This study investigates how��cerebrospinal fluid (CSF)—the fluid that surrounds and cushions the brain and spinal cord—moves within the brain, and how this movement is influenced by the body’s automatic (autonomic) functions, such as heart rate and breathing. CSF flow is important because it helps remove waste and maintain brain health. While previous research has linked CSF movement to sleep and brain activity, the researchers wanted to isolate the role of the��autonomic nervous system��(the system that controls involuntary processes like heartbeat and respiration).

To do this, they used fMRI scans to observe fluid-related signals in the brain while changing levels of carbon dioxide (CO₂) in participants’ blood—a method that affects blood vessel tone, breathing, and heart function without directly altering brain activity. They found that changes in CSF movement could not be explained simply by physical or mechanical factors. Instead, variations in��heart-rate variability��(natural fluctuations in the time between heartbeats) played a key role in driving slow CSF flow, independent of breathing. Additionally, changes in CO₂ levels mainly affected how frequently heart rate and breathing patterns fluctuated, rather than how strong those fluctuations were.

Overall, the findings suggest that CSF movement is strongly influenced by autonomic regulation, and that both higher and lower-than-normal CO₂ levels can disrupt this process. This highlights a new way to study and potentially control brain fluid dynamics—by adjusting CO₂ levels—without relying on sleep or direct neural activity, offering potential insights into brain health and disease.

��

Fig 1: The predictions of CSF flow dynamics across capnias is based on three different physiological pathways: vascular tone, sympathetic tone, and neuronal activity. According to the vascular-tone theory, CSF fluctuations should be maximal at normocapnia. According to the neuronal-activity theory, CSF fluctuations should be maximized at hypocapnia. Lastly, according to the sympathetic-tone theory, CSF fluctuations should be maximized at hypercapnia. These theories will be tested using empirical data involving different capnias, at which all three variables will be altered.

]]> /valiant/2026/03/26/neuroimaging-phewas-and-molecular-phenotyping-implicate-psmc3-in-alzheimers-disease/ Thu, 26 Mar 2026 18:43:37 +0000 /valiant/?p=6307 Xavier Bledsoe; Ting-Chen Wang; Yiyang Wu; Derek Archer; Hung Hsin Chen; Adam C. Naj; William S. Bush; Timothy J. Hohman; Logan Dumitrescu; Jennifer E. Below; Eric R. Gamazon (2026).��.��Alzheimer’s & Dementia, 22(2), e71217.��

This study looked at how genetic differences linked to Alzheimer’s disease (AD) may influence the brain, aiming to better understand how these genes actually lead to changes seen in patients. While previous research has identified many AD-related genes, it is still unclear how these genes affect brain structure and function. To explore this, the researchers used a functional genomics approach, meaning they examined how genetic variants influence gene activity (gene expression) and, in turn, brain features seen on imaging scans. They connected known AD genes to specific brain characteristics using a tool called the NeuroimaGene Atlas, and compared these predicted effects with real-world brain imaging data from patients. They also analyzed genetic covariance, which looks at how different traits share common genetic influences, to identify links between brain features and risk factors like family history of dementia.

The results suggest that a gene called PSMC3, which plays a role in breaking down unwanted or damaged proteins, may be important in the development of Alzheimer’s disease. Changes in AD-related genes were linked to differences in key brain areas involved in memory and thinking, such as the frontal cortex (important for decision-making and cognition), as well as changes in cerebrospinal fluid (the fluid surrounding the brain and spinal cord). The study also found shared genetic influences between Alzheimer’s risk and features of the hippocampus, a brain region critical for memory. Interestingly, higher activity of the PSMC3 gene was associated with better cognitive performance and lower levels of amyloid beta, a protein that builds up abnormally in Alzheimer’s disease. Overall, these findings help connect genetic risk factors to specific brain changes, offering a clearer picture of how Alzheimer’s disease develops and pointing to potential targets for future research and treatment.

FIGURE 1

Schematic overview of the analytical framework. A, Grid summarizing primary data resources integrated in the study. B, Directed acyclic graph illustrating TWAS analyses and downstream imputation of neuroimaging features via NeuroimaGene. C, Visualization of genetic covariance analyses comparing the genetic architecture of clinical AD and parental AD with neuroimaging-derived features. D, Logistic regression models evaluating associations between��neuroimaging features and parental AD status. E, Integration of clinical neuroimaging data linking brain features to AD status. F, Composite synthesis comparing the neuroimaging features obtained across transcriptomic, genetic covariance, parental history, and clinical approaches. AD, Alzheimer’s disease; Dx, diagnosis; TWAS, transcriptome-wide association study; UKBB, UK Biobank.

The brain undergoes major structural and functional changes from childhood through adolescence. Research suggests that neurodevelopment happens in a hierarchical way, meaning different brain regions and networks mature at different rates. However, less is known about how the brain’s intrinsic spatiotemporal propagations—patterns showing how activity spreads across the brain over time—develop during this period. This study examined how these activity patterns change from childhood to early adulthood.

Using a recently developed method that measures time-lagged dynamic propagations, the researchers analyzed how brain activity travels along three major axes of brain organization: the sensory–association (S-A) axis, which connects basic sensory regions to higher-order thinking areas; the task-positive to default network (TP-D) axis, which reflects shifts between attention-focused networks and the default mode network active during rest and internal thought; and the somatomotor–visual (SM-V) axis, which links movement and visual processing regions. The results showed that these propagation patterns gradually become more adult-like over development. As children mature, they spend more time in S-A and TP-D propagation states, while the occurrence of SM-V propagation states decreases.

Importantly, top-down propagations along the S-A axis—meaning activity flowing from higher-order cognitive regions to sensory regions—increased with age and were better predictors of cognitive performance than bottom-up propagations, which flow from sensory areas upward. These findings were replicated in two independent datasets, the Human Connectome Project Development cohort and the Nathan Kline Institute Rockland Sample, supporting the robustness and generalizability of the results. Overall, the study provides new insight into how large-scale functional brain dynamics develop during youth and how these changes support cognitive abilities.

Fig. 1: Spatiotemporal propagation patterns and their neurodevelopmental change from children to early adulthood.

A��The first three propagation patterns derived from the reference cohort (HCP-A), represent group-level reference propagation patterns. Each row displays a full propagation cycle for the recurring spatiotemporal patterns: sensorimotor to association (S-A), task-positive to default mode networks (TP-D), and somatomotor to visual networks (SM-V). The patterns are depicted through their temporal phase cycle, ranging from 0 to 2π.��B��Explained variance ratios of the first six propagation patterns from CPCA. The light blue line represents the youth cohort (HCP-D) and the dark line represents the reference adult cohort (HCP-A).��C��Between-cohort similarity matrix showing the pairwise Pearson’s correlation of the propagation patterns across youth (HCP-D) and adult (HCP-A) propagation patterns. We also confirmed cross-cohort similarity using HCP Young Adult cohort (N = 892, age 21-35, Figure.��)].��D��Reliability of propagation patterns, assessed by the discriminability for HCP-D and HCP-A cohorts.��E��Age-related similarity of propagation patterns to adult reference. Dots represent the spatial correlations of the propagation pattern between individuals in the youth cohort and the group-level adult reference. The regression line illustrates the developmental trend across age. Age effect was assessed using a Spearman correlation, with��p��values adjusted for multiple comparisons using the false-discovery-rate (FDR) correction. Significant age-related increases were observed for the S-A (pFDR <0.001), TP-D (pFDR <0.001) and SM-V (pFDR = 0.002) propagation patterns. Statistical significance is denoted by asterisks (*: pFDR <0.05).��F��Age prediction using the first three dynamic patterns. A combination of the first three dominant propagation patterns in the PLSR model predicts age with a Spearman’s correlation ρ of 0.80 and a mean absolute error (MAE) of 1.87 years.

In diffusion magnetic resonance imaging, or dMRI, an incomplete field of view (FOV) means that part of the brain is missing from the scan. This can seriously affect analyses of white matter connectivity, including tractography, which maps the pathways of nerve fiber bundles across the brain. Although previous studies have used deep generative models to estimate or “impute” the missing regions, it is still unclear how to best use additional information from paired multi-modality data, such as combining dMRI with structural T1-weighted (T1w) MRI, to improve the quality of reconstruction and support downstream analyses.

To address this, the researchers developed a new framework that imputes missing dMRI regions by integrating diffusion features from the acquired portion of the scan with information about the complete brain anatomical structure derived from paired imaging data. The idea is that using anatomical guidance from other modalities can improve how the missing diffusion signals are reconstructed. They tested the framework on two cohorts from different sites, including a total of 96 participants, and compared it with a baseline method that treated T1w and dMRI information equally without specifically leveraging their complementary roles.

The proposed framework significantly improved imputation quality, as measured by the angular correlation coefficient, and improved the accuracy of downstream tractography, as measured by the Dice score. These results suggest that carefully integrating paired multi-modality data leads to more accurate reconstruction of incomplete dMRI scans. By improving whole-brain tractography, this approach may reduce uncertainty in analyses of white matter bundles, particularly those relevant to neurodegenerative diseases.

Fig. 1.

Visualization (left) and histogram (right) of 103 real cases of dMRI scans with incomplete FOV that failed quality assurance. In the left figure, horizontal regions indicate the distribution of the incomplete part of FOV with an estimated position of a brain mask. The total cutoff distance from the incomplete FOV to the top of the brain is estimated using a corresponding and registered T1w image. Its histogram is presented in the right figure.

Multiple sclerosis (MS) causes lesions, or areas of damage, in the brain that can be seen on multicontrast magnetic resonance (MR) images. Automatically segmenting, or outlining, these lesions using deep learning (DL) can improve speed and consistency compared to manual tracing by experts. Although many DL methods perform well on data similar to what they were trained on, they often struggle when tested on new datasets from different hospitals or scanners, a problem known as poor out-of-domain generalization.

To address this issue, the researchers developed a new method called UNISELF. The goal of UNISELF is to achieve high segmentation accuracy within the original training domain while also performing well on data from different sources. UNISELF introduces a test-time self-ensembled lesion fusion strategy, which combines multiple predictions at test time to improve accuracy. It also uses test-time instance normalization (TTIN) of latent features, meaning it adjusts internal feature representations during testing to better handle domain shifts and missing input contrasts, such as when certain MR image types are unavailable.

The model was trained using data from the ISBI 2015 longitudinal MS segmentation challenge. On the official test dataset, UNISELF ranked among the top-performing methods. Importantly, when evaluated on out-of-domain datasets with different scanners, imaging protocols, and missing contrasts—including the MICCAI 2016 dataset, the UMCL dataset, and a private multisite dataset—UNISELF outperformed other benchmark models trained on the same ISBI data. These results suggest that UNISELF is both accurate and robust to real-world variations in MR imaging, making it a promising tool for automated MS lesion segmentation across diverse clinical settings.

Fig. 1.��An illustration of the spatial augmentation, network input, and network output during training in UNISELF.

This study followed cognitively normal adults over time to determine which magnetic resonance imaging (MRI) biomarkers best predict future cognitive impairment. Researchers examined 154 different MRI-based measurements in 509 participants from the Baltimore Longitudinal Study of Aging who were age 50 or older and cognitively normal at the start of the study. Participants underwent repeated cognitive testing and 3 Tesla (3T) MRI scans, including T1- and T2-weighted imaging to assess brain structure and diffusion tensor imaging (DTI) to measure white matter microstructural integrity. The analyses accounted for factors such as age and other confounders and also examined differences by sex and amyloid beta (Aβ) status, a biological marker associated with Alzheimer’s disease.

Over an average follow-up of 4.6 years, individuals who later developed cognitive impairment showed greater declines in white matter integrity compared to those who remained cognitively stable. These changes were especially pronounced in major white matter tracts, including the corpus callosum, cingulum bundle, and inferior fronto-occipital fasciculus, which are pathways that connect different brain regions. To a lesser extent, thinning and atrophy in the temporal lobe were also linked to later impairment. The associations between brain changes and future cognitive decline were stronger in men and in individuals who were amyloid-positive.

Overall, the findings suggest that early changes in white matter microstructure, as measured by DTI, are particularly sensitive indicators of future mild cognitive impairment (MCI) and dementia. Certain MRI metrics may therefore be especially useful for identifying risk in people who are still cognitively normal.

��

FIGURE 1

Study overview. Participants were selected from the BLSA neuroimaging substudy based on cognitively normal (CN) status and age 50 or older at baseline. The study data included longitudinal cognitive assessments, clinical diagnoses (Dx), 3T magnetic resonance imaging scans, and baseline plasma biomarkers related to Alzheimer’s disease and related dementias, specifically amyloid beta 42/40, collected between 2008 and 2019. The subsequently impaired (SI) group (also CN at baseline) included individuals who later developed mild cognitive impairment (MCI) or dementia or were “Impaired, not MCI/dementia.” Impairment onset dates ranged from 2012 to 2019 (≈1- to 9-year interval).

Disruptions in the brain’s fluid systems may contribute to the nerve cell damage seen in Huntington disease (HD). These fluid pathways are important for clearing waste, controlling inflammation, and distributing treatments delivered to the brain. In this study, we examined two key components of this system: the choroid plexus (ChP), which produces cerebrospinal fluid (CSF), and the parasagittal dural (PSD) space, a major CSF drainage pathway. We measured their size and function using advanced MRI techniques and analyzed how these measures relate to disease severity and genetic risk.

We studied 80 people with HD and 65 healthy controls. Compared with controls, people with HD had larger ChP and PSD structures and reduced blood flow through the ChP. Individuals with greater genetic risk (larger CAG repeat expansions) showed the largest changes, and these differences were linked to worse motor symptoms.

These results show that HD affects both the structure and function of key brain fluid pathways. Understanding these changes may help explain disease mechanisms and improve the delivery of treatments that rely on cerebrospinal fluid, highlighting the need for further research.

FIGURE 1

Cerebrospinal fluid (CSF) production and flow shown (A) schematically and on (B) axial, (C) sagittal, and (D) coronal��T₂-weighted MRI: (A) The bulk CSF flow pathway consists of (B) CSF production in the choroid plexus (ChP) complexes (yellow arrows), (C) flow through the cerebral aqueduct (orange arrow) at a typical peak velocity of 5–15 cm/s, outflow from the ventricular system via the foramina of Luschka and Magendie, circulation of the cerebrum and subarachnoid space, and uptake into the venous system via arachnoid granulations. Recent reports also suggest that CSF egress can occur (D) along conduits within the peri-sinus parasagittal dural space (green arrows).

New methods can estimate the age at which amyloid buildup first begins in the brain, known as estimated amyloid onset age or EAOA, using amyloid positron emission tomography, a brain imaging technique that detects amyloid plaques linked to Alzheimer’s disease. This study examined the genetic factors that influence EAOA to better understand the earliest biological changes in Alzheimer’s disease. Using harmonized amyloid PET data from 4,216 participants, researchers performed genome-wide survival analyses, tissue-specific gene expression studies, and genetic covariance analyses. They found that genetic variants in apolipoprotein E, or APOE, as well as ABCA7 and RASGEF1C, were linked to earlier amyloid onset. Individuals with the APOE ε4 ε4 and ε3 ε4 genotypes developed amyloid buildup about 6.3 and 5 years earlier than those with the ε3 ε3 genotype, while the ε2 variant appeared protective against early onset. A specific genetic variant called rs4147929, which affects how much ABCA7 is expressed in the brain, was associated with amyloid onset occurring about 4 years earlier and with lower ABCA7 expression, which in turn was linked to greater amyloid pathology seen at autopsy. The study also found shared genetic risk between earlier amyloid onset and several immune-related diseases. Together, these findings highlight APOE, ABCA7, and RASGEF1C as important genetic contributors to early amyloid accumulation and suggest potential biological targets for intervention at the earliest stages of Alzheimer’s disease.

FIGURE 1

A, Genome-wide survival analysis without��APOE��covariate. The��APOE��loci was the only significant loci (top SNV: rs429538; HR ��=�� 3.17,��p ��=�� 6.56��×��10⁻¹⁷⁵). A priori significance (5��×��10⁻⁸) is indicated by the red line. B, Genome-wide survival analysis with��APOE��ε2 and ε4 status included as covariates. Three loci passed significance rs4147929 (HR ��=�� 1.31,��p��=��2.87��×��10⁻⁸, minor allele��=��A, major allele��=��G, BP��=��1063444), rs3752246 (HR ��=�� 1.31,��p��=��3.76��×��10⁻⁸, minor allele��=��C, major allele��=��G, BP��=��1056493), and rs374637031 (HR��=��1.57,��p��=��4.95��×��10⁻⁸, minor allele��=��A, major allele��=��G, BP��=��106714314). rs4147929 is intronic in��ABCA7��on chromosome 19 while rs3752246 is a missense variant in��ABCA7. rs374637031 was not found in any SNV databases. A priori significance (5��×��10⁻⁸) is indicated by the red line.��APOE, apolipoprotein E; BP, base pair; HR, hazard ratio; SNV, single nucleotide variant.

As people age, having both thinking problems and movement problems at the same time—a pattern called a dual cognitive–motor deficit—is known to strongly increase the risk of developing dementia. However, it has not been clear how this combined deficit affects the brain’s structure, especially in vulnerable gray matter regions that are important for memory and movement. This study set out to better understand these brain changes.

The researchers studied 582 adults between the ages of 36 and 90 and grouped them into four categories: those with both cognitive and motor deficits, those with only cognitive deficits, those with only motor deficits, and a control group with neither. They examined brain tissue using advanced MRI techniques, including diffusion tensor imaging and mean apparent propagator imaging, which are well suited for detecting subtle microstructural changes in gray matter and white matter. In total, they analyzed 27 brain regions related to temporal (memory-related) and motor functions, as well as key white matter pathways.

The results showed that people with a dual cognitive–motor deficit had widespread microstructural changes in the brain. These alterations were not seen in individuals who had only cognitive deficits or only motor deficits once rigorous statistical corrections were applied. The observed changes are thought to reflect lower cellular density in temporal gray matter, reduced organization of nerve fibers, and possible loss of myelin in white matter tracts.

Together, these findings suggest that having combined cognitive and motor difficulties is linked to distinct and measurable changes in brain microstructure. Understanding these changes may help explain why this group is at particularly high risk for dementia and could support the development of earlier interventions aimed at slowing brain aging and delaying neurodegeneration.

FIGURE 1

Investigated regions of interest. Three-dimensional rendering of (A) temporal meta-ROIs and motor-related GM regions, and (B) associated WM tracts. A total of 27 ROIs were investigated in the current study. GM, gray matter; ROIs, regions of interest; WM, white matter.

Risk-taking behavior is a feature of many neuropsychiatric disorders, yet effective treatments are limited because we still don’t fully understand what is happening in the brain when people make risky choices. Scientists know that certain “reward circuitry” regions—such as the amygdala, orbitofrontal cortex, insula, and anterior cingulate—are involved, but the specific electrical activity that predicts risk-taking in these areas has not been well studied in humans. Identifying local field potential (LFP) frequency patterns linked to risk-taking could help guide future therapies.

In this study, eleven patients with hard-to-treat epilepsy, who already had stereotactic EEG electrodes implanted for medical reasons, took part in an experiment measuring brain activity in these reward-related regions. Each person played a simple gambling game in which they guessed whether a hidden playing card would be lower or higher than a visible one, choosing to bet either $5 or $20. While they made these decisions, researchers recorded their local field potentials—electrical signals generated by groups of neurons. The team used statistical models to look for specific changes in oscillatory power (brainwave activity across different frequencies) related to reward prediction error, which is the difference between expected and actual outcomes. They also calculated a “risk-taking value” for each trial based on the card number and the size of the bet, and identified which oscillatory patterns were linked to riskier choices.

The results showed clear time-frequency patterns associated with reward prediction error signals in both the amygdala and the orbitofrontal cortex, with several strong clusters of activity in each region. Risky choices themselves were predicted by increased oscillatory power in the theta-to-beta frequency range in the orbitofrontal cortex during card presentation, and by increased high-beta power in the insula. Further analysis pinpointed these signals to the lateral orbitofrontal cortex and the posterior insula. Activity in one insula cluster linked to risky decisions was also connected to a theta-alpha reward prediction error signal in the orbitofrontal cortex. Additionally, an amygdala reward prediction error signal was associated with how often participants chose the higher bet, and a lateral orbitofrontal cortex signal predicted high bets specifically in risky situations.

Overall, the study identifies distinct electrical activity patterns in key reward-related brain regions that predict when a person is about to make a risky decision. These oscillatory signatures could eventually serve as biomarkers—measurable indicators that help guide new treatments, including closed-loop neuromodulation, for disorders in which risk-taking becomes harmful.

Figure 1

Gambling task and behavioural data. (A) Behavioural task events and timing between events in seconds (mean ± standard deviation). (B) Mean response time (time from bet cue presentation to patient response) for each patient card number. Error bars represent standard error. (C) Mean per cent of trials on which subjects bet high ($20) for each patient card number. Asterisk indicates��P��< 0.05. Error bars represent standard error.