Amygdala size varies with stress perception

Abstract

Stress is inevitably linked to life. It has many and complex facets. Notably, perception of stressful stimuli is an important factor when mounting stress responses and measuring its impact. Indeed, moved by the increasing number of stress-triggered pathologies, several groups drew on advanced neuroimaging techniques to explore stress effects on the brain. From that, several regions and circuits have been linked to stress, and a comprehensive integration of the distinct findings applied to common individuals is being pursued, but with conflicting results. Herein, we performed a volumetric regression analysis using participants' perceived stress as a variable of interest. Data shows that increased levels of perceived stress positively associate with the right amygdala and anterior hippocampal volumes.

Keywords: Amygdala; FSL, FMRIB Software Library; FWE-R, Family-wise error rate; FreeSurfer; GM, Gray matter; Healthy subjects; M, Mean; PSS10, 10-items Perceived Stress Scale; Perceived stress; ROI, Region-of-interest; SD, Standard deviation; TFCE, Threshold-free cluster enhancement; VBM, Voxel-based morphometry; Voxel-based morphometry; WM, White matter; eTIV, Estimated total intracranial volume.

Fig. 1

Results from volumetric regression with PSS10 evaluated through FSL-VBM. A positive statistically significant association between PSS10 scores and two main subcortical clusters is observed. On the left, the biggest cluster is mainly composed of the right amygdala (peak), embracing part of the right hippocampus and a small portion of the right putamen. On the right, a smaller cluster is observed the in left hippocampus (peak) and left amygdala. Following the FSL-VBM pipeline, after brain extraction and segmentation, a study-specific GM template was created, upon which all GM images were registered. During this registration step, a compensation for the non-linear component of the transformation is introduced by the FSL-VBM protocol, which already adjusts for intracranial differences. Therefore, only age, sex, and MRI parameters were defined as covariates, and TFCE and FWE-R correction at a significance level α = 0.05 were used.

Fig. 2

Results from volumetric regression with PSS10 evaluated through FreeSurfer. A positive statistically significant association between PSS10 scores and right amygdala was observed, after correction for multiple comparisons. On the left, a representation of the 54_R Amygdala cluster from FreeSurfer subcortical regions labelling. On the right, a graphical representation of the model with significance between PSS10 scores and right amygdala volumes corrected for GM, and age/sex covariates; the equation represents the correlation between the PSS10 scores and corrected right amygdala volumes, and not the global model per se. Brain volumes were computed using FreeSurfer subcortical output and corrected for individual GM to best replicate the VBM regression analysis. Multilinear regression models with ROI volumes as dependent variables and PSS10 scores, age, and sex as independent variables were established. The models were computed using the regstats function in MATLAB and the Bonferroni-Holm correction for 14 multiple comparisons was used to calculate the corrected p-values. Statistical significance was established for α = 0.05.