Deep learning identifies brain structures that predict cognition and explain heterogeneity in cognitive aging

Increased grey matter volumes in the temporal lobe and its relationship with cognitive functioning in euthymic patients with bipolar disorder

BACKGROUND

Bipolar disorder (BD) is characterized by episodic mood dysregulation, although a significant portion of patients suffer persistent cognitive impairment during euthymia. Previous magnetic resonance imaging (MRI) research suggests BD patients may have accelerated brain aging, observed as lower grey matter volumes. How these neurostructural alterations are related to the cognitive profile of BD is unclear.

METHODS

We aim to explore this relationship in euthymic BD patients with multimodal structural neuroimaging. A sample of 27 euthymic BD patients and 24 healthy controls (HC) underwent structural grey matter MRI and diffusion-weighted imaging (DWI). BD patient's cognition was also assessed. FreeSurfer algorithms were used to obtain estimations of regional grey matter volumes. White matter pathways were reconstructed using TRACULA, and four diffusion metrics were extracted. ANCOVA models were performed to compare BD patients and HC values of regional grey matter volume and diffusion metrics. Global brain measures were also compared. Bivariate Pearson correlations were explored between significant brain results and five cognitive domains.

RESULTS

Euthymic BD patients showed higher ventricular volume (F(1, 46) = 6.04; p = 0.018) and regional grey matter volumes in the left fusiform (F(1, 46) = 15.03; pFDR = 0.015) and bilateral parahippocampal gyri compared to HC (L: F(1, 46) = 12.79, pFDR = 0.025/ R: F(1, 46) = 15.25, pFDR = 0.015). Higher grey matter volumes were correlated with greater executive function (r = 0.53, p = 0.008).

LIMITATIONS

We evaluated a modest sample size with concurrent pharmacological treatment.

CONCLUSIONS

Higher medial temporal volumes in euthymic BD patients may be a potential signature of brain resilience and cognitive adaptation to a putative illness neuroprogression. This knowledge should be integrated into further efforts to implement imaging into BD clinical management.

Digging into the intrinsic capacity concept: Can it be applied to Alzheimer's disease?

Historically, aging research has largely centered on disease pathology rather than promoting healthy aging. The World Health Organization's (WHO) policy framework (2015-2030) underscores the significance of fostering the contributions of older individuals to their families, communities, and economies. The WHO has introduced the concept of intrinsic capacity (IC) as a key metric for healthy aging, encompassing five primary domains: locomotion, vitality, sensory, cognitive, and psychological. Past AD research, constrained by methodological limitations, has focused on single outcome measures, sidelining the complexity of the disease. Our current scientific milieu, however, is primed to adopt the IC concept. This is due to three critical considerations: (I) the decline in IC is linked to neurocognitive disorders, including AD, (II) cognition, a key component of IC, is deeply affected in AD, and (III) the cognitive decline associated with AD involves multiple factors and pathophysiological pathways. Our study explores the application of the IC concept to AD patients, offering a comprehensive model that could revolutionize the disease's diagnosis and prognosis. There is a dearth of information on the biological characteristics of IC, which are a result of complex interactions within biological systems. Employing a systems biology approach, integrating omics technologies, could aid in unraveling these interactions and understanding IC from a holistic viewpoint. This comprehensive analysis of IC could be leveraged in clinical settings, equipping healthcare providers to assess AD patients' health status more effectively and devise personalized therapeutic interventions in accordance with the precision medicine paradigm. We aimed to determine whether the IC concept could be extended from older individuals to patients with AD, thereby presenting a model that could significantly enhance the diagnosis and prognosis of this disease.

Cortical thickness reveals sex differences in verbal and visuospatial memory

Although previous studies have reported the sex differences in behavior/cognition and the brain, the sex difference in the relationship between memory abilities and the underlying neural basis in the aging process remains unclear. In this study, we used a machine learning model to estimate the association between cortical thickness and verbal/visuospatial memory in females and males and then explored the sex difference of these associations based on a community-elderly cohort (n = 1153, age ranged from 50.42 to 86.67 years). We validated that females outperformed males in verbal memory, while males outperformed females in visuospatial memory. The key regions related to verbal memory in females include the medial temporal cortex, orbitofrontal cortex, and some regions around the insula. Further, those regions are more located in limbic, dorsal attention, and default-model networks, and are associated with face recognition and perception. The key regions related to visuospatial memory include the lateral prefrontal cortex, anterior cingulate gyrus, and some occipital regions. They overlapped more with dorsal attention, frontoparietal and visual networks, and were associated with object recognition. These findings imply the memory performance advantage of females and males might be related to the different memory processing tendencies and their associated network.

The volume and the distribution of premorbid white matter hyperintensities: Impact on post-stroke aphasia

White matter hyperintensities (WMH) are a radiological manifestation of progressive white matter integrity loss. The total volume and distribution of WMH within the corpus callosum have been associated with pathological cognitive ageing processes but have not been considered in relation to post-stroke aphasia outcomes. We investigated the contribution of both the total volume of WMH, and the extent of WMH lesion load in the corpus callosum to the recovery of language after first-ever stroke. Behavioural and neuroimaging data from individuals (N = 37) with a left-hemisphere stroke were included at the early subacute stage of recovery. Spoken language comprehension and production abilities were assessed using word and sentence-level tasks. Neuroimaging data was used to derive stroke lesion variables (volume and lesion load to language critical regions) and WMH variables (WMH volume and lesion load to three callosal segments). WMH volume did not predict variance in language measures, when considered together with stroke lesion and demographic variables. However, WMH lesion load in the forceps minor segment of the corpus callosum explained variance in early subacute comprehension abilities (t = -2.59, p = .01) together with corrected stroke lesion volume and socio-demographic variables. Premorbid WMH lesions in the forceps minor were negatively associated with early subacute language comprehension after aphasic stroke. This negative impact of callosal WMH on language is consistent with converging evidence from pathological ageing suggesting that callosal WMH disrupt the neural networks supporting a range of cognitive functions.

Quantitative estimate of cognitive resilience and its medical and genetic associations

BACKGROUND

We have proposed that cognitive resilience (CR) counteracts brain damage from Alzheimer's disease (AD) or AD-related dementias such that older individuals who harbor neurodegenerative disease burden sufficient to cause dementia remain cognitively normal. However, CR traditionally is considered a binary trait, capturing only the most extreme examples, and is often inconsistently defined.

METHODS

This study addressed existing discrepancies and shortcomings of the current CR definition by proposing a framework for defining CR as a continuous variable for each neuropsychological test. The linear equations clarified CR's relationship to closely related terms, including cognitive function, reserve, compensation, and damage. Primarily, resilience is defined as a function of cognitive performance and damage from neuropathologic damage. As such, the study utilized data from 844 individuals (age = 79 ± 12, 44% female) in the National Alzheimer's Coordinating Center cohort that met our inclusion criteria of comprehensive lesion rankings for 17 neuropathologic features and complete neuropsychological test results. Machine learning models and GWAS then were used to identify medical and genetic factors that are associated with CR.

RESULTS

CR varied across five cognitive assessments and was greater in female participants, associated with longer survival, and weakly associated with educational attainment or APOE ε4 allele. In contrast, damage was strongly associated with APOE ε4 allele (P value < 0.0001). Major predictors of CR were cardiovascular health and social interactions, as well as the absence of behavioral symptoms.

CONCLUSIONS

Our framework explicitly decoupled the effects of CR from neuropathologic damage. Characterizations and genetic association study of these two components suggest that the underlying CR mechanism has minimal overlap with the disease mechanism. Moreover, the identified medical features associated with CR suggest modifiable features to counteract clinical expression of damage and maintain cognitive function in older individuals.

Grey matter volume and CSF biomarkers predict neuropsychological subtypes of MCI

There is increasing evidence of different subtypes of individuals with mild cognitive impairment (MCI). An important line of research is whether neuropsychologically-defined subtypes have distinct patterns of neurodegeneration and cerebrospinal fluid (CSF) biomarker composition. In our study, we demonstrated that MCI participants of the ADNI database (N = 640) can be discriminated into 3 coherent neuropsychological subgroups. Our clustering approach revealed amnestic MCI, mixed MCI, and cluster-derived normal subgroups. Furthermore, classification modeling revealed that specific predictive features can be used to differentiate amnestic and mixed MCI from cognitively normal (CN) controls: CSF Aβ142 concentration for the former and CSF Aβ1-42 concentration, tau concentration as well as grey matter atrophy (especially in the temporal and occipital lobes) for the latter. In contrast, participants from the cluster-derived normal subgroup exhibited an identical profile to CN controls in terms of cognitive performance, brain structure, and CSF biomarker levels. Our comprehensive data analytics strategy provides further evidence that multimodal neuropsychological subtyping is both clinically and neurobiologically meaningful.

Decoding degeneration: the implementation of machine learning for clinical detection of neurodegenerative disorders

Machine learning represents a growing subfield of artificial intelligence with much promise in the diagnosis, treatment, and tracking of complex conditions, including neurodegenerative disorders such as Alzheimer's and Parkinson's diseases. While no definitive methods of diagnosis or treatment exist for either disease, researchers have implemented machine learning algorithms with neuroimaging and motion-tracking technology to analyze pathologically relevant symptoms and biomarkers. Deep learning algorithms such as neural networks and complex combined architectures have proven capable of tracking disease-linked changes in brain structure and physiology as well as patient motor and cognitive symptoms and responses to treatment. However, such techniques require further development aimed at improving transparency, adaptability, and reproducibility. In this review, we provide an overview of existing neuroimaging technologies and supervised and unsupervised machine learning techniques with their current applications in the context of Alzheimer's and Parkinson's diseases.