Refined Causal Association Between Physical Activity Intensity and Depression: A Comprehensive Two-sample Mendelian Randomization Study

doi:10.21203/rs.3.rs-5050662/v1

Depression remains a significant public health issue with high morbidity and mortality worldwide, making it crucial to investigate the role and causality of different types of physical activities in relieving depression. This study utilized a two-sample Mendelian randomization (MR) approach to evaluate the causal relationship between various physical activities, strenuous sports and walking for pleasure, and both general and major depression. We used summary statistics from large-scale genome-wide association studies involving European ancestry participants. Causal associations were estimated using the inverse-variance weighted model, with additional validation through extensive sensitivity analyses. Meta-analysis was conducted to comprehensively estimate the effects of multiple MR analyses. Results from single MR analyses and meta-analyses showed no significant causal effect of strenuous sports on depression (general depression: logOR=-0.1405, p = 0.2984; major depression: logOR = 0.0453, p = 0.9293). In contrast, walking for pleasure demonstrated a significant causal effect in reducing general depression (logOR=-0.5398, p = 0.0069) and major depression (logOR=-1.9786, p = 0.0103). Reverse MR analyses found no causal association between genetic liability to depression and the physical activities tested. Our findings underscore the significant role of walking for pleasure in alleviating and preventing both general and major depression. However, strenuous sports do not show a statistically significant causal effect. Further research is warranted to explore physical activities with intensities similar to walking for pleasure as potential preventive or rehabilitative measures in depression therapy, either alone or alongside psychotherapy or pharmacotherapy.

Depression, or depressive disorders, is a mood disorder that can cause a variety of emotional and physical problems and negatively impacts people’s ability to function for both work and daily activities. According to National Institute of Mental Health, there are two common types of depression, major depressive disorder, a persistent feeling of sadness and loss of interest for at least 2 weeks; and dysthymia (i.e., persistent depressive disorder), consisting of less severe symptoms of depression but last much longer compared to major depression, usually for at least 2 years. In 2017, depression was the third leading cause of non-fatal health loss worldwide. ¹ Of note, the number of years lived with disability due to depression increased more than 30% from 1990 to 2007 and 14% from 2007 to 2017, globally.¹ In the United State, approximately one in five of adults reported having ever been diagnosed with depression in lifetime.² Per the guideline of Clinical Practice Guideline for the Treatment of Depression Across Three Age Cohorts, by American Psychiatry Association, the recommended treatment options for management of depression include psychotherapy and pharmacotherapy (i.e., antidepressant medications).³ However, the reported remission rate of treating depression in clinical trials are only 22–40% with an even lower remission rate, around 10–30% reported by clinical practice report.⁴ Importantly, even with remission, patients with depression are at substantial risk of recurrence with 60% lifetime risk of recurrence after the first major depressive episode.⁵ Regardless of poor treatment outcome of depression, only approximately 50% of patients with depression seek or talk to a doctor or other professional about their symptoms of depressive disorders and even less percent of people takes a prescription medication for treating the symptoms in the whole patient population due to stigma, financial burden, concerns or intolerance to side effects by antidepressants, etc.⁶

Physical activity is a promising prevention and intervention approach for treating symptoms and reducing incidence of depression as it precludes from stigma, is free of adverse effects and no cost or much lower cost compared to the recommended standard psychotherapies and pharmacotherapies. For adult patients with depression for whom psychotherapy or pharmacotherapy is either ineffective or unacceptable, the guideline recommends exercise monotherapy.³ Recently, many studies have found that physical activity has beneficial effects to prevent the incidence of depression or improve depression symptoms.^7–9 A meta-analysis identified resistance exercise training significantly reduced depressive symptoms.¹⁰ Similarly, a systematic review reported a significant, inverse relationship between baseline physical activity and follow-up depression.⁷ However, observational cohort studies and relevant meta-analyses are unable to establish a causal inference between exposure and outcome due to potential confounding issues. Moreover, previous quantitative studies, primarily comprising observational inquiries, have not consistently aligned their conclusions regarding the observed association between engaging in physical activities and the incidence of depression.^11–13 Although most previous research supports the notion that exercise is associated with reduced odds of incident depression compared to non-exercising participants, the causal nature of this association remains unclear. Conventionally, randomized controlled trials (RCTs) are regarded as the gold standard for providing evidence on causal associations. However, implementing an RCT can be challenging or even impossible at times due to ethical restrictions. Mendelian randomization (MR) analysis overcomes the confounding issues, the major limitation of observational studies, by using genetic variants in non-experimental data to make causal inference about the effect of an exposure on the outcome, therefore, can be utilized to evaluate the causal relationship between the exposures and outcomes.

Choi et al. also reported a protective relationship between objectively assessed (e.g., accelerometer-based) physical activity and the risk for major depression using MR analysis.¹⁴ However, several questions remain. First, the study only assessed the relationship for major depression, not for general depression disorder, which represents persistent and chronic depression symptoms. Second, the study only evaluated accelerometer-based and self-reported physical activities and did not assess the relationship between different intensity of physical activity and depression. Aiming to accurately assess the causal effects of physical activities in various intensities on both general depression and major depression and to identify their specific associations, our research conducted a two-sample MR analysis using accessible genome-wide association study (GWAS) datasets. The exposures include two types of physical activities: strenuous sports and walking for pleasure. The outcome, depressive disorder, was classified into two categories: general depression and major depression. Furthermore, in the present study, the meta-analysis provided comprehensive evaluations of the causal effects through multiple MR analyses on the common pathway of causality by integrating these results.

The data utilized in this MR study were derived from publicly available summary-level GWAS datasets. All original studies obtained appropriate written consent from participants, which included provisions for data sharing and analyses. The current study has been reviewed and approved by the Institutional Review Board (IRB), ensuring compliance with ethical standards. Given that this research involves minimal risk and uses deidentified data, a waiver of consent was granted by the Institutional Review Board (IRB), in accordance with institutional policies and relevant guidelines. The summary-level GWAS datasets included a combined sample of 2,386,832 adult European participants, with corresponding GWAS sample sizes detailed in Table 1. All GWAS data for this study were analyzed between January 22 and April 28, 2024.

Table 1

Summary of GWAS datasets on exposures and outcomes
	Phenotype/Trait	GWAS ID	Consortium	Sample Size	Number of SNPs	PubMed ID
Exposures
	Types of physical activity in last 4 weeks: Strenuous sports	ukb-b-7663	UKBB	460,376	9,851,867	30305743
	Types of physical activity in last 4 weeks: Walking for pleasure (not as a means of transport)	ukb-b-7337	UKBB	460,376	9,851,867	30305743
Outcomes
	Major Depressive Disorder	ieu-a-1188	PGC	173,005	13,554,550	29700475
	Major Depressive Disorder	ebi-a-GCST90086060	IEU	56,637	14,462,316	33893285
	Major Depression	ieu-b-102	PGC	500,199	8,633,288	30718901
	Depressive Symptoms	ieu-b-4840	WithinFamily	47,517	8,281,749	35534559
	Depressive Symptoms	ieu-a-1000	SSGAC	161,460	6,524,475	27089181
	Depression	ebi-a-GCST90018833	IEU	449,414	24,184,163	34594039
	Depression (broad)	ebi-a-GCST005902	IEU	322,580	7,624,934	29662059
	Depression	finn-b-F5_DEPRESSIO	FinnGen	215,644	16,380,457	36653562

2.1 Study Design

The design of a Mendelian randomization study typically relies on observational datasets, where genetic variants, SNPs extracted from summarized GWAS datasets, are utilized as instrumental variables (IVs) to investigate potential causal relationships between the exposure of interest and the outcome. In our research, the entire study design for the two-sample Mendelian randomization (MR) analysis was developed as depicted in Fig. 1. Various types of physical activities undertaken in the last four weeks were considered as exposures, while general depression and major depression were regarded as the outcomes. To conduct the two-sample MR analysis, three core assumptions must be satisfied:

the single nucleotide polymorphisms (SNPs) used as instrumental variables (IV) should be strongly associated with the exposure of interest (the relevance assumption).
the IVs should exclusively influence the outcome through the exposure and not exhibit pleiotropic effects (the exclusion-restriction assumption).
the IVs should not be associated with any confounding factors related to the exposure-outcome relationship (the independence assumption).

2.2 Data Source

Genome-wide association study summary-level statistics for different types of physical activities taken in past four weeks, and general depression & major depression were accessible from UK Biobank (UKB) (https://biobank.ctsu.ox.ac.uk/), IEU OpenGWAS project (IEU) (https://gwas.mrcieu.ac.uk/), Finnish Genomics project (FinnGen) (https://public-metaresults-fg-ukbb.finngen.fi/), the Within family consortium (WithinFamily) (https://www.withinfamilyconsortium.com/), the Psychiatric Genomics Consortium (PGC) (https://www.med.unc.edu/pgc/), and the Social Science Genetic Association Consortium (SSGAC) (https://thessgac.org/). There were no samples overlaped among the outcome GWAS datasets (general depression & major depression). To avoid the bias of population heterogeneity, all participants of the summarized GWAS data were completely from the European adult population. Detailed information of summary-level GWAS datasets utilized in our two-sample MR analyses was summarized in Table 1.

2.3 Instrumental Variables Extraction

The extraction of IVs is a crucial step before carrying out the two-sample MR analyses. This is because SNPs as exposure proxies must ensure the valid causal inference. Therefore, the SNPs included in this research should satisfy three core assumptions for IVs (Fig. 1). First, the SNPs strongly associated with the two exposures were extracted at a genome-wide significance threshold of p-value < 5×10^− 8. Second, applying linkage disequilibrium (LD) clumping with PLINK (v1.8)¹⁵ ensured the selection of independent SNPs (LD threshold: LD r² < 0.001, LD region width > 10,000 kb), in which sequencing data of the 1000 Genomes Project Phase 3 was set to estimate LD. Then, the F-statistics, defined as R²(N-2)/(1-R²) (N: sample size of participants; R²: variance in the exposure explained by SNPs),¹⁶ were calculated for assessing the strength of each filtered SNP. The SNPs with the value of F-statistics less than 10 were considered the weak IVs and must be excluded.

To eliminate the horizontal pleiotropy bias and avoid violate the second assumption, all independent SNPs were verified for by using PhenoScanner online database of genotype-phenotype associations (https://www.phenoscanner.medschl.cam.ac.uk) (date checked March. 2024), identifying if selected SNPs had association with secondary phenotypes (p-value < 5×10^− 8). Palindromic SNPs with minor allele frequencies (MAF) above 0.42, or SNPs with allele mismatches, were omitted during data harmonization. Ultimately, a collection of genome-wide-significant SNPs as IVs were extracted from the summarized GWAS data of the exposure. Comprehensive information on IVs was presented in Supp. Tables 1 & 2.

Table 2

Summary of two-sample MR analyses
Exposure	Outcome	IVW Regression				Weighted Median			Power
Exposure	Outcome	logOR	SE	p-value	OR	logOR	SE	p-value	Power
ukb-b-7663	ieu-a-1188	0.5261	0.7668	0.4927	1.6922	1.1141	1.0234	0.2763	94.20%
	ebi-a-GCST90086060	-2.7890	2.5166	0.2677	0.0615	-3.3281	3.1010	0.2832	100.00%
	ieu-b-102	-0.1417	0.7107	0.8420	0.8679	-0.0460	0.6855	0.9465	12.20%
	ieu-b-4840	-1.5322	0.8833	0.0828	0.2161	-1.0990	0.9487	0.2467	100.00%
	ieu-a-1000	-0.2023	0.5242	0.6996	0.8169	-0.9227	0.5509	0.0940	41.60%
	ebi-a-GCST90018833	-0.0811	1.3236	0.9511	0.9221	-0.2055	1.7703	0.9076	5.00%
	ebi-a-GCST005902	-0.1153	0.1437	0.4224	0.8911	0.0003	0.1635	0.9986	98.50%
	finn-b-F5_DEPRESSIO	0.7175	1.0664	0.5011	2.0492	-0.2367	1.4402	0.8694	93.30%
ukb-b-7337	ieu-a-1188	-2.9363	0.7241	0.0001	0.0531	-3.8044	0.7527	4.316×10^− 07	100.00%
	ebi-a-GCST90086060	0.8385	1.2991	0.5186	2.3129	1.2453	1.6837	0.4595	100.00%
	ieu-b-102	-2.5333	0.3408	1.064×10^− 13	0.0794	-2.4096	0.4463	6.705×10^− 08	100.00%
	ieu-b-4840	0.2915	0.4111	0.4784	1.3384	-0.0476	0.4908	0.9227	100.00%
	ieu-a-1000	-0.8677	0.2460	0.0004	0.4199	-1.0306	0.2841	0.0003	100.00%
	ebi-a-GCST90018833	-0.9668	0.8006	0.2272	0.3803	-0.3126	1.0092	0.7567	100.00%
	ebi-a-GCST005902	-0.4609	0.0916	4.906×10^− 07	0.6307	-0.5267	0.1111	2.114×10^− 06	100.00%
	finn-b-F5_DEPRESSIO	-1.9266	0.7053	0.0063	0.1456	-1.5648	0.8955	0.0806	100.00%

2.4 Statistical Methods

The two-sample MR analyses were conducted by using the R package “TwoSampleMR”¹⁷ and “MR-PRESSO”.¹⁸ The priori statistics power for each MR analysis were calculated from the Shiny web application (https://sb452.shinyapps.io/power/).¹⁹

To measure the causal effect of two types of physical activities taken in last four weeks on general depression & major depression, two-sample MR analyses were mainly implemented by using the Inverse-Variance Weighted (IVW) approach,²⁰ and the weighted median approach.²¹ The MR-Egger regression approach,²² the MR-Pleiotropy RESidual Sum and Outlier (MR-PRESSO) approach¹⁸ were employed for detecting outliers, heterogeneity and horizontal pleiotropy. The IVW regression method provided robust and accurate estimates of the causal effect under the assumption of no horizontal pleiotropy. Consequently, the primary two-sample MR analyses relied on the IVW method. When heterogeneity is detected in the absence of pleiotropy, using the weighted median method can yield robust estimates in MR analysis.²¹

To assess the reliability of the results in two-sample MR analysis, it is crucial to conduct comprehensive sensitivity analysis, which helps to check and address underlying horizontal pleiotropy and heterogeneity. The MR-Egger intercept test can detect potential violations of the second assumption regarding horizontal pleiotropy with p-value below 0.05. Furthermore, MR-PRESSO method can discover the existence of overall horizontal pleiotropic effect by implementing Global tests and check the horizontal pleiotropic outliers by taking outlier tests. After excluding the outlier SNP with an individual p-value below 0.05 in the outlier test, the MR-PRESSO Global and outlier tests were repeated for the remaining SNPs until the Global test yielded statistical insignificance (p-value > 0.05). The subsequent two-sample MR analysis was carried out on the remaining SNPs. Cochran’s Q test detected the heterogeneity among SNPs in IVW regression, where Q p-value above 0.05 indicates that the null hypothesis of no heterogeneity cannot be rejected.²³ Leave-one-out analyses are a useful tool for detecting genetic variants that dominate the estimate of the causal effect. The removal of such SNPs can enhance the robustness of the findings in MR analyses. The reverse two-sample MR analyses followed the same procedure as the described above.

The pooled causal effects of two types of physical activities in the past four weeks on multiple GWAS datasets of general depression and major depression were estimated respectively through meta-analysis using the R package “metafor”.²⁴ Cochran’s Q statistics test and Higgins I² index statistics in meta-analysis were used to examine the heterogeneity among multiple estimates of causal effects of two exposures in each group of two-sample MR analyses. If the p-value in Q test was greater than 0.05 (I² < 50%), indicating no heterogeneity in meta-analysis, the fixed-effects model was deemed appropriate to use; otherwise, under the presence of heterogeneity in meta-analysis, the random-effects model was considered more suitable. Moreover, Egger’s test was used to evaluate publication bias for a meta-analysis.²²

3.1 Causal Effects of Strenuous Sports on General Depression & Major Depression

In the assessment of the causal effect of strenuous sports undertaken in the last four weeks on general depression as well as major depression, we utilized eight summary-level GWAS datasets as the outcome, comprising 3 datasets for major depression, 5 datasets for general depression. During three two-sample MR analyses with major depression as the outcome, the SNP rs7617323 was detected as an outlier in the GWAS dataset ieu-b-102 through the MR-PRESSO outlier test and was subsequently deleted from the collection of IVs (Supp.Table 3). The causal effects of taking strenuous sports in the past four weeks on major depression did not achieve statistical significance in the regression of the IVW model (OR = 1.6922 with p-value = 0.4927 for outcome GWAS dataset ieu-a-1188; OR = 0.0615 with p-value = 0.2677 for outcome GWAS datasets ebi-a-GCST90086060; OR = 0.8679 with p-value = 0.8420 for outcome GWAS datasets ieu-b-102). While, in five two-sample MR analyses with general depression as the outcome, there was no outlier detected. The results from IVW regression revealed there were no significant causal effects of taking strenuous sports in the last four weeks on general depression (OR = 0.2161 with p-value = 0.0828 for outcome GWAS dataset ieu-b-4840; OR = 0.8169 with p-value = 0.6996 for outcome GWAS dataset ieu-a-1000; OR = 0.9221 with p-value = 0.9511 for outcome GWAS dataset ebi-a-GCST90018833; OR = 0.8911 with p-value = 0.4224 for outcome GWAS dataset ebi-a-GCST005902; OR = 2.0492 with p-value = 0.5011 for outcome GWAS dataset finn-b-F5_DEPRESSIO) (Table 2).

None of the eight two-sample MR analyses with taking strenuous sports in the last four weeks as the exposure exhibited heterogeneity or horizontal pleiotropy (Supp.Table 3). The statistical power in the five two-sample MR analyses were sufficient (> 80%), excluding the MR analysis with outcome GWAS datasets ieu-b-102 (power:12.2%), ieu-a-1000 (power:41.6%), and ebi-a-GCST90018833 (power:5%) (Table 2).

3.2 Causal Effects of Walking for Pleasure on General Depression & Major Depression

In evaluating the causal effect of walking for pleasure in the last four weeks on major depression, outliers were identified in the two-sample MR analyses with the outcome GWAS dataset ieu-a-1188 (rs2670012, rs2910829, rs34224566, rs34898535) and ieu-b-102 (rs2910829, rs34898535, rs7133277), by taking the MR-PRESSO outlier test, and were subsequently removed from the collection of screened SNPs. The causal effects of walking for pleasure in the last four weeks on major depression were found to be statistically significant using the IVW regression method with two out of the three outcome GWAS datasets (OR = 0.0531 with p-value = 0.0001 for outcome GWAS dataset ieu-a-1188; OR = 0.0794 with p-value = 1.064×10^− 13 for outcome GWAS datasets ieu-b-102). However, the significant causal relationship was not observed in the two-sample MR analysis using the outcome GWAS dataset ebi-a-GCST90086060 (OR = 1.2991 with p-value = 0.5186).

In the five two-sample MR analyses with general depression as the outcome, the MR-PRESSO outlier test indicated that outliers exist in the collection of SNPs of two two-sample MR analyses with the outcome GWAS datasets ieu-b-4840 (rs11896330), ebi-a-GCST005902 (rs11648192, rs2910829, rs34224566, rs34898535, rs7207400), and finn-b-F5_DEPRESSIO (rs11648192, rs7133277) (Supp.Table 3). After removing outliers, the five two-sample MR analyses were conducted using IVW regression. Among them, three analyses revealed significant causal effects of walking for pleasure in the last four weeks on general depression (OR = 0.4199 with p-value = 0.0004 for outcome GWAS dataset ieu-a-1000; OR = 0.6307 with p-value = 4.906×10^− 7 for outcome GWAS dataset ebi-a-GCST005902; OR = 0.1456 with p-value = 0.0063 for outcome GWAS dataset finn-b-F5_DEPRESSIO). Nevertheless, the remaining two analyses did not demonstrate significant causality (OR = 1.3384 with p-value = 0.4784 for outcome GWAS dataset ieu-b-4840; OR = 0.3803 with p-value = 0.2272 for outcome GWAS dataset ebi-a-GCST90018833) (Table 2).

Heterogeneity tests and MR-Egger intercept tests confirmed the absence of heterogeneity and pleiotropy in all eight two-sample MR analyses with the exposure, walking for pleasure in last four weeks (Supp.Table 3). The statistical powers of them were consistently sufficient (> 80%).

3.3 Meta-analysis for multiple MR analyses

The causal effects of engaging in pleasurable walking over the last four weeks varied across different GWAS studies on the outcomes of general depression and major depression. Meta-analysis provided a comprehensive evaluation of this exposure by pooling the results from multiple two-sample MR analyses.

The overall causal effect of engaging strenuous sports in the last four weeks on major depression was not statistically significant (logOR = 0.0453, SE = 0.5104, p-val = 0.9293, OR = 1.0463). The significant heterogeneity was not observed in this meta-analysis (Q = 1.7308, p-val = 0.4209, I² = 0.00%) (Fig. 2). Similarly, the overall causal effect of engaging strenuous sports over the last four weeks on general depression were not statistically significant (logOR=-0.1405, SE = 0.1351, p-val = 0.2984, OR = 0.8689). There was no heterogeneity among multiple two-sample MR analyses in this meta-analysis (Q = 3.1762, p-val = 0.5288, I² = 0.00%), indicating consistency among the findings of all five two-sample MR analyses regarding the estimated causality. Notably, both two meta-analyses did not detect any publication bias in Egger’s tests (z = -1.0974, p-val = 0.2725 for meta-analysis in Fig. 2; z = -0.4224, p-val = 0.6727 for meta-analysis in Fig. 3).

The overall causal effect of walking for pleasure in the last four weeks on major depression was statistically significant (logOR=-1.9786, SE = 0.7711, p-val = 0.0103, OR = 0.1383). However, significant heterogeneity was observed in this meta-analysis (Q = 6.910, p-val = 0.032, I² = 71.06%), indicating inconsistency among the findings of all three two-sample MR analyses regarding the estimated causality (Fig. 4). Also, the overall causal effect of walking for pleasure over the last four weeks on general depression were statistically significant (logOR=-0.6051, SE = 0.2398, p-val = 0.0116, OR = 0.5460) (Fig. 5). There was heterogeneity observed among the multiple two-sample MR analyses in this meta-analysis (Q = 10.5521, p-val = 0.0321, I² = 62.09%). Notably, both two meta-analyses did not detect any publication bias in Egger’s tests (z = 1.4719, p-val = 0.1411 for meta-analysis in Fig. 4; z = -0.8970, p-val = 0.3697 for meta-analysis in Fig. 5).

3.4 Reverse Two-sample MR Analysis

Reverse two-sample MR analyses were also conducted for assessing the causal impacts of general depression & major depression on two types of physical activities in the last four weeks (strenuous sports & walking for pleasure). In the context of homogeneous MR analyses, no significant effects were detected in the reverse path of the two-sample MR analyses (Supplementary Table 4).

Depression is a chronic disease with high morbidity and mortality and results in significant impairment in day-to-day functioning. The mortality rate in patients with depressive disorder is approximately 40–150% higher than the general population.²⁵ The significant challenges to combat depression include high incidence of relapse and recurrence, low intention of professional help seek, and associated issues with the currently standard pharmacotherapy and psychotherapy such as high cost, accessibility, side effects. Due to lack of effectiveness of the currently recommended standard of care, participating in physical activities as a potential effective therapy or preventive measure for depression has gained substantial attention and emerging as an alternative therapy for treating depression.

A recently published paper by Noetel, M., et al., utilizing pooled data from 41 studies involving 2264 patients with depression, identified that aerobic exercise and resistance training have a large effect favoring exercise over control conditions. The majority of the 41 studies applied moderate to vigorous intensity exercises.¹² In terms of the association between various intensity of physical activities and depression, Loprinzi reported that both light and moderate-to-vigorous physical activities were inversely associated with depression after controlling for age, gender, body weight, and race, etc. using data from the 2005–2006 National Health and Nutrition Examination Survey in old adults (65 + years of age).²⁶ A small cohort study in 24 women diagnosed with major depression also observed exercise of any intensity significantly improved feelings of depression.²⁷ However, findings from these observational epidemiology studies are susceptible to bias due to potential confounding factors that may have impact on outcomes and/or reverse causation (i.e., where the outcome influences the exposure), which are well-known limitations of this type of studies.

In this study, we leveraged genetic IVs to comprehensively assess the causal association between two types of physical activities at different intensities (strenuous sports and walking for pleasure) and two distinct severities of depression (general depression and major depression). This research was conducted through MR analyses for single causal pathways and the meta-analysis for overall estimates of causal effect, allowing for a robust examination of causality. Choi et al. used bi-directional MR analysis to explore the association between physical activity and a reduced risk of depression.¹⁴ However, a notable limitation of their study is the lack of a specific definition for the exposure, accelerometer-based activity. It is unclear what types and intensity of the physical activities were included in their study as accelerometer-based activity. Despite evidence supporting the protective impact of accelerometer-based activity on major depressive disorder, their conclusion may not offer practical guidance for psychological therapy in individuals with major depression due to lack of information on the exposure, accelerometer-based activity. Our study revealed distinct causal effects between two types of physical activities on depression. Engaging in strenuous sports over the past four weeks did not yield a significant overall causal effect on alleviating either general depression or major depression. In contrast, walking for pleasure during the same period demonstrated a causal benefit in alleviating both general depression and major depression. This benefit was evident not only in most individual causal pathways but also in the overall estimate of causal effects on depressions, as assessment by meta-analysis. The non-significant effects observed in all inverse two-sample MR analyses consistently indicate that there is no statistically valid causal effect of the risk of two different types of depression (i.e., major depression and general depression) on the two types of physical activity (i.e., strenuous sports and walking for pleasure). This finding confirmed the unique direction of causality between physical activity and depression.

Supported by two-sample MR analyses from multiple GWAS datasets, the results in this research contribute to a deeper understanding of the potential causal relationship between exercise-taking and depression, emphasizing the role of physical activity at various intensities in preventing and treating depression. Previous epidemiological meta-analyses have confirmed the effective symptom relief for depression patients in observational studies.^12,28–30 However, these estimates reflect average effects across various types of physical activities without specific distinctions. According to the results in meta-analyses, our study discovered that, compared to engaging in high-intensity physical activity, such as strenuous sports, walking for pleasure, which represents a convenient form of light physical activity, was proved in our study to have a more effective impact on reducing both general depression and major depression. Therefore, engaging in walking for pleasure has the potential to serve as a therapeutic and protective measure for patients in alleviating general depression or major depression.

Our research is subject to several limitations. First, the MR analyses were exclusively conducted in European populations to mitigate bias from population stratification. However, caution must be exercised in generalizing our findings to other populations. Second, although most individual two-sample MR analyses, as well as overall effect, indicated that engaging in walking for pleasure had a statistically significant protective effect on the two types of depression, it is essential to conduct MR analyses across different age and gender groups if detailed demographic information is available in the future GWAS datasets. A more accurate causality estimation will be helpful in providing a specific guideline for making prevention and therapeutic strategies for different patients with depression.

Although addressing these limitations is urgent, this study integrated two-sample MR analyses with genetic IVs and meta-analysis using multiple cohort GWAS datasets to statistically provide robust evidence for identifying the potential causality between physical activity at different intensities and various types of depression. The evaluation of causal effects of different physical activities in alleviating depression carries significant implications for public health. This finding underscores the importance of promoting physical activity as a key component of preventive and therapeutic interventions for depression at a population level. Public health initiatives can now leverage this evidence to design and implement targeted programs aimed at encouraging regular exercise (e.g. walking for pleasure) to improve mental well-being and reduce the burden of depression-related disorders. From the perspective of psychiatric diseases, the recognition of taking proper physical activity as having a causal impact on reducing depression sheds new light on treatment strategies. Clinicians and mental health professionals can integrate exercise prescriptions into comprehensive treatment plans for patients with depression. This approach not only complements traditional pharmacological and psychotherapeutic interventions but also empowers individuals with depression by offering them a proactive and holistic approach to managing their mental health.

Our research is the first comprehensive study that evaluated the causal relationship between different intensity of physical activities and depression. This paper strengthened the evidence on causality supporting the role of physical activity in alleviating depression and major depressive disorder, while also revealing the impact of various types and intensities of physical activities on depression. The primary findings of this study endorse walking for pleasure over a four-week period as a cost-effective, and easily accessible approach in preventing the worsening of depression. However, engaging in strenuous sports does not exhibit a statistically significant causal effect in this regard. Physical activities with intensity similar to walking for pleasure can be further investigated as a preventive measure or a rehabilitation method in the therapy for general depression or major depression, either independently or in conjunction with psychotherapy or pharmacotherapy.

Author Contribution

JP designed the entire study, analyzed the data and wrote the main manuscript text and XW prepared literature review. All authors reviewed the manuscript.

Data Availability

All GWAS summary statistics analysed in this study are publicly available for download by qualified researchers (https://gwas.mrcieu.ac.uk/).

James SL, Abate D, Abate KH, Abay SM, Abbafati C, Abbasi N, Abbastabar H, Abd-Allah F, Abdela J, Abdelalim A, Abdollahpour I. Global, regional, and national incidence, prevalence, and years lived with disability for 354 diseases and injuries for 195 countries and territories, 1990-2017: a systematic analysis for the Global Burden of Disease Study 2017. The Lancet. 2018;392(10159):1789-858.
Lee B, Wang Y, Carlson SA, Greenlund KJ, Lu H, Liu Y, et al. National, State-Level, and County-Level Prevalence Estimates of Adults Aged ≥18 Years Self-Reporting a Lifetime Diagnosis of Depression - United States, 2020. MMWR Morb Mortal Wkly Rep. 2023;72(24):644-50.
Guideline Development Panel for the Treatment of Depressive Disease. Summary of the clinical practice guideline for the treatment of depression across three age cohorts. Am Psychol. 2022;77(6):770-80.
Mendlewicz J. Towards achieving remission in the treatment of depression. Dialogues Clin Neurosci. 2008;10(4):371-5.
Lye MS, Tey YY, Tor YS, Shahabudin AF, Ibrahim N, Ling KH, et al. Predictors of recurrence of major depressive disorder. PLoS One. 2020;15(3):e0230363.
Goodwin RD, Dierker LC, Wu M, Galea S, Hoven CW, Weinberger AH. Trends in U.S. Depression Prevalence From 2015 to 2020: The Widening Treatment Gap. Am J Prev Med. 2022;63(5):726-33.
Mammen G, Faulkner G. Physical activity and the prevention of depression: a systematic review of prospective studies. Am J Prev Med. 2013;45(5):649-57.
Dinas PC, Koutedakis Y, Flouris AD. Effects of exercise and physical activity on depression. Ir J Med Sci. 2011;180(2):319-25.
Martinsen EW. Physical activity in the prevention and treatment of anxiety and depression. Nord J Psychiatry. 2008;62 Suppl 47:25-9.
Gordon BR, McDowell CP, Hallgren M, Meyer JD, Lyons M, Herring MP. Association of Efficacy of Resistance Exercise Training With Depressive Symptoms: Meta-analysis and Meta-regression Analysis of Randomized Clinical Trials. JAMA Psychiatry. 2018;75(6):566-76.
Li J, Zhou X, Huang Z, Shao T. Effect of exercise intervention on depression in children and adolescents: A systematic review and network meta-analysis. J Affect Disord. 2023.
Noetel M, Sanders T, Gallardo-Gomez D, Taylor P, Del Pozo Cruz B, van den Hoek D, et al. Effect of exercise for depression: systematic review and network meta-analysis of randomised controlled trials. BMJ. 2024;384:e075847.
Schuch FB, Vancampfort D, Firth J, Rosenbaum S, Ward PB, Silva ES, et al. Physical Activity and Incident Depression: A Meta-Analysis of Prospective Cohort Studies. Am J Psychiatry. 2018;175(7):631-48.
Choi KW, Chen CY, Stein MB, Klimentidis YC, Wang MJ, Koenen KC, et al. Assessment of Bidirectional Relationships Between Physical Activity and Depression Among Adults: A 2-Sample Mendelian Randomization Study. JAMA Psychiatry. 2019;76(4):399-408.
Chang CC, Chow CC, Tellier LC, Vattikuti S, Purcell SM, Lee JJ. Second-generation PLINK: rising to the challenge of larger and richer datasets. Gigascience. 2015;4:7.
Burgess S, Thompson SG. Bias in causal estimates from Mendelian randomization studies with weak instruments. Stat Med. 2011;30(11):1312-23.
Hemani G, Zheng J, Elsworth B, Wade KH, Haberland V, Baird D, et al. The MR-Base platform supports systematic causal inference across the human phenome. Elife. 2018;7.
Verbanck M, Chen CY, Neale B, Do R. Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization between complex traits and diseases. Nat Genet. 2018;50(5):693-8.
Brion MJ, Shakhbazov K, Visscher PM. Calculating statistical power in Mendelian randomization studies. Int J Epidemiol. 2013;42(5):1497-501.
Burgess S, Butterworth A, Thompson SG. Mendelian randomization analysis with multiple genetic variants using summarized data. Genet Epidemiol. 2013;37(7):658-65.
Bowden J, Davey Smith G, Haycock PC, Burgess S. Consistent Estimation in Mendelian Randomization with Some Invalid Instruments Using a Weighted Median Estimator. Genet Epidemiol. 2016;40(4):304-14.
Bowden J, Davey Smith G, Burgess S. Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression. Int J Epidemiol. 2015;44(2):512-25.
Cohen JF, Chalumeau M, Cohen R, Korevaar DA, Khoshnood B, Bossuyt PM. Cochran's Q test was useful to assess heterogeneity in likelihood ratios in studies of diagnostic accuracy. J Clin Epidemiol. 2015;68(3):299-306.
Viechtbauer W. Conducting Meta-Analyses in R with the metafor Package. Journal of Statistical Software. 2010;36(3):1-48.
Cooney G, Dwan K, Mead G. Exercise for depression. JAMA. 2014;311(23):2432-3.
Kvam S, Kleppe CL, Nordhus IH, Hovland A. Exercise as a treatment for depression: A meta-analysis. J Affect Disord. 2016;202:67-86.
Schuch FB, Vancampfort D, Richards J, Rosenbaum S, Ward PB, Stubbs B. Exercise as a treatment for depression: A meta-analysis adjusting for publication bias. J Psychiatr Res. 2016;77:42-51.

No competing interests reported.

SupplementarymaterialsTablesFigures.docx

Refined Causal Association Between Physical Activity Intensity and Depression: A Comprehensive Two-sample Mendelian Randomization Study

Status:

Version 1

Abstract

Figures

1 INTRODUCTION

2 DATA & METHODS

2.1 Study Design

2.2 Data Source

2.3 Instrumental Variables Extraction

2.4 Statistical Methods

3 RESULTS

3.1 Causal Effects of Strenuous Sports on General Depression & Major Depression

3.2 Causal Effects of Walking for Pleasure on General Depression & Major Depression

3.3 Meta-analysis for multiple MR analyses

3.4 Reverse Two-sample MR Analysis

4 DISCUSSIONS

5 CONCLUSIONS

Declarations

Author Contribution

Data Availability

References

Additional Declarations

Supplementary Files

Status:

Version 1