Effects of lockdowns on neurobiological and psychometric parameters in unipolar depression during the COVID-19 pandemic

The study examines whether COVID-19 lockdowns and restriction measures led to neurobiological and psychometric changes in individuals with recurrent major depressive disorder (MDD) compared to healthy individuals. Motivated by evidence that outbreaks and quarantine can increase stress, anxiety, and depression, and that stress and traumatic events can affect brain structure and neuroplasticity, the authors hypothesized cumulative effects of lockdowns. Specifically, they expected reductions in grey matter in the amygdala, hippocampus, anterior cingulate cortex, and prefrontal cortex; decreases in serum BDNF levels; and increases in perceived stress and depressive symptoms over time, with stronger effects in recurrent MDD than in healthy participants. The study’s purpose is to longitudinally assess these adaptational processes during the pandemic’s restriction phases in Austria and to clarify potential vulnerability or resilience in MDD versus healthy individuals.

Prior work indicates that outbreaks and quarantine are associated with common and sometimes lasting psychological symptoms. Perceived stress during lockdown correlates with development of depressive symptoms in the general population. Diathesis-stress and differential susceptibility frameworks suggest heightened vulnerability in individuals with mental disorders. Neurobiological literature shows that traumatic life events can alter brain structure, including hippocampal and prefrontal regions, and that social isolation can reduce grey matter and attenuate BDNF. Some pandemic-era studies reported increased amygdala and other regional volumes post-lockdown in healthy subjects that waned over time, while meta-analyses found initial symptom increases early in the pandemic with later declines toward baseline. Cross-sectional imaging studies often show small structural differences in MDD versus controls, but longitudinal differences are less consistent. Thus, evidence on neurobiological changes linked specifically to COVID-19 restriction measures, especially in clinical populations, remains sparse and mixed.

Design: Prospective, naturalistic, quasi-experimental longitudinal study with three measurement time points during COVID-19 restriction phases in Austria (TP1: Sep 9–Nov 12, 2020; TP2: Dec 9–22, 2020; TP3: May 5–Jun 10, 2021). Participants: Recruited from prior departmental studies (NCT02810717, NCT02753738). Final analyzed sample: 28 healthy individuals (HI) and 18 patients with recurrent MDD. Inclusion/exclusion: Excluded suspected/confirmed SARS-CoV-2 infection, new severe somatic/neurological disorder (or any mental disorder for HIs), MRI contraindications, pregnancy/breast-feeding. Baseline included SCID-IV, physical exam, blood draw. Clinical care proceeded per clinical indication; psychopharmacology was changed in ten patients during the study; some received outpatient psychotherapy; no inpatient, ECT, or TMS during study. Psychometrics: At each time point, online BDI-II (last 2 weeks) and PSQ-20 (subscales: worries, tension, joy, demands; last 4 weeks) were administered via EvaSys. MRI acquisition: 3T Siemens MAGNETOM PRISMA, 64-channel head coil; T1-weighted MPRAGE (TE=1800 ms, TR=2.37 ms, 208 slices, 288×288 matrix, slice 0.85 mm, in-plane 1.15×1.15 mm). Processing: FreeSurfer 7.1 recon-all; 34 cortical ROIs per hemisphere and subcortical regions. Longitudinal processing with within-participant template; quality control via visual inspection. Total cortical volume and surface area corrected for eTIV; cortical thickness not corrected for eTIV. Serum BDNF: Serum collected, processed (centrifuged 1500×g, 15 min), stored at −80°C; measured using Biosensis Mature BDNF Rapid ELISA (dilution ~1:100), absorbance at 450 nm with 690 nm correction, duplicates averaged; intra-/inter-assay CVs assessed. Statistics: Linear mixed models (LMM) evaluated effects of time (3 levels), group (MDD vs HI), hemisphere, and ROI on cortical thickness, surface area, and subcortical volumes, with subject as random intercept; interaction terms up to four-way included then pruned if non-significant. Covariates: age, sex, medication status (HI, no medication change, medication change). A priori ROIs: amygdala, hippocampus, ACC (caudal, rostral), prefrontal cortex regions (superior frontal, rostral/caudal middle frontal, pars opercularis/triangularis/orbitalis, lateral/medial orbitofrontal, frontal pole). Exploratory analyses included all cortical ROIs and subcortical regions (thalamus, caudate, putamen, pallidum, hippocampus, amygdala, accumbens) plus cerebellar cortex. Additional LMMs for BDI-II, PSQ-20 subscales, and sBDNF with time and group as fixed factors, subject random intercept, adjusting for age, sex, and medication. Correlations: Pearson correlations between sBDNF and hippocampus/amygdala volumes and PFC/ACC cortical thickness at each time point; also correlated change scores (TP2−TP1, TP3−TP1, TP3−TP2) for sBDNF with grey matter changes. Correlated sBDNF with BDI-II and PSQ-20 subscales overall and by group. Non-parametric Mann–Whitney U tests used for non-normal psychometrics. Multiple testing: Bonferroni corrections applied separately for imaging models (a priori and exploratory), sBDNF, and psychometric scores. Software: SPSS v24 and jamovi v2.3.

- Sample: 28 healthy individuals (HI) and 18 recurrent MDD; sex not different between groups; MDD older (mean age 37±10.03) than HI (27.96±5.12). Two MRI measurements from MDD participants were excluded due to segmentation errors. - Structural MRI (a priori ROIs: ACC, PFC, hippocampus, amygdala): No main effect of group on cortical thickness (F=0.19, p>0.1), no main effect of time (F=0.29, p>0.1), and no time×group interaction (F=0.28, p>0.1). Similar null effects for surface area and for hippocampus/amygdala volumes (group F=0.78, p>0.1; time F=0.69, p>0.1; time×group F=0.05, p>0.1). - Exploratory imaging across all 34 cortical ROIs: No main effects of time (thickness F=0.16, p>0.1; surface area F=0.69, p>0.1), no group effects (F=0.03, p>0.1; F=1.28, p>0.1), and no time×group interactions (F=0.43, p>0.1; F=0.39, p>0.1). Subcortical volumes plus cerebellar cortex: no effects of time (F=0.06, p>0.1), group (F=2.59, p>0.1), or interaction (F=0.27, p>0.1). ROI-wise models confirmed null effects (all p>0.1). - Psychometrics: BDI-II showed a significant main effect of group (F=30.89, p<0.001) with higher scores in MDD, but no time effect (F=1.36, p>0.1) and no time×group interaction (F=0.16, p>0.1). PSQ-20 subscales: no time effects (all p>0.1) and no time×group interactions (all p>0.1). Group effects present for Worries (F=19.19, p<0.001), Tension (F=34.44, p<0.001), and Joy (F=12.05, p=0.001), but not for Demands (F=2.13, p=0.15). - sBDNF: No main effect of time (F=0.732, p>0.05) and no time×group interaction (F=0.633, p>0.05). Mean sBDNF (ng/ml): MDD 11.16±3.3 (TP1), 13.28±3.1 (TP2), 14.62±6.4 (TP3); HI 11.57±3.3 (TP1), 11.54±2.6 (TP2), 11.94±3.3 (TP3). No significant correlations between sBDNF and brain measures, or between sBDNF and BDI-II/PSQ-20 (r range approximately −0.38 to 0.27; all p>0.1).

Contrary to hypotheses, repeated COVID-19 lockdowns over approximately nine months did not produce detectable longitudinal changes in grey matter structure (cortical thickness, surface area, subcortical volumes), serum BDNF levels, or psychometric measures in either recurrent MDD patients or healthy individuals. Cross-sectional differences persisted, with MDD patients showing elevated depressive symptoms and higher stress (PSQ-20 worries, tension) and lower joy compared to healthy individuals, but these did not change over time. The absence of longitudinal effects may indicate that the restriction measures did not surpass a traumatic threshold sufficient to induce measurable structural neuroplastic changes, that individuals adapted over time, or that ongoing treatments (e.g., antidepressants, psychotherapy) mitigated neurobiological and symptom fluctuations. Differences from reports of volumetric changes post-lockdown in healthy samples may reflect timing, context (Austria’s moderate incidence and mortality), heterogeneity in lifestyle and stress exposure, and clinical treatment effects. Overall, findings align with meta-analyses showing initial pandemic-related symptom increases with subsequent stabilization, supporting a resilience or adaptation perspective rather than progressive deterioration.

This multimodal longitudinal study, the first in Austria to assess neurobiological and psychometric effects of multiple COVID-19 lockdowns in recurrent MDD and healthy individuals, found no significant longitudinal changes in brain structure, serum BDNF, or depressive/stress symptoms across three time points from late 2020 to mid-2021. Persistent group differences were observed (higher depression and stress, lower joy in MDD), but without temporal progression. Results suggest adaptation and resilience rather than cumulative adverse effects of repeated lockdowns on the measured parameters. These findings contribute to the evidence base for longitudinal mental health impacts of restriction measures and can inform future meta-analyses and studies on trajectories, moderators, and mechanisms of resilience.

- Naturalistic, observational design without a non-exposed control group and with first measurements approximately five months after the initial lockdown, potentially missing early adaptational neuroimmune/neuroplastic changes. - Specific clinical subgroup (recurrent MDD), limiting generalizability to first-episode or treatment-naïve patients. - Small sample size typical of neuroimaging studies may reduce power to detect subtle brain or BDNF changes. - Possible bias from ongoing clinical care and study-related contact, including access to specialized treatments and medication adjustments, which may have attenuated symptom variability. - Lack of detailed data on compliance with restriction measures and digital phenotype/lifestyle metrics (e.g., mobility, screen time), which could mediate effects. - Unclear peri- vs post-traumatic timing relative to the evolving pandemic; measurements may reflect adapted or adapting states. - Context-specific pandemic burden (moderate incidence/mortality in Austria) may limit generalizability to other regions with different pandemic trajectories.