Background

Major depressive disorder (MDD) is widely acknowledged as a significant global mental health challenge [1, 2]. With a lifetime prevalence of 10 percent, MDD ranks first in prevalence among all mental illnesses and is one of the leading contributors to the global morbidity burden [3]. MDD, a prototypical example of a severe mental illness, is characterized by a multifaceted etiology that encompasses both genetic and environmental determinants. In addition to exploring the biological mechanisms of MDD, addressing its long-term management and rehabilitation remains a top priority. The rising prevalence of depressive comorbidities in clinical practice complicates treatment and recovery [4]. These conditions lead to numerous adverse outcomes, including functional dependence [5], poor quality of life [6], and shortened life expectancy [7], thus, creating significant economic burdens for families and society. Further exploration is needed to understand the challenges posed by comorbid cases of MDD.

As the global standards of health care gradually improve, comorbid and multi-morbid case histories are becoming more prevalent. Epidemiological evidence suggests that 75 percent of people aged 65-74 years are multi-morbid, with an even higher prevalence of co-morbidity detected in cohorts older than 75 years [8]. Among adults aged 18-65 years, a population-based survey showed that the prevalence of comorbid states ranges from 7 to 35 percent [9]. Due to its high prevalence, MDD became one of the most common comorbid conditions among adults, especially in patients with cardiovascular diseases (e.g., coronary artery disease) [10], metabolic diseases (e.g., diabetes mellitus) [11], skin, muscle and connective tissue diseases (e.g., microvascular dysfunction) [12] as well as rheumatological and immunological diseases (e.g., arthritis) [13]. In addition, the results of a meta-analysis of 12 longitudinal studies and 16 cross-sectional surveys showed that older people with chronic conditions had a higher risk of depression when compared to those without chronic conditions (RR: 1.53, 95 percent confidence interval (CI): 1.20-1.97) [14]. On the other hand, numerous studies have confirmed that individuals with depression have a markedly increased risk of becoming comorbid with epilepsy [15], hypertensive disorders [16], thyroid dysfunction [17], and pain [18]. All indications point to the possibility of MDD interacting with other disorders, or even being bi-directionally associated with such. Nonetheless, the causal relationships between MDD and other conditions remain unclear.

The most common instrument for disentangling relationships between human conditions, so-called observational studies, faces challenges in evaluating the significance of uncovered effects due to the influence of confounding factors and reverse causation, which hinder directional inferences. In contrast, the Mendelian randomization (MR) analysis framework investigates causal links between different exposures and outcomes [19] by leveraging comprehensive sets of genetic variants, with an emphasis on that collected in GWAS databases. In recent years, discussions regarding causal associations between MDD and other diseases, such as schizophrenia [20], prostate cancer [21], inflammatory bowel disease [22], atopic diseases [23], and osteoporosis [24] have become widespread. However, numerous diseases have yet to be validated in relation to MDD. Bidirectional MR studies facilitate the development of clinical guidelines by discerning the causal pathophysiological drivers for MDD and its common co-morbidities.

To comprehensively explore the causal implications of genetic predisposition to MDD on the risks of developing other disorders, we compiled a list of disease phenotypes from the FinnGen database (version R9). As proxies for MDD, the genome-wide association study (GWAS) results from the Psychiatric Genomics Consortium (PGC) were extracted. Subsequently, two-sample bidirectional MR analyses were conducted to assess causal connections between MDD and a range of disease phenotypes at the genetic level.

Methods

Study design and data source

Figure 1 presents the study design. Phenome-wide, bidirectional two-sample MR analyses were combined to explore causal associations between MDD and the health conditions profiled in the FinnGen Biobank (R9). Three assumptions were made regarding these MR analyses. Firstly, the genetic variants employed as instrumental variables exhibited a close and direct relationship with the exposure. Secondly, these genetic variants should not be associated with any potential confounding factors, preserving the integrity of the analysis. The final assumption stated that the chosen genetic variants influenced the outcome exclusively through the mediation of known risk factors, ruling out any alternative pathways [25].

Fig. 1: Study design overview.
figure 1

A total of 877 phenotypes were included. With adjusted p-values at the significance level (FDR < 0.05), genetically predicted depression was associated with a total of 324 disease phenotypes. In turn, a total of 51 genetically predicted disease phenotypes were associated with depression. Among these, bidirectional causal associations with depression were detected for 15 disease phenotypes. MR, Mendelian randomization; FDR, False discovery rate.

Full size image

The dataset for MDD was sourced from the Psychiatric Genomics Consortium [26] and was originally combined by a meta-analysis of data from 807,553 individuals (246,363 cases and 561,190 controls) in the three largest genome-wide association studies of depression [27,28,29].

GWAS data for all other phenotypes were obtained from the R9 FinnGen database, with a total sample size of 377,277 individuals (166,407 men and 210,870 women), and a total of 2,272 disease phenotypes profiled [30]. We selected a total of 904 disease and health-related phenotypes with Ncase > 2000 (Table S2). We conducted a rigorous screening of the GWAS summary dataset to exclude a set of specific labels, such as depression-related medications (e.g., ANTIDEPRESSANTS), mode of reimbursements (e.g., HYPOTHY_REIMB), depression itself, or emotion-related phenotypes (e.g. F5_DEPRESSIO), descriptive non-disease traits (e.g., BMI_IRN), and some other conditions out of scope of this study (e.g., O15_DELIV_SPONT). In cases where traits overlapped, priority was given to retaining the trait with the largest sample size. Following the filtering process, a total of 877 phenotypes from 41 disease categories were retained (Table S1).

Two-sample MR analysis

The primary MR analyses were conducted utilizing the inverse variance weighting approach (IVW), supplemented by the weighted median (WM) and MR-Egger techniques. We used a non-zero intercept in this regression model as an indicator of imbalanced horizontal pleiotropy [31]. The validity of IVs was called into question when the MR Egger intercept significantly deviated from zero. Additionally, we evaluated the heterogeneity of the MR analyses using Cochran’s Q test and the I2 statistic, thereby enhancing the robustness of our investigation [32].

Significant correlations were identified by applying a stringent False Discovery Rate (FDR) threshold of less than 0.05, primarily based on the IVW method. To ensure the robustness of our IVs, we selected single nucleotide polymorphisms (SNPs) from the GWAS dataset based on their genome-wide significance (p < 5.00E−08). In cases of an absence of significant variants at the genome-wide level or when the number of suitable variants is limited to 10 or less, the thresholds for IVs selection within the MR pipeline should be relaxed to a p-value threshold of 1.00E−05. Subsequently, these SNPs were further refined within a 10 Mb window, employing a clumped r2 cut-off value of 0.001. For each MR analysis, we systematically removed SNPs that were not present in the resultant dataset, SNPs with intermediate allele frequencies, and palindromic SNPs. This rigorous curation of the IVs ensured the quality and reliability of our MR analyses.

All statistical analyses in this study were performed using the R software package TwoSampleMR (version 0.5.7).

Results

Causal effects of MDD on other disorders

With adjusted p-values at the significance level (FDR < 0.05), genetically predicted depression was associated with a total of 324 disease phenotypes (Fig. 2a, c & Table 1). In turn, a total of 51 genetically predicted disease phenotypes were associated with depression at FDR < 0.05 (Fig. 2b, c & Table 2). Among these, bidirectional causal associations with depression were detected for 15 genetically predicted disease phenotypes at FDR < 0.05 (Fig. 1).

Fig. 2: The Beta value plot of the 2-sample MR analysis.
figure 2

a Volcano plot of MDD on other diseases in the FinnGEN database. b Volcano plot of other diseases in the Finnish database on MDD. The Beta value is used as the X-axis and -lg(FDR-Value) is used as the Y-axis. The FDR-Values were cut off at the 1.00E-5 level. The red dotted line indicates the level of statistical significance (FDR <0.05). c Violin plot of the MR and reverse MR analysis. The Beta value is used as the Y-axis and groups are used as the X-axis. Group 1: the Beta value plot of MDD on other diseases in the FinnGEN database. Group 2: the Beta value plot of other diseases in the Finnish database on MDD. FDR false discovery rate.

Full size image
Table 1 Causal influences of the genetically predicted major depressive disorder on other diseases.
Full size table
Table 2 Causal effects of genetically components of various diseases on major depressive disorder.
Full size table

In this study, MR analyses yielded a total of 320 positive and 4 negative causal effects of the genetic predisposition to MDD on other disorders (Table 1). The majority of associations were associated with diseases of the musculoskeletal system and connective tissue (66/133), diseases of the digestive system (40/97), psychiatric and behavioral disorders (46/59), diseases of the genitourinary system (37/126), diseases of the respiratory system (44/69), diseases of the nervous system (18/40), endocrine nutritional and metabolic disorders (15/42), and diseases of the circulatory system (18/74). Among psychiatric disorders, MDD had the greatest causal impact on personality disorders (OR = 2.33~3.18), followed by a category of “any mental disorder or suicide (or attempt); psychiatric complications of pregnancy, partum or puerperium; nerve system disorders; hyperkinetic disorders; eating disorders; suicide or other intentional self-harm” (OR > 2.00). Another large causal effect of MDD was on fibromyalgia (OR = 1.70~3.19), followed by a total of 11 dermatoskeletal and connective tissue disorders including rheumatism, radiculopathy, trochanteric bursitis, ankylosing spondylitis, etc (OR > 1.50). Among digestive disorders, more pronounced causal effects of MDD were seen on dental caries and temporomandibular joint disorders (OR > 1.50). Among the categories of mental and behavioral disorders, the most significantly causally affected by MDD were emotionally unstable personality disorder (OR = 2.39~3.82), along with a group of nine other disorders (OR > 2.00), including mixed and other personality disorders, other and unspecified mood [affective] disorders, other anxiety disorders, and persistent mood disorders, etc. Four genitourinary disorders that were significantly affected by the genetic component of MDD were unspecified/other endometriosis, adenomyosis (Endometriosis of the uterus), neuromuscular dysfunction of the bladder, and inflammatory diseases of the uterus (OR > 1.50). Among the diseases of the nervous system, the genetic component of MDD affected six disorders: migraine without aura and triptan purchases, migraine without aura, other headache syndromes, migraine with aura and triptan purchases, brachial plexus disorders, and Other and unspecified nerve root and plexus disorders, also in other diseases (OR > 1.50). Among the endocrine and nutritional metabolism category, a total of five diseases were significantly and causally affected by MDD: ovarian dysfunction, familial hypercholesterolemia with ischemic heart disease, obesity, obesity due to excess calories, as well as disorders of lipoprotein metabolism and other lipemias (OR > 1.50). Among circulatory disease, only post-myocardial infarction status and non-ruptured cerebral aneurysm categories were significantly and causally affected by a genetic predisposition to MDD (OR > 1.50) (Fig. 2a and 3a).

Fig. 3: The OR value plot of the effects of other disorders and MDD.
figure 3

a The OR value plot of the effects of other disorders on MDD in the two-sample MR analysis. The causal impact of other disorders on depression is characterized by a wide range of disorders with strong and significant differences in their contributions, with the more prominent ones being emotionally unstable personality disorder (OR = 3.02), personality disorders (more control exclusions) (OR = 2.66), mixed and other personality disorders (OR = 2.46), any mental disorder, or suicide (or attempt), or psychic disorders complicating pregnancy, partum or puerperium or nerve system disorders (OR = 2.44), fibromyalgia (OR = 2.33), panic disorder (OR = 2.30), hyperkinetic disorders (OR = 2.22), other and unspecified mood [affective] disorders (OR = 2.17), other anxiety disorders (OR = 2.16). KRA_PSY_ANYMENTAL_SUICID_PREG_NERV_EXMORE: any mental disorder, suicide (or attempt), or psychic disorders complicating pregnancy, partum or puerperium or nerve system disorders (more control exclusions); F5_BEHEMOCHILD: behavioural and emotional disorders with onset usually occurring in childhood and adolescence; F5_OTHERSUB: mental and behavioral disorders due to multiple drug use and use of multiple drugs and use of other psychoactive substances; VWXY20_INTENTI_SELF_P_EXPOS_OTHER_UNSPE_CHEMIC_NOXIO_SUBST: intentional self-poisoning by and exposure to other and unspecified chemicals and noxious substances. The β-values were cut off at the 0.50 level. b The OR value plot of the effects of MDD on other disorders in the 2-sample MR analysis. The causal impact of depression on other disorders is characterized by a relatively limited variety of disorders with restricted and non-significant differences in contribution, with the more prominent ones being haemangioma and lymphangioma, any site (OR = 1.41), benign neoplasm of the eye and adnexa (OR = 1.31), and benign lipomatous neoplasm of other sites/unspecified (OR = 1.30).

Full size image

Causal effects of other diseases on MDD

A total of 48 positive and 3 negative causal correlations were identified in the MR analyses of the causal effects of other diseases on MDD (Table 2). Most associations are causally connected to neoplastic conditions associated with MDD (14/64), diseases of the nervous system (11/40), and certain infectious and parasitic diseases (5/31). Among the neoplastic conditions, the most prominent causal impacts on MDD were detected for hemangioma and lymphangioma, at any site (OR = 1.33 ~ 1.50). Among the neurological disorders having a causal impact on MDD were the inflammatory diseases of the central nervous system (OR = 1.13~1.36). Certain infectious and parasitic diseases have also exhibited a causal relationship with MDD, with a category “Other predominantly sexually transmitted diseases, not elsewhere classified” (OR = 1.19 ~ 1.38) (Figs. 2b and 3b).

MR sensitivity analyses

For each bidirectional MR analysis with significant causal associations revealed, we assessed the hypothesis that IVs exhibit no pleiotropy. The MDD dataset comprised 83 IVs, while other diseases had varying numbers of IVs, ranging from 10 to 241. Sensitivity analyses performed within the framework of 2-sample MR and the MR-Egger regression pointed to limited evidence of heterogeneity (I2 < 0.80). Additionally, there were some indications of pleiotropic effects, with intercept values falling below the 0.06 threshold (Tables S3). On the other hand, both the primary (IVW) and sensitivity analyses (WM and Egger) demonstrated concordant directional effects, pointing to the robustness of the primary results (Tables S4). Similarly, in the sensitivity analyses of the reverse 2-sample MR and MR-Egger regression, little heterogeneity was observed (I2 < 0.80), with no evidence of pleiotropic effects (intercepts < 0.03) (Table S5). The directions of the effects between the main and sensitivity analyses were consistent, and the main results were robust (Table S6).

Discussion

In this study, we report the results of phenome-wide MR analysis of a total of 877 non-depression-related disease phenotypes from the FinnGen database. In 2-sample MR analyses, a total of 324 genetic signatures of disease phenotypes were causally associated with MDD. Research indicates that MDD significantly influences various systemic diseases, encompassing conditions such as bone-muscle-skin-related diseases, oncological diseases, infectious diseases, and traumatic injuries. Additionally, it extends its causal impact to psycho-neurological disorders, cardio-cerebral vascular diseases, respiratory disorders, endocrine-related diseases, digestive disorders, and genitourinary disorders. Conversely, genetic liability to MDD causally affects infectious diseases, tumor-related disorders, traumatic injuries, and others, in addition to mental and behavioral disorders, neurological disorders, circulatory disorders, and reproductive disorders. In a reverse 2-sample MR analysis, the genetic susceptibility to 51 phenotypes was causally affected by MDD. Intersections of these two sets of causal associations produced a total of 15 disease phenotypes with a bidirectional causal link to MDD.

Among other disorders causally contributing to MDD, the most prominent were those affecting the musculoskeletal system and connective tissue. Many of these conditions are associated with a range of painful symptoms (e.g., fibromyalgia, low back pain, pain in joints, limb pain, and thoracic spine pain, etc.). Notably, depression is commonly associated with somatic symptoms. Likewise, pain clinics commonly encounter patients exhibiting depressive symptoms. One observational study involving 5,372 community-dwelling older adults found that individuals experiencing pain were notably more likely to exhibit depressive symptoms [33]. Another prospective survey conducted in the United States indicated that the presence of depressive symptoms predicted the future onset of chronic musculoskeletal pain [34]. Furthermore, a recent systematic review and meta-analysis, which included five cohort studies and six cross-sectional studies, demonstrated that back pain could be a predictor for depressive episodes (OR = 2.07) [35]. Depression is closely associated with somatic pain, whereas the link between migraine and depression in children is well documented [36]. Similarly, adults with migraine often present with anxiety and depressive symptoms [37]. Furthermore, depression is an independent risk factor for poor migraine prognosis [38]. This bidirectional relationship was further confirmed in the present study. Additionally, depression and pain share a descending pathway within the autonomic nervous system (ANS) [18]. Dysfunctions of ANS encompass the circulatory system (e.g., cardiac arrhythmias) [39], that of the digestive system (e.g., gastroesophageal reflux) [40], and metabolic disorders (e.g., obesity) [41]. This study effectively demonstrates the significant causal implications of genetic predisposition to these disorders on the development of depression. Together with pain, these ANS dysfunctions may have an impact on MDD through a shared mechanistic pathway, which is possibly related to pain.

Indeed, multiple lines of evidence suggest the potential for a bidirectional causal relationship between depression and pain. The present study found further evidence for one kind of pain-related condition, cervicobrachial syndrome. This finding, however, needs replication and validation through larger-scale longitudinal studies.

One of the possible mechanisms to connect digestive diseases and mental disorders is through changes in the niche ecology of the gut, which may involve the function or composition of bacteria, viral, fungal, archaea, and protozoa species [42]. Distinct species of individual human gut are shaped by a multitude of factors, including direct perturbations of the intestinal milieu as well as lifestyle, and dietary choices [43]. While the framework for studying the microbiota-gut-brain axis (MGBA) originated from observations of patients with inflammatory bowel disease and irritable bowel syndrome (IBS) [44, 45], it further expanded into the phenomenon of depression and other psychiatric conditions [46]. In particular, the impacts of gut microbes on depressive behavior have been validated in a well-established rat model [47]. The most likely culprit that connects the microbiota with the pathophysiology of depression is the microbial synthesis of serotonin and its role in ensuring the continuity of the gut barrier [48, 49]. The mechanistic connections between various phenotypes of digestive disorders and MDD, as found in this study, likely occur through variations in gut microbiota. Specifically, it typically initiates with abnormalities in genetic composition affecting homeostasis in the host, followed by changes in metabolite abundance, which leads to disruption of intestinal barrier integrity, translocation of toxic metabolites into the circulation, and contributes to chronic inflammatory responses that ultimately culminate in the development of depressive comorbidities [50]. Shared genetic profiles of depression and metabolic traits provide convincing evidence for the above theories [51, 52]. Furthermore, kynurenine levels have been strongly associated with the severity of depression, indicating its potential as a predictive marker for disease progression [53]. The involvement of metabolites in digestive disorders such as inflammatory bowel disease has been extensively described [54]. Collectively, these findings underscore the potential role of the gut microbiome and its metabolites as critical links between MDD and digestive disorders.

Currently, mental illnesses continue to be diagnosed symptomatically, with no objective genetic or soluble biomarker validated so far [55]. Biochemical boundaries between MDD and other mental or neurological disorders are not well-defined. The 5-hydroxytryptamine (5-HT) deficiency hypothesis of MDD put forward by Coppen in 1967 may also be relevant to other psychiatric conditions. Recently, this view has been strengthened in a large number of studies [56,57,58]. 5-HT, a monoamine neurotransmitter, is released in both the central nervous system (5%) and the peripheral nervous system (95%) [59]. Genetic variants and changes in expression levels of 5-HT receptors and transporters [60] as well as overactivity of presynaptic autoreceptors [61] are implicated in a wide range of psychiatric or neurological disorders, such as depression, panic disorder, schizophrenia, and migraine [62,63,64,65]. Association of depression with other psychiatric or neurological disorders at the genetic level has proceeded [66,67,68,69,70]. Accordingly, this study demonstrated that both psychiatric and neurological disorders have a causal influence on MDD, possibly through alterations in serotoninergic signaling. Some of these interactions were bidirectional, including that with other and unspecified nonorganic psychotic disorders; with delirium, not induced by alcohol and other psychoactive substances; with other and unspecified nerve root and plexus disorders; with migraine with aura and triptan purchases; with other neurological diseases, and with nerve, nerve root and plexus disorders.

Our observations are also in line with previous findings that other psychiatric comorbidity among mental disorders is common and that many of these disorders exhibit pairwise relationships [71, 72]. For example, an analysis based on 27 national communities involving 145,990 respondents showed that, after adjusting for comorbidities, the risk of developing generalized anxiety disorder was 55 times higher in the first year after developing major depression when compared to patients without comorbidities (hazard ratio (HR) = 55.1, 95% CI = 51.0–59.6); even 15 years later, patients with major depression still had a nearly 7-fold higher risk of developing generalized anxiety disorder (HR = 6.6, 95% CI = 5.7–7.7) [73]. Additionally, a majority of antidepressants and antipsychotics used in clinical practice for various psychiatric conditions target the same set of pathways [74].

Although typical infection is only a short-term pathological process, it may have a profound effect on MDD. Multiple meta-analyses have confirmed [75,76,77] that considerable consensus exists on the increase of pro-inflammatory cytokines and acute-phase proteins, in particular c-reactive protein (CRP), in individuals with MDD compared to healthy controls [78,79,80]. As an acute phase response protein, the high sensitivity of CRP is regularly used as a biomarker of infection in clinical practice [81]. Acute stress and/or traumatic brain injury increases blood-brain barrier (BBB) permeability [82] prompting CRP to cross the BBB [83], triggering an inflammatory response as well as emotional and behavioral disorders. Naturally, dysregulation of peripheral myeloid cells, pro-inflammatory cytokines, and complement pathways may be potential complementary mechanisms of action [84]. In addition, CNS inflammation can be driven by elevated levels of CRP and its pro-inflammatory activity through the activation of microglia and astrocytes [85]. In comparison, while chronic and low-grade inflammation has been observed in only a quarter of patients with MDD, pro-inflammatory signaling pathways are still considered to have a significant role in the development of MDD [86,87,88]. Indeed, the levels of acute phase proteins and various pro-inflammatory cytokines, including tumor necrosis factor (TNF)-α, interleukin (IL)-1β, and IL-6, in patients with depression are higher than in the general population [89, 90]. Pro-inflammatory pathways initiate at the inflammasome, which plays a key role in the relationship between inflammation and oxidative stress [91], damaging cellular and intracellular membranes and key biomolecules [92]. Similar damage has been reported in many diseases that are likely to be comorbid with depression,such as rheumatoid arthritis [93], psoriasis [94], cancer [95], type 2 diabetes [96], and coronary heart disease [97]. For all these diseases, their causal effects on MDD were confirmed. Among these, the association between asthma and depression has been widely explored [23, 98]. A large genome-wide cross-trait analysis revealed significantly elevated levels of IFN-γ, TNF-α, and IL-17A in individuals with asthma and psychological symptoms [99]. A meta-analysis of studies on depression and asthma suggests that the two conditions are comorbid due to the sharing of immune-inflammatory pathways [100]. Specifically, environmental factors trigger a chronic airway inflammatory response, characterized by elevated inflammatory markers, resulting in asthma attacks. Asthma attacks are associated with activation of NF-κB, which impacts 5-HT1A gene expression and contributes to the development of MDD [101]. On the other hand, the causal association of several inflammatory disorders (e.g., otitis externa, conjunctivitis, rhinitis, periodontitis, cystitis, and pelvic inflammatory diseases) with MDD, though rarely discussed in observational studies, has also been well-documented.

In depressed patients, the most prevalent comorbidity is type 2 diabetes, which shares with MDD both its genetic underpinning and pathophysiological pathways [102,103,104]. In Both conditions, the neuroendocrine function of the hypothalamic-pituitary-adrenal (HPA) axis is altered. In depression, overactivation of the HPA axis leads to elevation in the levels of cortisol [105] and promotes insulin resistance, thus, shifting affected individuals toward metabolic syndrome [106]. The findings of this study contribute to supporting a causal link between diabetes and MDD. However, no evidence of a bidirectional causal association was observed. Similarly, the HPA axis was shown to be involved in stroke [107], suicide [108], epilepsy [109], and substance abuse [110]. Although inconsistent findings have also been reported [111] for each of these conditions, their causal effects on depression were confirmed in the current work.

In this study, the most pronounced causal influence on MDD was observed for emotionally unstable personality disorder (OR = 3.02, 95%CI:2.39-3.82). The exploration of the association between personality traits and MDD dates back to Hippocrates’ humor theory. Among more modern, but still empirical models of personality are Eysenck’s three dimensions (i.e., neuroticism, psychoticism, and extraversion) and the Big Five model of personality (extraversion, neuroticism, conscientiousness, agreeableness, and openness) [112]. Notably, an in-depth discussion of the relationship between neuroticism, characterized by emotional instability, emotional impulsivity, and emotional hypersensitivity, and MDD has been initiated earlier [20, 113]. Also, an earlier Meta-analysis pooling 175 studies hinted toward a robust and significant correlation between MDD and high scoring for neuroticism (Cohen’s d = 1.54) [114]. Some other studies examined the relationship between the two, [115, 116] with available evidence showing similarities in the structural characteristics of the brain in neuroticism and depression reinforcing and strengthening these conclusions [117]. The current study reaffirms the robust causal effect of personality disorders on MDD.

Our previous studies have examined in depth the mechanisms of genetic-level interactions between insomnia and depression [118]. Insomnia and depression share a large amount of genetic variation and display bidirectional causal associations with one another [118]. The effects of the genetic susceptibility to insomnia on depression were distinctly higher than that of depression on insomnia.

The present study indicated that the effects of depression on other disorders were more prominent than the reverse when considered genetically, mainly in the form of a broader spectrum of diseases, and a higher magnitude of the the effect sizes. The genetic variance profile, derived from the multigene overlap analysis, provides a compelling explanation for this phenomenon. Our previous study showed the presence of about 158,000 causal genetic variants in MDD, whereas cardiovascular disease exhibits a relatively lower number of causal variants [119]. For instance, atrial fibrillation is associated with around 0.5 million causal variants, whereas heart failure has 28,000 causal variants. In short, the greater number of genetic variants influencing depression implies a higher susceptibility to more potent causal effects compared to other disorders.

The comprehensive results of our study showed a substantial degree of overlap with the findings from the recently published large-scale meta-analysis [120]. An in-depth exploration of the complex pathophysiological mechanisms underlying depression, encompassing genetic, psychological, social, and biological factors, unveils its intricate interaction with various coexisting conditions (Fig. 4). Psychosocial factors should not be underestimated [121, 122]. Education has been shown to be a strong predictor of depression [123]. Additionally, personal income significantly influences depression risk, with low-income individuals facing a much higher likelihood of depression compared to those with higher incomes [124]. Notably, these factors also contribute to other conditions, such as type 2 diabetes [125], hypertension [126], and stroke [127], which share common comorbidities with depression. One possible explanation is that less-educated or lower-income people are more likely to be under greater social stress, which would probably lead to changes in neurotransmitter and hormone levels, ultimately contributing to comorbid depression. The most effective strategy for addressing the comorbidities of depression should rely on integrated interventions that include psychological, social, behavioral, and pharmacological dimensions.

Fig. 4: The depression and its comorbid conditions: shared pathophysiology.
figure 4

The numerous risk factors involved such as genetics, childhood trauma, individual traits, stress, socially undesirable environments, and unhealthy behavioral patterns may contribute to co-morbid depression in those with medical disorders. The potential link between depression and medical disorders is established through one or more pathway mechanisms, encompassing behavioral, biological, and psychological factors. The figure was drawn by Figdraw.

Full size image

The primary strength of our study lies in utilizing data from the newly released Finnish database to integrate extensive disease phenotypes, revealing comorbidities associated with depression through hypothesis-free MR-PheWAS analyses. Additionally, upon identifying robust causal associations between MDD and other disorders at the genetic level, this study pinpointed disease phenotypes exhibiting bi-directional causal influences on MDD through inverse MR analysis. Several limitations of the MR approach need acknowledgment. Firstly, although pleiotropy is a concern, the impact of this bias should be minimal due to the overall consistency observed in the sensitivity analyses and the fact that the regression intercepts are all small in the MR-Egger analysis. Secondly, the filtering criteria were relaxed to obtain a larger number of IVs, which may have introduced bias into the results. Thirdly, caution should be exercised when applying the conclusions of this study to other populations, as the data were derived from the GWAS database for European ancestry. Fourth, some conditions with relatively small sample sizes remained within the scope of analysis, potentially increasing the risk of random error in the results. Thus, further validation through preclinical and clinical studies is necessary.

Conclusions

This phenome-wide MR study reveals significant bidirectional causal relationships between MDD and a variety of disease phenotypes. These results provide valuable insights into the genetic underpinnings of MDD and its comorbidities, offering a foundation for future research and potential clinical interventions. In the course of clinical practice, physicians should be adept at identifying and promptly diagnosing and treating comorbid depression.