Predicting Future Recurrence in Affective Disorders from Brain-Age Gaps in a 9-Year Follow-Up

595 

Abstract Submission 

Nils Winter¹, Katharina Förster², Thomas Frodl³, Klaus Berger⁴, Philipp Kanske⁵, Jan Ernsting⁶, Ramona Leenings⁷, Carlotta Barkhau⁷, Maximilian Konowski¹, Lukas Fisch⁷, Daniel Emden⁷, Udo Dannlowski⁷, Tim Hahn⁷, Dominik Grotegerd⁸

¹University of Münster, Münster, North-Rhine Westphalia, ²University of Dresden, Dresden, Saxony, ³RWTH Aachen University, Aachen, North-Rhine Westphalia, ⁴Institute of Epidemiology and Social Medicine, University of Münster, Münster, NRW, Germany, ⁵Clinical Psychology and Behavioral Neuroscience, Technische Universität Dresden, Dresden, Germany, ⁶University of Münster, Münster, NRW, ⁷Institute for Translational Psychiatry, Münster, North Rhine Westphalia, ⁸Institute for Translational Psychiatry, University of Münster, Münster, North Rhine-Westphalia

Nils Winter  
University of Münster
Münster, North-Rhine Westphalia

Katharina Förster  
University of Dresden
Dresden, Saxony

Thomas Frodl  
RWTH Aachen University
Aachen, North-Rhine Westphalia

Klaus Berger  
Institute of Epidemiology and Social Medicine, University of Münster
Münster, NRW, Germany

Philipp Kanske  
Clinical Psychology and Behavioral Neuroscience, Technische Universität Dresden
Dresden, Germany

Jan Ernsting  
University of Münster
Münster, NRW

Ramona Leenings  
Institute for Translational Psychiatry
Münster, North Rhine Westphalia

Carlotta Barkhau  
Institute for Translational Psychiatry
Münster, North Rhine Westphalia

Maximilian Konowski  
University of Münster
Münster, North-Rhine Westphalia

Lukas Fisch  
Institute for Translational Psychiatry
Münster, North Rhine Westphalia

Daniel Emden  
Institute for Translational Psychiatry
Münster, North Rhine Westphalia

Udo Dannlowski  
Institute for Translational Psychiatry
Münster, North Rhine Westphalia

Tim Hahn  
Institute for Translational Psychiatry
Münster, North Rhine Westphalia

Dominik Grotegerd  
Institute for Translational Psychiatry, University of Münster
Münster, North Rhine-Westphalia

Affective disorders contribute immensely to the global burden of disease worldwide (Murray et al., 2012; Paykel et al., 2005). Recently, a novel multivariate biomarker has emerged in the field of neuroimaging that aims at quantifying the age-associated biological changes that occur in the brain (Cole et al., 2019). The underlying hypothesis of the brain age prediction paradigm is that this brain age gap (BAG) may serve as a marker of disease risk and there are a number of studies emphasizing an association between brain age gaps and clinically relevant variables (Bittner et al., 2021; Elliott et al., 2019). We investigated associations of the brain age gap with disease course over nine years in patients with affective disorders in a long-term prospective design.

At two time-points, we acquired T1-weighted MRI images (mean [SD] follow-up period 8.98 [2.20] years) of patients with affective disorders (N = 38) and healthy controls (HC: N = 37) at two sites (Dublin, UK; Münster, Germany). Using a publicly available, uncertainty-aware brain age prediction model trained on a sample of over 10,000 individuals of the German National Cohort (GNC), we estimated individual BAG at two time-points (baseline and follow-up) using gray matter segments derived from MRI images (Hahn et al., 2022). In short, in contrast to existing brain age models, the MCCQR-NN model provides accurate estimations of predictive uncertainty in high-dimensional neuroimaging data while ensuring state-of-the-art model performance. It is therefore especially suited for the detection of subtle brain age changes, e.g., in clinical cohorts. Employing linear-mixed-effects models, we tested main effects of diagnosis at follow-up and hospitalizations during follow-up on BAG at baseline and follow-up, as well as the interaction of diagnosis and hospitalization with time respectively. In an exploratory analysis, we tested if BAG at baseline was predictive of further hospitalizations during the nine-year follow-up using logistic regression and 10-fold nested cross-validation. All brain age predictions and machine learning analyses were made using the Python package PHOTONAI (Leenings et al., 2021).

MDD patients showed a larger BAG compared to HC (MDD>HC: p=.039, MDD vs. BD: n.s.), while BD patients only showed a tendency for a larger BAG (p=.066). In the Münster subsample (N=52), patients with hospitalizations showed a higher BAG compared to patients without hospitalizations (p=.001). No significant group-by-time interaction could be detected.
However, an increased baseline BAG was linked to the number of hospitalizations during follow-up (p=.018). Employing machine learning to predict hospitalization based on the baseline BAG resulted in a classification accuracy of 64.3%, yet this did not reach statistical significance.

Using a state-of-the-art brain age prediction model trained on T1-weighted MRI images of 10,000 individuals, we calculated and compared brain age gaps in a sample of 75 patients with affective disorders and healthy controls, measured at two time points with a mean follow-up length of nine years. Our results show that BAG did not change over time as a function of patients' course of disease. The present study rather suggests that a higher estimation of biological ageing (higher BAG) predicts future hospitalizations. Therefore, BAG may indicate a patient's vulnerability to future recurrence.

Psychiatric (eg. Depression, Anxiety, Schizophrenia) ¹

Classification and Predictive Modeling ²

Multivariate Approaches

Anatomical MRI

Affective Disorders

Aging

Machine Learning

MRI

Psychiatric Disorders