Hormones inside the early adolescent female brain: a multimodality brain MRI study
Poster No:
2405
Submission Type:
Abstract Submission
Authors:
Muskan Khetan1, Ye Tian2, Nandita vijayakumar3, Sarah Whittle4
Institutions:
1University of Melbourne, Melbourne, VIC, 2University of Melbourne, Melbourne, Victoria, 3Deakin university, Melbourne, Victoria, 4University of Melbourne, Melbourne , Victoria
First Author:
Co-Author(s):
Introduction:
During adolescence, about one-third of young people are diagnosed with internalizing disorders like depression and anxiety, with these disorders being more common in females than males. The sex difference in incidence rates may be related to pubertal factors, such as the influence of pubertal hormones on brain development. While some studies have examined the association between pubertal hormone levels and brain structure in adolescents, more research is needed to test associations between these hormone levels and both brain structure and function in adolescents, particularly during early adolescence when hormone levels begin to surge.
Methods:
We used the Adolescent Brain Cognitive Development (ABCD) study baseline and second follow-up datasets (age 9-12 years) for female participants (N ~ 2204). Then, using elastic-net regression, we investigated the associations between pubertal hormone levels (oestradiol (E2), testosterone, Dehydroepiandrosterone (DHEA)) and brain structure (gray matter volumes, sulcal depth, cortical thickness and white matter microstructure) and function (resting-state connectivity, emotional n-back task-related function). Structure and function data were extracted from all regions across the whole brain (using the FreeSurfer Destrieux parcellation) and white matter tracts (using AtlasTrack).
For analysis (Figure1), data was randomly partitioned into training and test data. Then, using the MATLAB package GLMnet, a ten-fold, cross-validated elastic-net regression was applied to the training set for feature selection for each modality separately. From each modality, selected features that explained variance in hormones in the test data were included for further analysis. Next, within 10- iterations all significant features from each modality were modeled together in the same multi-modality model. The average r-square (R2) and correlation (r) for the fitted model were computed. To focus on the most important features, we interpreted only the top 20% fitted features.
For analysis (Figure1), data was randomly partitioned into training and test data. Then, using the MATLAB package GLMnet, a ten-fold, cross-validated elastic-net regression was applied to the training set for feature selection for each modality separately. From each modality, selected features that explained variance in hormones in the test data were included for further analysis. Next, within 10- iterations all significant features from each modality were modeled together in the same multi-modality model. The average r-square (R2) and correlation (r) for the fitted model were computed. To focus on the most important features, we interpreted only the top 20% fitted features.
·Flow-chart describing the uni and multi-modality machine learning framework; blue colour boxes showing unimodal, green boxes showing multimodal, orange boxes are common steps for both uni & multimodal
Results:
In the multi-modality analysis, R2 and r values were 0.025 and 0.14 (E2), 0.18 and 0.46 (testosterone), and 0.036 and 0.32 (DHEA). From structural and functional features, we found that structural features, especially cortical thickness and sulcal depth, in addition to resting state functional connectivity, were among the most important features for all three hormones. The structure of frontal cortical regions (particularly middle frontal and subcallosal cortices) were negatively associated with E2, while parietal and temporal cortical structural properties were positively associated with testosterone and DHEA. Subcortical volumes were positively associated with testosterone (e.g., hippocampus) and DHEA (e.g., amygdala). In addition, resting-state connectivity between subcortical regions (such as thalamus or caudate) and cortical networks (such as default mode networks or fronto-parietal network) explained more variance in the hormones as compared to inter and intra cortical networks.
Conclusions:
Associations between E2 and frontal structure, and testosterone and DHEA and temporal, parietal and subcortical regions, suggests a differential role of pubertal hormones in the development of brain regions underlying social, cognitive and emotional processes. The involvement of connectivity between subcortical regions and cortical networks further indicates the contribution of these hormones in organising cognitive and emotional brain networks. These findings are similar to studies that reported alterations in these brain regions in individuals with internalising symptoms. Thus, our findings indicate that more research is needed to understand the functional and behavioural implications of these hormone-brain associations.
Disorders of the Nervous System:
Psychiatric (eg. Depression, Anxiety, Schizophrenia)
Emotion, Motivation and Social Neuroscience:
Emotional Learning
Lifespan Development:
Early life, Adolescence, Aging 2
Modeling and Analysis Methods:
Other Methods
Novel Imaging Acquisition Methods:
Multi-Modal Imaging 1
Keywords:
Affective Disorders
Emotions
FUNCTIONAL MRI
Statistical Methods
STRUCTURAL MRI
Structures
Sub-Cortical
Thalamus
WHITE MATTER IMAGING - DTI, HARDI, DSI, ETC
Other - Pubertal hormones
1|2Indicates the priority used for review
Provide references using author date format
J.A. Silvers. (2017). vlPFC–vmPFC–amygdala interactions underlie age-related differences in cognitive regulation of emotion. Cereb. cortex, 27 (7), pp. 3502-3514
McLean, C. P. (2009). Brave men and timid women? A review of the gender differences in fear and anxiety. Clinical psychology review, 29(6), 496-505.
M.M. Herting.(2014). The role of testosterone and oestradiol in brain volume changes across adolescence: a longitudinal structural MRI study. Hum. Brain Mapp., 35 (11), pp. 5633-5645
Orlhac F (2022). A Guide to ComBat Harmonization of Imaging Biomarkers in Multicenter Studies. J Nucl Med, 63(2):172-179
Peper, J. S.(2009). Cerebral white matter in early puberty is associated with luteinizing hormone concentrations. Psychoneuroendocrinology, 34(5), 705-711.
Yee-Ming Chan. (2019) Effects of the Timing of Sex-Steroid Exposure in Adolescence on Adult Health Outcomes, The Journal of Clinical Endocrinology & Metabolism, Volume 104, Issue 10, Pages 4578–4586.
Vijayakumar, N. (2019). Puberty and the human brain: Insights into adolescent development. Neuroscience and Biobehavioral Reviews, 92, 417-436.