Age-related changes in brain structure result from a complex interplay among various neurobiological processes, which may contribute to more complex trajectories than what can be described by simple linear or quadratic models. We used a nonparametric smoothing spline approach to delineate cross-sectionally estimated age trajectories of the volume of 17 neuroanatomic structures in 1100 healthy adults (18-94 years). Accelerated estimated decline in advanced age characterized some structures, for example hippocampus, but was not the norm. For most areas, 1 or 2 critical ages were identified, characterized by changes in the estimated rate of change. One-year follow-up data from 142 healthy older adults (60-91 years) confirmed the existence of estimated change from the cross-sectional analyses for all areas except 1 (caudate). The cross-sectional and the longitudinal analyses agreed well on the rank order of age effects on specific brain structures (Spearman ρ = 0.91). The main conclusions are that most brain structures do not follow a simple path throughout adult life and that accelerated decline in high age is not the norm of healthy brain aging.

Subjective memory impairment (SMI) is a common risk factor for Alzheimer's disease, with few established options for treatment. Here we investigate the effects of two months episodic memory training on regional brain atrophy in 19 memory clinic patients with SMI. We used a sensitive longitudinal magnetic resonance imaging protocol and compared the patients with 42 matched healthy volunteers randomly assigned to a group performing the same training, or a no-training control group. Following intervention, the SMI sample exhibited structural gray matter volume increases in brain regions encompassing the episodic memory network, with cortical volume expansion of comparable extent as healthy training participants. Further, we found significant hippocampal volume increases in the healthy training group but not in the SMI group. Still, individual differences in left hippocampal volume change in the patient group were related to verbal recall improvement following training. The present results reinforce earlier studies indicating intact brain plasticity in aging, and further suggest that training-related brain changes can be evident also in the earliest form of cognitive impairment.

Objective

HIV infection and aging are both associated with neurodegeneration. However, whether the aging process alone or other factors associated with advanced age account for the progression of neurodegeneration in the aging HIV-positive (HIV+) population remains unclear.

Methods

HIV+ (n = 70) and HIV-negative (HIV-, n = 34) participants underwent diffusion tensor imaging (DTI) and metrics of microstructural properties were extracted from regions of interest (ROIs). A support vector regression model was trained on two independent datasets of healthy adults across the adult life-span (n = 765, Cam-CAN = 588; UiO = 177) to predict participant age from DTI metrics, and applied to the HIV dataset. Predicted brain age gap (BAG) was computed as the difference between predicted age and chronological age, and statistically compared between HIV groups. Regressions assessed the relationship between BAG and HIV severity/medical comorbidities. Finally, correlation analyses tested for associations between BAG and cognitive performance.

Results

BAG was significantly higher in the HIV+ group than the HIV- group F (1, 103) = 12.408, p = .001). HIV RNA viral load was significantly associated with BAG, particularly in older HIV+ individuals (R²  = 0.29, F(7, 70) = 2.66, p = .021). Further, BAG was negatively correlated with domain-level cognitive function (learning: r = -0.26, p = .008; memory: r = -0.21, p = .034).

Conclusions

HIV infection is associated with augmented white matter aging, and greater brain aging is associated with worse cognitive performance in multiple domains.

Many brain structures show a complex, non-linear pattern of maturation and age-related change. Often, quadratic models (beta(0) + beta(1)age + beta(2)age(2) + epsilon) are used to describe such relationships. Here, we demonstrate that the fitting of quadratic models is substantially affected by seemingly irrelevant factors, such as the age-range sampled. Hippocampal volume was measured in 434 healthy participants between 8 and 85 years of age, and quadratic models were fit to subsets of the sample with different age-ranges. It was found that as the bottom of the age-range increased, the age at which volumes appeared to peak was moved upwards and the estimated decline in the last part of the age-span became larger. Thus, whether children were included or not affected the estimated decline between 60 and 85 years. We conclude that caution should be exerted in inferring age-trajectories from global fit models, e.g. the quadratic model. A nonparametric local smoothing technique (the smoothing spline) was found to be more robust to the effects of different starting ages. The results were replicated in an independent sample of 309 participants.

Background

Hippocampal dysfunction and volume reductions have been reported in patients with schizophrenia and bipolar disorder. The hippocampus consists of anatomically distinct subfields. We investigated to determine whether in vivo volumes of hippocampal subfields differ between clinical groups and healthy control subjects.

Methods

Clinical examination and magnetic resonance imaging were performed in 702 subjects (patients with schizophrenia spectrum [n = 210; mean age, 32.0 ± 9.3 (SD) years; 59% male], patients with bipolar spectrum [n = 192; mean age, 35.5 ± 11.5 years; 40% male] and healthy control subjects [n = 300; mean age, 35.3 ± 9.9 years; 53% male]). Hippocampal subfield volumes were estimated with FreeSurfer. General linear models were used to explore diagnostic differences in hippocampal subfield volumes, covarying for age, intracranial volume, and medication. Post hoc analyses of associations to psychosis symptoms (Positive and Negative Syndrome Scale) and cognitive function (verbal memory [California Verbal Learning Test, second edition] and IQ [Wechsler Abbreviated Scale of Intelligence]) were performed.

Results

Patient groups had smaller cornu ammonis (CA) subfields CA2/3 (left, p = 7.2 × 10(-6); right, p = 2.3 × 10(-6)), CA4/dentate gyrus (left, p = 1.4 × 10(-5); right, p = 2.3 × 10(-6)), subiculum (left, p = 3.7 × 10(-6); right, p = 2.8 × 10(-8)), and right CA1 (p = .006) volumes than healthy control subjects, but smaller presubiculum volumes were found only in patients with schizophrenia (left, p = 6.7 × 10(-5); right, p = 1.6 × 10(-7)). Patients with schizophrenia had smaller subiculum (left, p = .035; right, p = .031) and right presubiculum (p = .002) volumes than patients with bipolar disorder. Smaller subiculum volumes were related to poorer verbal memory in patients with bipolar disorder and healthy control subjects and to negative symptoms in patients with schizophrenia.

Conclusions

Hippocampal subfield volume reductions are found in patients with schizophrenia and bipolar disorder. The magnitude of reduction is greater in patients with schizophrenia, particularly in the hippocampal outflow regions presubiculum and subiculum.

Regional brain morphology has a complex genetic architecture, consisting of many common polymorphisms with small individual effects. This has proven challenging for genome-wide association studies (GWAS). Due to the distributed nature of genetic signal across brain regions, multivariate analysis of regional measures may enhance discovery of genetic variants. Current multivariate approaches to GWAS are ill-suited for complex, large-scale data of this kind. Here, we introduce the Multivariate Omnibus Statistical Test (MOSTest), with an efficient computational design enabling rapid and reliable inference, and apply it to 171 regional brain morphology measures from 26,502 UK Biobank participants. At the conventional genome-wide significance threshold of α = 5 × 10^-8, MOSTest identifies 347 genomic loci associated with regional brain morphology, more than any previous study, improving upon the discovery of established GWAS approaches more than threefold. Our findings implicate more than 5% of all protein-coding genes and provide evidence for gene sets involved in neuron development and differentiation.

An amendment to this paper has been published and can be accessed via a link at the top of the paper.

The thickness of the cerebral cortical sheet and its surface area are highly heritable traits thought to have largely distinct polygenic architectures. Despite large-scale efforts, the majority of their genetic determinants remain unknown. Our ability to identify causal genetic variants can be improved by employing brain measures that better map onto the biology we seek to understand. Such measures may have fewer variants but with larger effects, that is, lower polygenicity and higher discoverability. Using Gaussian mixture modeling, we estimated the number of causal variants shared between mean cortical thickness and total surface area, as well as the polygenicity and discoverability of regional measures. We made use of UK Biobank data from 30 880 healthy White European individuals (mean age 64.3, standard deviation 7.5, 52.1% female). We found large genetic overlap between total surface area and mean thickness, sharing 4016 out of 7941 causal variants. Regional surface area was more discoverable (P = 2.6 × 10-6) and less polygenic (P = 0.004) than regional thickness measures. These findings may serve as a roadmap for improved future GWAS studies; knowledge of which measures are most discoverable may be used to boost identification of genetic predictors and thereby gain a better understanding of brain morphology.

BACKGROUND:Restriction spectrum imaging (RSI) is a recently introduced magnetic resonance imaging diffusion technique. The utility of RSI in multiple sclerosis (MS) is unknown. OBJECTIVE:To investigate the association between RSI-derived parameters and neurological disability in MS. METHODS:Seventy-seven relapsing-remitting MS patients were scanned with RSI on a 3-T scanner. RSI-derived parameters: fast and slow apparent diffusion coefficient (sADC), fractional anisotropy, restricted fractional anisotropy, neurite density (ND), cellularity, extracellular water fraction, and free water fraction, were obtained in white matter lesions (WML) and normal appearing white matter (NAWM). Patients were divided into three groups according to their expanded disability status scale (EDSS): with minimal, low, and substantial disability (<2.5, 2.5-3, and >3, respectively). Group comparisons and correlation analyses were performed. RESULTS:All tested RSI-derived parameters differed between WML and NAWM ( p < 0.001 for all pairwise comparisons). The sADC in WML showed largest difference across disability subgroups (analysis of variance (ANOVA): F = 5.1, η2 = 0.12, p = 0.008). ND in NAWM showed strongest correlation with disability (ϱ = -0.39, p < 0.001). CONCLUSION:The strongest correlation with EDSS of ND obtained in NAWM indicates that processes outside lesions are important for disability in MS. Our study suggests that RSI-derived parameters may help understand the "clinico-radiological paradox" and improve disease monitoring in MS.

Deep learning methods hold strong promise for identifying biomarkers for clinical application. However, current approaches for psychiatric classification or prediction do not allow direct interpretation of original features. In the present study, we introduce a sparse deep neural network (DNN) approach to identify sparse and interpretable features for schizophrenia (SZ) case-control classification. An L₀ -norm regularization is implemented on the input layer of the network for sparse feature selection, which can later be interpreted based on importance weights. We applied the proposed approach on a large multi-study cohort with gray matter volume (GMV) and single nucleotide polymorphism (SNP) data for SZ classification. A total of 634 individuals served as training samples, and the classification model was evaluated for generalizability on three independent datasets of different scanning protocols (N = 394, 255, and 160, respectively). We examined the classification power of pure GMV features, as well as combined GMV and SNP features. Empirical experiments demonstrated that sparse DNN slightly outperformed independent component analysis + support vector machine (ICA + SVM) framework, and more effectively fused GMV and SNP features for SZ discrimination, with an average error rate of 28.98% on external data. The importance weights suggested that the DNN model prioritized to select frontal and superior temporal gyrus for SZ classification with high sparsity, with parietal regions further included with lower sparsity, echoing previous literature. The results validate the application of the proposed approach to SZ classification, and promise extended utility on other data modalities and traits which ultimately may result in clinically useful tools.