[Paper Review] More Alike than Different: Quantifying Deviations of Brain Structure and Function in Major Depressive Disorder across Neuroimaging Modalities

Nils R. Winter, Ramona Leenings+29 more|arXiv (Cornell University)|Dec 20, 2021

Functional Brain Connectivity Studies7 citations

TL;DR

This large-scale, multimodal neuroimaging study across 1,809 participants reveals that brain structure and function in Major Depressive Disorder (MDD) patients and healthy controls show remarkably small univariate differences, with effect sizes ranging from partial η² = 0.004 to 0.017 and classification accuracy only slightly above chance (54–55%). The findings challenge the clinical and theoretical relevance of univariate case-control neuroimaging in MDD, advocating instead for multivariate, theory-driven, and ecologically valid approaches.

ABSTRACT

Introduction: Identifying neurobiological differences between patients suffering from Major Depressive Disorder (MDD) and healthy individuals has been a mainstay of clinical neuroscience for decades. However, recent meta- and mega-analyses have raised concerns regarding the replicability and clinical relevance of brain alterations in depression. Methods: Here, we systematically investigate healthy controls and MDD patients across a comprehensive range of modalities including structural magnetic resonance imaging (MRI), diffusion tensor imaging, functional task-based and resting-state MRI under near-ideal conditions. To this end, we quantify the upper bounds of univariate effect sizes, predictive utility, and distributional dissimilarity in a fully harmonized cohort of N=1,809 participants. We compare the results to an MDD polygenic risk score (PRS) and environmental variables. Results: The upper bound of the effect sizes range from partial eta squared = .004 to .017, distributions overlap between 89% and 95%, with classification accuracies ranging between 54% and 55% across neuroimaging modalities. This pattern remains virtually unchanged when considering only acutely or chronically depressed patients. Differences are comparable to those found for PRS, but substantially smaller than for environmental variables. Discussion: We provide a large-scale, multimodal analysis of univariate biological differences between MDD patients and controls and show that even under near-ideal conditions and for maximum biological differences, deviations are extremely small and similarity dominates. We sketch an agenda for a new focus of future research in biological psychiatry facilitating quantitative, theory-driven research, an emphasis on computational psychiatry and multivariate machine learning approaches, as well as the utilization of ecologically valid phenotyping.

Motivation & Objective

• To systematically quantify univariate neurobiological differences between MDD patients and healthy controls across multiple neuroimaging modalities under near-ideal, harmonized conditions.
• To assess the upper bounds of effect sizes, predictive utility, and distributional dissimilarity in MDD neuroimaging data.
• To compare the magnitude of neuroimaging-derived differences with those from polygenic risk scores (PRS) and environmental variables.
• To evaluate whether subgroup analyses (e.g., acute vs. chronic depression) or increased methodological homogeneity alter the observed similarity between MDD and control groups.
• To challenge the clinical and theoretical relevance of current univariate case-control neuroimaging approaches in MDD and advocate for a paradigm shift in biological psychiatry.

Proposed method

• Conducted a fully harmonized, large-scale cohort study with N = 1,809 participants, including both MDD patients and healthy controls.
• Employed multiple neuroimaging modalities: structural MRI, diffusion tensor imaging (DTI), task-based fMRI, and resting-state fMRI.
• Calculated univariate effect sizes using partial eta-squared (partial η²) to quantify group differences across modalities.
• Assessed distributional overlap between MDD and control groups using statistical measures of distributional dissimilarity.
• Evaluated classification accuracy via multivariate models to estimate predictive power of neuroimaging features.
• Compared neuroimaging effect sizes with those from MDD polygenic risk scores (PRS) and environmental variables to contextualize biological relevance.

Experimental results

Research questions

• RQ1 What are the upper bounds of univariate effect sizes in brain structure and function between MDD patients and healthy controls across multiple neuroimaging modalities?
• RQ2 How much do the distributions of neuroimaging features in MDD patients and healthy controls overlap, and what does this imply for individual-level classification?
• RQ3 How does the predictive accuracy of neuroimaging features compare to that of polygenic risk scores (PRS) and environmental variables in distinguishing MDD from controls?
• RQ4 Do subgroup analyses (e.g., acute vs. chronic depression) or increased methodological homogeneity alter the observed similarity between MDD and control groups?
• RQ5 To what extent do current univariate neuroimaging studies in MDD provide clinically relevant or theoretically meaningful biological differences?

Key findings

• The upper bound of univariate effect sizes across neuroimaging modalities ranged from partial η² = 0.004 to 0.017, indicating very small group-level differences.
• Distributional overlap between MDD patients and healthy controls ranged from 89% to 95%, indicating that the majority of individual neuroimaging profiles were indistinguishable between groups.
• Classification accuracy for distinguishing MDD patients from controls was only 54–55% across all neuroimaging modalities, barely exceeding chance performance.
• These results remained virtually unchanged when analyzing only acutely or chronically depressed patients, suggesting that the small effect sizes are robust across clinical subtypes.
• Neuroimaging-derived differences were substantially smaller than those associated with environmental variables and comparable in magnitude to polygenic risk scores (PRS), indicating limited clinical utility.
• The study concludes that the small effect sizes observed in prior univariate neuroimaging studies of MDD are not primarily due to small sample sizes or methodological heterogeneity, but reflect an inherent neurobiological similarity between MDD patients and healthy individuals.