Major Depressive Disorder (MDD) is a prevalent condition significantly impacting quality of life and psychosocial functioning. While factors like poverty and family history increase MDD risk, leisure/social and physical activities have emerged as potentially controllable risk factors. Meta-analyses consistently show that physical activity is associated with a reduced risk of MDD. Similarly, studies suggest that leisure activities and strong social relationships are linked to lower depressive symptoms and prevalence of MDD. However, the bidirectional relationship between activity and MDD, where MDD symptoms might lead to decreased participation in activities, remains unclear. Traditional observational studies are limited in disentangling this complex relationship. Given the link between physical activity, obesity, and MDD, obesity traits could be potential mediators in this association. Mendelian Randomization (MR), using genetic variants as instrumental variables, offers a method to infer causal relationships, accounting for confounding and reverse causation. Previous MR studies have shown a protective relationship between activity and MDD, but the role of inactivity remains under-explored. This study aims to examine the bidirectional causal effects between MDD and different types of leisure/social and physical activity, and inactivity, while investigating the mediating effects of obesity measures and brain imaging phenotypes.

Existing research indicates a strong correlation between physical activity and reduced MDD risk, independent of age and geographic location. Meta-analyses of prospective cohort studies have consistently supported this protective effect. Furthermore, engagement in leisure activities and the presence of strong social support networks have been associated with lower rates of depressive symptoms and MDD. Conversely, MDD is often characterized by reduced energy levels and a loss of interest in previously enjoyable activities, suggesting a potential feedback loop where MDD leads to decreased social and physical activity. However, observational studies cannot definitively establish causality due to the presence of confounding factors and potential reverse causation. This necessitates the use of methodologies such as Mendelian Randomization to infer causal relationships.

This two-sample Mendelian Randomization (MR) study leveraged publicly available genome-wide association study (GWAS) summary statistics. The data included information on MDD, various types of leisure/social activities (including ‘leisure/social inactivity’), physical activities (including ‘physical inactivity’), body mass index (BMI), body fat percentage (BFP), and brain imaging-derived phenotypes (IDPs). The study employed a rigorous quality control process for selecting single nucleotide polymorphisms (SNPs) as instrumental variables (IVs), including criteria for association strength, linkage disequilibrium, minor allele frequency, and the absence of association with plausible confounders (like alcohol consumption and smoking). Weak instrument bias was assessed using F-statistics. The primary analysis used inverse-variance weighted (IVW) regression, complemented by MR-Egger regression, weighted median, simple mode, and weighted mode methods. Sensitivity analyses were conducted to assess robustness and pleiotropy, including random-effects IVW, MR-Egger intercept test, leave-one-out analysis, and MR-PRESSO to detect and correct for outlier SNPs. To account for overlapping samples in some datasets, the MR-lap package was employed. Two-step MR analyses were performed to assess mediation effects of obesity-related measures (BMI, BFP) and IDPs on the association between activity/inactivity and MDD.

The MR analysis revealed a protective causal effect of several types of activity, including attending sports clubs/gyms, strenuous sports, heavy DIY, and other exercises, on MDD risk. Conversely, leisure/social and physical inactivity showed a significant causal association with increased MDD risk. The odds ratios for the effect of inactivity on MDD were substantial. Bidirectional causal relationships were identified for several activities, meaning that these activities reduced MDD risk, but conversely, having MDD reduced the likelihood of engaging in those activities. Two-step MR analysis indicated partial mediation of the effect of inactivity on MDD risk through increased BMI and BFP. Analysis with IDPs as mediators showed some IDPs played mediating or masking roles in the relationship between inactivity and MDD. Specifically, certain brain imaging measures appeared to either partially mediate the effect of inactivity on MDD or mask the overall effect, depending on the measure and type of inactivity.

The findings support a bidirectional causal relationship between social/physical activity and MDD. Engagement in various activities has a protective effect against MDD, while MDD itself leads to reduced participation in those activities. This suggests a potential vicious cycle where inactivity increases the risk of MDD, and MDD further reduces motivation for activity. The mediating roles of BMI/BFP and specific brain imaging phenotypes highlight potential biological pathways connecting inactivity and MDD. The results corroborate previous research demonstrating the protective effects of physical activity against depression and extend these findings by examining the causal role of inactivity and the potential mediating mechanisms. The study's strength lies in its rigorous use of MR methods and comprehensive analysis of multiple activity types, including inactivity.

This study provides strong evidence for a bidirectional causal association between social/physical activity and MDD. Engaging in regular physical and social activities reduces MDD risk, while MDD itself contributes to reduced activity levels. The mediating effects of obesity and brain imaging phenotypes offer valuable insights into the underlying mechanisms. These findings support public health initiatives promoting physical and social activities for MDD prevention and management. Future research should focus on longitudinal studies to track the trajectory of activity and MDD over time, using objective measures of activity to further refine these findings. Investigating the underlying biological pathways could pave the way for novel therapeutic approaches targeting the activity-MDD relationship.

The study acknowledges several limitations. While SNPs were strongly associated with the exposures, they are not perfect proxies for the actual behaviors. The use of self-reported activity data might introduce subjective bias. The generalizability of findings is limited by the predominantly European ancestry population in the datasets. The study relied on relaxing statistical thresholds in certain instances to ensure sufficient SNPs for analyses. The potential for overlapping samples between some datasets is acknowledged, and mitigating steps were taken, such as the use of strong instruments, subgroup analyses, and MR-lap for bias correction. The use of binary activity variables limits the precision in determining risk parameters.