Risk refers to a specific form of uncertainty when information about the probability of each possible outcome is known (von Neumann and Morgenstern, 1944). Risk preference, also known as “risk attitude”, “risk tolerance” and “sensitivity to risk”, is usually referred to as the propensity to engage in risky activities (Dohmen et al., 2011; Hertwig et al., 2019). People with a high risk preference tend to engage in more risky behaviors, including automobile speeding, alcoholism, and promiscuity, which not only result in significant economic losses but also impact social order and public safety (Bolton, 1992; Karlsson Linnér et al., 2019; Moss, 2013). Conversely, individuals with a lower risk preference are more inclined toward a safe outcome than a risky but potentially better outcome (Anderson and Mellor, 2008).

Current evidence indicates a close association of risk preference with major depressive disorder (MDD) and bipolar disorder (BP). Chong Chen et al. reported that individuals with increasing depressive symptoms exhibited enhanced risk aversion in the experience-based task but not the description-based task compared to those with decreasing depressive symptoms (Chen et al., 2022a). Moreover, Xinlei Ji et al. reported greater activation in the left dorsolateral prefrontal cortex in MDD patients with suicide attempts, which means that suicide attempt patients would be hypersensitive to loss, contributing to conservative decision-making when faced with substantial and extreme choices (Ji et al., 2021). Conversely, Qinyu Liu et al. discovered that MDD patients exhibited higher risk preference and lower ability to learn and adapt from within-task observations than healthy control individuals (HCs) (Liu et al., 2022). In contrast to MDD, BP patients often exhibit consistently elevated risk preferences, including increased risky staying and risky shifting, which contribute to their riskier decisions in the Iowa Gambling Task (van Enkhuizen et al., 2014). This finding was also confirmed in animal experiments conducted by Jared W Young et al., who observed increased risk-taking behavior in dopamine transporter knockdown mice, which were considered a BD manic model (Young et al., 2011).

Overall, despite some studies exploring the relationship between risk preference and emotional disorders, the results have been heterogeneous and unable to determine causality. In addition, there are several concerns regarding this relationship. First, prior studies often considered risk preference as a part of the neurocognitive process, rather than an independent factor, in the exploration of the relationship between risk preference and mood disorders (Milienne-Petiot et al., 2017b; van Enkhuizen et al., 2014). Second, the results regarding the link between risk preference and mood disorders are inconsistent in prior studies. Third, previous studies have often been influenced by small sample sizes, high heterogeneity, and the presence of behavioral, social, and genetic confounding factors.

Mendelian randomization (MR) is a popular method to explore the mutual causality between diseases and risk factors using genetic variants and large samples, with the advantage of avoiding conventional confounding biases in observational studies (Sekula et al., 2016). Because genetic variants are assigned randomly before birth, they are relatively independent of environmental factors; thus, they can be identified before disease onset, with minimization of residual confounding and reverse causality, which are limitations to conventional observational studies. As an extension to the MR approach, two-sample MR analysis allows the direct use of the summary statistics of genome-wide association studies (GWASs) without the need to analyze data at an individual level. At the same time, two-sample MR could combine SNP exposure and SNP outcome into a single causal estimate from independent GWAS, which could also avoid the classic problems of observational research. Therefore, exploring the bidirectional relationship between risk preference and mood disorders using a large GWAS cohort not only helps us further understand the causal link between risk preference and mood disorders but may also help identify potential markers for the early identification of mood disorders.

The aim of this study was to examine the bidirectional relationships between risk preference and mood disorders. Regarding the disparity in prior findings, we posed the following hypothesis: risk preferences vary among individuals with different mood disorders. To verify our hypothesis, we conducted a two-sample bidirectional MR analysis with the use of publicly available GWAS data to assess the potential causal association between risk preference and mood disorders.