Type 2 diabetes is causally associated with depression: a Mendelian randomization analysis

Abstract

Type 2 diabetes (T2D) has been associated with a high prevalence of depression. We aimed to determine the causal relation by performing a Mendelian randomization (MR) study using 34 T2D risk genetic variants validated in East Asians as the instrumental variable (IV). An MR analysis was performed involving 11 506 participants from a large longitudinal study. The T2D genetic risk score (GRS) was built using the 34 typical T2D common variants. We used T2D_GRS as the IV estimator and performed inverse-variance weighted (IVW) and Egger MR analysis. The T2D_GRS was found to be associated with depression with an OR of 1.21 (95% CI: 1.07–1.37) after adjustments for age, sex, body mass index, current smoking and drinking, physical activity, education, and marital status. Using T2D_GRS as the IV, we similarly found a causal relationship between genetically determined T2D and depression (OR: 1.84, 95% CI: 1.25–2.70). Though we found no association between the combined effect of the genetic IVs for T2D and depression with Egger MR (OR: 0.95, 95% CI: 0.42–2.14), we found an association for T2D and depression with IVW (OR: 1.75, 95% CI: 1.31–2.46) after excluding pleiotropic SNPs. Overall, the MR analyses provide evidence inferring a potential causal relationship between T2D and depression.

Keywords

causal modeling / depression / Mendelian randomization / type 2 diabetes

Introduction

Depression is a common mental health disorder that causes various symptoms, including a persistent feeling of sadness, tearfulness, loss of interest, tiredness, reduced appetite, and frequent or recurrent thoughts of death [¹,²]. Today, depression has become a major public health problem and the leading cause of ill health and disability worldwide. As reported by the World Health Organization (WHO), more than 300 million people are living with depression worldwide, and at least 18% increment was found between 2005 and 2015, reaching up to 54 million in China [³]. Known risk factors that are likely to develop or trigger depression include aging, female gender, childhood adversity, history of other mental health disorders, traumatic or stressful events, serious or chronic illness, medical treatments, and diabetes status [⁴]. Meanwhile, diabetes has also become an epidemic and health threat worldwide, including China [⁵,⁶]. Depression, in combination with diabetes, will result in serious morbidities, mortalities, and decreased health-related quality of life [³,⁷,⁸].

A bidirectional relationship of type 2 diabetes (T2D) with depression has been consistently reported in numerous epidemiological studies [⁹,¹⁰]. Systematic reviews and meta-analyses indicated that T2D was associated with a high risk of depression [⁹,¹¹,¹²]. Performing with a large nationwide Chinese population of 229 047 middle-aged and elderly participants (age≥40 years), our previous observational analysis suggested that a higher prevalence of depression was observed among patients with previously diagnosed diabetes than those with normal glucose regulation [¹³]. However, reasons for the reciprocal link between T2D and depression remain unclear. Shared environmental factors (for example, unhealthy lifestyles and increased obesity risk) have been hypothesized to underlie depression and T2D [⁹,¹¹]. The prior studies mainly used cross-sectional study designs, which are likely to be influenced by various potential biases, such as confounders or reverse causation; thus, whether T2D plays a causal role in the development of depression remains undiscerned.

The Mendelian randomization (MR) analysis has recently gained considerable interest as a novel and powerful approach to investigate causal relationships of metabolic exposure factors, biomarkers, and cardiovascular diseases [¹⁴,¹⁵]. Based on the theory of random assortment of genetic variants at the time of gamete formation, MR is considered parallel to the randomized controlled trial and avoids the potential problems of traditional observational studies, which are clearly regarded as key strengths [¹⁶]. The causal relationship of T2D with depression was rarely examined using MR analysis.

Therefore, we performed an MR analysis in the present study to investigate the potential causal relationship of T2D with depression in a large Chinese population (n = 11 506) with a major strength of high homogeneity. The T2D genetic risk score (GRS) calculated on the basis of 34 T2D variants identified and established in East Asians was used as the IV throughout the analysis [¹⁷,¹⁸]. We further performed the sensitivity analyses, which included taking the weighted and un-weighted T2D_GRS and excluding the variants with pleiotropic effects as the IVs. Then, we used the recently developed inverse-variance weighted (IVW) and Egger MR analysis, which reduces inflation of a causal effect estimate due to measured and unmeasured pleiotropy.

Materials and methods

Participants

The participants were community-dwelling adults (aged≥40 years) from two adjacent communities at Baoshan district (Shanghai, China) which belonged to a large longitudinal study in 2011 and 2013. Details of this study had been previously provided [¹⁹]. Except for participants who had unavailable information on mental health or genotype (three or more genetic variants), a total of 11 506 participants were included in our study. The present study was approved by the Institutional Review Board of Ruijin Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, and all participants involved in the study provided written informed consent prior to participation.

Anthropometric information and biochemical measurements

We used a standard questionnaire to acquire information regarding lifestyle factors (such as physical activity, alcohol drinking, and tobacco smoking habits), psychological status, education, marital status, disease, and medical history. Anthropometric information of each participant was measured by the trained investigators. Body mass index (BMI) was calculated by using the following formula: BMI= weight (kilograms) divided by height (squared meters). After at least 10 min rest in a quiet room, blood pressure was measured in triplicate on the non-dominant arm and averaged in each participant.

Venous blood samples were collected after a minimum of 10 h overnight fast for biochemical analyses. Each participant also underwent a 75 g oral glucose tolerance test. Details of anthropometric and biochemical measurements, including plasma glucose and lipids profile, were available in previous studies [¹⁵,¹⁹].

Definitions

Depression was screened by using a practical and standard questionnaire, namely, Patient Health Questionnaire 9 (PHQ–9). The PHQ–9 contains nine items corresponding to criteria provided by the Diagnostic and Statistical Manual of Mental Disorders, 4th Edition (DSM–IV) for diagnosis of major depression and was widely used in primary health centers for screening of depression [¹³,²⁰–²²]. Each of the nine items is scored as 0, 1, 2, or 3, which corresponded to the level of agreement with the response item of “not at all,” “several days,” “more than half the days,” or “nearly every day.” The total summed score of the nine items ranged from 0 to 27, in which the PHQ–9 score of 5–27 represented depression while 5–9 and 10–27 denoted sub-threshold and probable depression, respectively [¹³,²⁰].

We defined T2D according to the diagnostic criteria of the 1999 WHO: fasting plasma glucose (FPG)≥7.0 mmol/L or 2 h PG≥11.1 mmol/L or previously diagnosed T2D or use of relevant medication. The homeostasis model assessment of insulin resistance index (HOMA-IR) was calculated as fasting insulin (mIU/mL) × fasting glucose (mmol/L) / 22.5 [²³]. The homeostasis model assessment of b cell function or insulin secretion (HOMA-b) was calculated by using the following formula: HOMA-b = [20×fasting insulin (mIU/mL)] / [fasting glucose (mmol/L) − 3.5] [²⁴]. The systolic blood pressure (SBP)≥140 mmHg or diastolic blood pressure (DBP)≥90 mmHg or use of anti-hypertension medication was defined as hypertension. The definition of current smoking, current drinking status, and physical activity at leisure time is referred to in previous studies [¹⁵,²⁵]. Education status was defined as low, intermediate, and high, corresponding to the education time of no more than 6 years, 7−9 years, and≥10 years. Marital status was categorized into single or not.

Genotyping

Single-nucleotide polymorphisms (SNPs) were genotyped by the mass spectrometer of MALDI-TOF and Mass Array Type 4.0 software (CapitalBio Corporation, Beijing, China), which has previously been provided in detail [²⁶]. We carefully selected 34 genome-wide significant (P≤5×10⁻⁸) SNPs, which were either discovered in Europeans and replicated in East Asians [²⁷,²⁸] or those identified and validated in meta-analysis, including GWASs from East Asians [²⁹]. Except for CDC123/CAMK1D rs10906115 and rs12779790 (r² = 0.055), all SNPs have no linkage disequilibrium. We calculated the weighted T2D_GRS as a weighted sum of each allele’s effect size (b-coefficients), as reported in a GWAS meta-analysis [²⁹]. An un-weighted T2D_GRS was a simple and un-weighted count of the number of T2D risk alleles in each participant. The average value of weighted T2D_GRS was 34.52 points (SD, 3.90) and that of un-weighted T2D_GRS was 35.89 points (SD, 3.62).

Additionally, we performed the relationships of each variant with metabolic traits (Table 3) and found that 16 SNPs had potential pleiotropic effects on confounding traits (BMI, SBP, DBP, and lipids). A total of eight SNPs (specifically, rs780094, rs9356744, rs864745, rs7903146, rs2237892, rs17817449, rs4430796, and rs3786897), which are listed in the GWAS catalog, are associated with other confounder traits, such as waist-to-hip ratio, BMI, obesity, and urate level (http://www.ebi.ac.uk/gwas/). We then created the weighted and un-weighted T2D_GRS_{/concluding 14 SNPs} using SNPs after excluding those with potential pleiotropic effects from our study sample and the GWAS Catalog.

Statistical analysis

We performed statistical analyses using SAS version 9.4 (SAS Institute, Cary, NC, USA). Linear regression analyses and c² tests were performed to assess trends across the T2D_GRS quartiles for continuous and categorical variables, respectively. Variables in non-normal distribution, including triglycerides (TG), HOMA-IR, and HOMA-b, were normalized by logarithmic transformation before performing statistical analyses. We used multivariable-adjusted linear regression models to evaluate the association of each SNP with T2D related traits. Multivariable logistic regression analyses were performed to assess the relationships of SNPs or GRSs and the presence of T2D or depression.

In the MR analysis, we used the IV estimators to measure the strength of the causal relationship between T2D and depression. The T2D_GRS was chosen as the IV. The IV estimate of causal OR was derived by using the Wald-type estimator [³⁰] and then exponentiating to express as an OR. The computational formula was OR_IV = exp (ln (OR_{GRS-depression}) / ln(OR_GRS-T2D)), in which ln(OR_{GRS-depression}) was {\mathrm{\beta }}_{\mathrm{G}\mathrm{R}\mathrm{S}-\mathrm{d}\mathrm{e}\mathrm{p}\mathrm{r}\mathrm{e}\mathrm{s}\mathrm{s}\mathrm{i}\mathrm{o}\mathrm{n}} and ln(OR_GRS-T2D) was {\mathrm{\beta }}_{\mathrm{G}\mathrm{R}S-\mathrm{T}2\mathrm{D}}.

The IV estimates were adjusted for age (years), sex, BMI (kg/m²), current smoking (yes or no), current drinking (yes or no), physical activity (mild, moderate or vigorous), hypertension (yes or no), TG (mmol/L), total cholesterol (mmol/L), education, and marital status. Finally, we performed the Egger MR regression by running the IVW model with the intercept unconstrained; the intercepts can be regarded as metrics of average pleiotropy effect across the chosen variants [³¹]. A two-sided P value of less than 0.05 was considered statistically significant.

Results

Genetic loci and associations with diabetes and depression

Among the 11 506 participants, the average age was 63.3 years (SD, 13.6), 4090 (35.5%) were men, 2860 (24.9%) were T2D patients, and 290 (2.5%) were with depression (242 sub-threshold depression and 48 probable depression) (Table 1). As expected, T2D related traits, such as FPG, 2h PG, and log₁₀HOMA-IR, and the number of T2D patients are positively related to T2D_GRS, while log₁₀HOMA-b showed inverse association.

The information of each individual SNP and its association with diabetes and depression are shown in Table 2. None of each individual SNP was found to be associated with depression after adjustment for age, sex, and BMI. We also examined the relationships of each SNP with metabolic traits (specifically: BMI, SBP, DBP, total cholesterol [TC], and log₁₀TG) to determine the pleiotropy of the SNPs (Table 3). A total of 16 SNPs showed significant associations with the T2D related traits, and were considered to have pleiotropic effects (Table 2 and Table 3).

Association of depression with T2D_GRS and T2D

The theoretically observed risk of depression as a function of T2D_GRS and T2D status is shown in Table 4. Each 1 SD (3.62 points) increment in un-weighted GRS was associated with an 18% increased odds of depression in an adjusted age, sex, and BMI model (95% confidence interval [CI]: 1.05–1.33, P = 0.006). The results did not considerably change after further adjustments for current smoking, drinking status (yes or no), physical activity (mild, moderate or vigorous), education, marital status, hypertension (yes or no), TG, and TC in model 2 (OR= 1.21, 95% CI: 1.07−1.37, P = 0.002). Similar results were found in model 3 after adjustments for FPG, 2 h PG, and T2D status based on model 2 (OR= 1.17, 95% CI: 1.03−1.32). The association of un-weighted GRS_{/concluding 14 SNPs} and the odds of depression had similar results.

Each 1 SD (3.90 points) increment in weighted GRS was associated with a 14% increased odds of depression (95% CI: 1.01−1.29, P = 0.03), with multivariable adjustments for age, BMI, smoking, drinking status, hypertension, physical activity, education, marital status, TG, and TC. After further adjustments for FPG, 2 h PG, and T2D status in model 3, we did not find a significant association between the GRS and depression (OR= 1.10, 95% CI: 0.97−1.25, P = 0.13). However, we observed a significant association of the GRS and depression only in T2D individuals and not in those without T2D to further detect the associations of present depression with T2D_GRS according to T2D status (Table 5). In addition, a similar association was confirmed between GRS_{/concluding 14 SNPs} and large odds of depression with an OR of 1.16 (95% CI: 1.02−1.31) after multivariable adjustment (Table 4).

In the association analyses, present T2D status was associated with high odds of depression with an OR of 1.37 (95% CI: 1.05−1.78) in model 1. Further adjustment for other confounding factors (based on model 1) did not alter the significant association (OR= 1.45, 95% CI: 1.10−1.92). Additionally, in the association analyses, multiple T2D_GRSs were consistently associated with T2D with ORs of 1.36 (95% CI:1.30−1.42) for un-weighted GRS, 1.33 (95% CI: 1.27−1.40) for weighted GRS, 1.15 (95% CI: 1.10−1.21) for un-weighted GRS_{/concluding 14 SNPs,} and 1.16 (95% CI: 1.11−1.21) for weighted GRS_{/concluding 14 SNPs} (all P<0.05, data not shown).

MR analysis for the association of T2D with depression

We further performed a standard MR analysis to estimate the causal effect of genetically determined T2D on depression (Table 6). Consistently significant associations of genetically determined T2D with depression were observed in the analyses using multiple T2D_GRSs as the IVs. Using the un-weighted GRS as the IVs to estimate the causal effect of T2D on depression yielded an OR of 1.84 (95% CI: 1.25−2.70, P = 0.0003). The causal OR of genetically determined T2D for depression was 1.57 (95% CI: 1.04−2.37, P = 0.02) using weighted GRS. After excluding the pleiotropic SNP demonstrating a statistical association with pleiotropic traits, we found that the IV estimates for causal effect of T2D on depression were also statistically significant (Table 6). Though no association was found between the combined effect of the genetic IVs for T2D and depression with Egger MR (OR: 0.95; 95% CI: 0.42−2.14), we found associations for T2D with IVW (OR= 1.75, 95% CI: 1.31−2.46) after excluding the possible pleiotropic effects of SNPs on depression.

Discussion

In the present study of 11 506 community-dwelling adults in China, we demonstrated that the T2D_GRS based on 34 specific loci of East Asians was significantly associated with large odds of depression. The MR analyses also suggested a potential evidence of a causal relationship between T2D and depression.

Evidence from epidemiology studies has suggested a positive association between T2D and depression [⁹–¹²]. A meta-analysis including 11 studies showed that compared with non-diabetic controls, T2D patients had a 24% increased risk of triggering depression [¹¹]. Another systematic review concluded that the prevalence of depression was twice as common in diabetic patients as in the general population, affecting 10%–20% of adults with diabetes [⁴]. Despite the observational studies, the MR analyses of T2D with depression are limited. Two previous studies have investigated diabetes and major depression using T2D associated variants in European populations [³²,³³]. Samaan et al. investigated the association of 20 T2D predisposing SNPs analyzed individually or together as a T2D_GRS with major depressive disorder [³²]. Another study tested the potential causal association of 11 T2D SNPs with major depressive disorder by using an MR analysis [³³]. Both studies found that T2D might not be causally related to depression, which was inconsistent with findings from previous observational studies. The authors speculated that the relation between T2D and incident risk of depression might be influenced by the strength or direction of environmental factors or unmeasured confounders rather than a shared biological origin [³²,³³]. However, these studies notably had diverse ethnic group and varying methods. In the present MR analysis, we found a significant association between genetically determined T2D and depression, indicating a potential causality. The results from our study support those obtained from previous observational findings. Several other previous MR studies focused on the causal relationships between BMI or obesity, dietary or lifestyle factors, such as cigarette taking and alcohol drinking, with depression [³⁴–³⁶]. Nevertheless, neither of these factors strongly indicate their causal association with depression. However, this finding might suggest a straight link between T2D and depression not attributed to the confounding effect. Though validation in other populations is merited and encouraged, our findings suggest that a better treatment and control of T2D may be beneficial for prevention of depression.

Our study met several assumptions, allowing us to perform a high-quality MR study. First, the IV was reliably associated with the exposures of interest. All candidate variants selected in our study have been validated to strongly relate with T2D in large meta-analysis of GWASs in East Asians, and most of these variants can be replicated in the present study. Second, the IV was independent of any confounders. We checked for pleiotropic association with major known confounders, such as BMI, blood pressure, and lipids profile, to limit the potential influence of pleiotropic effects of the IV. We also searched the GWAS Catalog website excluding SNPs reported to be associated with other metabolic traits and created a T2D_GRS excluding those SNPs with pleiotropic effect. The significant association between the T2D_GRS comprised non-pleiotropic SNPs, and depression was likely to be interpreted as an evidence of the true causal relationship of T2D with depression, which is independent of potential confounders. Additionally, the causal relationship remained consistent in the sensitivity analysis. In the association analyses, we found a significant relation between T2D_GRS and depression after adjustment for BMI, lifestyle factors, education, marital status, hypertension, and lipids. However, after further adjustment for glycemic traits or present T2D status, this significance attenuated, which might infer that the relationship was mediated by hyperglycemia. Moreover, stratified analyses based on T2D status showed that a significant relationship of the GRS and depression is more observed in T2D individuals compared with those without T2D. Although Egger MR analysis confirmed the presence of unmeasured pleiotropy when using T2D_GRS concluding 34 SNPs as IV, the causal estimate from IVW was again directionally consistent with conventional MR. Moreover, the intercept of MR–Egger was insignificant after excluding pleiotropic variants, which indicated the absence of directional pleiotropy. Third, the IV was independent of the outcome. The proxy SNPs does not have a direct effect on the outcome or any other mediated effects other than through the exposure of interest. In our study, none of the SNPs used in the creation of T2D_GRS was associated with depression. However, until the underlying mechanisms linking all SNPs (or GRS) and depression are comprehensively understood, this assumption remains largely untestable. The mechanisms underlining the associations of T2D with risk of depression remain unclear. A previous study indicated that T2D and depression might share some common biological pathways, including innate inflammatory response, hypothalamic−pituitary−adrenal (HPA) axis, circadian rhythms, and insulin resistance [³⁷].

The major strengths of our study included its well-defined community setting, the utilization of genetic and MR approaches, and a T2D_GRS well representing genetic risk of T2D as the IV. However, several weaknesses of our study should be acknowledged. First, only typical T2D common variants were included to calculate the T2D_GRS in our study, which was considered to represent limited diabetes heritability. We failed to assess the probable contribution of rare or low-frequency SNPs. Second, we used the PHQ–9 questionnaire rather than a comprehensive psychiatric evaluation to define depression. However, the use of a brief rating scale to screen for depression in primary care patients is supported by previous studies and suggested by the United States Preventive Services Task Force [³⁸]. In a systematic review synthesizing the evidence for screening of depression in people with diabetes, the PHQ–9 proved to be the best validated compared with other questionnaires [³⁹]. Therefore, the use of the PHQ–9 can be a practical and time-consuming diagnostic procedure. Third, though this study provided an evidence for a seemly causal effect of lifetime exposure of hyperglycemia on depression, we cannot account for the effect of acute changes. Fourth, the genetic variants used in this study were robustly associated with T2D in East Asians; it may be questioned to generalize the finding to other ethnic groups. Finally, though this study is a longitudinal design study, we only had one-time information for the diagnosis of depression so far, hindering us to further analyze the contribution of T2D on incident risk of depression and validate the findings observed in the present study.

Overall, we found that genetically determined T2D was associated with the presence of depression. The MR analysis provided evidence for potential causal relationship between T2D and depression. The psychiatric symptoms shall be paid more attention, and better therapy and control for diabetes would be beneficial for depression prevention.

References

[1] Whiteford HA, Degenhardt L, Rehm J, Baxter AJ, Ferrari AJ, Erskine HE, Charlson FJ, Norman RE, Flaxman AD, Johns N, Burstein R, Murray CJ, Vos T. Global burden of disease attributable to mental and substance use disorders: findings from the Global Burden of Disease Study 2010. Lancet 2013; 382(9904):1575–1586

[2] Ferrari AJ, Charlson FJ, Norman RE, Patten SB, Freedman G, Murray CJ, Vos T, Whiteford HA. Burden of depressive disorders by country, sex, age, and year: findings from the global burden of disease study 2010. PLoS Med 2013; 10(11): e1001547

[3] World Health Organization. Depression and Other Common Mental Disorders: Global Health Estimates. 2017. Available from

[4] Roy T, Lloyd CE. Epidemiology of depression and diabetes: a systematic review. J Affect Disord 2012; 142(Suppl): S8–S21

[5] Nanditha A, Ma RC, Ramachandran A, Snehalatha C, Chan JC, Chia KS, Shaw JE, Zimmet PZ. Diabetes in Asia and the Pacific: implications for the Global Epidemic. Diabetes Care 2016; 39(3): 472–485

[6] Xu Y, Wang L, He J, Bi Y, Li M, Wang T, Wang L, Jiang Y, Dai M, Lu J, Xu M, Li Y, Hu N, Li J, Mi S, Chen CS, Li G, Mu Y, Zhao J, Kong L, Chen J, Lai S, Wang W, Zhao W, Ning G; 2010 China Noncommunicable Disease Surveillance Group. Prevalence and control of diabetes in Chinese adults. JAMA 2013; 310(9): 948–959

[7] Black SA, Markides KS, Ray LA. Depression predicts increased incidence of adverse health outcomes in older Mexican Americans with type 2 diabetes. Diabetes Care 2003; 26(10): 2822–2828

[8] Anderson BJ, Edelstein S, Abramson NW, Katz LE, Yasuda PM, Lavietes SJ, Trief PM, Tollefsen SE, McKay SV, Kringas P, Casey TL, Marcus MD. Depressive symptoms and quality of life in adolescents with type 2 diabetes: baseline data from the TODAY study. Diabetes Care 2011; 34(10): 2205–2207

[9] Mezuk B, Eaton WW, Albrecht S, Golden SH. Depression and type 2 diabetes over the lifespan: a meta-analysis. Diabetes Care 2008; 31(12): 2383–2390

[10] Golden SH, Lazo M, Carnethon M, Bertoni AG, Schreiner PJ, Diez Roux AV, Lee HB, Lyketsos C. Examining a bidirectional association between depressive symptoms and diabetes. JAMA 2008; 299(23): 2751–2759

[11] Nouwen A, Winkley K, Twisk J, Lloyd CE, Peyrot M, Ismail K, Pouwer F; European Depression in Diabetes (EDID) Research Consortium. Type 2 diabetes mellitus as a risk factor for the onset of depression: a systematic review and meta-analysis. Diabetologia 2010; 53(12): 2480–2486

[12] Rotella F, Mannucci E. Diabetes mellitus as a risk factor for depression. A meta-analysis of longitudinal studies. Diabetes Res Clin Pract 2013; 99(2): 98–104

[13] Sun JC, Xu M, Lu JL, Bi YF, Mu YM, Zhao JJ, Liu C, Chen LL, Shi LX, Li Q, Yang T, Yan L, Wan Q, Wu SL, Liu Y, Wang GX, Luo ZJ, Tang XL, Chen G, Huo YN, Gao ZN, Su Q, Ye Z, Wang YM, Qin GJ, Deng HC, Yu XF, Shen FX, Chen L, Zhao LB, Wang TG, Lai SH, Li DH, Wang WQ, Ning G; Risk Evaluation of Cancers in Chinese Diabetic Individuals: A Longitudinal (REACTION) Study group. Associations of depression with impaired glucose regulation, newly diagnosed diabetes and previously diagnosed diabetes in Chinese adults. Diabet Med 2015; 32(7): 935–943

[14] Harrison SC, Holmes MV, Humphries SE. Mendelian randomisation, lipids, and cardiovascular disease. Lancet 2012; 380(9841): 543–545

[15] Xu M, Huang Y, Xie L, Peng K, Ding L, Lin L, Wang P, Hao M, Chen Y, Sun Y, Qi L, Wang W, Ning G, Bi Y. Diabetes and risk of arterial stiffness: a Mendelian randomization analysis. Diabetes 2016; 65(6): 1731–1740

[16] Lawlor DA, Harbord RM, Sterne JA, Timpson N, Davey Smith G. Mendelian randomization: using genes as instruments for making causal inferences in epidemiology. Stat Med 2008; 27(8): 1133–1163

[17] Palmer TM, Lawlor DA, Harbord RM, Sheehan NA, Tobias JH, Timpson NJ, Davey Smith G, Sterne JA. Using multiple genetic variants as instrumental variables for modifiable risk factors. Stat Methods Med Res 2012; 21(3): 223–242

[18] Liu S, Song Y. Building genetic scores to predict risk of complex diseases in humans: is it possible? Diabetes 2010; 59(11): 2729–2731

[19] Bi Y, Lu J, Wang W, Mu Y, Zhao J, Liu C, Chen L, Shi L, Li Q, Wan Q, Wu S, Yang T, Yan L, Liu Y, Wang G, Luo Z, Tang X, Chen G, Huo Y, Gao Z, Su Q, Ye Z, Wang Y, Qin G, Deng H, Yu X, Shen F, Chen L, Zhao L, Zhang J, Sun J, Dai M, Xu M, Xu Y, Chen Y, Lai S, Bloomgarden ZT, Li D, Ning G. Cohort profile: risk evaluation of cancers in Chinese diabetic individuals: a longitudinal (REACTION) study. J Diabetes 2014; 6(2): 147–157

[20] Kroenke K, Spitzer RL, Williams JB. The PHQ-9: validity of a brief depression severity measure. J Gen Intern Med 2001; 16(9): 606–613

[21] Löwe B, Kroenke K, Herzog W, Gräfe K. Measuring depression outcome with a brief self-report instrument: sensitivity to change of the Patient Health Questionnaire (PHQ-9). J Affect Disord 2004; 81(1): 61–66

[22] Pignone MP, Gaynes BN, Rushton JL, Burchell CM, Orleans CT, Mulrow CD, Lohr KN. Screening for depression in adults: a summary of the evidence for the U.S. Preventive Services Task Force. Ann Intern Med 2002; 136(10): 765–776

[23] Matthews DR, Hosker JP, Rudenski AS, Naylor BA, Treacher DF, Turner RC. Homeostasis model assessment: insulin resistance and β-cell function from fasting plasma glucose and insulin concentrations in man. Diabetologia 1985; 28(7): 412–419

[24] Cersosimo E, Solis-Herrera C, Trautmann ME, Malloy J, Triplitt CL. Assessment of pancreatic β-cell function: review of methods and clinical applications. Curr Diabetes Rev 2014; 10(1): 2–42

[25] Hagströmer M, Oja P, Sjöström M. The International Physical Activity Questionnaire (IPAQ): a study of concurrent and construct validity. Public Health Nutr 2006; 9(6): 755–762

[26] Xu M, Bi Y, Huang Y, Xie L, Hao M, Zhao Z, Xu Y, Lu J, Chen Y, Sun Y, Qi L, Wang W, Ning G. Type 2 diabetes, diabetes genetic score and risk of decreased renal function and albuminuria: a Mendelian randomization study. EBioMedicine 2016; 6: 162–170

[27] Cho YS, Lee JY, Park KS, Nho CW. Genetics of type 2 diabetes in East Asian populations. Curr Diab Rep 2012; 12(6): 686–696

[28] Kato N. Insights into the genetic basis of type 2 diabetes. J Diabetes Investig 2013; 4(3): 233–244

[29] Cho YS, Chen CH, Hu C, Long J, Ong RT, Sim X, Takeuchi F, Wu Y, Go MJ, Yamauchi T, Chang YC, Kwak SH, Ma RC, Yamamoto K, Adair LS, Aung T, Cai Q, Chang LC, Chen YT, Gao Y, Hu FB, Kim HL, Kim S, Kim YJ, Lee JJ, Lee NR, Li Y, Liu JJ, Lu W, Nakamura J, Nakashima E, Ng DP, Tay WT, Tsai FJ, Wong TY, Yokota M, Zheng W, Zhang R, Wang C, So WY, Ohnaka K, Ikegami H, Hara K, Cho YM, Cho NH, Chang TJ, Bao Y, Hedman ÅK, Morris AP, McCarthy MI; DIAGRAM Consortium; MuTHER Consortium,Takayanagi R, Park KS, Jia W, Chuang LM, Chan JC, Maeda S, Kadowaki T, Lee JY, Wu JY, Teo YY, Tai ES, Shu XO, Mohlke KL, Kato N, Han BG, Seielstad M. Meta-analysis of genome-wide association studies identifies eight new loci for type 2 diabetes in east Asians. Nat Genet 2012; 44(1): 67–72

[30] Balding DJ. A tutorial on statistical methods for population association studies. Nat Rev Genet 2006; 7(10): 781–791

[31] 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): 512–525

[32] Samaan Z, Garasia S, Gerstein HC, Engert JC, Mohan V, Diaz R, Anand SS, Meyre D. Lack of association between type 2 diabetes and major depression: epidemiologic and genetic evidence in a multiethnic population. Transl Psychiatry 2015; 5(8): e618

[33] Clarke TK, Obsteter J, Hall LS, Hayward C, Thomson PA, Smith BH, Padmanabhan S, Hocking LJ, Deary IJ, Porteous DJ, McIntosh AM. Investigating shared aetiology between type 2 diabetes and major depressive disorder in a population based cohort. Am J Med Genet B Neuropsychiatr Genet 2017; 174(3): 227–234

[34] Wium-Andersen MK, Ørsted DD, Nordestgaard BG. Tobacco smoking is causally associated with antipsychotic medication use and schizophrenia, but not with antidepressant medication use or depression. Int J Epidemiol 2015; 44(2): 566–577

[35] Wium-Andersen MK, Ørsted DD, Tolstrup JS, Nordestgaard BG. Increased alcohol consumption as a cause of alcoholism, without similar evidence for depression: a Mendelian randomization study. Int J Epidemiol 2015; 44(2): 526–539

[36] Hartwig FP, Bowden J, Loret de Mola C, Tovo-Rodrigues L, Davey Smith G, Horta BL. Body mass index and psychiatric disorders: a Mendelian randomization study. Sci Rep 2016; 6(1): 32730

[37] Moulton CD, Pickup JC, Ismail K. The link between depression and diabetes: the search for shared mechanisms. Lancet Diabetes Endocrinol 2015; 3(6): 461–471

[38] U.S. Preventive Services Task Force. Screening for depression: recommendations and rationale. Ann Intern Med 2002; 136(10): 760–764

[39] Kroenke K, Spitzer RL, Williams JB, Löwe B. The Patient Health Questionnaire Somatic, Anxiety, and Depressive Symptom Scales: a systematic review. Gen Hosp Psychiatry 2010; 32(4): 345–359

RIGHTS & PERMISSIONS

Higher Education Press and Springer-Verlag GmbH Germany, part of Springer Nature