Constructing personalized characterizations of structural brain aberrations in patients with dementia using explainable artificial intelligence.

Deep learning approaches for clinical predictions based on magnetic resonance imaging data have shown great promise as a translational technology for diagnosis and prognosis in neurological disorders, but its clinical impact has been limited. This is partially attributed to the opaqueness of deep learning models, causing insufficient understanding of what underlies their decisions. To overcome this, we trained convolutional neural networks on structural brain scans to differentiate dementia patients from healthy controls, and applied layerwise relevance propagation to procure individual-level explanations of the model predictions. Through extensive validations we demonstrate that deviations recognized by the model corroborate existing knowledge of structural brain aberrations in dementia. By employing the explainable dementia classifier in a longitudinal dataset of patients with mild cognitive impairment, we show that the spatially rich explanations complement the model prediction when forecasting transition to dementia and help characterize the biological manifestation of disease in the individual brain. Overall, our work exemplifies the clinical potential of explainable artificial intelligence in precision medicine.

Fetal influence on the human brain through the lifespan.

Human fetal development has been associated with brain health at later stages. It is unknown whether growth in utero, as indexed by birth weight (BW), relates consistently to lifespan brain characteristics and changes, and to what extent these influences are of a genetic or environmental nature. Here we show remarkably stable and lifelong positive associations between BW and cortical surface area and volume across and within developmental, aging and lifespan longitudinal samples (N = 5794, 4-82 y of age, w/386 monozygotic twins, followed for up to 8.3 y w/12,088 brain MRIs). In contrast, no consistent effect of BW on brain changes was observed. Partly environmental effects were indicated by analysis of twin BW discordance. In conclusion, the influence of prenatal growth on cortical topography is stable and reliable through the lifespan. This early-life factor appears to influence the brain by association of brain reserve, rather than brain maintenance. Thus, fetal influences appear omnipresent in the spacetime of the human brain throughout the human lifespan. Optimizing fetal growth may increase brain reserve for life, also in aging.

Subtypes of brain change in aging and their associations with cognition and Alzheimer's disease biomarkers.

Structural brain changes underly cognitive changes in older age and contribute to inter-individual variability in cognition. Here, we assessed how changes in cortical thickness, surface area, and subcortical volume, are related to cognitive change in cognitively unimpaired older adults using structural magnetic resonance imaging (MRI) data-driven clustering. Specifically, we tested (1) which brain structural changes over time predict cognitive change in older age (2) whether these are associated with core cerebrospinal fluid (CSF) Alzheimer's disease (AD) biomarkers phosphorylated tau (p-tau) and amyloid-ß (Aß42), and (3) the degree of overlap between clusters derived from different structural features. In total 1899 cognitively healthy older adults (50 - 93 years) were followed up to 16 years with neuropsychological and structural MRI assessments, a subsample of which (n = 612) had CSF p-tau and Aß42 measurements. We applied Monte-Carlo Reference-based Consensus clustering to identify subgroups of older adults based on structural brain change patterns over time. Four clusters for each brain feature were identified, representing the degree of longitudinal brain decline. Each brain feature provided a unique contribution to brain aging as clusters were largely independent across modalities. Cognitive change and baseline cognition were best predicted by cortical area change, whereas higher levels of p-tau and Aß42 were associated with changes in subcortical volume. These results provide insights into the link between changes in brain morphology and cognition, which may translate to a better understanding of different aging trajectories.

No phenotypic or genotypic evidence for a link between sleep duration and brain atrophy.

Short sleep is held to cause poorer brain health, but is short sleep associated with higher rates of brain structural decline? Analysing 8,153 longitudinal MRIs from 3,893 healthy adults, we found no evidence for an association between sleep duration and brain atrophy. In contrast, cross-sectional analyses (51,295 observations) showed inverse U-shaped relationships, where a duration of 6.5 (95% confidence interval, (5.7, 7.3)) hours was associated with the thickest cortex and largest volumes relative to intracranial volume. This fits converging evidence from research on mortality, health and cognition that points to roughly seven hours being associated with good health. Genome-wide association analyses suggested that genes associated with longer sleep for below-average sleepers were linked to shorter sleep for above-average sleepers. Mendelian randomization did not yield evidence for causal impacts of sleep on brain structure. The combined results challenge the notion that habitual short sleep causes brain atrophy, suggesting that normal brains promote adequate sleep duration-which is shorter than current recommendations.

Hippocampal-cortical functional connectivity during memory encoding and retrieval.

Memory encoding and retrieval are critical sub-processes of episodic memory. While the hippocampus is involved in both, less is known about its connectivity with the neocortex during memory processing in humans. This is partially due to variations in demands in common memory tasks, which inevitably recruit cognitive processes other than episodic memory. Conjunctive analysis of data from different tasks with the same core elements of encoding and retrieval can reduce the intrusion of patterns related to subsidiary perceptual and cognitive processing. Leveraging data from two large-scale functional resonance imaging studies with different episodic memory tasks (514 and 237 participants), we identified hippocampal-cortical networks active during memory tasks. Whole-brain functional connectivity maps were similar during resting state, encoding, and retrieval. Anterior and posterior hippocampus had distinct connectivity profiles, which were also stable across resting state and memory tasks. When contrasting encoding and retrieval connectivity, conjunctive encoding-related connectivity was sparse. During retrieval hippocampal connectivity was increased with areas known to be active during recollection, including medial prefrontal, inferior parietal, and parahippocampal cortices. This indicates that the stable functional connectivity of the hippocampus along its longitudinal axis is superposed by increased functional connectivity with the recollection network during retrieval, while auxiliary encoding connectivity likely reflects contextual factors.

Matrix metalloproteinases are associated with brain atrophy in cognitively unimpaired individuals.

Matrix metalloproteinases (MMPs) and their tissue inhibitors (TIMPs) have been linked to age-related neurodegeneration and Alzheimer's disease (AD), but their role in normal aging is poorly understood. We used linear mixed models to determine if baseline or rate of yearly change in cerebrospinal fluid (CSF) levels of MMP-2; MMP-3; MMP-10; TIMP-123 (composite of TIMP-1, TIMP-2, and TIMP-3); or TIMP-4 predicted changes in bilateral entorhinal cortex thickness, hippocampal volume, or lateral ventricle volume in cognitively unimpaired individuals. We also assessed effects on the CSF AD biomarkers amyloid-ß42 and phosphorylated tau181. Low baseline levels of MMP-3 predicted larger ventricle volumes and more entorhinal cortex thinning. Increased CSF MMP-2 levels over time predicted more entorhinal thinning, hippocampal atrophy, and ventricular expansion, while increased TIMP-123 over time predicted ventricular expansion. No MMP/TIMPs predicted changes in CSF AD biomarkers. Notably, we show for the first time that longitudinal increases in MMP-2 and TIMP-123 levels may predict age-associated brain atrophy. In conclusion, MMPs and TIMPs may play a role in brain atrophy in cognitively unimpaired aging.

Associations of neuroinflammatory IL-6 and IL-8 with brain atrophy, memory decline, and core AD biomarkers - in cognitively unimpaired older adults.

Concentrations of pro-inflammatory cytokines -interleukin-6 (IL-6) and interleukin-8 (IL-8) - are increased with age and in Alzheimer's disease (AD). It is not clear whether concentrations of IL-6 and IL-8 in the central nervous system predict later brain and cognitive changes over time nor whether this relationship is mediated by core AD biomarkers. Here, 219 cognitively healthy older adults (62-91 years), with baseline cerebrospinal fluid (CSF) measures of IL-6 and IL-8 were followed over time - up to 9 years - with assessments that included cognitive function, structural magnetic resonance imaging, and CSF measurements of phosphorylated tau (p-tau) and amyloid-ß (Aß-42) concentrations (for a subsample). Higher baseline CSF IL-8 was associated with better memory performance over time in the context of lower levels of CSF p-tau and p-tau/Aß-42 ratio. Higher CSF IL-6 was related to less CSF p-tau changes over time. The results are in line with the hypothesis suggesting that an up-regulation of IL-6 and IL-8 in the brain may play a neuroprotective role in cognitively healthy older adults with lower load of AD pathology.

Tracing the development and lifespan change of population-level structural asymmetry in the cerebral cortex.

Cortical asymmetry is a ubiquitous feature of brain organization that is subtly altered in some neurodevelopmental disorders, yet we lack knowledge of how its development proceeds across life in health. Achieving consensus on the precise cortical asymmetries in humans is necessary to uncover the developmental timing of asymmetry and the extent to which it arises through genetic and later influences in childhood. Here, we delineate population-level asymmetry in cortical thickness and surface area vertex-wise in seven datasets and chart asymmetry trajectories longitudinally across life (4-89 years; observations = 3937; 70% longitudinal). We find replicable asymmetry interrelationships, heritability maps, and test asymmetry associations in large-scale data. Cortical asymmetry was robust across datasets. Whereas areal asymmetry is predominantly stable across life, thickness asymmetry grows in childhood and peaks in early adulthood. Areal asymmetry is low-moderately heritable (max h2SNP ~19%) and correlates phenotypically and genetically in specific regions, indicating coordinated development of asymmetries partly through genes. In contrast, thickness asymmetry is globally interrelated across the cortex in a pattern suggesting highly left-lateralized individuals tend towards left-lateralization also in population-level right-asymmetric regions (and vice versa), and exhibits low or absent heritability. We find less areal asymmetry in the most consistently lateralized region in humans associates with subtly lower cognitive ability, and confirm small handedness and sex effects. Results suggest areal asymmetry is developmentally stable and arises early in life through genetic but mainly subject-specific stochastic effects, whereas childhood developmental growth shapes thickness asymmetry and may lead to directional variability of global thickness lateralization in the population.

Sex differences in multilayer functional network topology over the course of aging in 37543 UK Biobank participants.

Aging is a major risk factor for cardiovascular and neurodegenerative disorders, with considerable societal and economic implications. Healthy aging is accompanied by changes in functional connectivity between and within resting-state functional networks, which have been associated with cognitive decline. However, there is no consensus on the impact of sex on these age-related functional trajectories. Here, we show that multilayer measures provide crucial information on the interaction between sex and age on network topology, allowing for better assessment of cognitive, structural, and cardiovascular risk factors that have been shown to differ between men and women, as well as providing additional insights into the genetic influences on changes in functional connectivity that occur during aging. In a large cross-sectional sample of 37,543 individuals from the UK Biobank cohort, we demonstrate that such multilayer measures that capture the relationship between positive and negative connections are more sensitive to sex-related changes in the whole-brain connectivity patterns and their topological architecture throughout aging, when compared to standard connectivity and topological measures. Our findings indicate that multilayer measures contain previously unknown information on the relationship between sex and age, which opens up new avenues for research into functional brain connectivity in aging.

Is Short Sleep Bad for the Brain? Brain Structure and Cognitive Function in Short Sleepers.

Many sleep less than recommended without experiencing daytime sleepiness. According to prevailing views, short sleep increases risk of lower brain health and cognitive function. Chronic mild sleep deprivation could cause undetected sleep debt, negatively affecting cognitive function and brain health. However, it is possible that some have less sleep need and are more resistant to negative effects of sleep loss. We investigated this using a cross-sectional and longitudinal sample of 47,029 participants of both sexes (20-89 years) from the Lifebrain consortium, Human Connectome project (HCP) and UK Biobank (UKB), with measures of self-reported sleep, including 51,295 MRIs of the brain and cognitive tests. A total of 740 participants who reported to sleep <6 h did not experience daytime sleepiness or sleep problems/disturbances interfering with falling or staying asleep. These short sleepers showed significantly larger regional brain volumes than both short sleepers with daytime sleepiness and sleep problems (n = 1742) and participants sleeping the recommended 7-8 h (n = 3886). However, both groups of short sleepers showed slightly lower general cognitive function (GCA), 0.16 and 0.19 SDs, respectively. Analyses using accelerometer-estimated sleep duration confirmed the findings, and the associations remained after controlling for body mass index, depression symptoms, income, and education. The results suggest that some people can cope with less sleep without obvious negative associations with brain morphometry and that sleepiness and sleep problems may be more related to brain structural differences than duration. However, the slightly lower performance on tests of general cognitive abilities warrants closer examination in natural settings.SIGNIFICANCE STATEMENT Short habitual sleep is prevalent, with unknown consequences for brain health and cognitive performance. Here, we show that daytime sleepiness and sleep problems are more strongly related to regional brain volumes than sleep duration. However, participants sleeping ≤6 h had slightly lower scores on tests of general cognitive function (GCA). This indicates that sleep need is individual and that sleep duration per se is very weakly if at all related brain health, while daytime sleepiness and sleep problems may show somewhat stronger associations. The association between habitual short sleep and lower scores on tests of general cognitive abilities must be further scrutinized in natural settings.

Genetic architecture of brain age and its causal relations with brain and mental disorders.

The difference between chronological age and the apparent age of the brain estimated from brain imaging data-the brain age gap (BAG)-is widely considered a general indicator of brain health. Converging evidence supports that BAG is sensitive to an array of genetic and nongenetic traits and diseases, yet few studies have examined the genetic architecture and its corresponding causal relationships with common brain disorders. Here, we estimate BAG using state-of-the-art neural networks trained on brain scans from 53,542 individuals (age range 3-95 years). A genome-wide association analysis across 28,104 individuals (40-84 years) from the UK Biobank revealed eight independent genomic regions significantly associated with BAG (p < 5 × 10-8) implicating neurological, metabolic, and immunological pathways - among which seven are novel. No significant genetic correlations or causal relationships with BAG were found for Parkinson's disease, major depressive disorder, or schizophrenia, but two-sample Mendelian randomization indicated a causal influence of AD (p = 7.9 × 10-4) and bipolar disorder (p = 1.35 × 10-2) on BAG. These results emphasize the polygenic architecture of brain age and provide insights into the causal relationship between selected neurological and neuropsychiatric disorders and BAG.

Sustained upregulation of widespread hippocampal-neocortical coupling following memory encoding.

Systems consolidation of new experiences into lasting episodic memories involves hippocampal-neocortical interactions. Evidence of this process is already observed during early post-encoding rest periods, both as increased hippocampal coupling with task-relevant perceptual regions and reactivation of stimulus-specific patterns following intensive encoding tasks. We investigate the spatial and temporal characteristics of these hippocampally anchored post-encoding neocortical modulations. Eighty-nine adults participated in an experiment consisting of interleaved memory task- and resting-state periods. We observed increased post-encoding functional connectivity between hippocampus and individually localized neocortical regions responsive to stimuli encountered during memory encoding. Post-encoding modulations were manifested as a nearly system-wide upregulation in hippocampal coupling with all major functional networks. The configuration of these extensive modulations resembled hippocampal-neocortical interaction patterns estimated from active encoding operations, suggesting hippocampal post-encoding involvement exceeds perceptual aspects. Reinstatement of encoding patterns was not observed in resting-state scans collected 12 h later, nor when using other candidate seed regions. The similarity in hippocampal functional coupling between online memory encoding and offline post-encoding rest suggests reactivation in humans involves a spectrum of cognitive processes engaged during the experience of an event. There were no age effects, suggesting that upregulation of hippocampal-neocortical connectivity represents a general phenomenon seen across the adult lifespan.

Brain aging differs with cognitive ability regardless of education.

Higher general cognitive ability (GCA) is associated with lower risk of neurodegenerative disorders, but neural mechanisms are unknown. GCA could be associated with more cortical tissue, from young age, i.e. brain reserve, or less cortical atrophy in adulthood, i.e. brain maintenance. Controlling for education, we investigated the relative association of GCA with reserve and maintenance of cortical volume, -area and -thickness through the adult lifespan, using multiple longitudinal cognitively healthy brain imaging cohorts (n = 3327, 7002 MRI scans, baseline age 20-88 years, followed-up for up to 11 years). There were widespread positive relationships between GCA and cortical characteristics (level-level associations). In select regions, higher baseline GCA was associated with less atrophy over time (level-change associations). Relationships remained when controlling for polygenic scores for both GCA and education. Our findings suggest that higher GCA is associated with cortical volumes by both brain reserve and -maintenance mechanisms through the adult lifespan.

Relationship between cerebrospinal fluid neurodegeneration biomarkers and temporal brain atrophy in cognitively healthy older adults.

It is unclear whether cerebrospinal fluid (CSF) biomarkers of neurodegeneration predict brain atrophy in cognitively healthy older adults, whether these associations can be explained by phosphorylated tau181 (p-tau) and the 42 amino acid form of amyloid-ß (Aß42) biomarkers, and which neural substrates may drive these associations. We addressed these questions in 2 samples of cognitively healthy older adults who underwent longitudinal structural MRI up to 7 years and had baseline CSF levels of heart-type fatty-acid binding protein (FABP3)=, total-tau, neurogranin, and neurofilament light (NFL) (n = 189, scans = 721). The results showed that NFL, total-tau, and FABP3 predicted entorhinal thinning and hippocampal atrophy. Brain atrophy was not moderated by Aß42 and the associations between NFL and FABP3 with brain atrophy were independent of p-tau. The spatial pattern of cortical atrophy associated with the biomarkers overlapped with neurogenetic profiles associated with expression in the axonal (total-tau, NFL) and dendritic (neurogranin) components. CSF biomarkers of neurodegeneration are useful for predicting specific features of brain atrophy in older adults, independently of amyloid and tau pathology biomarkers.

Deep neural networks learn general and clinically relevant representations of the ageing brain.

The discrepancy between chronological age and the apparent age of the brain based on neuroimaging data - the brain age delta - has emerged as a reliable marker of brain health. With an increasing wealth of data, approaches to tackle heterogeneity in data acquisition are vital. To this end, we compiled raw structural magnetic resonance images into one of the largest and most diverse datasets assembled (n=53542), and trained convolutional neural networks (CNNs) to predict age. We achieved state-of-the-art performance on unseen data from unknown scanners (n=2553), and showed that higher brain age delta is associated with diabetes, alcohol intake and smoking. Using transfer learning, the intermediate representations learned by our model complemented and partly outperformed brain age delta in predicting common brain disorders. Our work shows we can achieve generalizable and biologically plausible brain age predictions using CNNs trained on heterogeneous datasets, and transfer them to clinical use cases.

Whole-brain connectivity during encoding: age-related differences and associations with cognitive and brain structural decline.

There is a limited understanding of age differences in functional connectivity during memory encoding. In the present study, a sample of cognitively healthy adult participants (n = 488, 18-81 years), a subsample of whom had longitudinal cognitive and brain structural data spanning on average 8 years back, underwent functional magnetic resonance imaging while performing an associative memory encoding task. We investigated (1) age-related differences in whole-brain connectivity during memory encoding; (2) whether encoding connectivity patterns overlapped with the activity signatures of specific cognitive processes, and (3) whether connectivity associated with memory encoding related to longitudinal brain structural and cognitive changes. Age was associated with lower intranetwork connectivity among cortical networks and higher internetwork connectivity between networks supporting higher level cognitive functions and unimodal and attentional areas during encoding. Task-connectivity between mediotemporal and posterior parietal regions-which overlapped with areas involved in mental imagery-was related to better memory performance only in older age. The connectivity patterns supporting memory performance in older age reflected preservation of thickness of the medial temporal cortex. The results are more in accordance with a maintenance rather than a compensation account.

Reduced Hippocampal-Striatal Interactions during Formation of Durable Episodic Memories in Aging.

Encoding of durable episodic memories requires cross-talk between the hippocampus and multiple brain regions. Changes in these hippocampal interactions could contribute to age-related declines in the ability to form memories that can be retrieved after extended time intervals. Here we tested whether hippocampal-neocortical- and subcortical functional connectivity (FC) observed during encoding of durable episodic memories differed between younger and older adults. About 48 younger (20-38 years; 25 females) and 43 older (60-80 years; 25 females) adults were scanned with fMRI while performing an associative memory encoding task. Source memory was tested ~20 min and ~6 days postencoding. Associations recalled after 20 min but later forgotten were classified as transient, whereas memories retained after long delays were classified as durable. Results demonstrated that older adults showed a reduced ability to form durable memories and reduced hippocampal-caudate FC during encoding of durable memories. There was also a positive relationship between hippocampal-caudate FC and higher memory performance among the older adults. No reliable age group differences in durable memory-encoding activity or hippocampal-neocortical connectivity were observed. These results support the classic theory of striatal alterations as one cause of cognitive decline in aging and highlight that age-related changes in episodic memory extend beyond hippocampal-neocortical connections.

Individual variations in 'brain age' relate to early-life factors more than to longitudinal brain change.

Brain age is a widely used index for quantifying individuals' brain health as deviation from a normative brain aging trajectory. Higher-than-expected brain age is thought partially to reflect above-average rate of brain aging. Here, we explicitly tested this assumption in two independent large test datasets (UK Biobank [main] and Lifebrain [replication]; longitudinal observations ≈ 2750 and 4200) by assessing the relationship between cross-sectional and longitudinal estimates of brain age. Brain age models were estimated in two different training datasets (n ≈ 38,000 [main] and 1800 individuals [replication]) based on brain structural features. The results showed no association between cross-sectional brain age and the rate of brain change measured longitudinally. Rather, brain age in adulthood was associated with the congenital factors of birth weight and polygenic scores of brain age, assumed to reflect a constant, lifelong influence on brain structure from early life. The results call for nuanced interpretations of cross-sectional indices of the aging brain and question their validity as markers of ongoing within-person changes of the aging brain. Longitudinal imaging data should be preferred whenever the goal is to understand individual change trajectories of brain and cognition in aging.

Scientists who study the brain and aging are keen to find an effective way to measure brain health, which could help identify people at risk for dementia or memory problems. One popular marker is 'brain age'. This measurement uses a brain scan to estimate a person's chronological age, then compares the estimated brain age to the person's actual age to determine whether their brain is aging faster or slower than expected for their age. However, since brain age relies on one brain scan taken at one point in time, it is not clear whether it really measures brain aging or if it might capture brain differences that have been present throughout the individual's life. Studies comparing individual brain scans over several years would be necessary to know for sure. Now, Vidal-Piñeiro et al. show that the brain-age measurement does not reflect faster brain aging. In the experiments, the researchers compared repeated brain scans of thousands of individuals over 40 years of age. The experiments showed that deviations from normative brain age detected in a single scan reflected early life differences more than changes in the brain over time. For example, people with older-looking brains were more likely to have had a low birth weight or to have a combination of genes associated with having an older looking brain. Vidal-Piñeiro et al. show that brain age mostly reflects a pre-existing brain condition rather than brain aging. The experiments also suggest that genetics and early brain development likely have a strong impact on brain health throughout life. Future studies trying to test or develop brain-aging measurements should use serial measurements to track brain changes over time.