Investigating bidirectional causal relationships between gut microbiota and insomnia

Abstract

Background

Although studies in recent years have explored the impact of gut microbiota on various sleep characteristics, the interaction between gut microbiota and insomnia remains unclear.

Aims

We aimed to evaluate the mutual influences between gut microbiota and insomnia.

Methods

We conducted Mendelian randomisation (MR) analysis using genome-wide association studies datasets on insomnia (N=386 533), gut microbiota data from the MiBioGen alliance (N=18 340) and the Dutch Microbiome Project (N=8208). The inverse variance weighted (IVW) technique was selected as the primary approach. Then, Cochrane’s Q, Mendelian randomization-Egger (MR-Egger) and MR Pleiotropy RESidual Sum and Outlier test (MR-PRESSO) tests were used to detect heterogeneity and pleiotropy. The leave-one-out method was used to test the stability of the MR results. In addition, we performed the Steiger test to thoroughly verify the causation.

Results

According to IVW, our results showed that 14 gut bacterial taxa may contribute to the risks of insomnia (odds ratio (OR): 1.01 to 1.04), while 8 gut bacterial taxa displayed a protective effect on this condition (OR: 0.97 to 0.99). Conversely, reverse MR analysis showed that insomnia may causally decrease the abundance of 7 taxa (OR: 0.21 to 0.57) and increase the abundance of 12 taxa (OR: 1.65 to 4.43). Notably, the genus Odoribacter showed a significant positive causal relationship after conducting the Steiger test. Cochrane’s Q test indicated no significant heterogeneity between most single-nucleotide polymorphisms. In addition, no significant level of pleiotropy was found according to MR-Egger and MR-PRESSO.

Conclusions

Our study highlighted the reciprocal relationships between gut microbiota and insomnia, which may provide new insights into the treatment and prevention of insomnia.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• An imbalance of gut microbiota can affect the pathophysiology of the brain through the gut–brain axis. Although studies have explored the effects of gut microbiota on sleep characteristics, the interactions between different gut microbiota taxa and insomnia remain unclear.

WHAT THIS STUDY ADDS

• Specific gut microbiota are causally associated with insomnia. Divergent microbiota may exert differential effects on insomnia.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• Our study offers preliminary evidence supporting a causal effect between gut microbiota and insomnia. This enriches our understanding of this connection and furnishes novel perspectives for microbiome-centred treatment strategies for insomnia.

INTRODUCTION

Insomnia is an extremely common clinical condition, characterised by difficulty falling asleep or poor sleep quality, and accompanied by symptoms affecting daily life, such as irritability or fatigue when awake.¹ It is estimated that the incidence of insomnia is about 10%–20%, with about half the cases reflecting a chronic course of disease.² Importantly, insomnia may lead to the development of other medical and mental disorders, which collectively result in increased mortality and enormous social costs.³ When compared with individuals with normal patterns of sleep, non-depressed individuals with insomnia have double the risk of developing depression.⁴ The causes and pathophysiology of insomnia involve the interaction of genetic, environmental, behavioural and physiological factors that collectively lead to excessive wakefulness. Both episodic and chronic insomnia may be triggered by unrelated illnesses, family problems, work, school or other sources of stress, which, speaking generally, negatively contribute to the quality of an individual’s sleep.⁵ Insomnia may present as a first symptom of many disorders, but it also frequently coexists with mental and physical health conditions. Changes in the composition and function of the gut microbiota are commonly noted in patients with insomnia.⁶ However, their causal connections to insomnia phenotypes are far from being clear.

The human gut microbiota consists of various microorganisms including bacteria, fungi and viruses that reside in the gastrointestinal tract.⁷ These bacteria are usually described using the taxonomic units of phylum, order, class, family, genus, species and strain. Representatives of the gut microbiota of healthy adults include the phyla Firmicutes, Bacteroidetes, Actinobacteria and Proteobacteria, with Firmicutes and Bacteroidetes appearing to be jointly dominant (up to 90%).⁸ The gut microbiota is dynamic; it changes with age, dietary habits and stress, as well as with pregnancy in women.⁷ When the diversity and composition of the gut microbiota are disturbed, one or more pathological states of the gut may develop, with a subsequent increase in the risk of systemic diseases such as allergies, obesity, metabolic syndrome and autoimmune diseases.

The gut microbiota plays a crucial role in the development of the enteric nervous system (ENS), which regulates intestinal peristalsis.⁹ In germ-free mice, sympathetic neurons exhibit signs of increased c-FOS activity and overactivation, while exposure to normal microbiota limits the activation of these neurons by producing short-chain fatty acids (SCFAs) such as butyric acid. Additionally, the microbiota influences the ENS through 5-hydroxytryptamine (5-HT or serotonin)-dependent circuits.¹⁰ Wang et al discovered that sleep deprivation, as well as sleep duration restriction and fragmentation, may induce changes in the gut microbiota composition.¹¹ In mice with antibiotic-induced microbiota depletion, Ogawa et al observed a significant decrease in theta wave power density during rapid eye movement (REM) sleep.¹² The ENS and the central nervous system are connected primarily through the vagus nerve, forming the gut–brain axis.¹³ This axis serves as a crucial communication pathway through which gut microbiota dysbiosis exerts a significant impact on brain pathophysiology and promotes a host of neuropsychiatric conditions, such as mild cognitive impairment, depression and Alzheimer’s disease.14,17 Given this connection, bidirectional relationships between insomnia and the diversity and composition of the gut microbiota may be expected.

Mendelian randomisation (MR) studies are similar to those with randomised treatment and control arm assignments. That is because of the random nature of the selection of the genetic variants, which recombine to form individual diploid genomes. MR avoids some of the problems that plague traditional observational studies by minimising the effects of confounding factors and reverse causation.¹⁸ This analytical approach is now widely used to infer causality from a genetic perspective, and its scope of application encompasses both somatic19,22 and mental disorders.23,25 In this study, we used a two-sample MR (TSMR) analysis to explore the causal connections between individual constituents of the gut microbiota and insomnia.

METHODS

Data sources

This study employed an MR analysis. It used a genome-wide association study (GWAS) dataset of insomnia²⁶ combined with two gut microbiota datasets from the MiBioGen alliance²⁷ and the Dutch Microbiome Project.²⁸ The MiBioGen alliance (https://www.mibiogen.org) includes GWAS summary statistics from 18 340 participants, covering a total of 211 taxa across 35 families, 20 orders, 16 classes, 9 phyla and 131 microbial genera. The Dutch subset data mainly stems from the Dutch Microbiome Project (https://www.ebi.ac.uk/gwas/), which investigated the gut microbiome composition and function in 8208 individuals. We utilised GWAS data for 207 microbial taxa (5 phyla, 10 classes, 13 orders, 26 families, 48 genera and 105 species), excluding relevant metabolic pathway sections. The data for insomnia were sourced from a GWAS meta-analysis by Jansen et al²⁶ including 386 533 subjects (109 402 cases and 277 131 controls) and 23andMe. Because of the access constraints previously reported for 23andMe samples, summary statistics of insomnia were downloaded from the UK Biobank (UKB) (N=386 533, Ncase=109 402, Ncontrol=277 131). Details about the UKB are available on its website (https://www.ukbiobank.ac.uk/). All participants were of European descent. Ethical approval was obtained in all original studies (study information outlined in online supplemental table S1).

Instrumental variables

We selected candidate instrumental variables (IVs) based on a genome-wide significance threshold of p<1.0×10⁻⁵. For reverse MR analysis, we applied a more stringent screening criterion of p<5×10^–8 for single-nucleotide polymorphisms (SNPs). SNPs were selected using the 1000 Genomes Project Phase 3 (European) reference panel.²⁹ To ensure the independence of the IVs, we performed pruning with an r² threshold of 0.01 within a 10 Mb window. This strategy helped us to ascertain that the IVs included in the analysis were independent, thereby improving the precision and robustness of the MR results.

Statistical analysis

In this study, we utilised three different models from the TwoSampleMR package in R software to investigate causal relationships.³⁰ The primary method employed was the inverse variance weighted (IVW) approach, which improves the accuracy of causal relationship estimates by assigning weights to the IVs based on their precision, thus helping to minimise random errors. The IVW model assumes that all IVs are valid and that their effects on the outcome are mediated solely through exposure. To verify the robustness of our findings, we also applied the weighted median (WM) and MR-Egger methods as supplementary techniques. The WM method gives more weight to IVs with greater precision and provides reliable causal estimates even when some IVs are invalid. Conversely, the MR-Egger model helps to detect and adjust for biases arising from the use of invalid IVs.^{31 32}

MR-Egger and MR Pleiotropy RESidual Sum and Outlier (MR-PRESSO) tests were used to test horizontal pleiotropy and outliers. MR-Egger was specifically applied to preliminarily identify the existence of horizontal pleiotropy. When the p value was greater than 0.05, it was interpreted as no significant horizontal pleiotropy. Compared with MR-Egger, MR-PRESSO has higher accuracy and is useful in identifying horizontal pleiotropy and outliers.³³ If an outlier SNP was found (p<0.05), the causal effects were re-estimated using the remaining SNPs after removing the outliers. Cochrane’s Q test and I² statistics (p<0.05 and I²>0.25) were performed to assess potential heterogeneity. A leave-one-out (LOO) sensitivity analysis was conducted to identify outliers and evaluate the stability of the results. IVW-based p<0.05 was interpreted as a significant correlation between particular gut microbiota taxa and insomnia. For each dataset, we aligned the exposure and outcome effects allele by allele for each pair of exposure and outcome datasets, obtaining the variance effect and standard error. Furthermore, MR Steiger tests were performed on the gut microbiota with bidirectional causal associations to insomnia, which confirmed the directionality of the results. A p value of less than 0.05 was considered statistically significant.³⁴ The flowchart of the current study is depicted in figure 1.

Figure 1. Study flowchart. Dutch, Dutch Microbiome Project; GWAS, genome-wide association study; IV, instrumental variable; IVW, inverse variance weighted; MiBioGen, MiBioGen alliance; MR, Mendelian randomisation; MR-PRESSO, MR Pleiotropy RESidual Sum and Outlier; SNP, single-nucleotide polymorphism; WM, weighted median.

RESULTS

IV selection

We conducted a screening of IVs for a total of 211 bacterial taxa from the MiBioGen consortium and a total of 207 bacterial taxa from the Dutch Microbiome Project. Following this, LOO sensitivity analysis and MR-PRESSO-based outlier detection were performed for IVs meeting the significance threshold of p<1×10^-5. After these quality control steps, a total of 4089 IVs were retained for the final evaluation. In reverse MR analysis, a similar rigorous screening process was applied, resulting in the inclusion of a total of 5118 IVs. Detailed descriptors of these SNPs can be found in online supplemental file 2.

TSMR analysis

In MR analysis with the IVW method, a total of 22 bacteria clades (groups of organisms that include a common ancestor and all its descendants) conferred a causal effect on insomnia (table 1, figure 2A, online supplemental figure S1). TSMR results from MiBioGen suggest that genera Clostridium innocuum group, Prevotella 7, Lachnoclostridium, ParaPrevotella, Family XIII AD3011 group, Rikenellaceae RC9 gut group and Parabacteroides, class Negativicutes and order Selenomonadales may increase the risks of insomnia (odds ratio (OR): 1.01 to 1.04). In contrast, the genera Coprococcus1 and Lactococcus, family Actinomycetaceae and order Actinomycetales were likely to reduce the risks of this condition (OR: 0.97 to 0.99). Similarly executed analysis of the Dutch dataset suggested that genera Pseudoflavonifractor and Anaerotruncus, species Veillonella unclassified, Bacteroides massiliensis and B. faecis are associated with increased risks of insomnia (OR: 1.01 to 1.02), while genus Odoribacter, family Clostridiaceae and species Ruminococcus torques and R. lactaris contribute to the reduction of insomnia risks (OR: 0.97 to 0.98). However, after false discovery rate (FDR) correction, the uncovered causal association remained significant only for the genus Clostridium innocuum group (FDR=0.007).

Table 1. Mendelian randomisation analysis revealing causal effects of the gut microbiota on insomnia.

Exposure	Source	Outcome	IV (n)	OR (95% CI)	P value	FDR
Genus Clostridium innocuum group	MiBioGen	Insomnia	9	1.03 (1.02 to 1.05)	<0.001	0.007
Genus Prevotella 7	MiBioGen	Insomnia	10	1.02 (1.01 to 1.04)	0.001	0.104
Genus Lachnoclostridium	MiBioGen	Insomnia	14	1.04 (1.01 to 1.07)	0.002	0.147
Genus Paraprevotella	MiBioGen	Insomnia	13	1.02 (1.01 to 1.04)	0.007	0.354
Genus Family XIIIAD3011 group	MiBioGen	Insomnia	13	1.04 (1.01 to 1.07)	0.009	0.354
Genus Rikenellaceae RC9 gut group	MiBioGen	Insomnia	12	1.01 (1.00 to 1.03)	0.015	0.458
Genus Parabacteroides	MiBioGen	Insomnia	10	1.04 (1.01 to 1.07)	0.016	0.458
Genus Coprococcus 1	MiBioGen	Insomnia	13	0.97 (0.95 to 1.00)	0.024	0.493
Class Negativicutes	MiBioGen	Insomnia	11	1.04 (1.00 to 1.07)	0.025	0.493
Order Selenomonadales	MiBioGen	Insomnia	11	1.04 (1.00 to 1.07)	0.025	0.493
Family Actinomycetaceae	MiBioGen	Insomnia	5	0.97 (0.95 to 1.00)	0.039	0.635
Order Actinomycetales	MiBioGen	Insomnia	5	0.97 (0.95 to 1.00)	0.039	0.635
Genus Lactococcus	MiBioGen	Insomnia	11	0.99 (0.97 to 1.00)	0.042	0.635
Genus Odoribacter	Dutch	Insomnia	10	0.97 (0.96 to 0.99)	0.002	0.403
Family Clostridiaceae	Dutch	Insomnia	7	0.98 (0.96 to 0.99)	0.004	0.403
Genus Pseudoflavonifractor	Dutch	Insomnia	8	1.02 (1.01 to 1.03)	0.007	0.464
Species Veillonella unclassified	Dutch	Insomnia	9	1.01 (1.00 to 1.03)	0.013	0.500
Genus Anaerotruncus	Dutch	Insomnia	6	1.01 (1.00 to 1.02)	0.014	0.500
Species Ruminococcus torques	Dutch	Insomnia	7	0.98 (0.96 to 1.00)	0.016	0.500
Species Bacteroides massiliensis	Dutch	Insomnia	6	1.02 (1.00 to 1.04)	0.018	0.500
Species Ruminococcus lactaris	Dutch	Insomnia	5	0.98 (0.96 to 1.00)	0.026	0.659
Species Bacteroides faecis	Dutch	Insomnia	13	1.01 (1.00 to 1.02)	0.038	0.839

CI, confidence interval; Dutch, Dutch Microbiome Project; FDR, false discovery rate; IV, instrumental variable; MiBioGen, MiBioGen alliance; OR, odds ratio.

Figure 2. Causal effects between the gut microbiota and insomnia. (A) Causal effects of the gut microbiota on insomnia. (B) Causal effects of insomnia on the gut microbiota. Beta, effect size; CI, confidence interval; Dutch, Dutch Microbiome Project; MiBioGen, MiBioGen alliance; IV (n), number of instrumental variables; OR, odds ratio.

Reverse MR analysis of gut microbiota datasets highlighted the causal positive influence of insomnia on the abundance of 12 taxa (OR: 1.65 to 4.43), as well as the negative effects on 7 taxa (OR: 0.21 to 0.57) (table 2, figure 2B, online supplemental figure S1). Notably, the result of Cochran’s IVW Q test revealed heterogeneity in the instrumental variables, supporting the abundance of the genus Sutterella (Q p=0.035). In addition, as presented in online supplemental tables S2 and S3, the Steiger test showed a positive causal relationship between the genus Odoribacter and insomnia.

Table 2. Mendelian randomisation analysis revealing causal effects of insomnia on the gut microbiota.

Exposure	Outcome	Source	IV (n)	OR (95% CI)	P-value	FDR
Insomnia	Family Alcaligenaceae	MiBioGen	13	0.48 (0.28 to 0.80)	0.005	0.513
Insomnia	Class Betaproteobacteria	MiBioGen	13	0.50 (0.30 to 0.82)	0.006	0.513
Insomnia	Order Burkholderiales	MiBioGen	13	0.51 (0.31 to 0.85)	0.009	0.513
Insomnia	Class Coriobacteriia	MiBioGen	13	1.65 (1.10 to 2.47)	0.016	0.513
Insomnia	Family Coriobacteriaceae	MiBioGen	13	1.65 (1.10 to 2.47)	0.016	0.513
Insomnia	Order Coriobacteriales	MiBioGen	13	1.65 (1.10 to 2.47)	0.016	0.513
Insomnia	Phylum Verrucomicrobia	MiBioGen	13	1.72 (1.06 to 2.79)	0.027	0.646
Insomnia	Genus Parasutterella	MiBioGen	13	0.57 (0.34 to 0.94)	0.029	0.646
Insomnia	Genus Lachnospiraceae UCG-001	MiBioGen	13	0.53 (0.30 to 0.94)	0.030	0.646
Insomnia	Genus Sutterella	MiBioGen	13	0.50 (0.26 to 0.95)	0.034	0.663
Insomnia	Genus Odoribacter	Dutch	13	2.37 (1.23 to 4.56)	0.010	0.860
Insomnia	Species Streptococcus thermophilus	Dutch	13	0.21 (0.06 to 0.73)	0.014	0.860
Insomnia	Species Holdemania filiformis	Dutch	13	4.43 (1.29 to 15.27)	0.018	0.860
Insomnia	Family Burkholderiales noname	Dutch	13	2.71 (1.06 to 6.97)	0.038	0.860
Insomnia	Species Odoribacter splanchnicus	Dutch	13	2.10 (1.04 to 4.23)	0.038	0.860
Insomnia	Genus Burkholderiales noname	Dutch	13	2.70 (1.05 to 6.94)	0.039	0.860
Insomnia	Species Burkholderiales bacterium 1-1-47	Dutch	13	2.69 (1.05 to 6.92)	0.040	0.860
Insomnia	Species Lachnospiraceae bacterium 7-1-58FAA	Dutch	13	1.97 (1.02 to 3.81)	0.043	0.860
Insomnia	Genus Subdoligranulum	Dutch	13	1.87 (1.01 to 3.45)	0.045	0.860

CI, confidence interval; Dutch, Dutch Microbiome Project; FDR, false discovery rate; IV, instrumental variable; MiBioGen, MiBioGen alliance; OR, odds ratio.

The effect directions of the WM were consistent with the IVW estimate. No evidence of pleiotropy was found according to the MR-Egger regression intercept analysis and the MR-PRESSO test (online supplemental tables S4 and S5). Finally, in the LOO sensitivity analysis, positive causal relationships exhibited high overall robustness. Specifically, the following taxa showed stable positive causal relationships with insomnia: genera Clostridium innocuum group, Prevotella 7, Lachnoclostridium, Paraprevotella, Family XIII AD 3011 group, Rikenellaceae RC 9 gut group, Parabacteroides, Odoribacter and Pseudoflavonifractor, family Clostridiaceae and species Veillonella unclassified. In a reverse analysis, the family Alcaligenaceae, class Betaproteobacteria, order Burkholderiales, genus Odoribacter and species Streptococcus thermophilus demonstrated relationships that were robust and causal. However, LOO analysis on other datasets identified certain SNPs that may influence causal effects, thus calling for caution when interpreting these results (online supplemental figures S2–S5).

DISCUSSION

Main findings

Our study revealed causal associations between gut microbiota and insomnia. We identified a total of 14 and 8 bacterial taxa, respectively, as positively and negatively correlated with insomnia. In addition, there were reverse effects of insomnia correlated with 19 identified microbial taxa. We also discovered bidirectional causal relationships between the genus Odoribacter sourced from the Dutch Microbiome Project and insomnia. Following the execution of the Steiger test, these positive causal relationships were confirmed as being significant.

In recent years, some studies have delved into the influence of gut microbiota on mental illness.35,37 The majority of these studies were performed using one microbial dataset. Here, we explored two datasets, the Dutch dataset and the MiBioGen dataset, which showed a non-negligible degree of overlap with 71 microbial taxa in common. Our study’s results agree with many previous studies, which have suggested an interaction between insomnia and the gut microbiota,^{11 38} mostly through the so-called ‘gut–brain axis’. For instance, Li et al have described associations of order Selenomonadales, class Negativicutes and genera Prevotella 7, Clostridium innocuum group and Lachnoclostridium with a higher risk of insomnia, which is consistent with our results.³⁹ However, in the same work, genera Coprococcus1 and Family XIIIAD3011 group, both highlighted by our study, were not associated with the risks of insomnia. This discrepancy may be explained by the differences in input datasets. While Li et al extracted their microbial data from the MiBioGen, their insomnia dataset was limited to the UKB only. Notably, another MR study, which also utilised insomnia data from the UKB, confirmed that the genus Family XIIIAD3011 group increased the risk of insomnia, consistent with our results.⁴⁰ On behalf of the genus Coprococcus 1, we should add that it has been confirmed as the producer of both serotonin and SCFAs, which have been found to be deficient in depression and other mental disorders.^{41 42} The role of these molecules in the phenotypes of insomnia requires further investigation.

Unfortunately, the heterogeneity of the computational methods employed in previous studies precluded direct comparison of the results presented here and those obtained previously. As we employed a unified analytic framework, the results of our study may be more representative than those obtained in previous studies. When inferring the bidirectional causal relationship between exposure and outcome, we adopted the Steiger test, which enhances the robustness of causal inference.

It is widely discussed that the gut microbiome regulates the gut–brain axis through endocrine, neuronal and immune signalling pathways.⁴³ In addition to synthesising vitamin K and the group B vitamins, the gut microbiota also produces other metabolites such as SCFAs as well as the bioactive derivatives of tryptophan and other amino acids.⁴⁴ In particular, the genus Clostridium innocuum group may synthesise acetate via either the acetyl-CoA pathway or the Wood–Ljungdahl pathway, while the genus Coprococcus 1 makes propionate via the succinate pathway. Additionally, the genus Clostridium innocuum group is capable of metabolising tryptophan by its tryptophanase.⁴⁵

Tryptophan, acting as the precursor of L-5-hydroxytryptophan (5-HTP),⁴⁶ can generate serotonin (5-HT) via a cascade of reactions; Subsequently, 5-HT undergoes further conversion into its derivative with somnogenic properties, melatonin (N-acetyl-5-methoxytryptamine).⁴⁷ Both serotonin and melatonin aid in establishing sleep patterns and participating in the regulation of emotions, cognition, response to reward, learning, memory and many other psychophysiological processes.⁴⁸

SCFAs are widely recognised as the main microbial metabolites derived from dietary fibres. Among these, butyrate, propionate and acetate have been shown to activate a variety of G protein-coupled receptors in human cells.⁴⁹ SCFA receptors GPR43 and GPR41, now referred to as FFAR2 and FFAR3,⁵⁰ appear to be highly expressed in the brain. The interaction between SCFAs and certain gut–brain pathways may directly or indirectly regulate neurological function, learning, memory and emotion.⁴⁹ Interestingly, in infants, higher faecal concentrations of propionate are associated with longer uninterrupted sleep duration,⁵¹ while in elderly individuals with insomnia symptoms, higher acetate, butyrate and propionate concentrations are associated with lower sleep quality.⁵² In rats, non-REM sleep may be improved by tributyrin, a precursor of butyrate.⁵³

The intestine is the place in the body with the highest number of immune cells. In this milieu, immune cells are constantly exposed to various antigens and possible immune stimuli. SCFAs generated by certain bacteria of the genus Clostridium in the colon are capable of affecting the immune balance of the intestine in a dramatic way.⁵⁴

Previous studies have shown that the relative abundance of Bacteroides, Coprococcus and Prevotella differs significantly in patients with major depressive disorder, bipolar disorder and schizophrenia from that in control subjects.⁵⁵ The abundance of Coprococcus in the faecal microbiota of Parkinson’s disease patients is also significantly lower than that of controls.⁵⁶ The increase in the abundance of Prevotella in the colonic mucosal area is associated with both local and systemic diseases,⁵⁷ including periodontitis, bacterial vaginosis, rheumatoid arthritis, metabolic disorders and low-grade systemic inflammation. Some Prevotella strains are clinically important as pathogenic factors promoting chronic inflammation. Microbial communities enriched in the genera Lachnoclostridium, Coprococcus 1 and Lachnospiraceae UCG-001 are associated with symptoms of depression.⁴¹ These previously discovered links with depression may also be related to the occurrence and development of insomnia.

Conversely, established insomnia may lead to changes in the microbial community by enhancing the expression of virulence genes, which may be induced by stress mediators released by the ENS.³⁸ Elevated levels of interleukin-6 (IL-6) have been observed in primary insomnia patients.⁵⁸ Transplanting human gut microbiota from sleep-deprived individuals into germ-free recipient mice induces an increase in IL-6 levels in the serum.⁵⁹ The levels of certain neurotransmitters, such as serotonin and dopamine, may also change in insomnia and, in turn, influence the gut microbiota that releases these neurotransmitters. Researchers also reported that certain neuronal communities within the depleted-neuron mice can modulate the composition of the gut microbiota.⁶⁰

Overall, the intertwined effects of insomnia on gut microbiota and vice versa represent a complex bidirectional relationship involving immune regulation, inflammatory response, release of neurotransmitters, and other molecular and cellular pathways.

Mental and metabolic disorders are closely related to the disruption of the biological clock. It was reported that dysregulation of the circadian rhythm of microbial hosts can significantly affect microbial oscillations. When circadian rhythms are disrupted, brain signals and the intestinal circadian clock also become disrupted, which may lead to imbalances in gut microbiota, bacterial migration, invasion and inflammation, thereby increasing the risks of metabolic disturbance. Similarly, the disruption of gut microbiota can lead to changes in the peripheral and intestinal clocks, and microbiota-derived metabolites and SCFAs can affect the host’s biological clock.⁶¹ This complexity urges further exploration of the causal connections between gut microbiota and insomnia.

Limitations

This study used MR analysis, which helped to avoid confounding factors and reverse causality. While multiple sensitivity analyses ensured the robustness of the results, there are still limitations that have to be acknowledged. As all the study participants were of European descent, caution should be exercised when generalising the results to other populations, as the microbiome composition varies between different ethnicities and geographical backgrounds. Due to differences in available data sources, uncontrollable confounding factors may not be excluded. TSMR analysis relies on specific assumptions, especially the assumption that no gene–environment interaction exists between the samples. In the study of microbiome diversity, environmental factors such as diet, lifestyle and geographical differences impact the composition of the microbiome, and gene–environment interactions cannot be ignored. MR only considers genetic components that contribute to gut microbiota composition and insomnia, so caution should be exercised when interpreting the results. Gut microbiota may be influenced by other factors such as dietary habits and health status, which also have genetic components. Currently, we cannot conclude whether genetic factors are related to these confounding factors or the abundance of bacterial taxa. Finally, although our study identified specific gut microbiota associated with the risk of insomnia, this mainly provides predictive insights. Future research should focus on elucidating the underlying mechanisms, such as those involving SCFAs and serotonin pathways. Concomitantly, randomised controlled trials assessing microbiome-targeted interventions, such as probiotics, prebiotics or faecal microbiota transplantation, would offer critical translational insights into therapeutic modulation of the gut–brain axis in insomnia.

Implications

Our study offers preliminary evidence supporting a causal effect between insomnia and gut microbiota, providing valuable insights for the future development of microbiome-inspired treatment plans for insomnia.

Biography

Shangyun Shi is currently pursuing a master's degree at Nanjing Medical University in China. Her primary research interests lie in the genetic mechanisms underlying mental disorders. This involves conducting genetic association studies and bioinformatics analyses using large-scale genetic datasets. Under the guidance of her tutor and the research team, who have delved deeply into Mendelian randomization (MR) and related genetic analyses, she has actively participated in the research work. The team has published dozens of articles in this field. Their research topics cover a wide range, including post-traumatic stress disorder and major depressive disorder, the relationship between depression and insomnia, as well as the connection between the gut microbiome and Alzheimer's disease.

Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information.