Intracranial aneurysms (IAs) are widespread life-threatening diseases (Vlak et al., 2011). Rupture of IAs results in aneurysmal subarachnoid hemorrhage (aSAH), a form of stroke that accounts for 5% of all strokes. Half of aSAH patients are younger than 55 years and a third of the patients die within the first few days to weeks of bleeding. Most survivors suffer from long-term disability or cognitive impairment (Nieuwkamp et al., 2009).

Among patients with IAs, psychiatric traits usually co-exist. Observational studies revealed IAs and psychiatric traits like insomnia (Colledge et al., 2017; Zhang et al., 2020), mood instability (McGowan et al., 2021), anxiety disorder (Lambiase et al., 2014; Lemos et al., 2020; Zhai et al., 2020), major depressive disorder (MDD) (Henderson et al., 2013; Tang et al., 2020; Zhai et al., 2020), subjective wellbeing (Tabernero et al., 2019), attention deficit/hyperactivity disorder (ADHD) (Jeong et al., 2015), autism spectrum disorder (ASD), bipolar disorder (BIP) (Prieto et al., 2014), schizophrenia (SCZ) (Liang et al., 2016), and neuroticism (Marijnissen et al., 2014) were bidirectionally correlated. For example, Marijnissen et al. found depression was a risk factor for stroke, including IAs, via a 9-year cohort study (Marijnissen et al., 2014). Inversely, among the 200 unruptured IAs treated by endovascular intervention, 31 (15.5%) had depression and 34 (17.0%) had anxiety after discharge (Zhai et al., 2020). Another example, a meta-analysis revealed insomnia might increase the risk of future IAs formation (He et al., 2017), while the reverse relationship between insomnia and IAs was also reported (Colledge et al., 2017). However, there was insufficient evidence on whether these psychiatric traits causally lead to IAs or vice versa, owing to possible residual confounding and reverse causation bias in observational researches (van’t Hof et al., 2017).

Causal inference between diseases and complex traits could be conducted under the development of Mendelian randomization (MR) and genome-wide association studies (GWAS). MR is a causal inference method for estimating the modifiable exposure (for example, insomnia) to an outcome (for example, IAs) via taking genetic variants as instrumental variables (IVs) for the exposure (Davey Smith and Ebrahim, 2003). MR method decreases residual confounding since the IVs are initially realigned in a random manner without connection to environmental factors, behaviors, and self-selected lifestyle factors. Besides, the MR method conquers reverse causality since IVs are determined irrespective of disease progression or development. Here, we conducted bi-directional two-sample MR analyses using GWAS summary data of IAs and the ten above-mentioned psychiatric traits to infer their causality. Clarifying such causal relationships from the genetic perspective may have important implications for primary prevention strategies in IA patients.

In this study, we firstly investigated the causal relationships between the ten psychiatric traits and IAs via two-sample bi-directional MR analyses. Our results demonstrated that insomnia had a risk effect on IAs and mood instability displayed a suggestive risk effect on IAs, whereas the other eight psychiatric traits had no significant effect on IAs. In the reverse MR analyses, no evidence implicated IAs as the cause of insomnia, MDD, subjective wellbeing, ADHD, ASD, BIP, SCZ, and neuroticism. Our instrumental variables were strong enough to avoid weak instrument bias according to the F statistics. The Cochran’s Q-test and scatter plots revealed null heterogeneity or pleiotropic effect.

Emerging evidence from prospective researches indicated that insomnia was correlated with increased risk of cerebral vascular disease (Wu et al., 2014; Zheng et al., 2019), including IAs. A recent meta-analysis including 23 cohorts demonstrated that insomnia significantly increased cardio-cerebral vascular events (He et al., 2017). Bakker et al. used 376 UKB phenotypes as exposure, including insomnia, to assess the risk factor of IAs in Europeans (Bakker et al., 2020). There were 20 valid SNPs used for generalized SMR (GSMR) analysis of insomnia on IAs, with the p = 0.0514. Here, we utilized summary data of insomnia not only from UKB but also 23andMe, expanding the valid SNPs to 159. Moreover, our four MR approaches fitted well in the linear relation between the SNPs effect on IAs and insomnia. Although inconsistent with Bakker’s result, our result indicated that insomnia had a causal risk effect on IAs. The causal relationship between insomnia and IAs reinforces the notion that prevention and early diagnosis of insomnia may help prevent IAs. However, the precise mechanism linking insomnia to IAs is unclear. Some of the proposed pathophysiological mechanisms may shed light on how insomnia might predispose an individual to IAs. Insomnia symptoms may alter cerebrovascular health through elevated circulating catecholamine (Irwin et al., 1999), sympathetic nervous activity (Dettoni et al., 2012), inflammation (Grandner M. A. et al., 2013; Grandner M. et al., 2013; Ferrie et al., 2013), endocrine or metabolic dysregulation (Okun, 2011; Westerlund et al., 2013). Clinical researches revealed that insufficient sleep might affect the levels of circulating catecholamine that influence response to emotional or physical stress (Irwin et al., 1999). Further study is needed to unravel the complex connection between insomnia and the development of IAs.

It has been reported that stroke was positively associated with anxiety (Lambiase et al., 2014), depression (Surtees et al., 2008; Henderson et al., 2013; Barlinn et al., 2015), and BIP (Prieto et al., 2014). A prospective cohort of 6,019 participants revealed higher levels of anxiety symptoms correlated with increased risk of occasional stroke (Lambiase et al., 2015). Are these psychiatric traits the risk factors of IAs (one of the main event of stroke)? However, Our results demonstrated anxiety disorders, MDD, BIP were not causally related to the development of IAs. The causal risk effect of mood instability on IAs was nominal significant according to the RAPS method (p = 0.047, 0.005 < p < 0.05). Although the p-values from IVW, BWMR, and MR.Corr methods were >0.05, similar effect sizes were obtained. Therefore, mood instability may be a potential risk factor for IAs. Further studies with larger sample sizes and additional mood instability genetic instruments are required to disentangle causality.

Several randomized controlled trials (RCT) reported no significant association was observed between subjective wellbeing and cardiovascular events (Lyall et al., 2018; O'Connor et al., 2020). Consistent with these observational studies, we found that subjective wellbeing had a null causal effect on IAs.

Whether there was reverse causality between psychiatric traits (depression etc.) and stroke had been debated previously (Begovac et al., 2008; Brunner et al., 2014; Lemos et al., 2020). Here, we did not detect significant genetic evidence regarding the causality of IAs to the eight psychiatric traits (insomnia, MDD, subjective wellbeing, ADHD, ASD, BIP, SCZ, and neuroticism).

More caution is needed about applying the effect size obtained by the MR method since the OR implies the mean impact of lifetime exposure. Insomnia may differ sharply across time periods. Therefore, the risk effect of insomnia on IAs is time-dependent.

Strengths of our study included the two-sample bi-directional MR analyses and the use of summary-level data from thus far the biggest sample numbers. Hence, the possible effect of reverse causality and conventional confounders could be reduced. In addition, we conducted comprehensive analyses including four MR approaches and the heterogeneity tests to avoid potential pleiotropic effects. However, it should be noted that our results were based on European populations. Thus, we should be cautious about applying this conclusion to non-Europeans since a distinct environment may significantly impact psychiatric traits and IAs.

Several limitations of the present study should also be recognized. Firstly, there was sample overlapping between the exposure and outcome because of the summary level data. Secondly, we did not conduct MR analyses of IAs on anxiety disorders and mood instability because we could only get access to significantly associated SNPs of the summary data for anxiety disorders and mood instability. Thirdly, the follow-up analysis to interpret the biological significance of the current result was lacking.

In conclusion, we performed two-sample bi-directional MR analyses between psychiatric traits and IAs based on the large-scale GWAS summary statistics. Our findings provided strong evidence for a causal risk effect of insomnia on IAs and suggestive evidence for mood instability as a causal risk effect on IAs. There was no genetic support for a causal effect of the other eight psychiatric traits (anxiety disorder, MDD, subjective wellbeing, ADHD, ASD, BIP, SCZ, and neuroticism) on IAs. In the reverse MR analyses, no genetic evidence implicated IAs as the cause of insomnia, MDD, subjective wellbeing, ADHD, ASD, BIP, SCZ, and neuroticism.