUN-/OLIG2- nuclei (including astrocytes) were subjected to snRNA-seq resulting in a transcriptomic dataset of 629,755 astrocyte nuclei, which were subsequently examined for common and region-specific AD related changes. Clustering within each brain region identified likely homeostatic and reactive subclusters, as well as six other clusters which we describe as “intermediate” subclusters. Homeostatic and reactive astrocytes were most abundant in V1 (59%) and EC (21%), respectively, and their transcriptomic profiles were anticorrelated. Reactive astrocytes were enriched in cytoskeleton ( GFAP MAP2 , MAP7 , MAPB1 , MAPT ), extracellular matrix ( CD44 , LAMA1 , TNC , VCAN ), chaperones ( CRYAB , HSPB1 , HSPB8 ), and oxidative stress/antioxidant ( MAOB , MT1X , MT2A , SOD2 ) genes, whereas homeostatic subclusters in trophic factor ( EGFR , PTN ) and glutamate metabolism ( GLUL , GRIA2 , GRM3 , SLC1A2 ) genes. The proportion of intermediate astrocyte subclusters was lowest in EC (25%) and highest in BA46 (41%), and their transcriptome correlated weakly with that of homeostatic and reactive astrocytes, representing an apparent gradient between homeostatic and reactive. Notably, these intermediate subclusters demonstrated the most heterogeneity among brain regions. Our astrocyte snRNA-seq dataset encompassing five regions of control and AD brains revealed homeostatic, reactive, and previously uncharacterized intermediate astrocytic states with distinct transcriptomic profiles for each brain region, suggesting complex, region-specific responses to AD pathology.

Many drugs that target amyloid-β (Aβ) in Alzheimer disease (AD) have failed to demonstrate clinical efficacy. However, four anti-Aβ antibodies have been shown to mediate the removal of amyloid plaque from brains of patients with AD, and the FDA has recently granted accelerated approval to one of these, aducanumab, using reduction of amyloid plaque as a surrogate end point. The rationale for approval and the extent of the clinical benefit from these antibodies are under intense debate. With the aim of informing this debate, we review clinical trial data for drugs that target Aβ from the perspective of the temporal interplay between the two pathognomonic protein aggregates in AD - Aβ plaques and tau neurofibrillary tangles - and their relationship to cognitive impairment, highlighting differences in drug properties that could affect their clinical performance. On this basis, we propose that Aβ pathology drives tau pathology, that amyloid plaque would need to be reduced to a low level (~20 centiloids) to reveal significant clinical benefit and that there will be a lag between the removal of amyloid and the potential to observe a clinical benefit. We conclude that the speed of amyloid removal from the brain by a potential therapy will be important in demonstrating clinical benefit in the context of a clinical trial.

Tau pathology is known as a primary driver of neurodegeneration in Alzheimer’s disease (AD). Understanding its underlying molecular mechanism is critical in expanding our knowledge of AD pathogenesis and developing novel AD therapeutic strategies. However, interrogating tau induced neurotoxicity mechanisms has been difficult due to heterogeneous susceptibility of neurons to tau pathology. Here, we aim to identify the most vulnerable neuronal subpopulation to tau pathology in AD and reveal more clear molecular mechanisms of tau induced neuro-toxicity/degeneration by analyzing gene expression changes in the vulnerable population. We performed single nuclei RNA sequencing and tau biochemistry from same tissue blocks of the same AD patients (5 brain regions of 32 AD donors with various Braak stages). About 1.5x10 6 neurons were enriched in total by NeuN antibody-based flow cytometry sorting and these cells were clustered into 15 neuronal subpopulations based on their similarity in gene expression. We tested for association between relative neuronal population abundance and tau pathology readouts (phospho-T231 ELISA, HT7-HT7 SIMOA and HEK seeding). Based on the strength of correlation, we identified the neuronal subpopulation that reduces relative abundance in association with its tau pathology. We identified a tau vulnerable neuronal population that showed strong negative correlation between its relative abundance and tau pathology readouts in BA20 and BA46. The population was one of largest excitatory subpopulation distinguished by marker genes, CBLN2 and LINC00507. This outcome was supported by multiple published transcriptomics studies and histologically validated by multiplexed in situ hybridization and immunohistochemistry. Differential gene expression analysis of the vulnerable neuronal population identified a list of genes that potentially links tau pathology and neuronal death. This study with large number of captured neurons for single cell transcriptomics and quantitative tau pathology readout for direct comparison to transcriptomics data enabled the discovery of a vulnerable neuronal subpopulation in more precise manner and revealed genes related to tau associated neurotoxicity more clearly. This result serves as a great starting point to further interrogate fundamental mechanisms of tau-driven neurodegeneration in AD and accelerate therapeutic target and biomarker discovery.

Altered cerebral vasculature is recognized as a key feature of Alzheimer’s disease (AD) neuropathology, with current evidence suggesting a diverse collection of changes including aberrant angiogenesis, vascular pruning, inflammation, senescence, and other remodeling events. Previous studies examining the endothelial cell transcriptome in AD have been limited to 1-2 brain regions and hampered by low numbers of isolated endothelial cell nuclei, potentially missing relevant regional heterogeneity. We hypothesized that such regional heterogeneity in endothelial cell gene expression in the normal aging brain may contribute to the varied vascular responses to AD neuropathology. Nuclei were isolated from five brain areas in subjects with increasing AD pathology (n=32 donors). The five brain areas chosen represent the hierarchical accumulation of tau pathology in AD: entorhinal cortex > inferior temporal gyrus > prefrontal cortex > secondary visual cortex > primary visual cortex. All subjects had detailed p-tau, in vitro tau seeding, and amyloid-beta plaque measurements available. Nuclei were FACS sorted to exclude NeuN+ and Olig2+ cells, and the resulting neuron- and oligodendrocyte-depleted fraction was subject to single-nucleus RNAseq. Endothelial cell nuclei were identified based on high von Willebrand factor expression and filtered to exclude contaminating glia, yielding a dataset comprised of 51,586 total nuclei. Transcriptomic analysis confirmed that endothelial cells share a core set of commonly upregulated genes across the five brain regions compared to other cell types. However, there is also regional heterogeneity with 193-311 differentially upregulated genes expressed in endothelial cells within each brain area relative to all other areas. Compared to normal aging brain, endothelial cells from AD donors downregulated 962 genes and upregulated 936 genes that are enriched in vasculogenesis, blood-brain barrier maintenance, senescence, and leukocyte interactions. Further analysis indicates regional differences in upregulation of these pathways and highlights the relationship to disease burden. This dataset demonstrates a previously unrecognized regional heterogeneity in endothelial cell transcriptome in the normal aging human brain. These regional differences may impact the endothelial cell response to the local AD pathology. This dataset will inform ongoing research efforts to unravel the molecular underpinnings of AD-induced endothelial cell dysfunction.

Rare sequence variants in the microglial cell surface receptor TREM2 have been shown to increase the risk for Alzheimer’s disease (AD). Disease-linked TREM2 mutations seem to confer a partial loss of function, and increasing TREM2 cell surface expression and thereby its function(s) might have therapeutic benefit in AD. However, druggable targets that could modulate microglial TREM2 surface expression are not known. To identify such targets, we conducted a screen of small molecule compounds with known pharmacology using human myeloid cells, searching for those that enhance TREM2 protein at the cell surface. Inhibitors of the kinases MEK1/2 displayed the strongest and most consistent increases in cell surface TREM2 protein, identifying a previously unreported pathway for TREM2 regulation. Unexpectedly, inhibitors of the downstream effector ERK kinases did not have the same effect, suggesting that noncanonical MEK signaling regulates TREM2 trafficking. In addition, siRNA knockdown experiments confirmed that decreased MEK1 and MEK2 were required for this recruitment. In iPSC-derived microglia, MEK inhibition increased cell surface TREM2 only modestly, so various cytokines were used to alter iPSC microglia phenotype, making cells more sensitive to MEK inhibitor-induced TREM2 recruitment. Of those tested, only IFN-gamma priming prior to MEK inhibitor treatment resulted in greater TREM2 recruitment. These data identify the first known mechanisms for increasing surface TREM2 protein and TREM2-regulated function in human myeloid cells and are the first to show a role for MEK1/MEK2 signaling in TREM2 activity.

Alzheimer's disease (AD) is a common neurodegenerative disease with poor prognosis. New options for drug discovery targets are needed. We developed an imaging based arrayed CRISPR method to interrogate the human genome for modulation of in vitro correlates of AD features, and used this to assess 1525 human genes related to tau aggregation, autophagy and mitochondria. This work revealed (I) a network of tau aggregation modulators including the NF-κB pathway and inflammatory signaling, (II) a correlation between mitochondrial morphology, respiratory function and transcriptomics, (III) machine learning predicted novel roles of genes and pathways in autophagic processes and (IV) individual gene function inferences and interactions among biological processes via multi-feature clustering. These studies provide a platform to interrogate underexplored aspects of AD biology and offer several specific hypotheses for future drug discovery efforts.

The proliferation and activation of microglia, the resident macrophages in the brain, is a hallmark of many neurodegenerative diseases such as Alzheimer's disease (AD) and prion disease. Colony stimulating factor 1 receptor (CSF1R) is critically involved in regulating microglial proliferation, and CSF1R blocking strategies have been recently used to modulate microglia in neurodegenerative diseases. However, CSF1R is broadly expressed by many cell types and the impact of its inhibition on the innate immune system is still unclear. CSF1R can be activated by two independent ligands, CSF-1 and interleukin 34 (IL-34). Recently, it has been reported that microglia development and maintenance depend on IL-34 signaling. In this study, we evaluate the inhibition of IL-34 as a novel strategy to reduce microglial proliferation in the ME7 model of prion disease. Selective inhibition of IL-34 showed no effects on peripheral macrophage populations in healthy mice, avoiding the side effects observed after CSF1R inhibition on the systemic compartment. However, we observed a reduction in microglial proliferation after IL-34 inhibition in prion-diseased mice, indicating that microglia could be more specifically targeted by reducing IL-34. Overall, our results highlight the challenges of targeting the CSF1R/IL34 axis in the systemic and central compartments, important for framing any therapeutic effort to tackle microglia/macrophage numbers during brain disease.

Alzheimer's disease (AD) neuropathology is extremely heterogeneous, and the evolution from preclinical to mild cognitive impairment until dementia is driven by interacting genetic/biological mechanisms not fully captured by current clinical/research criteria. We characterized the heterogeneous "construct" of AD through a cerebrospinal fluid biomarker-guided stratification approach. We analyzed 5 validated pathophysiological cerebrospinal fluid biomarkers (A beta(1-42), t-tau, -p-tau(181), NFL, YKL-40) in 113 participants (healthy controls [N = 20], subjective memory complainers [N = 36], mild cognitive impairment [N = 20], and AD dementia [N = 37], age: 66.7 +/- 10.4, 70.4 +/- 7.7, 71.7 +/- 8.4, 76.2 +/- 3.5 years [mean +/- SD], respectively) using Density-Based Spatial Clustering of Applications with Noise, which does not require a priori determination of the number of clusters. We found 5 distinct clusters (sizes: N = 38, 16, 24, 14, and 21) whose composition was independent of phenotypical groups. Two clusters showed biomarker profiles linked to neurodegenerative processes not associated with classical AD-related pathophysiology. One cluster was characterized by the neuroinflammation biomarker YKL-40. Combining nonlinear data aggregation with informative biomarkers can generate novel patient strata which are representative of cellular/molecular pathophysiology and may aid in predicting disease evolution and mechanistic drug response. (C) 2019 The Authors. Published by Elsevier Inc.