key: cord-0296526-03lj0z3k
authors: Zhao, H.; Han, X.; Li, L.; Zhang, X.; Liao, Y.; Zhang, H.; Li, W.; Shi, J.; Lai, W.; Wang, W.; McIntyre, R. S.; Teopiz, K. M.; guo, l.; Lu, C.
title: Investigating Causal Associations of Diet-Derived Circulating Antioxidants with Risk of Six Major Mental Disorders: A Mendelian Randomization Study
date: 2022-05-16
journal: nan
DOI: 10.1101/2022.05.11.22274935
sha: 51bdb81b736616711e18bfe131e1d9e6b6807c9d
doc_id: 296526
cord_uid: 03lj0z3k

Background: Observational studies have suggested associations between circulating antioxidant levels and many mental disorders , but evidence from randomized controlled trials (RCTs) is lacking and causal inferences have not been confirmed. The aim of this study was to explore whether genetically predicted diet-derived circulating antioxidants were causally associated with the risk of major mental disorders using Mendelian randomization (MR). Methods and findings: We performed 2-sample MR analyses of summary-level genetic data to explore whether diet-derived circulating antioxidants [e.g., vitamins E (- and {gamma}-tocopherol), ascorbate, retinol, {beta}- carotene, and lycopene ], assessed by absolute circulating antioxidant s and relative circulating antioxidant metabolites, were causally associated with the risk of six major mental disorders , including major depressive disorder (MDD), schizophrenia (SCZ), bipolar disorder (BIP) , autism spectrum disorder (ASD), attention-deficit/hyperactivity disorder (ADHD), and post-traumatic stress disorder (PTSD). The inverse-variance weighted method was adopted as primary MR analyses and five additional MR methods (likelihood-based MR, MR-Egger, weighted median, penalized weighted median, and MR-PRESSO) and different outcome databases were used for sensitivity analyses. We found suggestive evidence that genetically predicted higher absolute circulating -tocopherol levels marginally reduced the risk of SCZ , with the odds ratio (OR) per unit increase in log-transformed -tocopherol values was 0.71 [95% confidence interval (CI) 0.54 to 0.94; P = 0.016]. However, after adjusting for multiple testing (threshold of P < 0.008), we found no significant evidence that genetically predicted higher diet-derived absolute circulating antioxidant levels and antioxidant metabolites concentrations were significantly causally associated with the six-foregoing major mental disorders . Conclusions: Overall, our study does not support significant causal associations of genetically predicted diet-derived circulating antioxidants with the risk of major mental disorders . Therefore, simply taking antioxidants to increase blood antioxidants levels is unlikely to have a significant protective effect on the prevention of most mental disorders.

Author summary 58 Why was this study done? 59 Some observational studies have reported that diet-derived circulating antioxidants are associated with 60 a reduced risk of major mental disorders; however, these studies are susceptible to uncertain temporal 61 relationships, insufficient sample sizes, or potential confounding factors, and thus it remains unclear 62 whether these associations are accurate. 63 To our knowledge, there are no randomized clinical trials published to date on this topic. 64 Since oxidative stress is closely related to the occurrence of mental diseases, if diet-derived circulating 65 antioxidants can reduce the risk of major mental disorders, it will be an interesting target as primary 66 prevention of mental disorders.

What did the researchers do and find? 68 We performed a Mendelian randomization study design to explore whether genetically predicted diet-69 derived circulating antioxidants [e.g., vitamins E (α-and γ-tocopherol), ascorbate, retinol, β-carotene, 70 and lycopene], assessed by absolute circulating antioxidants and relative circulating antioxidant 71 metabolites, were causally associated with the risk of six major mental disorders, including major 72 depressive disorder, schizophrenia, bipolar disorder, autism spectrum disorder, attention-73 deficit/hyperactivity disorder, and post-traumatic stress disorder. 74 Overall, our study provides suggestive evidence that genetically predicted higher absolute α-75 tocopherol levels may be causally associated with a reduced risk of schizophrenia. However, our study 76 did not find genetically predicted significant causal associations of dietary antioxidants with major 77 mental disorders after correction for multiple testing.

What do these findings mean?

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 16, 2022. ; https://doi.org/10.1101/2022.05.11.22274935 doi: medRxiv preprint 5 79 Our findings suggest for healthy adults without nutritional deficiency, simply taking antioxidants to 80 increase blood antioxidants levels is unlikely to have a significant protective effect on the prevention 81 of most mental disorders. 82 In the future, large-scale GWASs are needed to further validate our current findings, especially the 83 suggestive protective effect of higher α-tocopherol levels on schizophrenia, by utilizing additional 84 genetic variants and more samples.

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 16, 2022. 86 With the acceleration of the pace of life, the intensification of competitive pressure, and the influence 87 of COVID-19, people's psychological pressure is increasing, and the prevalence of mental disorders is 88 increasing year by year [1] . Major intervention studies in healthy people cannot be conducted without sufficient evidence because of the 112 potential for unknown risks and harms to the subjects. Moreover, intervention trials are often limited 113 by timing, dosage, duration, use of natural or synthetic antioxidants, and the uncertainty of the onset 114 time and long-term progression of mental disorders. Therefore, the causal relationship between diet-115 derived antioxidants and the risk of major mental disorders remains unclear.

Mendelian randomization (MR), which uses genetic variants of the exposure as instrumental 117 variables to minimize measurement errors, confounding, and reversed causation, can provide reliable 118 estimation of the causal association between exposure and outcome under specific assumptions [22] .

With the development of sequencing technology, many large-scale genome-wide association studies 120 (GWAS) data on diet-derived antioxidants and major mental disorders have been published, which 121 provides an opportunity for 2-sample MR analyses using summary statistics from separate studies to 122 substantially increase the statistical power by combining data from multiple sources.

In this study, we used a 2-sample MR analysis to assess the causal associations between 124 genetically predicted diet-derived circulating antioxidants and major mental disorders, including MDD, 125 SCZ, BIP, ASD, ADHD, and PTSD.

Overall study design 128 The study herein used 2-sample MR analyses of summary level genetic data to investigate whether . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted May 16, 2022. ; https://doi.org/10.1101/2022.05.11.22274935 doi: medRxiv preprint 8 129 diet-derived circulating antioxidants, including vitamins E (α-and γ-tocopherol), ascorbate, retinol, 130 β-carotene, and lycopene, were causally associated with the risk of six major mental disorders, 131 including MDD, SCZ, BIP, ASD, ADHD, PTSD. We considered the following two phenotypes for . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) For genetic instrumental variables of circulating antioxidant metabolites, genetically predicted α-152 tocopherol, γ-tocopherol, ascorbate, and retinol were extracted from the metabolite GWAS analysis at 153 suggestive genome-wide significance level (P < 1e-5). A total of 11 SNPs for α-tocopherol (n = 7,725), 154 13 SNPs for γ-tocopherol (n = 6,226), and 14 SNPs for ascorbate (n = 2,085) were derived from 7,824 To further assess the robustness of our findings, a series of sensitivity analyses were performed. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 16, 2022. ; https://doi.org/10.1101/2022.05.11.22274935 doi: medRxiv preprint 11 195 PRESSO) was performed, which detects and corrects the effects from outliers (No. SNPs > 3) [43] . 196 Second, we also performed scatter, forest, and leave-one-out plots to detect high influence points. Bonferroni-corrected threshold of P < 0.008 (α = 0.05/6 outcomes) as significant evidence of 203 associations, and a P-value between 0.05 and 0.008 was considered suggestive evidence of 204 associations. All statistical analysis were performed using R version 4.1.0.

Strength of genetic instruments 207 The summary information of GWAS for diet-derived absolute circulating antioxidants and metabolites 208 and major mental disorders is shown in Table 1 and S1 and S2 Tables. The genetic instruments of α-209 tocopherol, ascorbate, and retinol were available both as absolute circulating antioxidants and 210 metabolites. Variance explained by the genetic instruments ranged from 1.7% to 30.1% for absolute 211 circulating antioxidants (all F statistic >10) and from 6.8% to 21.7% for antioxidants metabolites (all 212 F statistic >10). The statistical power in MR study suggested that for most analyses we had the adequate 213 statistical power to identify even modest causality (S3 Table) . The raw data information on the effect 214 estimation for the associations of selected SNPs with antioxidants and with major mental disorders is 215 given in S4 and S5 Tables. 216  Table 1 . The summary of instrumental variables for diet-derived absolute circulating antioxidants and . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. 1-3 Figure) . The leave-one-out test also suggested that the results were 248 robust even though rs964184 might have been eliminated due to its association with cholesterol and 249 triglycerides. In addition, we further confirmed that higher absolute diet-derived α-tocopherol levels 250 marginally reduced risk of SCZ by outcome from another GWAS from all-European population (OR 251 0.67, 95% CI 0.45 -1.00, P = 0.051, nearly reaching suggestive evidence, S6 Table) . 252 For robust null results in the primary analysis (IVW method), similar OR estimates as in the 253 primary MR analysis but with lower precision were obtained from MR-Egger, weighted median, and 254 penalized weighted median analysis. The MR-PRESSO method identified outlier SNP for absolute . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 16, 2022. ; https://doi.org/10.1101/2022.05.11.22274935 doi: medRxiv preprint 14 255 ascorbate on SCZ and BIP, for α-tocopherol metabolite on BIP, and for retinol metabolite on SCZ. 256 Outlier-correction did not materially change the OR estimates. No outlier SNPs were identified in the 257 MR-PRESSO analysis for the other outcomes (Figure 4) . 258 

In this study, we investigated the causal associations between diet-derived circulating antioxidants and 260 the risk of six major mental disorders based on MR analyses. The genetic variation of circulating 261 antioxidants was evaluated as authentic absolute blood levels and relative metabolites concentrations 262 as instrumental variables, and comparable results were obtained. We only found suggestive evidence 263 that genetically predicted higher absolute circulating α-tocopherol levels marginally reduced the risk 264 of SCZ. However, we found no significant evidence that genetically predicted higher diet-derived 265 antioxidants were significantly causally associated with the six major mental disorders after correction 266 for multiple testing. studies are consistent with our research. Our research also suggested that higher absolute α-tocopherol 275 levels marginally reduced the risk of SCZ. Although we found that the direction of association with 276 SCZ risk was consistent in α-tocopherol and γ-tocopherol metabolites, these associations did not reach . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 16, 2022. 

However, there was no significant causal relationship between genetically predicted higher diet-298 derived antioxidants with the major mental disorders. This seems to violate the oxidative stress . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 16, 2022. ; https://doi.org/10.1101/2022.05.11.22274935 doi: medRxiv preprint 16 299 hypothesis, but circulating antioxidant levels may not represent antioxidant capacity, and that 300 increasing the level of an antioxidant in blood through simple nutritional intake or supplements does 301 not necessarily produce additional antioxidative effects [56] . Therefore, for healthy adults without 302 nutritional deficiency, dietary supplements that are to increase antioxidant concentrations in blood is 303 unlikely to have a significant protective effect on the prevention of most mental disorders.

There are three main strengths in the present study. First, the MR design of two independent 305 samples based on genetic instrumental variables reduces the possibility of subjects exposed to 306 unnecessary risks and hazards in RCT study, and supplements the genetic theoretical basis of dietary 307 antioxidants and major mental disorders. Second, we used genetic instrument variables based on two 308 phenotypes, including absolute blood levels of antioxidants measured quantitatively by targeted and 309 relative metabolite concentrations detected by untargeted metabolomics. Untargeted metabolomics is 310 primarily used for broad coverage of metabolites, but its measurement accuracy is lower than targeted 311 quantitative measurements due to increased false positive rates. Importantly, although there is some 312 measurement bias for relative antioxidants metabolite concentrations, the direction of effect for the 313 suggestive findings at the absolute levels is consistent, which also adds to the reliability of our results. However, some limitations should also be considered with respect to the interpretation of the 317 results. First, our analyses are based on published summary data rather than individual data, therefore, 318 we were unable to test nonlinear causal relationships between antioxidant levels and risk for select 319 mental disorders. Second, although an inherent limitation of MR analyses is that there may be potential 320 polymorphic effects, the MR-Egger intercept test in our study has shown no significant pleiotropy, and . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Overall, our study provides suggestive evidence that genetically predicted higher absolute α-336 tocopherol levels may be causally associated with a reduced risk of SCZ. However, our study did not 337 find genetically predicted significant causal associations of dietary antioxidants with major mental 338 disorders after correction for multiple testing. Therefore, for healthy adults without nutritional 339 deficiency, simply taking antioxidants to increase blood antioxidants levels is unlikely to have a 340 significant protective effect on the prevention of most mental disorders. However, the findings do not 341 exclude the possibility that there may be smaller effects than we could not detect. In the future, large-342 scale GWASs are needed to further validate our current findings by utilizing additional genetic variants . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The author(s) received no specific funding for this work. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. 

Global burden of 369 diseases and injuries in 204 countries and 424 territories, 1990-2019: a systematic analysis for the Global Burden of Disease Study

Mental Disorders Collaborators. Global, regional, and national burden of 12 mental disorders in 204 427 countries and territories, 1990-2019: a systematic analysis for the Global Burden of Disease Study

Association of 430 Oxidative Stress with Psychiatric Disorders

Oxidative Stress, and 433 Antioxidants: Back and Forth in the Pathophysiology of Chronic Diseases

Possible connections among job stress, 436 depressive symptoms, lipid modulation and antioxidants

Oxidative stress, anti-oxidants and the 439 cross-sectional and longitudinal association with depressive symptoms: results from the CARDIA study

Antioxidant vitamin intakes and risk of 442 depression, anxiety and stress among female adolescents

Oxidative stress, α-tocopherol, ascorbic acid and reduced 445 glutathione status in schizophrenics

Nutritional status, food intake and 448 cardiovascular disease risk in individuals with schizophrenia in southern Brazil: a case-control study

Elevated serum MDA and depleted 451 non-enzymatic antioxidants, macro-minerals and trace elements are associated with bipolar disorder

Comparison of nutritional 454 status between children with autism spectrum disorder and typically developing children in the Mediterranean 455

Feeding behavior and dietary intake of male 458 children and adolescents with autism spectrum disorder: A case-control study

Nutrient intake, dietary patterns, 461 and anthropometric variables of children with ADHD in comparison to healthy controls: a case-control study

the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted

Dietary fatty acid and antioxidant intake in community-dwelling patients 466 suffering from schizophrenia

Adults 20 Years or Older Based on National Health and Nutrition Examination Survey

Dietary carotenoids intake and depressive symptoms in US adults, NHANES 2015-2016

Nutritional status of 474 children with autism spectrum disorders (ASDs): a case-control study

Oxidative stress and immune 477 aberrancies in attention-deficit/hyperactivity disorder (ADHD): a case-control comparison

Role of antioxidants in generalised anxiety 480 disorder and depression

The impact of omega-3 fatty acids, vitamins E and C 483 supplementation on treatment outcome and side effects in schizophrenia patients treated with haloperidol: an 484 open-label pilot study

Strengthening the 487 Reporting of Observational Studies in Epidemiology Using Mendelian Randomization: The STROBE-MR Statement

Genome-wide association study identifies 492 common variants associated with circulating vitamin E levels

Plasma Vitamin C and Type 2 Diabetes

Association Study and Mendelian Randomization Analysis in European Populations

International meta-analysis of PTSD 528 genome-wide association studies identifies sex-and ancestry-specific genetic risk loci

LDlink: a web-based application for exploring population-specific haplotype structure 531 and linking correlated alleles of possible functional variants

Mendelian randomization with Egger pleiotropy correction and weakly informative 534

Bayesian priors

Sensitivity Analyses for Robust Causal Inference from 536

Mendelian Randomization Analyses with Multiple Genetic Variants

Constrained maximum likelihood-based Mendelian randomization robust to both 539 correlated and uncorrelated pleiotropic effects

Consistent Estimation in Mendelian Randomization with Some 542 Invalid Instruments Using a Weighted Median Estimator

Detection of widespread horizontal pleiotropy in causal relationships 545 inferred from Mendelian randomization between complex traits and diseases

Biological insights from 108 schizophrenia-associated genetic loci

Calculating statistical power in Mendelian randomization studies

A five-year follow-up study of antioxidants, oxidative stress 552 and polyunsaturated fatty acids in schizophrenia

International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted

GC-MS based metabolomics identification of possible novel biomarkers 555 for schizophrenia in peripheral blood mononuclear cells

Comparison of chemometric 558 strategies for potential exposure marker discovery and false-positive reduction in untargeted metabolomics: 559 application to the serum analysis by LC-HRMS after intake of Vaccinium fruit supplements

Diet-Derived Circulating Antioxidants and Risk of Coronary Heart 562 Disease: A Mendelian Randomization Study

Vitamin E and essential polyunsaturated fatty acids 565 supplementation in schizophrenia patients treated with haloperidol

Free radicals, antioxidant defense systems, and schizophrenia

Free radical pathology and antioxidant defense in schizophrenia: a review

Oxidative stress in schizophrenia: pathogenetic and therapeutic implications

Effects of nutrients (in food) on the structure and function of the nervous system: update on dietary 575 requirements for brain. Part 1: micronutrients

Vitamin E and its function in membranes

the author/funder, who has granted medRxiv a license to display the preprint in perpetuity

The authors thank the staff and participants of the PGC (Mental Genomics Consortium) for making 376 the GWAS data publicly available. 

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 16, 2022. ; https://doi.org/10.1101/2022.05.11.22274935 doi: medRxiv preprint . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted May 16, 2022. ; https://doi.org/10.1101/2022.05.11.22274935 doi: medRxiv preprint