Usemann, P. | Search bioRxiv

bioRxiv2024

Gray matter correlates of childhood maltreatment in the context of major depression: searching for replicability in a multi-cohort brain-wide association study of 3225 adults

Childhood maltreatment effects on cerebral gray matter have been frequently discussed as a neurobiological pathway for depression. However, localizations are highly heterogeneous, and recent reports have questioned the replicability of mental health neuroimaging findings. Here, we investigate the replicability of gray matter correlates of maltreatment (measured retrospectively via the Childhood Trauma Questionnaire) across three large adult cohorts (total N=3225). Pooling cohorts revealed maltreatment-related gray matter reductions, with most extensive effects when not controlling for depression diagnosis (maximum partial R2=.022). However, none of these effects significantly replicated across cohorts. Non-replicability was consistent across a variety of maltreatment subtypes and operationalizations, as well as subgroup analyses with and without depression, and stratified by sex. In this work we show that there is little evidence for the replicability of gray matter correlates of childhood maltreatment, when adequately controlling for psychopathology. This underscores the need to focus on replicability research in mental health neuroimaging.

neuroscience↗

bioRxiv2024

Brain Structural Correlates of an Impending Initial Major Depressive Episode

BackgroundNeuroimaging research has yet to elucidate, whether reported gray matter volume (GMV) alterations in major depressive disorder (MDD) exist already before the onset of the first episode. Recruitment of presently healthy individuals with a known future transition to MDD (converters) is extremely challenging but crucial to gain insights into neurobiological vulnerability. Hence, we compared converters to patients with MDD and sustained healthy controls (HC) to distinguish pre-existing neurobiological markers from those emerging later in the course of depression. MethodsCombining two clinical cohorts (n=1709), voxel-wise GMV of n=45 converters, n=748 patients with MDD, and n=916 HC were analyzed in regions-of-interest approaches. By contrasting the subgroups and considering both remission state and reported recurrence at a 2-year clinical follow-up, we stepwise disentangled effects of 1) vulnerability, 2) the acute depressive state, and 3) an initial vs. a recurrent episode. ResultsAnalyses revealed higher amygdala GMV in converters relative to HC (pTFCE-FWE=.037, d=0.447) and patients (pTFCE-FWE=.005, d=0.508), remaining significant when compared to remitted patients with imminent recurrence. Lower GMV in the dorsolateral prefrontal cortex (pTFCE-FWE<.001, d=0.188) and insula (pTFCE-FWE=.010, d=0.186) emerged in patients relative to HC but not to converters, driven by patients with acute MDD. ConclusionBy examining one of the largest available converter samples in psychiatric neuroimaging, this study allowed a first determination of neural markers for an impending initial depressive episode. Our findings suggest a temporary vulnerability, which in combination with other common risk factors might facilitate prediction and in turn improve prevention of depression.

neuroscience↗

bioRxiv2023

Cross-validation for the estimation of effect size generalizability in mass-univariate brain-wide association studies

IntroductionStatistical effect sizes are systematically overestimated in small samples, leading to poor generalizability and replicability of findings in all areas of research. Due to the large number of variables, this is particularly problematic in neuroimaging research. While cross-validation is frequently used in multivariate machine learning approaches to assess model generalizability and replicability, the benefits for mass-univariate brain analysis are yet unclear. We investigated the impact of cross-validation on effect size estimation in univariate voxel-based brain-wide associations, using body mass index (BMI) as an exemplary predictor. MethodsA total of n=3401 adults were pooled from three independent cohorts. Brain-wide associations between BMI and gray matter structure were tested using a standard linear mass-univariate voxel-based approach. First, a traditional non-cross-validated analysis was conducted to identify brain-wide effect sizes in the total sample (as an estimate of a realistic reference effect size). The impact of sample size (bootstrapped samples ranging from n=25 to n=3401) and cross-validation on effect size estimates was investigated across selected voxels with differing underlying effect sizes (including the brain-wide lowest effect size). Linear effects were estimated within training sets and then applied to unseen test set data, using 5-fold cross-validation. Resulting effect sizes (explained variance) were investigated. ResultsAnalysis in the total sample (n=3401) without cross-validation yielded mainly negative correlations between BMI and gray matter density with a maximum effect size of R2p=.036 (peak voxel in the cerebellum). Effects were overestimated exponentially with decreasing sample size, with effect sizes up to R2p=.535 in samples of n=25 for the voxel with the brain-wide largest effect and up to R2p=.429 for the voxel with the brain-wide smallest effect. When applying cross-validation, linear effects estimated in small samples did not generalize to an independent test set. For the largest brain-wide effect a minimum sample size of n=100 was required to start generalizing (explained variance >0 in unseen data), while n=400 were needed for smaller effects of R2p =.005 to generalize. For a voxel with an underlying null effect, linear effects found in non-cross-validated samples did not generalize to test sets even with the maximum sample size of n=3401. Effect size estimates obtained with and without cross-validation approached convergence in large samples. DiscussionCross-validation is a useful method to counteract the overestimation of effect size particularly in small samples and to assess the generalizability of effects. Train and test set effect sizes converge in large samples which likely reflects a good generalizability for models in such samples. While linear effects start generalizing to unseen data in samples of n>100 for large effect sizes, the generalization of smaller effects requires larger samples (n>400). Cross-validation should be applied in voxel-based mass-univariate analysis to foster accurate effect size estimation and improve replicability of neuroimaging findings. We provide open-source python code for this purpose (https://osf.io/cy7fp/?view_only=a10fd0ee7b914f50820b5265f65f0cdb).

neuroscience↗