Introduction

Depressive disorders are highly prevalent. One-year occurrence of clinical depression (major depressive disorder) is approximately 6% and the risk of depression during the whole life is 15–18%, while subclinical depression is even more frequent compared to major depressive disorder1,2. Additionally, coronavirus pandemic had an impact on mental health3. Depression is almost twice as common in women than men and can present as complex manifestations including affected mood, cognitive disturbance, and somato-vegetative dysfunction1,2. Typical, melancholic depressive symptoms are manifested by loss of appetite and/or weight, while individuals with atypical depressive symptoms can present opposite symptoms, such as increased appetite and/or weight gain1,4.

A meta-analysis which included 183 studies examined the association between the indices of body weight and depression. The authors showed that both underweight and obesity significantly increased the risk of depression5. A possible U-shaped relationship between body mass index (BMI) and depression can indicate heterogeneity within this disease. The association between obesity and depression is bidirectional. A meta-analysis of longitudinal studies showed that obesity at baseline increased the risk of onset of depression at follow-up, while depression increased the odds for developing obesity later in life6. Metabolic and neurobiological mechanisms by which obesity is connected with depression are complex and little known. According to the current literature, psychiatric consequences of obesity can derive from diet rich in sugar and saturated fat, lack of physical activity, and body composition changes7.

According to the literature, depression is also associated with body composition8,9. Previous studies showed that abdominal fat accumulation has a particularly negative impact on depressive symptoms in women9,10,11. Lee et al. observed that depressive mood is related with increased visceral fat mass, but not with subcutaneous adipose tissue11. Everson-Rose et al. examined middle-aged women and showed that visceral fat volume is higher in women with depression, particularly in overweight and obese women12. Guedes et al. demonstrated a significant inverse correlation between total fat-free mass and severity of depressive symptoms in individuals with metabolic syndrome13. Adipose and muscle tissues secrete adipokines, myokines and cytokines which can pass through blood–brain barrier and impact brain structures14.

Previous studies showed that depressive symptoms can be connected with changes in dimensions and function of frontal lobe, hippocampus, stratum, insula and amygdala1,15,16. These brain structures are responsible for emotion, memory, motivation, attention, executive function, regulation of systemic metabolism, and behaviour related with food intake7,15,16,17. Different clinical or biological characteristics of depression could be related to specific brain structure alterations18.

According to the literature, obesity might be related to the changes in brain structure volumes and thickness19. Data from the United Kingdom Biobank study (18.7% participants with BMI ≥ 30 kg/m2) showed that higher BMI was related to lower grey matter volume. Obesity was associated with lower volume of putamen, pallidum, and nucleus accumbens regions20. Raji et al. observed that elevated abdominal fat (both subcutaneous and visceral) predicted lower brain volume21. Another study from United Kingdom Biobank analysed the relationship between fat distribution and brain volume. Central obesity, defined by waist-hip ratio, dual-energy X-ray absorptiometry (DXA) scans, and abdominal magnetic resonance imaging (MRI), negatively correlated with grey matter volume22. Kilgour et al. analysed 53 articles and observed that there can be a positive association between volumes of selected grey matter regions (right temporal lobe and bilateral ventromedial prefrontal cortex) and muscle size23.

There is limited data concerning brain structure volumes in subjects with normal-/underweight and depressive symptoms. Additionally, few studies analysed the differences in brain structure volumes between women with low body weight or high body weight and depressive disorders.

The present study aimed to examine the association of specific brain structures volumes with depressive symptoms and body composition in women with obesity/overweight and normal-/underweight women. We hypothesized that different brain structures can be associated with depressive symptoms severity and body composition in women with BMI ≥ 25 kg/m2 and with BMI < 25 kg/m2.

Materials and methods

Study population

The study was a part of the Bialystok PLUS population study. Ethical approval for the study was obtained from the Ethics Committee of the Medical University of Bialystok, Poland (approval number: R-I-002/108/2016). All procedures performed in the study were in accordance with the Declaration of Helsinki and all participants gave written informed consent. The study recruitment was described previously24. Overall, 1134 individuals were recruited and examined as a population cohort of Białystok PLUS study between August 2017 and February 2022. The flow of study participants into this analytic sample is described in Fig. 1. Men were excluded from this study due to distinct clinical manifestation of depressive symptoms compared to women. People above 70 years of age were not included in this analysis due to higher prevalence of cognitive disorders in this age group than in younger subjects. Women aged 20–70 years were included in this study. Finally, taking into account the exclusion criteria, we included 265 women. The exclusion criteria were: 1/history of stroke, 2/transient ischemic attack, 3/multiple sclerosis, 4/epilepsy, 5/dementia, 6/Parkinson’s disease, 7/schizophrenia, 8/bipolar disorder, 9/use of antidepressive or antipsychotic medications, 10/decompensated hypothyroidism or hyperthyroidism (thyroid-stimulating hormone, TSH, < 0.1 or > 5 µU/ml), 11/type 1 diabetes, 12/oral glucocorticosteroid use, 13/acute infection (high-sensitivity C-reactive protein, hsCRP, ≥ 10 mg/l), 14/alcohol abuse (≥ 8 points in Alcohol Use Disorder Identification Test, AUDIT), 15/illicit or recreational drug use in last year, 16/incomplete Beck Depression Inventory (BDI), 17/contraindications or lack of consent for MRI.

Figure 1
figure 1

Selection of the study groups from the population cohort study.

Full size image

Depressive symptoms assessment

To assess depressive symptoms in the present study, we used BDI test, which is a commonly available self-report rating inventory for diagnostic screening. The study participants completed the Polish version of the BDI, which measured the severity of depressive symptoms within the preceding month. In our previous studies, the cut-off point for the presence of depressive symptoms was ≥ 10 points in BDI (≥ 21 points in BDI for clinical depressive symptoms and 10–20 BDI score for subclinical depressive symptoms)25,26. According to the published data, subclinical depressive symptoms are associated with structural brain changes similar to those in clinical depression27.

Anthropometric and sociodemographic parameters

All study participants underwent general physical examination. Anthropometric measurements including height and weight were taken. BMI was calculated as body weight in kilograms divided by height in meters squared. Obesity was recognized at BMI ≥ 30 kg/m2, overweight 25–29.99 kg/m2, normal weight 18.5–24.99 kg/m2, underweight < 18.5 kg/m2.

We used self-report questionnaires to collect data on illicit or recreational drugs. Evaluation of alcohol consumption and alcohol-related problems was done by the self-report version of a 10-item screening test developed by the World Health Organization—AUDIT. All individuals received AUDIT questionnaires and were asked to complete the survey. Body composition was analysed using DXA (Lunar iDXA, GE Healthcare, Chicago, Illinois, United States) at the Clinical Research Centre, Medical University of Bialystok. Body composition analyses were performed and controlled by qualified staff members of Bialystok PLUS study. The equipment was calibrated before each examination. Study participants were positioned on the examination table in a supine position, with their feet secured with an adjustable strap and their arms by their side. Using this method, body composition including body fat and fat-free tissue was estimated. For each area of the body (trunk, arms, and legs), DXA assessed fat-free mass and fat mass with the precision (coefficient of variation) of 2.0 and 8.0%, respectively. In our study, we used measurements of total fat mass (kg), total fat-free mass (kg), percentage of android fat (%, android fat mass divided by total fat mass and multiplied by 100), percentage of gynoid fat, (%, gynoid fat mass divided by total fat mass and multiplied by 100), percentage of visceral fat, (%, visceral fat mass divided by total fat mass and multiplied by 100).

Brain imaging and data processing

MRI scans were acquired with a 3.0 T Siemens Biograph nMR scanner. Brain analysis was conducted with the Portable Batch System server. Morphometric analysis of brain structure was conducted with Freesurfer version 7.2.0 (available at https://surfer.nmr.mgh.harvard.edu/) using a recon-all stream28,29. Freesurfer includes modules to segment cortical and subcortical brain structures. In this study, we analysed the volumes of depression-related brain structures: superior frontal gyrus, middle frontal gyrus (caudal + rostral middle frontal), anterior cingulate gyrus (caudal + rostral anterior cingulate), insula, hippocampus, amygdala, and nucleus accumbens. The means of each brain structure’s volumes were analysed (right + left/2)30,31,32,33,34,35. Additionally, we calculated estimated total intracranial volume (eTIV). The reconstruction and segmentation were visually inspected.

Statistical analysis

Statistical analyses were performed using Statistica 13.0 (Statsoft, OK, USA) and STATA 16 (StataCorp, TX, USA). The variables were tested for normal distribution using the Shapiro–Wilk test. Due to the non-normal distribution of data, all values were expressed as median and interquartile range. The comparisons between two groups were performed using Mann–Whitney U test for continuous variables and Chi-squared test for nominal variables. Multivariate linear regression was used to assess the relationship between brain structure volumes and BDI score after controlling for age and eTIV separately in women with obesity/overweight and in women with normal-/underweight. Multivariate linear regression was used to evaluate the association between brain structure volumes and body composition (total fat mass, total fat-free mass, percentage of android fat, percentage of gynoid fat, percentage of visceral fat). The level of significance was set at p < 0.05.

Results

The study sample consisted of 265 women with median age 47 years and BMI 24.79 kg/m2. The analysed population was divided according to BMI into a group of women with obesity/overweight (n = 131) and with normal-/underweight (n = 134). Median age and BMI of women with BMI ≥ 25 kg/m2 were 53 years and 28.91 kg/m2, whereas median age and BMI of women with BMI < 25 kg/m2 were 41 years and 22.00 kg/m2. The group of women with obesity/overweight had higher age and BMI compared to women with normal-/underweight (p < 0.001 and p < 0.001). Additionally, the group of women with BMI ≥ 25 kg/m2 had higher total fat mass, total fat-free mass, percentage of android and visceral fat and lower percentage of gynoid fat compared to women with BMI < 25 kg/m2 (all p < 0.001). The presence of depressive symptoms (BDI ≥ 10 points) was comparable in the group with obesity/overweight and normal-/underweight women (28.24% vs. 27.61%, p = 0.909) (Table 1). Only nucleus accumbens volume was significantly decreased in women with obesity/ overweight compared to normal-/underweight group (p = 0.041) (Table 1).

Table 1 The characteristics of total population, women with overweight/obesity, and women with normal-/ underweight.
Full size table

Analysis of the groups of women with BMI ≥ 25 kg/m2

In the group of women with BMI ≥ 25 kg/m2, nucleus accumbens volume was inversely associated with BDI score as continuous variable after adjustment for age and eTIV (Table 2). In this group, 37 women had BDI score ≥ 10 points. We used multivariate regression analysis with nucleus accumbens volume as a dependent variable and the presence of depressive symptoms (BDI score ≥ 10) as an independent variable, adjusted for age and eTIV. We observed an inverse relationship between the volume of nucleus accumbens and the presence of depressive symptoms (≥ 10 points in BDI) (β = −0.142, p = 0.031) (Supplementary Table S1). In the group of women with BMI ≥ 25 kg/m2, 6 women had BDI score ≥ 21 and 31 subjects 10–20 points. To assess distinct relations between nucleus accumbens volume and clinical or subclinical depressive symptoms, in the next step we performed multivariate regression analysis with the presence of depressive symptoms divided into clinical depressive symptoms (BDI score ≥ 21) and subclinical depressive symptoms (BDI score 10–20) as an independent variable, adjusted for age and eTIV. We observed an inverse relationship between the volume of nucleus accumbens and the presence of ≥ 21 points in BDI (β = −0.213, p = 0.008) but not a significant association between the volume of nucleus accumbens and the presence of BDI score 10–20 (β = −0.140, p = 0.083) (Supplementary Table S2).

Table 2 Multivariate regression analysis results with each brain structure volume as a dependent variable and BDI score as an independent variable in the group of women with BMI ≥ 25 kg/m2 (n = 131).
Full size table

In addition, we used multivariate linear regression with nucleus accumbens volume as a dependent variable and each body composition parameter and BDI score as independent variables, adjusting each model for age and eTIV. In this group, nucleus accumbens volume was inversely associated with the percentage of visceral fat and BDI score (Table 3, Supplementary Tables S3–S8).

Table 3 The associations of nucleus accumbens volume (dependent variable) with body composition and BDI score (independent variables) in women with BMI ≥ 25 kg/m2 (n = 131).
Full size table

Analysis of the groups of women with BMI < 25 kg/m2

In the group of women with BMI < 25 kg/m2, insula volume was inversely associated with BDI score as continuous variable after adjustment for age and eTIV (Table 4). In this group, 37 women had BDI score ≥ 10 points. We used multivariate regression analysis with insula volume as a dependent variable and age, eTIV, and the presence of depressive symptoms (BDI score ≥ 10) as independent variables. We observed an inverse relationship between the volume of insula and the presence of ≥ 10 points in BDI (β = −0.198, p = 0.016) (Supplementary Table S9). In this group of women with BMI < 25 kg/m2, 6 women had BDI score ≥ 21 and 31 subjects 10–20 points. Similar to the analysis described above, we then performed multivariate regression analysis with insula volume as a dependent variable and the presence of depressive symptoms divided into clinical depressive symptoms (BDI score ≥ 21) and subclinical depressive symptoms (BDI score 10–20) as an independent variable, adjusted for age and eTIV. We observed an inverse relationship between the volume of insula and the presence of ≥ 21 points in BDI (β = −0.155, p = 0.019), but not a significant association with the presence of BDI score 10–20 (β = −0.099, p = 0.131) (Supplementary Table S10).

Table 4 Multivariate regression analysis results with each brain structure volume as a dependent variable and BDI score as an independent variable in the group of women with BMI < 25 kg/m2 (n = 134).
Full size table

In the next step, we used multivariate linear regression with insula volume as a dependent variable and each body composition parameter and BDI score as independent variable, adjusting each model for age and eTIV. In normal-/underweight women, insula volume was positively connected only with total fat-free mass and negatively with BDI score (Table 5, Supplementary Tables S11–S16).

Table 5 The associations of insula volume (dependent variable) with body composition and BDI score (independent variables) in women with BMI < 25 kg/m2 (n = 134).
Full size table

Discussion

The main finding of our study was that the associations between brain structure volumes, body composition and depressive symptoms are distinct in women with BMI ≥ 25 kg/m2 and BMI < 25 kg/m2. Depressive symptoms in women with obesity/overweight can be associated with nucleus accumbens volume, while in women with normal-/underweight they show an association with insula volume. In women with obesity/overweight, nucleus accumbens volume was inversely associated with the content of visceral fat and the depressive symptoms’ severity, while in women with normal-/underweight, insula volume was positively associated with total fat-free mass and negatively with the depressive symptoms’ severity.

The present study was performed in a population of women with a wide range of BMI from mild underweight to severe obesity. In agreement with previous observations from general population, higher prevalence of obesity or overweight was shown in elderly compared to young women36. In our study, obesity/overweight was associated with decreased nucleus accumbens volume. García-García et al. analysed the dataset of United Kingdom Biobank (participant ages range from 40 to 80 years), finding a negative association between BMI and nucleus accumbens volume37. This brain structure is a region in the ventral striatum which receives dopaminergic inputs from the ventral tegmental area in the midbrain and belongs to mesolimbic dopamine system, which is a part of brain reward circuit38. A number of studies concern the relationship between nucleus accumbens volume and obesity or depression. However, the data regarding women with these both diseases is limited. Our analysis demonstrated that BDI score was inversely connected with nucleus accumbens volume only in women with obesity/overweight. Blunted sensitivity of brain reward circuit in people with depression can lead to compensating mechanisms, such as quick rewarding by eating food rich in fat and sugar7,38. In consequence, atypical depressive symptoms, such as increased appetite and weight gain, can be observed in women with obesity/overweight.

According to the published data, depressive disorders in people with obesity are associated with inflammation in the brain: the activation of microglial cells, an increase in the production of proinflammatory cytokines in the brain, release of reactive oxygen species and, ultimately, neurotoxic effects in the mesolimbic system7,39,40. Peripheral low-grade inflammation is also observed in people with chronic psychological stress and depression38,41. Visceral fat accumulation is also associated with persistent low-grade inflammation40. In our previous study, we observed increased mass of visceral fat in women with depressive symptoms25. In this study, percentage of visceral fat and the presence of depressive symptoms were inversely connected with nucleus accumbens volume in women with obesity/overweight. The results of a large population-based cohort study from the United Kingdom Biobank showed a negative association between central obesity and volume of nucleus accumbens in people with obesity22. Impaired blood–brain barrier integrity in mood disorders and obesity can promote increased permeation of peripheral immune cells from visceral fat to brain and elicit sustained neuroinflammatory actions that ultimately reduce volume of structures associated with depressive-like behaviour, such as nucleus accumbens7,38. In a previous study, we observed that in women with depressive symptoms, excess visceral fat deposits can be connected with insulin resistance25. Insulin modulates dopamine release, and insulin receptors are expressed in structures of mesolimbic system. According to the published data, peripheral insulin resistance can expand to central nervous system and lead to brain insulin resistance42,43. The combination of depressive symptoms and excess visceral fat might possibly influence the volume of nucleus accumbens in women with obesity/overweight.

Only in the group of women with normal-/underweight, we observed an inverse association between the volume of insula and BDI score. Insula has a key role in cognitive and emotional functions, motivational processes and interoception (neural mapping of body states)17. The insula-frontal functional connectivity plays a role in cognitive processes during depression17. Our findings add new information to data from literature and they suggest that insula volume might be connected with total fat-free mass and the depressive symptoms severity only in women with normal-/underweight. We suspect that one possible explanation of our results is a tendency to being physically inactive in persons with depression1. Killgory et al. found that total number of minutes of weekly physical exercise was positively associated with grey matter volume within a region of the posterior left insula44. Peter et al. showed that aerobic capacity positively correlated with grey matter density in the right anterior insula45. One explanation of this phenomenon could be the secretion capacity of skeletal muscles14. Myokines are secreted from muscle cells in response to muscle contractions14. According to the published studies, myokines penetrate the blood–brain barrier to enhance brain-derived neurotrophic factor (BDNF) production14. Mood disorders are connected with decreased BDNF level in central nervous system1. Serum BDNF positively correlated with cortical thickness and volume in multiple brain regions in minor depression46. Additionally, we observed a positive relationship between plasma BDNF and insulin sensitivity in previous studies47. Another possible explanation of our results in the group of women with normal-/underweight is the observation regarding frequent occurrence of eating disorders in people with depression48. Curzio et al. analysed brain structure alterations in adolescents with anorexia nervosa and observed reduced grey matter volumes in left insula and both frontal lobes in this group compared to healthy controls49. The observed brain structure alterations in the group of women with normal-/underweight may be related with typical melancholic symptoms of depression in this group. The presence of melancholic depression in women can cause BMI decrease in long-term observation50. Based on the results of the previous studies, melancholic depression is associated with a reduction of insula volume and caudal anterior cingulate cortex thickness17,51.

Lamers et al. observed different roles of hypothalamic–pituitary–adrenal axis function, inflammation, and metabolic syndrome in melancholic versus atypical depression. Individuals with melancholic depression had higher level of awakening saliva cortisol and diurnal cortisol slope compared with persons with atypical depression and with controls. People with atypical depression had significantly higher levels of inflammatory markers in blood (C-reactive protein, interleukin-6, tumor necrosis factor-α), BMI and waist circumference than persons with melancholic depression and controls4. In women with obesity/overweight and normal-/ underweight, different depressive symptoms can be dominant. Atypical depressive symptoms can mainly be observed in women with obesity/overweight, while melancholic depressive symptoms in women with normal-/underweight4,7,50. The results of a previous study showed that visceral fat volume was connected with depression, but this relationship was weaker in normal-/underweight women compared to overweight and obese women12. This might explain why we did not observe a significant association between nucleus accumbens volume and depressive symptoms severity in women with BMI < 25 kg/m2. Women with normal-/underweight and depressive symptoms can present hypercortisolemia, which is associated with loss of muscle mass. Depression and obesity cooccurrence is often defined by atypical features, which is mostly associated with normal or decreased cortisol levels7. However, the discussed reasons could not fully elucidate the pathomechanism of this different relationship in the study groups. Further studies will be needed to understand the associated mechanism.

There are several limitations to the present study. The main limitation is a small sample size, which did not permit us to separate four BMI groups (underweight, normal weight, overweight, obese women). Additionally, we used BDI questionnaires to define the group of women with depressive symptoms and did not perform a comprehensive psychiatric evaluation. Moreover, in our analyses we only used measurements from structural, and not functional, magnetic resonance imaging. The main strength of our study is the analysis of women without antidepressive and/or antipsychotic drugs.

In conclusion, depressive symptoms might be associated with nucleus accumbens volume in overweight/obese women, while in normal-/ underweight women—with alterations in insula volume. These relationships can be connected with body composition. Further structural and functional research is needed to investigate which pathway disturbance can be responsible for depressive symptoms in women with underweight, normal weight, overweight and obesity. In view of the current data, few studies have investigated an association between low body weight and depressive symptoms. The understanding of depressive disorders in different groups of women can help choose proper medication or develop targeted therapies.