Introduction: Major depression (MD) is more common amongst women than men, and MD episodes have been associated with fluctuations in reproductive hormones amongst women. To investigate biological underpinnings of heterogeneity in MD, the associations between depression, stratified by sex and including perinatal depression (PND), and blood biomarkers, using UK Biobank (UKB) data, were evaluated, and extended to include the association of depression with biomarker polygenic scores (PGS), generated as proxy for each biomarker. Method: Using female (N = 39,761) and male (N = 38,821) UKB participants, lifetime MD and PND were tested for association with 28 blood biomarkers. A GWAS was conducted for each biomarker and genetic correlations with depression subgroups were estimated. Using independent data from the Australian Genetics of Depression Study, PGS were constructed for each biomarker, and tested for association with depression status (n [female cases/controls] = 9,006/6,442; n [male cases/controls] = 3,106/6,222). Regions of significant local genetic correlation between depression subgroups and biomarkers highlighted by the PGS analysis were identified. Results: Depression in females was significantly associated with levels of twelve biomarkers, including total protein (OR = 0.90, CI = [0.86, 0.94], p = 3.9 × 10−6) and vitamin D (OR = 0.94, CI = [0.90, 0.97], p = 2.6 × 10−4), and PND with five biomarker levels, also including total protein (OR = 0.88, CI = [0.81, 0.96], p = 4.7 × 10−3). Depression in males was significantly associated with levels of eleven biomarkers. In the independent Australian Genetics of Depression Study, PGS analysis found significant associations for female depression and PND with total protein (female depression: OR = 0.93, CI = [0.88, 0.98], p = 3.6 × 10−3; PND: OR = 0.91, CI = [0.86, 0.96], p = 1.1 × 10−3), as well as with vitamin D (female depression: OR = 0.93, CI = [0.89, 0.97], p = 2.0 × 10−3; PND: OR = 0.92, CI = [0.87, 0.97], p = 1.4 × 10−3). The male depression sample did not report any significant results, and the point estimate of total protein (OR = 0.98, CI = [0.92–1.04], p = 4.7 × 10−1) did not indicate any association. Local genetic correlation analysis highlighted significant genetic correlation between PND and total protein, located in 5q13.3 (rG = 0.68, CI = [0.33, 1.0], p = 3.6 × 10−4). Discussion and Conclusion: Multiple lines of evidence from genetic analysis highlight an association between total serum protein levels and depression in females. Further research involving prospective measurement of total protein and depressive symptoms is warranted.

The lifetime risk of depression, estimated to be approximately 10.6% globally, with higher rates in high-income countries [1], has stimulated a search for both diagnostic and treatment biomarkers [2]. Hundreds of potential biomarkers have been implicated [3], related to growth factors [4, 5], inflammation [6‒8], oxidative stress [9], neurotrophy [10], and endocrine levels [11], although many studies are underpowered and/or access a limited number of biomarkers. Large biobanks such as the UK Biobank (UKB) provide unprecedented power to investigate the association between a range of biomarkers and depression. Furthermore, the availability of genetic data allows for the generation of PGS that can be used as a proxy for biomarker levels in cohorts where biomarkers have not been directly measured [12].

The heterogeneity of major depression (MD) complicates the identification of associated biomarkers [2]. A potential major source of heterogeneity is difference between men and women [13, 14]. Depression is approximately 50% more common amongst women than men [15] and is particularly common at times of change in reproductive hormones, including onset of menstruation, the perinatal period, and menopause. However, whether MD heterogeneity between women and men is due to genetic heterogeneity is contentious. A Swedish family study found higher heritability for women compared to men [16], but an analysis of between sex heterogeneity using four large data sets (PGC29, GERA, UKB, and iPSYCH) found no evidence of genetic heterogeneity [17]. Even if the genetic risk factors are similar between men and women, biomarker associations may be different. Biomarkers may be categorised as either trait or state, depending on whether levels represent a stable attribute that creates a specific vulnerability to a disorder and persists before diagnosis, during the disorder and after remission (trait), or whether levels are transiently high or low and indicate onset of the disorder (state) [2].

Using the trait model, we sought to extend previous phenotypic investigations of difference in biomarkers between men and women [13, 14, 18, 19] by estimating the association and genetic correlation between 28 commonly measured blood biomarkers and lifetime depression in women compared to men, and also to investigate the extent to which there is evidence of involvement of biomarkers specifically in PND, a reproductive-related depressive episode in women. To do this, we firstly tested the sex-stratified association of lifetime MD with 28 potential biomarkers, measured in UKB. This data collection in a very large sample also provides genetic data that can be used as a proxy for lifelong high or low levels of biomarkers that may make individuals more susceptible to disease. For each biomarker, we conducted a separate GWAS in female and male participants, and then estimated its genetic correlation with female or male depression, respectively. To test for genetic vulnerability to depression, we generated biomarker PGS using the GWAS summary statistics and tested for association with female or male depression and PND in the Australian Genetics of Depression Study.

All genetic results are based on the GRCh37 (hg19) genome assembly.

Biomarkers

The UKB is a community-based cohort established to investigate risk factors for the major diseases of middle and old age [20]. Upon recruitment, saliva samples were taken for genetic sampling, and blood samples were used for measurement of 30 biochemical markers (online suppl. Table S1; for all online suppl. material, see https://doi.org/10.1159/000538058) (https://www.ukbiobank.ac.uk/enable-your-research/about-our-data/biomarker-data). Two biomarkers (oestradiol and rheumatoid factor), which had over 70% of results below the reportable range, were excluded from the analysis. Measures of apolipoprotein B, cholesterol, and low-density lipoprotein cholesterol (LDL) were adjusted for the use of statins [12]. To avoid violating assumptions of the regression model, the log of the biomarker measure (phenotype) was used for all analyses of the remaining 28 biomarkers.

The correlation of each biomarker with all other biomarkers was estimated to quantify independence. For all analyses, “female biomarkers” refers to biomarkers measured using female participants and “male biomarkers” refers to biomarkers measured using male participants.

Depression Cases and Controls

UK Biobank

Only participants who had completed the follow-up Mental Health Questionnaire (MHQ) [21] (n = 177,366), which included the Composite International Diagnostic Interview – Short Form (CIDI-SF) [22] (measuring lifetime depression), were eligible for inclusion. Depression cases were participants of European-ancestry who met the DSM-5 [23] criteria for lifetime depression (n [female] = 14,877, n [male] = 7,546). Perinatal depression (PND) cases (n = 2,085) were parous female depression cases who stated that their depression was “probably related to childbirth” (data field 20445). Controls (n [female] = 24,764, n [male] = 31,275) were participants who did not have current depression, responded negatively to MHQ screening criteria for lifetime depression (data field 20441: “prolonged loss of interest in normal activities” and data field 20446: “prolonged feelings of sadness or depression”), had no history of depression or bipolar disorder according to the derived data field 20126, had not seen a medical professional about their mental health, did not meet criteria for post-traumatic stress disorder using the post-traumatic stress disorder Check List – civilian Short version (PCL-S) [24] (data fields 20494–20498), and answered “no” to the question “Have you ever been prescribed medication for unusual or psychotic experiences?” (data field 20466). PND controls met the additional criteria of having given birth at least once (n = 18,186).

Australian Genetics of Depression Study

The Australian Genetics of Depression Study recruited a total of 20,689 participants (aged between 18 and 90 years; 75% female) primarily through a media campaign (86%). Full details of the recruitment strategy have been described elsewhere [25]. Participants completed an online questionnaire, and those who consented to provide a DNA sample were mailed a saliva kit. The online questionnaire included a compulsory module that assessed self-reported psychiatric history and clinical depression using the Composite Interview Diagnostic Interview Short Form (CIDI-SF) [22]. For the September 2018 data freeze, a total of 15,807 participants (11,537 women) had returned DNA samples and were genotyped. Depression cases were women and men of European ancestry who fulfilled DSM criteria for MD or who reported a previous diagnosis of depression (n [female] = 9,006, n [male] = 3,106). Women reporting symptoms of depression around childbirth were invited to complete the Lifetime Edinburgh Postnatal Depression Scale (EPDS) [26, 27]. Of 5,848 genotyped women with at least one live birth, 3,804 (65%) met criteria for PND (either a score ≥13 on the Lifetime EPDS [27], or previous diagnosis of PND by a health professional, or a period of depression of at least 2 weeks that occurred perinatally).

Control participants for the Australian Genetics of Depression Study are taken from the QSkin study [28] which is a population cohort study of risk factors for skin cancer in QLD, Australia. QSkin recruited 43,794 volunteer participants through mailout to a random sample from the compulsory Queensland electoral register in 2011. A follow-up study included questions on lifetime diagnosis of mental health as well as a request for a DNA sample, resulting in 17,965 genotyped participants. MD controls (n [female] = 6,462, n [male] = 6,222) were women and men of European ancestry who did not report a history of mental illness; PND controls (n = 6,134) were also parous.

Choice of Covariates

In order to investigate potential covariates to be included in the analysis, linear regression was used to estimate the association of each biomarker with age, assessment centre, batch, fasting time, month and time of day of assessment, and dilution factor (hereafter referred to as standard covariates). Given the possible effect of BMI on some biomarkers, the association of each biomarker with BMI was also estimated using linear regression. In addition, for female participants, the association of current use of hormone replacement therapy and menstrual status (having ceased menstruating due to natural menopause or other factors) on biomarkers was estimated. The R package “relaimpo” [29] was used to calculate the proportion of variance explained by these covariates (R2), and the contribution of each covariate to R2, using the “lmg” [30] approach that assumes possible correlation amongst covariates and computes the relative importance of each covariate using its unweighted average in all possible orderings of covariates in the linear model equation [29]. Each biomarker was also regressed on forty genetic principal components and the proportion of variance explained by all principal components was added to the total variance explained by all covariates.

Association between Lifetime Depression and Biomarkers

Univariate logistic regression was used to test the association of depression with each standardised biomarker in males and females separately. The standard covariates were included in the analysis, as well as BMI, and, for females, current use of hormone replacement therapy (HRT) and menstrual status, all of which had a strong association with many biomarkers.

Since biomarkers are correlated with each other [31], multiple regression was used to find the independent effect of each biomarker. To avoid issues of collinearity, for each cluster of biomarkers with correlation coefficient |r| >0.7, only the most significant biomarker in the cluster (based on the univariate regression analysis) was used in the multiple regression analysis [32]. The analysis was repeated for UKB PND cases and controls, including number of births as an additional covariate.

Genome-Wide Association Study of Biomarkers in the UKB

A separate GWAS for males and females of genetically inferred European ancestry was conducted for each biomarker using common SNPs (Minor Allele Frequency >0.01). For each biomarker, fastGWA [33] (https://yanglab.westlake.edu.cn/software/gcta/index.html#fastGWA), an efficient tool for mixed linear model analysis of biobank-scale data, was used to estimate the linear association of the log of the biomarker as a continuous phenotype with each SNP. The standard covariates and forty genetic principal components were included, as well as menstruation status for GWAS of female participants. Since inclusion of BMI as a covariate may introduce collider bias (online Supplementary Note: Methods; online suppl. Fig. S1), multi-trait Conditional and Joint Analysis (mtCOJO) [34], found in simulations to be robust to collider bias, was used to conduct post-GWAS conditioning on BMI (online Supplementary Note). For females, post-GWAS conditioning also included current HRT use. Individual level genetic data were used to estimate the SNP-based heritability (h2SNP) of each biomarker using the Haseman-Elston method implemented in GCTA [35, 36] (https://yanglab.westlake.edu.cn/software/gcta/index.html#Haseman-Elstonregression).

GWAS of UKB Female Depression and PND

The GWAS for the UKB PND cohort was conducted using SAIGE [37], which infers and accounts for relatedness in the data through a mixed model analysis, as well as an unbalanced case/control ratio through a saddlepoint approximation, but for the larger female and male depression cohorts the GWASs were constructed using fastGWA-GLMM-binary [38] (https://yanglab.westlake.edu.cn/software/gcta/index.html#fastGWA-GLMM), a more efficient tool for large binary data sets that also uses the saddlepoint approximation to adjust for inflation in test statistics due to case/control imbalance. Potential confounding due to any remaining population stratification was addressed through the addition of ten genetic principal components to the analysis.

Estimation of Genetic Correlation between Biomarkers and Lifetime Depression and PND

The genetic correlation of each biomarker with two depression groups: either female or male depression (for female or male biomarkers respectively) was estimated using LDSC [39]. In addition, the genetic correlation of female biomarkers with PND was also estimated. A conservative Bonferroni correction (n = 28) was applied to the results.

Association of Polygenic Scores with Depression

Although used extensively to measure cross-trait associations [40], genetic correlation may not always be accurate [41, 42]. Two associated traits may have a negligible genetic correlation if their correlation is positive in some areas of the genome and negative in others [41]. Small samples may also lack power to reach significance [42]. To complement the estimation of the genetic correlation, we therefore used polygenic scores (PGS) for each biomarker to estimate the corresponding associated liability to depression.

SBayesR [43], a Bayesian method that uses summary statistics to reweight SNP effects, accounting for LD, under the assumption that the SNP effects are drawn from a mixture of four normal distributions, was used to generate SNP weights for PGS. For each biomarker, PGS were then constructed using Plink 2.0 with the SBayesR reweighted effect sizes for all Australian Genetics of Depression Study cases and controls. Using the mean and standard deviation of controls (stratified by sex), all PGS were standardised to a mean of zero and standard deviation of one. Logistic regression (both univariate and multiple regression) was used to find the association of each standardised biomarker PGS with depression in males and females separately, including age and ten principal components as covariates. The same procedure estimated the association of PND with each female biomarker, with the number of live births also included as an added covariate in all analyses. Bonferroni correction (n = 28) was applied to the results.

Local Genetic Correlations

Local analysis in (co)variant association (LAVA) [44] was used to calculate local genetic correlations of female depression and PND with biomarkers with significant associations for the PGS analysis in contrast to the genetic correlation analysis. For each independent locus that reached genome-wide significance for the relevant biomarker (FUMA analysis [45]: p < 5.0 × 10−8; r2 threshold = 0.1; max distance between LD blocks = 1000 kb), univariate analysis was used to estimate the SNP-based heritability (h2observed) for both the biomarker and the depression group. Where h2observed was significant for both biomarker and the depression group, local genetic correlation was estimated, across the locus, using bivariate analysis. Bonferroni correction was applied to results. For any nominally significant (p < 0.05) SNP-dense loci (number of SNPs >1,000), further analysis was implemented, through univariate and bivariate analysis of a sliding 100 kb window across the locus.

Online supplementary Table S1 provides the list of 28 biomarkers used in this analysis, including the number of female and male participants in the UKB who had a usable phenotype for each biomarker, with their mean age at measurement, as well the number of these participants who met criteria for female/male depression or PND cases or controls.

Contribution of Covariates to Variance Explained

Figure 1 illustrates the contribution of each covariate to variance explained in each biomarker in males and females, expressed in absolute and proportional terms. Online supplementary Table S2 provides the variance explained by each covariate for the biomarker phenotype, and online supplementary Table S3 provides more detail of variance explained by age and BMI, and (for women) menstruation status, including effect size, SE, and p values. The difference in variance explained by age and BMI between men and women (female h2 – male h2) is also listed in online supplementary Table S2. Total variance explained by ageing and BMI showed most difference between women and men for C-reactive protein (CRP) (difference in variance explained = 11.6%), and urate (difference in variance explained = 10.1%) (online suppl. Table S2). For women, the full set of covariates explained most variance (> twenty percent) for CRP, cystatin C, sex hormone binding globulin (SHBG) and urate, and least variance for lipoprotein A and total protein. For men, variance explained by covariates was highest for SHBG (18%) and lowest for lipoprotein A (0.2%). Of all covariates, BMI explained most variance, particularly for women, explaining over 10 percent of variance for CRP, SHBG, urate, high-density lipoprotein (HDL), and triglycerides, whilst, for men, BMI explained more than 10 percent of variance for CRP, HDL, and alanine aminotransferase (ALT). For women, age explained greater than 5 percent of variance for cystatin C, urea, cholesterol, glycated haemoglobin (glyHg), LDL, apoB, insulin-like growth factor 1 (IGF_1), and alkaline phosphatase (ALP). For men, age explained greater than 5 percent of variance for three biomarkers: cystatin C, albumin, and SHBG. Of other covariates, more than 5 percent of variance was explained by menstruation status for ALP and cholesterol and by the month of assay for vitamin D. The large proportion of variance explained by BMI in many phenotypes motivated the mtCOJO analysis.

Fig. 1.

Variance explained for each biomarker by each of 10 covariates, with 40 principal components aggregated to form an 11th covariate. Left side: female participants; right side: male participants. For both female and male participants, Top: Biomarker variance explained by each covariate is illustrated as a proportion of all covariates; Bottom: Biomarker variance is illustrated as a proportion of variance explained by all covariates for the individual biomarker. Maximum variation explained by all covariates for any biomarker is 0.26 for C-reactive protein (CRP) (For female participants).

Fig. 1.

Variance explained for each biomarker by each of 10 covariates, with 40 principal components aggregated to form an 11th covariate. Left side: female participants; right side: male participants. For both female and male participants, Top: Biomarker variance explained by each covariate is illustrated as a proportion of all covariates; Bottom: Biomarker variance is illustrated as a proportion of variance explained by all covariates for the individual biomarker. Maximum variation explained by all covariates for any biomarker is 0.26 for C-reactive protein (CRP) (For female participants).

Close modal

Correlation between Covariates

The phenotypic correlation matrix for the biomarker traits is illustrated using a heatmap in Figure 2, with details of high correlations (Pearson correlation coefficient |r| > 0.5), including significance levels, provided in online supplementary Table S4. As illustrated in Figure 2, several biomarkers are strongly correlated with one another, which resulted in the creation of four clusters of biomarkers with |r| > 0.7 for the purposes of multiple regression: direct and total bilirubin: r = 0.91, p < 0.005 (total bilirubin cluster); apolipoprotein A (apoA) and HDL cholesterol: r = 0.91, p ≤ 0.005 (apoA cluster); apoB, LDL, and cholesterol (apoB cluster); and aspartate aminotransferase (AST) and ALT: r = 0.73, p < 0.005) (AST cluster). For the multiple regression analysis, only the most highly significant biomarker (based on univariate regression results) in these clusters was used, and less significant biomarkers in the cluster were omitted. All other biomarkers, with |r| <0.5, were assumed to be orthogonal for the purposes of multiple regression.

Fig. 2.

Heatmap of correlation matrix of 28 biomarkers using phenotype data from UKB. Four clusters with |r| > 0.7 are highlighted: direct and total bilirubin (total bilirubin cluster); apolipoprotein A (apoA) and high-density lipoprotein cholesterol (HDL) (apoA cluster); apolipoprotein B, LDL, and cholesterol (apoB cluster); and aspartate aminotransferase (AST) and alanine aminotransferase (ALT) (AST cluster).

Fig. 2.

Heatmap of correlation matrix of 28 biomarkers using phenotype data from UKB. Four clusters with |r| > 0.7 are highlighted: direct and total bilirubin (total bilirubin cluster); apolipoprotein A (apoA) and high-density lipoprotein cholesterol (HDL) (apoA cluster); apolipoprotein B, LDL, and cholesterol (apoB cluster); and aspartate aminotransferase (AST) and alanine aminotransferase (ALT) (AST cluster).

Close modal

Association of Biomarkers with Lifetime Depression

Univariate Regression Analysis

Figure 3 illustrates the association of all biomarkers found to be significantly associated with lifetime depression in men and/or women, or lifetime PND in women, using univariate regression analysis, where OR refers to the odds of having lifetime depression, compared to the odds of not having this disorder, per standard deviation increase in the log of the biomarker. Full details for all biomarkers are provided in online supplementary Table S5 for female and male depression and in online supplementary Table S6 for PND. As illustrated in Figure 3, the small size of the PND sample created large confidence intervals compared to male and female depression groups. Low levels of apoA, HDL, IGF_1 and total bilirubin, and high levels of ALP, apoB, gamma-glutamyl transpeptidase (GGT) and triglycerides are associated with all three groups, with high levels of triglycerides ranked highest for all three (female dep: OR = 1.11, CI = [1.08–1.14], p = 3.4e−14; male dep: OR = 1.19, CI = [1.16–1.22], p = 1.7e−42; PND: OR = 1.16, CI = [1.1–1.23], p = 1.4 × 10−8). Low levels of vitamin D, urea, total protein, and creatinine, and high levels of cystatin C were significantly associated with both female and male depression, though not with PND. High levels of LDL were associated with female but not male depression. Low levels of testosterone, direct bilirubin, SHBG, and albumin, and high levels of CRP, ALT, glyHg, and AST were associated with depression in men only. High levels of CRP, ALT, and AST were also associated with PND.

Fig. 3.

Univariate regression analysis: The association of each biomarker (female and male) with female depression and male depression respectively, and the association of female biomarkers with PND. The small size of the PND sample created large confidence intervals compared to male and female depression groups. High level of triglycerides was most significantly associated with all three depression groups.

Fig. 3.

Univariate regression analysis: The association of each biomarker (female and male) with female depression and male depression respectively, and the association of female biomarkers with PND. The small size of the PND sample created large confidence intervals compared to male and female depression groups. High level of triglycerides was most significantly associated with all three depression groups.

Close modal

Multiple Regression Analysis

Figure 4 illustrates the odds ratios of all biomarkers found to be significantly associated with either female or male depression, or PND, using a single model of all biomarkers with covariates, excluding highly correlated biomarkers to avoid collinearity. Full details of multiple regression analyses for all biomarkers are provided in online supplementary Tables S5, S6. Significantly associated with all three depression groups were low levels of total protein (female dep: OR = 0.9, CI = [0.86–0.94], p = 3.9 × 10−6; male dep: OR = 0.92, CI = [0.88–0.96], p = 7.9 × 10−5; PND: OR = 0.88, CI = [0.81–0.96], p = 4.7 × 10−3); and high levels of triglycerides (female dep: OR = 1.08, CI = [1.04–1.13], p = 3.3 × 10−4; male dep: OR = 1.08, CI = [1.04–1.12], p = 2.2 × 10−4; PND: OR = 1.16, CI = [1.07–1.27], p = 4.8 × 10−4). Also significantly associated with both female and male depression were low levels of creatinine, total bilirubin, urea, and vitamin D, and high levels of cystatin C and phosphate. Low levels of the Apo A cluster and IGF_1, and high levels of calcium were significantly associated with depression in women and PND but not male depression, whilst low levels of urate and high levels of ALP and glyHg were associated with depression in males only.

Fig. 4.

Multiple regression Analysis: The association of biomarkers with three depression groups using a single fitted model of all biomarkers (excluding biomarkers that would introduce collinearity) and covariates. Low levels of total protein ranked highest in significant association with female depression.

Fig. 4.

Multiple regression Analysis: The association of biomarkers with three depression groups using a single fitted model of all biomarkers (excluding biomarkers that would introduce collinearity) and covariates. Low levels of total protein ranked highest in significant association with female depression.

Close modal

Heritability and Genetic Correlation

SNP-based heritability, estimated using the HE method implemented by GCTA, is described in the online Supplementary Note, with details provided in online supplementary Table S7. Using summary statistics for each biomarker (after removing the genetic effects of BMI and, for females, HRT), genetic correlation was calculated for each biomarker with each of three depression groups, using LDSC. Six biomarkers were significantly genetically correlated with female depression, as illustrated in Figure 5, which includes corresponding genetic correlations of biomarkers with PND and male depression for comparison. Total bilirubin was significantly genetically correlated with both female depression and PND (female depression: rG = −0.12 [0.04], p = 7.0 × 10−3; PND: rG = −0.21 [0.10], p = 4.3 × 10−2). Details for all biomarkers are provided in online supplementary Table S8.

Fig. 5.

Genetic correlation of biomarkers with three depression groups: female biomarkers with female depression and PND, and male biomarkers with male depression. Each biomarker illustrated is significantly genetically correlated with female depression, and total bilirubin is also significantly genetically correlated with PND. The small size of the PND sample resulted in lack of power and large standard errors. No significant results were obtained for male depression.

Fig. 5.

Genetic correlation of biomarkers with three depression groups: female biomarkers with female depression and PND, and male biomarkers with male depression. Each biomarker illustrated is significantly genetically correlated with female depression, and total bilirubin is also significantly genetically correlated with PND. The small size of the PND sample resulted in lack of power and large standard errors. No significant results were obtained for male depression.

Close modal

PGS Analysis

Significant results for the PGS analysis in the independent AGDS cohort are illustrated in Figure 6 with details of all results provided in online supplementary Table S9 for female and male depression and online supplementary Table S10 for PND. Lifetime female depression and PND were both significantly associated with PGS for total protein and vitamin D using multiple regression analysis, but no male biomarker PGS were significantly associated with male depression. Total protein was not ranked highly for males, and its point estimate (OR = 0.98, CI = [0.92–1.04], p = 4.7 × 10−1) did not indicate any suggestive association.

Fig. 6.

Forest plot illustrating the association of biomarker PGS with female depression or PND, using univariate or multiple regression analysis. Only significant results are depicted. Odds ratio represents the odds of female depression or PND status per standard deviation increase in standardised biomarker PGS.

Fig. 6.

Forest plot illustrating the association of biomarker PGS with female depression or PND, using univariate or multiple regression analysis. Only significant results are depicted. Odds ratio represents the odds of female depression or PND status per standard deviation increase in standardised biomarker PGS.

Close modal

Local Genetic Correlation

To further investigate significant PGS results, particularly in the case of total protein, which was not significantly globally genetically correlated with any depression group, and vitamin D, which was not significantly genetically correlated with PND, LAVA [44] was used to estimate local genetic correlations between total protein/vitamin D and female depression/PND. Results that achieved significance after Bonferroni correction, including those from the fine-grained sliding window analysis, are reported in Table 1. All results are provided in online supplementary Tables S12–S14.

Table 1.

Details of local genetic correlation analysis for results significant after correction for multiple testing

Protein/depression groupLocusStart; StopNum. SNPsRho (CI)R2p valueAdjusted pIncluded genes
Total protein/PND 5q13.3 75867877 76454345 1,416 0.68 (0.33, 1.0) 0.46 3.6e−04 1.0e−02 IQGAP2, NCRUPAR, F2RL1, S100Z, CRHBP, AGGF1, ZBED3, ZBED3-AS1 
5q13.3 75867877 75967876 377 1.0 (0.44, 1.0) 1.00 1.6e−03 1.5e−03 IQGAP2 
5q13.3 75967877 76067876 233 0.88 (0.47, 1.0) 0.77 7.0e−04 1.4e−03 IQGAP2, NCRUPAR, F2RL1 
Vitamin D/PND 3p12.1 85308140 85408139 189 1.0 (−1.0, −0.43) 1.0 2.6e−03 1.0e−02 CADM2 
Protein/depression groupLocusStart; StopNum. SNPsRho (CI)R2p valueAdjusted pIncluded genes
Total protein/PND 5q13.3 75867877 76454345 1,416 0.68 (0.33, 1.0) 0.46 3.6e−04 1.0e−02 IQGAP2, NCRUPAR, F2RL1, S100Z, CRHBP, AGGF1, ZBED3, ZBED3-AS1 
5q13.3 75867877 75967876 377 1.0 (0.44, 1.0) 1.00 1.6e−03 1.5e−03 IQGAP2 
5q13.3 75967877 76067876 233 0.88 (0.47, 1.0) 0.77 7.0e−04 1.4e−03 IQGAP2, NCRUPAR, F2RL1 
Vitamin D/PND 3p12.1 85308140 85408139 189 1.0 (−1.0, −0.43) 1.0 2.6e−03 1.0e−02 CADM2 

Results for fine-grained analysis are shown in bold.

For female total protein, 157 independent genome-wide significant loci were annotated by FUMA. Of these 157 loci, univariate analysis reported 30 to have significant SNP-based heritability (h2observed) for both total protein and PND. Bivariate analysis of these 30 loci reported three with significant local genetic correlation of total protein with PND, one of which, the 5q13.3 genomic region, survived Bonferroni correction (n = 30) (Table 1; online suppl. Table S11). Further fine-grained analysis, using a sliding 100 kb window across the locus, implemented for three nominally significant SNP-dense loci (number of SNPs >1,000) reported two regions with significant h2observed for both total protein and PND, located sequentially within the 5q13.3 locus and implicating IQ motif-containing GTPase activating protein 2 (IQGAP2) (Table 1; online suppl. Table S12).

For female depression, of the 157 loci, local genetic correlations with total protein were found for 3 loci with significant h2observed for both total protein and female depression, one of which was nominally significant, but not surviving Bonferroni correction (n = 3). Top ranked in significance was the 5p15.33 locus including telomerase reverse transcriptase (TERT) (rG = −0.26, CI = [−0.52, 1.0], p = 4.7 × 10−2) (online suppl. Table S13). No results met criteria for sliding window analysis. A sensitivity analysis of local genetic correlation between total protein and male depression, using loci with genome-wide significance for total protein measured in male participants, returned no significant results (online suppl. Table S13).

Results for the local genetic correlation analysis of vitamin D with PND are reported in online supplementary Table S14. Significant h2observed for both vitamin D and PND were reported for 10 of 38 genome-wide significant loci. Of these, local genetic correlation analysis found a significant result for two loci located in 4q13.3 and 3p12.1. The sliding window analysis returned one result that survived Bonferroni correction, located within the 3p12.1 genomic location and intronic to the cell adhesion molecule 2 (CADM2) (Table 1; online suppl. Table S14). For female depression, significant h2observed were reported for 2 loci, neither of which were significant (online suppl. Table S14).

In this study, we first used recorded levels of common biomarkers, available through UKB, to test for association of these biomarkers with lifetime depression in males and females separately. We evaluated if there were specific biomarker associations with PND which has been hypothesised to be related to reproductive hormones. We investigated whether polygenic scores for biomarker levels are associated with depression or PND in an independent cohort. Our analyses build on the groundwork established by a recent comprehensive study of UKB biomarkers [12], and given sexual heterogeneity in depression we have now extended this analysis through the comparison of the association of depression with biomarkers and covariates measured separately for men and women. To our knowledge, this is now the largest study of the sex-stratified genetic association of common biomarkers with depression. In particular, the specific focus on PND provides the opportunity to identify biomarkers genetically associated with a reproductive-related depression disorder.

Lifetime depression was associated with a total of 23 of 28 biomarkers, supporting previous research, which has found depression to be associated with many biomarkers [2, 5, 46‒64]. Twelve and eleven of these remained significant after adjusting for all other biomarkers for female and male depression respectively, with female depression most highly associated with lower levels of total protein and male depression with lower levels of vitamin D. The association of female depression/PND with total protein was supported by the PGS analysis. In general, our results did not support heterogeneity between depression in the perinatal period and more general lifetime female depression, since no biomarkers differentiate the two disorders, although genes that contribute to biomarker levels may vary across the disorders. In addition, many biomarkers were significantly associated with both male and female depression.

The significant genetic correlation of both female depression and PND with total bilirubin supported previous literature that has reported an association of depression with biomarkers of oxidative stress [54, 61]. The previously reported association with vitamin D [65‒67] was also supported by a significant negative genetic correlation with female depression, although its genetic correlation with PND and male depression was not significant. Although neither total protein nor vitamin D achieved significant global genetic correlation with PND, both female depression and PND were associated with polygenic risk scores that predict these biomarkers in the independent AGDS sample, and local genetic correlation analysis highlighted genomic regions that reported significant genetic correlation.

Whilst there seems to be little research regarding the relationship of depression with serum total protein, two studies support this finding: an early analysis of total protein and total protein fractions found significant differences between patients with MD and controls, with MD associated with significantly lower levels of total protein, albumin, and gamma globulin, and higher levels of alpha-1 and alpha-2 globulin [68], and a more recent study reported depression to be significantly associated with low total protein levels [61]. Our results also indicated that a low level of total protein is associated with genetic liability to female depression including PND. Fine-grained local genetic correlation analysis implicated IQGAP2, for which increased expression has been associated with immunosuppression and poor prognosis in lymphoma [69] but improved prognosis in breast cancer [70]. Nominally significant local negative genetic correlation for total protein with female depression was found for the 5p15.33 genomic region, which contains TERT, associated with the length of telomeres and associated genomic stability [71]. In contrast to these significant findings, no regions with significant genetic correlation were found for total protein and male depression.

The association between vitamin D deficiency and depression, including PND has been well established [46, 65‒67, 72‒74]. However, a causal association between vitamin D and depression has not been found, and evidence from recent studies do not support any causal influence of vitamin D on psychiatric disorders [65, 75, 76] but may reflect lifestyle choices (e.g., less time out of doors), or melanin-rich skin in a predominantly European population. A pleiotropic association [65, 75] is implied through the nominally significant negative local genetic correlation with PND in the 3p12.1 genomic region, intronic to CADM2, which has been identified as highly pleiotropic, associated with both sunseeking behaviour [77] and numerous psychiatric disorders [78, 79].

Limitations

This study has several limitations, of which the first is the timing of biomarker measurement, which did not occur during depression onset, but often late in life. We overcame this through a focus on lifetime depression, and, for Australian Genetics of Depression data, using a measure of lifetime PND [27]. We also controlled for menopause and current use of HRT in the female analysis, as well as BMI (for both female and male analyses), which has a strong effect on many biomarkers and, for many women and men, may have changed since depression onset. A second major limitation is the small size of many of the samples, which led to large standard errors in analyses and non-significant results. Nevertheless, the sex-stratified approach is a strength of this study, given that sex-differentiated research may be crucial to any analysis of the disparity in depression prevalence between males and females. To our knowledge, this is also the largest study of the association of PND with biomarkers, and, although results need to be substantiated through further research, including non-European participants, provides insights into the aetiology of this disorder.

Multiple analyses support an association between total protein levels and history of depression in women, and also provide support for the use of PGS as a proxy for biomarker levels. PGS results generally did not support a distinct aetiology for PND, since biomarkers associated with PND were also associated with more general female MD. This analysis provides evidence of specific genomic regions significantly associated with total protein level that also increase vulnerability to depression, including PND, for females, but not for males.

We thank the participants and investigators of the UKB study for their invaluable contributions to this work. The analysis of UKB data was approved by the Human Research Ethical Conduct (UQ) (Project 2020 HE002938: Statistical Methods and Algorithms for Analysis Of and Application to Genetic Data Sets), but individual patient consent was not required as this study was based on publicly available data.

For Australian Genetics of Depression Study data, all study protocols were approved by the QIMR Berghofer Medical Research Institute Human Research Ethics Committee (Ref 2118). The protocol for approaching participants through the DHS, enrolling them in the study, and consenting for all phases of the study (including invitation to future related studies) and accessing MBS and PBS records was approved by the Ethics Department of the Department of Human Services. Written and informed patient consent for participation in the study was obtained. We thank all Australian Genetics of Depression Study participants for their invaluable contributions to this work.

All authors declare that they have no relevant financial or non-financial interests to disclose.

The Australian Genetics of Depression Study was primarily funded by National Health and Medical Research Council (NHMRC) of Australia Grant 1086683. This work was further supported by NHMRC Grants 1145645, 1078901 and 1087889, and generously supported by a donation from the Axelsen family. The QSkin study is supported by National Health and Medical Research Council of Australia Grants APP1073898 and APP1058522. J.K. has been supported by a UQ Research Training Program scholarship, and E.M.B. received funding from an NHMRC Centre for Research Excellence (APP1198304) and the University of Queensland Health Research Accelerator Program. S.M.-B. receives sponsored research grants from Sage Therapeutics. D.C.W. is supported by a Research Fellowship from the National Health and Medical Research Council of Australia (APP1155413).

J.K. and E.M.B. designed the study and drafted the manuscript, and J.K. analysed the data. N.R.W., S.M.-B., J.M., I.B.H., D.C.W., C.M.O., S.E.M., N.G.M., and J.G. revised the manuscript for intellectual content. All authors have read and approve of the final version.

UKB data are available for public access. Summary statistics for the GWAS of each biomarker, for females and males separately, are available from the NHGRI-EBI Catalog of human genome-wide association studies (https://www.ebi.ac.uk/gwas/). The individualised PGS calculated for participants from the Australian Genetics of Depression study are not available due to privacy reasons. The underlying variant PGS for all biomarkers, calculated from each GWAS using SBayesR, contain no identifying data and are available from the corresponding author upon reasonable request.

1.
Bromet
EJ
,
Andrade
LH
,
Bruffaerts
R
,
Williams
DR
.
Major depressive disorder
.
Mental Disorders Around the World2017
; p.
41
56
.
2.
Hacimusalar
Y
,
Esel
E
.
Suggested biomarkers for major depressive disorder
.
Noro Psikiyatr Ars
.
2018
;
55
(
3
):
280
90
.
3.
Strawbridge
R
,
Young
AH
,
Cleare
AJ
.
Biomarkers for depression: recent insights, current challenges and future prospects
.
Neuropsychiatr Dis Treat
.
2017
;
13
:
1245
62
.
4.
Labandeira-Garcia
JL
,
Costa-Besada
MA
,
Labandeira
CM
,
Villar-Cheda
B
,
Rodriguez-Perez
AI
.
Insulin-like growth factor-1 and neuroinflammation
.
Front Aging Neurosci
.
2017
;
9
:
365
.
5.
Kuang
WH
,
Dong
ZQ
,
Tian
LT
,
Li
J
.
IGF-1 defends against chronic-stress induced depression in rat models of chronic unpredictable mild stress through the PI3K/Akt/FoxO3a pathway
.
Kaohsiung J Med Sci
.
2018
;
34
(
7
):
370
6
.
6.
Osimo
EF
,
Pillinger
T
,
Rodriguez
IM
,
Khandaker
GM
,
Pariante
CM
,
Howes
OD
.
Inflammatory markers in depression: a meta-analysis of mean differences and variability in 5,166 patients and 5,083 controls
.
Brain Behav Immun
.
2020
;
87
:
901
9
.
7.
Guo
X
,
Jiang
K
.
Is depression the result of immune system abnormalities
.
Shanghai Arch Psychiatry
.
2017
;
29
(
3
):
171
3
.
8.
Reay
WR
,
Kiltschewskij
DJ
,
Geaghan
MP
,
Atkins
JR
,
Carr
VJ
,
Green
MJ
, et al
.
Genetic estimates of correlation and causality between blood-based biomarkers and psychiatric disorders
.
Sci Adv
.
2022
;
8
(
14
):
eabj8969
.
9.
Bhatt
S
,
Nagappa
AN
,
Patil
CR
.
Role of oxidative stress in depression
.
Drug Discov Today
.
2020
;
25
(
7
):
1270
6
. .
10.
Almeida
FB
,
Barros
HMT
,
Pinna
G
.
Neurosteroids and neurotrophic factors: what is their promise as biomarkers for major depression and PTSD
.
Int J Mol Sci
.
2021
;
22
(
4
):
1758
. .
11.
Chavez-Castillo
M
,
Nunez
V
,
Nava
M
,
Ortega
A
,
Rojas
M
,
Bermudez
V
, et al
.
Depression as a neuroendocrine disorder: emerging neuropsychopharmacological approaches beyond monoamines
.
Adv Pharmacol Sci
.
2019
;
2019
:
7943481
. .
12.
Sinnott-Armstrong
N
,
Tanigawa
Y
,
Amar
D
,
Mars
N
,
Benner
C
,
Aguirre
M
, et al
.
Author correction: genetics of 35 blood and urine biomarkers in the UK biobank
.
Nat Genet
.
2021
;
53
(
11
):
1622
.
13.
Ramsey
JM
,
Cooper
JD
,
Bot
M
,
Guest
PC
,
Lamers
F
,
Weickert
CS
, et al
.
Sex differences in serum markers of major depressive disorder in The Netherlands study of depression and anxiety (NESDA)
.
PLoS One
.
2016
;
11
(
5
):
e0156624
.
14.
Jentsch
MC
,
Burger
H
,
Meddens
MBM
,
Beijers
L
,
van den Heuvel
ER
,
Meddens
MJM
, et al
.
Gender differences in developing biomarker-based major depressive disorder diagnostics
.
Int J Mol Sci
.
2020
;
21
(
9
):
3039
.
15.
World Health Organisation
. Depressive disorder (depression) 2023. Available from: https://www.who.int/news-room/fact-sheets/detail/depression.
16.
Kendler
KS
,
Ohlsson
H
,
Lichtenstein
P
,
Sundquist
J
,
Sundquist
K
.
The genetic epidemiology of treated major depression in Sweden
.
Am J Psychiatry
.
2018
;
175
(
11
):
1137
44
.
17.
Trzaskowski
M
,
Mehta
D
,
Peyrot
WJ
,
Hawkes
D
,
Davies
D
,
Howard
DM
, et al
.
Quantifying between-cohort and between-sex genetic heterogeneity in major depressive disorder
.
Am J Med Genet B Neuropsychiatr Genet
.
2019
;
180
(
6
):
439
47
.
18.
Hou
R
,
Westbury
L
,
Fuggle
N
,
Cooper
C
,
Dennison
E
.
Immune-endocrine biomarkers associated with mental health: a 9-year longitudinal investigation from the Hertfordshire Ageing Study
.
Brain Behav Immun
.
2022
;
101
:
146
52
.
19.
Jha
MK
,
Minhajuddin
A
,
Chin-Fatt
C
,
Greer
TL
,
Carmody
TJ
,
Trivedi
MH
.
Sex differences in the association of baseline C-Reactive Protein (CRP) and acute-phase treatment outcomes in major depressive disorder: findings from the EMBARC study
.
J Psychiatr Res
.
2019
;
113
:
165
71
.
20.
Sudlow
C
,
Gallacher
J
,
Allen
N
,
Beral
V
,
Burton
P
,
Danesh
J
, et al
.
UK Biobank: an open access resource for identifying the causes of a wide range of complex diseases of middle and old age
.
PLoS Med
.
2015
;
12
(
3
):
e1001779
.
21.
Davis
KAS
,
Coleman
JRI
,
Adams
M
,
Allen
N
,
Breen
G
,
Cullen
B
, et al
.
Mental health in UK Biobank: development, implementation and results from an online questionnaire completed by 157 366 participants: a reanalysis
.
BJPsych Open
.
2020
;
6
(
2
):
e18
.
22.
Kessler
RC
,
Andrews
G
,
Mroczek
D
,
Ustun
B
,
Wittchen
H-U
.
The world health organization Composite International Diagnostic Interview Short-Form (CIDI-SF)
.
Int J Methods Psychiatr Res
.
1998
;
7
(
4
):
171
85
.
23.
American Psychiatric Association
.
Diagnostic and statistical manual of mental disorders
.
Arlington
; VA2013.
24.
Lang
AJ
,
Wilkins
K
,
Roy-Byrne
PP
,
Golinelli
D
,
Chavira
D
,
Sherbourne
C
, et al
.
Abbreviated PTSD Checklist (PCL) as a guide to clinical response
.
Gen Hosp Psychiatry
.
2012
;
34
(
4
):
332
8
.
25.
Byrne
EM
,
Kirk
KM
,
Medland
SE
,
McGrath
JJ
,
Colodro-Conde
L
,
Parker
R
, et al
.
Cohort profile: the Australian genetics of depression study
.
BMJ Open
.
2020
;
10
(
5
):
e032580
.
26.
Cox
JL
,
Holden
JM
,
Sagovsky
R
.
Detection of postnatal depression. Development of the 10-item Edinburgh postnatal depression scale
.
Br J Psychiatry
.
1987
;
150
:
782
6
.
27.
Meltzer-Brody
S
,
Boschloo
L
,
Jones
I
,
Sullivan
PF
,
Penninx
BW
.
The EPDS-Lifetime: assessment of lifetime prevalence and risk factors for perinatal depression in a large cohort of depressed women
.
Arch Womens Ment Health
.
2013
;
16
(
6
):
465
73
.
28.
Olsen
CM
,
Green
AC
,
Neale
RE
,
Webb
PM
,
Cicero
RA
,
Jackman
LM
, et al
.
Cohort profile: the QSkin sun and health study
.
Int J Epidemiol
.
2012
;
41
(
4
):
929
i
.
29.
Gromping
U
.
Relative importance for linear regression in R: the package relaimpo
.
J Stat Softw
.
2006
;
17
(
1
):
1
27
.
30.
Lindeman
RH
,
Merenda
PF
,
Gold
RZ
.
Introduction to bivariate and multivariate analysis
.
Glenview
,
Ill: Scott, Foresman
;
1980
.
31.
Chen
PY
,
Cripps
AW
,
West
NP
,
Cox
AJ
,
Zhang
P
.
A correlation-based network for biomarker discovery in obesity with metabolic syndrome
.
BMC Bioinformatics
.
2019
;
20
(
Suppl 6
):
477
.
32.
Dormann
CF
,
Elith
J
,
Bacher
S
,
Buchmann
C
,
Carl
G
,
Carré
G
, et al
.
Collinearity: a review of methods to deal with it and a simulation study evaluating their performance
.
Ecography
.
2013
;
36
(
1
):
27
46
.
33.
Jiang
L
,
Zheng
Z
,
Qi
T
,
Kemper
KE
,
Wray
NR
,
Visscher
PM
, et al
.
A resource-efficient tool for mixed model association analysis of large-scale data
.
Nat Genet
.
2019
;
51
(
12
):
1749
55
.
34.
Zhu
Z
,
Zheng
Z
,
Zhang
F
,
Wu
Y
,
Trzaskowski
M
,
Maier
R
, et al
.
Causal associations between risk factors and common diseases inferred from GWAS summary data
.
Nat Commun
.
2018
;
9
(
1
):
224
.
35.
Yang
J
,
Lee
SH
,
Goddard
ME
,
Visscher
PM
.
GCTA: a tool for genome-wide complex trait analysis
.
Am J Hum Genet
.
2011
;
88
(
1
):
76
82
.
36.
Yang
J
,
Zeng
J
,
Goddard
ME
,
Wray
NR
,
Visscher
PM
.
Concepts, estimation and interpretation of SNP-based heritability
.
Nat Genet
.
2017
;
49
(
9
):
1304
10
.
37.
Zhou
W
,
Nielsen
JB
,
Fritsche
LG
,
Dey
R
,
Gabrielsen
ME
,
Wolford
BN
, et al
.
Efficiently controlling for case-control imbalance and sample relatedness in large-scale genetic association studies
.
Nat Genet
.
2018
;
50
(
9
):
1335
41
.
38.
Jiang
L
,
Zheng
Z
,
Fang
H
,
Yang
J
.
A generalized linear mixed model association tool for biobank-scale data
.
Nat Genet
.
2021
;
53
(
11
):
1616
21
.
39.
Bulik-Sullivan
B
,
Finucane
HK
,
Anttila
V
,
Gusev
A
,
Day
FR
,
Loh
P-R
, et al
.
An atlas of genetic correlations across human diseases and traits
.
Nat Genet
.
2015
;
47
(
11
):
1236
41
. .
40.
Cross-Disorder Group of the Psychiatric Genomics Consortium
;
Lee
SH
,
Ripke
S
,
Neale
BM
,
Faraone
SV
,
Purcell
SM
, et al
.
Genetic relationship between five psychiatric disorders estimated from genome-wide SNPs
.
Nat Genet
.
2013
;
45
(
9
):
984
94
.
41.
Kraft
P
,
Chen
H
,
Lindstrom
S
.
The use of genetic correlation and mendelian randomization studies to increase our understanding of relationships between complex traits
.
Curr Epidemiol Rep
.
2020
;
7
(
2
):
104
12
.
42.
Ni
G
,
Moser
G
;
Schizophrenia Working Group of the Psychiatric Genomics Consortium
,
Wray
NR
,
Lee
SH
.
Estimation of genetic correlation via linkage disequilibrium score regression and genomic restricted maximum likelihood
.
Am J Hum Genet
.
2018
;
102
(
6
):
1185
94
.
43.
Lloyd-Jones
LR
,
Zeng
J
,
Sidorenko
J
,
Yengo
L
,
Moser
G
,
Kemper
KE
, et al
.
Improved polygenic prediction by Bayesian multiple regression on summary statistics
.
Nat Commun
.
2019
;
10
(
1
):
5086
.
44.
Werme
J
,
van der Sluis
S
,
Posthuma
D
,
de Leeuw
CA
.
An integrated framework for local genetic correlation analysis
.
Nat Genet
.
2022
;
54
(
3
):
274
82
.
45.
Watanabe
K
. Functional mapping and annotation of genome-wide association studies 2019 [Available from: https://fuma.ctglab.nl/.
46.
Aghajafari
F
,
Letourneau
N
,
Mahinpey
N
,
Cosic
N
,
Giesbrecht
G
.
Vitamin D deficiency and antenatal and postpartum depression: a systematic review
.
Nutrients
.
2018
;
10
(
4
):
478
.
47.
Al-Dujaili
AH
,
Al-Hakeim
HK
,
Twayej
AJ
,
Maes
M
.
Total and ionized calcium and magnesium are significantly lowered in drug-naive depressed patients: effects of antidepressants and associations with immune activation
.
Metab Brain Dis
.
2019
;
34
(
5
):
1493
503
.
48.
Ambrus
L
,
Westling
S
.
Inverse association between serum albumin and depressive symptoms among drug-free individuals with a recent suicide attempt
.
Nord J Psychiatry
.
2019
;
73
(
4–5
):
229
32
.
49.
Amini
S
,
Amani
R
,
Jafarirad
S
,
Cheraghian
B
,
Sayyah
M
,
Hemmati
AA
.
The effect of vitamin D and calcium supplementation on inflammatory biomarkers, estradiol levels and severity of symptoms in women with postpartum depression: a randomized double-blind clinical trial
.
Nutr Neurosci
.
2022
;
25
(
1
):
22
32
. .
50.
Beijers
L
,
Wardenaar
KJ
,
Bosker
FJ
,
Lamers
F
,
van Grootheest
G
,
de Boer
MK
, et al
.
Biomarker-based subtyping of depression and anxiety disorders using Latent Class Analysis. A NESDA study
.
Psychol Med
.
2019
;
49
(
4
):
617
27
.
51.
Chen
Z
,
Shen
X
,
Tian
K
,
Liu
Y
,
Xiong
S
,
Yu
Q
, et al
.
Bioavailable testosterone is associated with symptoms of depression in adult men
.
J Int Med Res
.
2020
;
48
(
8
):
300060520941715
.
52.
Gao
J
,
Xu
W
,
Han
K
,
Zhu
L
,
Gao
L
,
Shang
X
.
Changes of serum uric acid and total bilirubin in elderly patients with major postischemic stroke depression
.
Neuropsychiatr Dis Treat
.
2018
;
14
:
83
93
.
53.
Huang
Y
,
Huang
W
,
Wei
J
,
Yin
Z
,
Liu
H
.
Increased serum cystatin C levels were associated with depressive symptoms in patients with type 2 diabetes
.
Diabetes Metab Syndr Obes
.
2021
;
14
:
857
63
.
54.
Jayanti
S
,
Moretti
R
,
Tiribelli
C
,
Gazzin
S
.
Bilirubin and inflammation in neurodegenerative and other neurological diseases
.
Neuroimmunol Neuroinflamm
.
2020
;
2020
:
2020
.
55.
Kim
WJ
,
Kim
HR
,
Song
JS
,
Choi
ST
.
Low levels of serum urate are associated with a higher prevalence of depression in older adults: a nationwide cross-sectional study in Korea
.
Arthritis Res Ther
.
2020
;
22
(
1
):
104
.
56.
Levada
OA
,
Troyan
AS
,
Pinchuk
IY
.
Serum insulin-like growth factor-1 as a potential marker for MDD diagnosis, its clinical characteristics, and treatment efficacy validation: data from an open-label vortioxetine study
.
BMC Psychiatry
.
2020
;
20
(
1
):
208
.
57.
Li
H
,
Wang
A
,
Qi
G
,
Guo
J
,
Li
X
,
Wang
W
, et al
.
Cystatin C and risk of new-onset depressive symptoms among individuals with a normal creatinine-based estimated glomerular filtration rate: a prospective cohort study
.
Psychiatry Res
.
2019
;
273
:
75
81
.
58.
Maddock
RJ
,
Moses
J
,
Roth
W
,
King
R
,
Murchison
A
,
Berger
P
.
Serum phosphate and anxiety in major depression
.
Psychiatry Res
.
1987
;
22
(
1
):
29
36
.
59.
Meng
X
,
Huang
X
,
Deng
W
,
Li
J
,
Li
T
.
Serum uric acid a depression biomarker
.
PLoS One
.
2020
;
15
(
3
):
e0229626
.
60.
Mrug
S
,
Orihuela
C
,
Mrug
M
,
Sanders
PW
.
Sodium and potassium excretion predict increased depression in urban adolescents
.
Physiol Rep
.
2019
;
7
(
16
):
e14213
.
61.
Peng
YF
,
Xiang
Y
,
Wei
YS
.
The significance of routine biochemical markers in patients with major depressive disorder
.
Sci Rep
.
2016
;
6
:
34402
.
62.
Sadeghi
M
,
Roohafza
H
,
Afshar
H
,
Rajabi
F
,
Ramzani
M
,
Shemirani
H
, et al
.
Relationship between depression and apolipoproteins A and B: a case-control study
.
Clinics
.
2011
;
66
(
1
):
113
7
.
63.
Schmitz
N
,
Deschenes
S
,
Burns
R
,
Smith
KJ
.
Depressive symptoms and glycated hemoglobin A1c: a reciprocal relationship in a prospective cohort study
.
Psychol Med
.
2016
;
46
(
5
):
945
55
.
64.
Wang
H
,
Huang
B
,
Wang
W
,
Li
J
,
Chen
Y
,
Flynn
T
, et al
.
High urea induces depression and LTP impairment through mTOR signalling suppression caused by carbamylation
.
EBioMedicine
.
2019
;
48
:
478
90
.
65.
Arathimos
R
,
Ronaldson
A
,
Howe
LJ
,
Fabbri
C
,
Hagenaars
S
,
Hotopf
M
, et al
.
Vitamin D and the risk of treatment-resistant and atypical depression: a Mendelian randomization study
.
Transl Psychiatry
.
2021
;
11
(
1
):
561
.
66.
Cuomo
A
,
Giordano
N
,
Goracci
A
,
Fagiolini
A
.
Depression and Vitamin D deficiency: causality, assessment and clinical practice implications
.
Neuropsychiatry
.
2017
;
07
(
05
):
606
14
.
67.
Menon
V
,
Kar
SK
,
Suthar
N
,
Nebhinani
N
.
Vitamin D and depression: a critical appraisal of the evidence and future directions
.
Indian J Psychol Med
.
2020
;
42
(
1
):
11
21
.
68.
Maes
M
,
Wauters
A
,
Neels
H
,
Scharpe
S
,
Van Gastel
A
,
D’Hondt
P
, et al
.
Total serum protein and serum protein fractions in depression: relationships to depressive symptoms and glucocorticoid activity
.
J Affect Disord
.
1995
;
34
(
1
):
61
9
.
69.
Tang
T
,
Wang
J
,
Zhang
L
,
Cheng
Y
,
Saleh
L
,
Gu
Y
, et al
.
IQGAP2 acts as an independent prognostic factor and is related to immunosuppression in DLBCL
.
BMC Cancer
.
2021
;
21
(
1
):
603
.
70.
Kumar
D
,
Patel
SA
,
Hassan
MK
,
Mohapatra
N
,
Pattanaik
N
,
Dixit
M
.
Reduced IQGAP2 expression promotes EMT and inhibits apoptosis by modulating the MEK-ERK and p38 signaling in breast cancer irrespective of ER status
.
Cell Death Dis
.
2021
;
12
(
4
):
389
.
71.
Dogan
F
,
Forsyth
NR
.
Telomerase regulation: a role for epigenetics
.
Cancers
.
2021
;
13
(
6
):
1213
.
72.
Cuomo
A
,
Maina
G
,
Bolognesi
S
,
Rosso
G
,
Beccarini Crescenzi
B
,
Zanobini
F
, et al
.
Prevalence and correlates of vitamin D deficiency in a sample of 290 inpatients with mental illness
.
Front Psychiatry
.
2019
;
10
:
167
.
73.
Jani
R
,
Knight-Agarwal
CR
,
Bloom
M
,
Takito
MY
.
The association between pre-pregnancy body mass index, perinatal depression and maternal vitamin D status: findings from an Australian cohort study
.
Int J Womens Health
.
2020
;
12
:
213
9
.
74.
Cai
N
,
Revez
JA
,
Adams
MJ
,
Andlauer
TFM
,
Breen
G
,
Byrne
EM
, et al
.
Minimal phenotyping yields GWAS hits of low specificity for major depression
.
bioRxiv preprint
.
2019
;
2018
.
75.
Revez
JA
,
Lin
T
,
Qiao
Z
,
Xue
A
,
Holtz
Y
,
Zhu
Z
, et al
.
Genome-wide association study identifies 143 loci associated with 25 hydroxyvitamin D concentration
.
Nat Commun
.
2020
;
11
(
1
):
1647
.
76.
Rahman
ST
,
Waterhouse
M
,
Romero
BD
,
Baxter
C
,
English
DR
,
Almeida
OP
, et al
.
Effect of vitamin D supplementation on depression in older Australian adults
.
Int J Geriatr Psychiatry
.
2023
;
38
(
1
):
e5847
.
77.
Sanna
M
,
Li
X
,
Visconti
A
,
Freidin
MB
,
Sacco
C
,
Ribero
S
, et al
.
Looking for sunshine: genetic predisposition to sun seeking in 265,000 individuals of European ancestry
.
J Invest Dermatol
.
2021
;
141
(
4
):
779
86
.
78.
Sanchez-Roige
S
,
Jennings
MV
,
Thorpe
HHA
,
Mallari
JE
,
van der Werf
LC
,
Bianchi
SB
, et al
.
CADM2 is implicated in impulsive personality and numerous other traits by genome- and phenome-wide association studies in humans and mice
.
Transl Psychiatry
.
2023
;
13
(
1
):
167
.
79.
Pasman
JA
,
Chen
Z
,
Smit
DJA
,
Vink
JM
,
Van Den Oever
MC
,
Pattij
T
, et al
.
The CADM2 gene and behavior: a phenome-wide scan in UK-biobank
.
Behav Genet
.
2022
;
52
(
4–5
):
306
14
.