Brain age to identify structural alterations in schizophrenia unrelated with aging
Poster No:
649
Submission Type:
Abstract Submission
Authors:
Alejandro Roig-Herrero1, Rafael Navarro-González1, Álvaro Planchuelo-Gómez1, Santiago Aja-Fernández1, Juan Calabia del Campo2, Vicente Molina-Rodríguez1, Rodrigo De Luis-García1
Institutions:
1Universidad de Valladolid, Valladolid, Valladolid, 2Hospital Clínico Universitario de Valladolid, Valladolid, Valladolid
First Author:
Co-Author(s):
Introduction:
Neuroimaging has consistently revealed significant changes in the brain structure of schizophrenia patients. Some of these changes seem to be present even before illness onset [1, 2], while others become more pronounced years after disease onset [3].
In the brain age paradigm, a machine learning model is trained to predict a person's age based on brain imaging data. The difference between the predicted age and the actual chronological age, known as the brain age gap, is considered a marker of brain health [4].
Since increased brain age has been consistently found in schizophrenia across several studies [5, 6], a fundamental question naturally arises: are changes in the gray matter in schizophrenia the result of an accelerated brain aging process, or are they (at least in part) the result of a fundamentally different process? Our aim is to shed light on this question.
In the brain age paradigm, a machine learning model is trained to predict a person's age based on brain imaging data. The difference between the predicted age and the actual chronological age, known as the brain age gap, is considered a marker of brain health [4].
Since increased brain age has been consistently found in schizophrenia across several studies [5, 6], a fundamental question naturally arises: are changes in the gray matter in schizophrenia the result of an accelerated brain aging process, or are they (at least in part) the result of a fundamentally different process? Our aim is to shed light on this question.
Methods:
67 chronic schizophrenia patients (30 females) and 97 healthy controls (43 females) underwent brain MRI acquisition including T1-weighted images using a Philips Achieva 3T MRI unit (Philips Healthcare, Best, the Netherlands) with a 32-channel head coil in the MRI facility at the Universidad de Valladolid (Spain). Acquisition parameters are available elsewhere [7].
Following the image acquisition, FastSurfer was employed to extract a total of 1,479 features [8]. Fastsurfer uses Deep Learning to perform brain segmentation based on the Desikan-Killiany atlas [9]. In this study 289 were considered, describing the volume of cortical and subcortical gray matter regions and white matter regions from the atlas, as well as the surface, thickness and curvature of the cortical regions.
In parallel, a brain age prediction model was trained from 2,771 structural T1w MRI scans from different studies and databases. Details about this brain age model are available elsewhere [10].
The Shapiro-Wilk test and Levene's test for equality of variances were used to assess normality and homogeneity of variances in age and total intracranial volume. To test for significant differences, a t-test was used if the null hypothesis in the Shapiro-Wilk and Levene tests was not rejected; otherwise, the Wilcoxon rank-sum test was employed. To test for sex-related significant differences, a Fisher exact test was used.
Brain age differences between groups were assessed with an analysis of covariance (ANCOVA), including age, total intracranial volume and sex as covariates.
With regard to the 289 morphological features described before, an ANCOVA was performed, with age, total intracranial volume and sex as covariates. Next, the same analysis was repeated, but considering the estimated brain age as covariate instead of the chronological age. Bonferroni correction was applied for multiple comparisons after setting the level of statistical significance at P<0.05.
Following the image acquisition, FastSurfer was employed to extract a total of 1,479 features [8]. Fastsurfer uses Deep Learning to perform brain segmentation based on the Desikan-Killiany atlas [9]. In this study 289 were considered, describing the volume of cortical and subcortical gray matter regions and white matter regions from the atlas, as well as the surface, thickness and curvature of the cortical regions.
In parallel, a brain age prediction model was trained from 2,771 structural T1w MRI scans from different studies and databases. Details about this brain age model are available elsewhere [10].
The Shapiro-Wilk test and Levene's test for equality of variances were used to assess normality and homogeneity of variances in age and total intracranial volume. To test for significant differences, a t-test was used if the null hypothesis in the Shapiro-Wilk and Levene tests was not rejected; otherwise, the Wilcoxon rank-sum test was employed. To test for sex-related significant differences, a Fisher exact test was used.
Brain age differences between groups were assessed with an analysis of covariance (ANCOVA), including age, total intracranial volume and sex as covariates.
With regard to the 289 morphological features described before, an ANCOVA was performed, with age, total intracranial volume and sex as covariates. Next, the same analysis was repeated, but considering the estimated brain age as covariate instead of the chronological age. Bonferroni correction was applied for multiple comparisons after setting the level of statistical significance at P<0.05.
Results:
No differences were found in age, intracranial volume or sex between the two groups.
An increased brain age gap was found in the schizophrenia group with respect to healthy controls (7.4 years ± 9.8 vs -1.3 years ± 8.7, P<10-8, see Figure 1A).
Figure 1B graphically compares the p-values obtained using both corrections. When correcting for chronological age, statistically significant differences were found in 31 imaging features (P<0.000173), generally indicating lower volume and cortical thickness across widespread brain regions. However, when correcting for brain age, significant differences were found in only three imaging features (see Figure 2).
An increased brain age gap was found in the schizophrenia group with respect to healthy controls (7.4 years ± 9.8 vs -1.3 years ± 8.7, P<10-8, see Figure 1A).
Figure 1B graphically compares the p-values obtained using both corrections. When correcting for chronological age, statistically significant differences were found in 31 imaging features (P<0.000173), generally indicating lower volume and cortical thickness across widespread brain regions. However, when correcting for brain age, significant differences were found in only three imaging features (see Figure 2).
Conclusions:
Our results suggest that most structural variations in schizophrenia patients correlate with increased brain aging, yet a subset of these differences appears unrelated to that process. Focusing on these distinct changes is vital for comprehending the schizophrenia brain, offering potential insights into subgroups, treatment effects or prognostic indicators.
Disorders of the Nervous System:
Psychiatric (eg. Depression, Anxiety, Schizophrenia) 1
Lifespan Development:
Aging 2
Modeling and Analysis Methods:
Other Methods
Novel Imaging Acquisition Methods:
Anatomical MRI
Keywords:
Aging
MRI
Schizophrenia
Other - Brain Age
1|2Indicates the priority used for review
Provide references using author date format
[2] Fusar-Poli, P. et al. (2014), ‘Neuroanatomical markers of genetic liability to psychosis and first episode psychosis: a voxelwise meta-analytical comparison’, The World Journal of Biological Psychiatry, 15(3), 219-228.
[3] Olabi, B. et al. (2011), ‘Are there progressive brain changes in schizophrenia? A meta-analysis of structural magnetic resonance imaging studies’, Biological Psychiatry, 70(1), 88-96.
[4] Franke, K. et al. (2010), ‘Estimating the age of healthy subjects from T1-weighted MRI scans using kernel methods: exploring the influence of various parameters’, Neuroimage, 50(3), 883-892.
[5] Koutsouleris, N. et al. (2014), ‘Accelerated brain aging in schizophrenia and beyond: a neuroanatomical marker of psychiatric disorders’, Schizophrenia Bulletin, 40(5), 1140-1153.
[6] Constantinides, C. et al. (2023), ‘Brain ageing in schizophrenia: evidence from 26 international cohorts via the ENIGMA Schizophrenia consortium’, Molecular Psychiatry, 28(3), 1201-1209.
[7] Lubeiro, A. et al. (2017), ‘Biological and cognitive correlates of cortical curvature in schizophrenia’, Psychiatry Research: Neuroimaging, 270, 68-75.
[8] Henschel, L. et al. (2020), ‘Fastsurfer-a fast and accurate deep learning based neuroimaging pipeline’, NeuroImage, 219, 117012.
[9] Desikan, R. S. et al. (2006), ‘An automated labeling system for subdividing the human cerebral cortex on MRI scans into gyral based regions of interest’. Neuroimage, 31(3), 968-980.
[10] Navarro-González, R. et al. (2023), ‘Increased MRI-based Brain Age in chronic migraine patients’, The Journal of Headache and Pain, 24(1), 133.