Melatonin receptor 1B gene rs10830963 polymorphism, depressive symptoms and glycaemic traits

Abstract

Background: The association between depression and type 2 diabetes is bidirectional. Underlying biological determinants remain elusive. We examined whether a common melatonin receptor 1B gene diabetes risk variant rs10830963 influenced the associations between depressive symptoms and glycaemic traits.

Materials: The Prevalence, Prediction and Prevention of Diabetes-Botnia Study participants (n = 4,455) with no diabetes who underwent an oral glucose tolerance test were genotyped for rs10830963 and completed the Mental Health Inventory on depressive symptoms.

Results: The rs10830963 did not influence significantly the associations between depressive symptoms and glycaemic traits. Yet, the addition of each copy of the minor G allele of the rs1080963 and higher depressive symptoms were both, independent of each other, associated significantly with higher glucose response (glucose area under the curve), higher insulin resistance (Insulin Sensitivity Index) and lower insulin secretion (Disposition Index). Depressive symptoms, but not rs1080963, were also significantly associated with higher fasting insulin, insulin area under the curve and insulin resistance (Homeostasis Model Assessment, Homeostasis Model Assessment-2); rs1080963, but not depressive symptoms, was significantly associated with higher fasting glucose and lower Corrected Insulin Response.

Conclusions: Our study shows that the diabetes risk variant rs10830963 does not contribute to the known comorbidity between depression and type 2 diabetes.

Key messages

• The association between depression and type 2 diabetes is bidirectional.

• We tested whether a common variant rs10830963 in the gene encoding Melatonin Receptor 1B influences the known association between depressive symptoms and glycaemic traits in a population-based sample from Western Finland.

• The MTNR1B genetic diabetes risk variant rs10830963 does not contribute to the known comorbidity between depression and type 2 diabetes.

• Depressive symptoms and rs10830963 are associated with a worse glycaemic profile independently of each other.

Keywords:

• Depression
• diabetes
• glycaemic traits
• insulin sensitivity and resistance
• melatonin
• MTNR1B
• psychological aspects

Introduction

Several studies have reported on a bidirectional association between depression and type 2 diabetes (T2D) [1,2]. Epidemiological studies have shown that nearly 31% of individuals with T2D also suffer from depression and its subclinical symptoms [2,3]. On the other hand, depression and its subclinical symptoms increase the risk of T2D [4,5] and are associated with insulin resistance and decreased insulin secretion in some [6–8] but not all studies [9]. While the mechanisms linking these two disorders remain poorly understood, they have been suggested to share common underlying biological determinants.

In addition to the cytokine-mediated inflammatory response and hypothalamic-pituitary-adrenocortical (HPA) axis dysregulation, circadian disruption has been implicated in both T2D [10,11] and depression [12]. Melatonin is a neurohormone with an important role in circadian regulation. It is secreted from the pineal gland after being formed from tryptophan via acetylation and subsequent methylation of the neurotransmitter serotonin. Melatonin secretion is minimal during the day but is dramatically increased during the night. Melatonin exerts its effects via two membrane receptor isoforms (MTNR1B and MTNR1A). These receptors are found in the brain, suprachiasmatic nucleus, retina, pancreatic β-cells and α-cells and numerous other human tissues [13,14].

When melatonin binds to its receptors, it activates Gi and Gq proteins which, in turn, inhibit cAMP and cGMP pathways that result in decreased insulin secretion [15]. Melatonin may also play a role in depression [16]. Some evidence suggests that bright light therapy in the morning, used in resetting the circadian melatonergic rhythm, may be beneficial in the treatment of seasonal and non-seasonal depression [17]. Recent studies have also shown that a novel antidepressant, agomelatine, which acts on the melatonergic (MTNR1B and MTNR1A agonists) and serotonergic (5HT_2B and 5HT_2C antagonist) receptors, may be useful in the treatment of depression [18].

Against this background, we tested whether a common variant rs10830963 in the gene encoding MTNR1B influences the association between depressive symptoms and glycaemic traits in a population-based sample from Western Finland. This variant is associated with increased diabetes risk, fasting glucose and impaired insulin secretion in genome-wide association studies [19,20] and with decreased insulin secretion in a smaller subsample of our population-based study with no T2D [21].

Materials

Study design and participants

The population-based PPP-Botnia has been described in detail elsewhere [22]. In brief, of 9,518 invited individuals aged 17 to 78 years residing in Western Finland, 5,208 (2,443 men and 2,765 women) (55.0%) participated [22]. Of the study participants, 5,205 (99.9%) underwent a 75-g 2-hr oral glucose tolerance test (OGTT) and 5,016 (96.3%) returned a psychological questionnaire including questions on depressive symptoms. Of them, a total of 4,724 (90.7%) individuals were genotyped for rs10830963, and 4,455 (85.5%) of them did not have T2D. They formed the analytic sample of our study.

All participants provided written informed consent. The study protocol was approved by the Ethics Committee of the Helsinki University Central Hospital.

Genotyping of MTNR1B variants

Genotyping was performed either by mass spectrometry or by allelic discrimination method, as described in detail elsewhere [23]. The frequency of the risk allele G was 31%, and genotype frequencies were in HWE.

Glycaemic traits

The subjects participated in an OGTT by ingesting 75 g of glucose after a 12-h overnight fast. During the OGTT, venous samples for glucose and insulin were drawn at 0, 30 and 120 min. Plasma glucose was measured with the glucose dehydrogenase method (HemoCue, Ängelholm, Sweden) and serum insulin by a fluoroimmunoassay (Delphia; Perkin-Elmer Finland, Turku, Finland).

Homeostatic model assessment for Insulin Resistance (HOMA-IR) [24], HOMA2-IR using online HOMA calculator (https://www.dtu.ox.ac.uk/homacalculator/) and Insulin Sensitivity Index (ISI) [25] were used as indices of insulin resistance [26]. Disposition Index (DI) [27] and Corrected Insulin Response (CIR) [26] were used as indices of insulin secretion. The following formulas were used to calculate these indices: HOMA-IR = Fasting insulin [mU/l]×fasting glucose [mmol/l]/22.5; ISI =10000/√ (fasting glucose [mmol/l]×fasting insulin [mU/l]) × (mean OGTT glucose [mmol/l]×mean OGTT insulin [mU/l]); CIR = (100 × insulin [mU/l] at 30 min)/((glucose [mmol/l] at 30 min) × (glucose [mmol/l] at 30 min–3.89 mmol/l)); DI = CIR × ISI. Area Under the Curve (AUC) for insulin and glucose were calculated as follows: AUC insulin =15 × fasting insulin [mU/l] + 15 × insulin [mU/l] at 30 min +45 × insulin [mU/l] at 30 min +45 × insulin [mU/l] at 120 min; AUC glucose = 15 × fasting glucose +15 × glucose [mmol/l] at 30 min +45 × glucose [mmol/l] at 30 min +45 × glucose [mmol/l] at 120 min.

Depressive symptoms

Depressive symptoms were recorded using the five-item Mental Health Inventory (MHI-5) derived from the SF-36 [28]. The MHI-5 questions (feeling nervous, feeling down in the dumps, feeling downhearted and blue, feeling calm and peaceful [reverse scored] and feeling happy [reverse scored]), are rated on a six-point scale ranging from all the time (1) to none of the time (6) during the past four weeks. A sumscore of these items is transformed into a scale that ranges from 0 to 100 [29]. To facilitate interpretation, we reverse scored the sumscore. Hence, a higher sumscore in our study reflects higher depressive symptoms. By using previously established cut-off of 60 for the non-reversed sumscore, we identified individuals with clinically relevant depressive symptoms [30].

Covariates

Covariates included sex, age (years), season of OGTT date (dark [November–March]/light [April–October]), body mass index (BMI; kg/m²; calculated from measured weight [kg] and height [m] during clinical examination), self-reported educational attainment (basic or less or missing, upper secondary school, lower tertiary school, upper tertiary school), reported current smoking status (yes/no or former), alcohol consumption (yes/no or former) and physical activity (yes or no based upon leisure-time activity; if participants performed more than 30-minute physical activity three or more times per week with an intensity resulting in breathlessness, sweating, or both, they were considered to perform regular exercise (yes). If participates performed less or performed no activity, they were considered to do little or no exercise (no)).

Statistical analysis

IBM SPSS version 23.0 was used for data analyses [31]. We used multiple linear regression analysis to test whether the MNTR1B rs10830963 genotype influenced the associations between depressive symptoms and glycaemic traits by including the main effects of rs10830963 and depressive symptoms and their interaction into the regression equation with indices of glycaemic traits as the outcomes. Before proceeding to the interaction tests, we used multiple linear regression analyses to also test whether the rs10830963 and depressive symptoms were associated with the glycaemic traits and whether these effects were independent of each other. In these analyses, we used depressive symptoms as both continuous and dichotomized at the clinical cut-off. All associations were tested in the presence of covariates. In our sample, the glycaemic traits under the four main categories (glucose and insulin concentrations, insulin resistance, insulin secretion) were significantly interrelated (Pearson’s r for fasting glucose and glucose AUC = 0.49, 95% CI 0.47–0.51; fasting insulin and AUC insulin = 0.50, 95% CI 0.47–0.52; HOMA-IR, HOMA2-IR and ISI=|0.53–0.99|, 95% CI |0.51–0.99|; DI and CIR = 0.76, 95% CI 0.75–0.77). However, to decrease the likelihood of type 1 error, we used the false discovery rate (FDR) [32] procedure to account for multiple testing setting the false discovery rate across 12 tests at two-tailed alpha level 0.05 (two main predictors, their interaction and four main outcome categories).

Results

The characteristics of the sample are presented in Table 1. Table 1 also compares the analytic sample of our study with the excluded sample. In comparison with the excluded sample, individuals in the analytic sample were more often women, younger, had lower BMI, more often had higher education, less often smoked, more often used alcohol and reported lower depressive symptoms. They also had lower glucose, insulin, HOMA-IR, HOMA2-IR and higher CIR values; the groups did not differ in 30-min insulin, ISI and DI (Table 1). The groups did not differ in physical activity, rs10830963 genotype frequencies and in season of the testing date (Table 1). In our analytic sample, rs10830963 was not significantly associated with depressive symptoms (Supplemental Figure 1).

Table 1. Sample characteristics.

Characteristic	Excluded sample Nmax = 753 N (%) or mean (SD; 95% CI)	Analytic sample Nmax = 4,455 N (%) or mean (SD; 95% CI)	P*
Body mass index (kg/m^2)	28.25 (5.04; 27.89–28.61)	26.22 (4.28; 26.09–26.35)	<.001
Sex (male/female)	389 (51.7%)/364 (48.3%)	2,054 (46.1.0%)/2,401 (53.9%)
Age (years)	57.57 (14.66; 56.53–58.62)	48.20 (15.45; 47.75–48.65)
Education
Basic or less or missing	350 (49.6%)	1,194 (27.5%)
Upper secondary	226 (32.1%)	1,730 (39.9%)
Lower tertiary	63 (8.9%)	708 (16.3%)
Upper tertiary	66 (9.4%)	704 (16.3%)
Smoking (yes/no or former)	257 (42.5%)/347 (57.5%)	1,392 (37.4%)/2,333 (62.6%)	.015
Alcohol consumption (yes/no of former)	535 (73.0%)/198 (27.0%)	3,696 (83.8%)/716 (16.2%)	<.001
Exercise (yes [≥3 times a week]/no [<3 times a week])	403 (56.0%)/317 (44.0%)	2,394 (54.8%)/1,978 (45.2%)	.544
Glucose (mmol/l)
Fasting	6.28 (1.88; 6.14–6.41)	5.27 (.55; 5.25–5.29)	<.001
30 min	9.64 (2.35; 9.46–9.84)	8.26 (1.61; 8.21–8.31)	<.001
120 min	7.74 (3.99; 7.43–8.04)	5.21 (1.56; 5.16–5.25)	<.001
area under the curve^a	1,015.13 (308.66; 991.61–1,038.64)	808.84 (141.20; 804.67–813.00)	<.001
Insulin (mU/l)
Fasting	9.34 (10.32; 8.59–10.08)	6.78 (7.11; 6.57–6.99)	<.001
30 min	57.00 (41.04; 53.84–60.16)	59.41 (38.12; 58.28–60.54)	.296
120 min	49.48 (54.49; 45.26–53.69)	32.90 (35.47; 31.85–33.96)	<.001
area under the curve^b	5,729.93 (4,199.34; 5,049.68–5,248.47)	5,149.07 (3,321.39; 5,049.68–5,248.47)	<.001
Insulin Resistance (IR)
Homeostasis Model Assessment-IR^c	2.83 (4.29; 2.52–3.14)	1.61 (1.77; 1.56–1.66)	<.001
Homeostasis Model Assessment-2-IR	1.36 (1.05; 1.27–1.44)	0.96 (0.60; .95–.98)	<.001
Insulin Sensitivity Index^d	7.58 (5.60; 7.14–8.02)	9.24 (5.63; 9.08–9.41)	.436
Insulin secretion
Disposition Index^e	1,104.13 (1,833.76; 959.90–1,248.36)	1,603.16 (1,692.28; 1,551.90–1,654.42)	.580
Corrected Insulin Response^f	143.26 (157.53; 131.03–155.50)	195.79 (180.66; 190.36–201.22)	.029
Depressive symptoms (0–100)^g
	21.55 (16.12; 20.22–22.89)	20.44 (14.41; 20.01–20.86)	<.001
Clinically relevant depressive symptoms (yes/no)	83 (14.8%)/478 (85.2%)	511 (11.5%)/3,944 (88.5%)	.022
Oral Glucose Tolerance Test conducted during the dark season (November–March)/light season	401 (53.3%)/351 (46.7%)	2,315 (52.0%)/2,137 (48.0%)	.512
MTNR1B rs10830963			.222
CC	214 (47.6%)	2,157 (48.4%)
CG	180 (40.0%)	1,858 (41.7%)
GG	56 (12.4%)	440 (9.9%)

^a AUC insulin (Area Under the Curve for Insulin) = 15 × fasting insulin [mU/l] + 15 × insulin [mU/l] at 30 min + 45 × insulin [mU/l] at 30 min +45 × insulin [mU/l] at 120 min.

^b AUC glucose (Area Under the Curve for Glucose) = 15 × fasting glucose + 15 × glucose [mmol/l] at 30 min + 45 × glucose [mmol/l] at 30 min + 45 × glucose [mmol/l] at 120 min.

^c HOMA-IR = Fasting insulin [mU/l]×fasting glucose [mmol/l]/22.5.

^d ISI =10000/√ (fasting glucose [mmol/l]×fasting insulin [mU/l]) × (mean OGTT glucose [mmol/l] × mean OGTT insulin [mU/l]).

^e DI = CIR × ISI.

^f CIR = (100 × insulin [mU/l] at 30 min)/((glucose [mmol/l] at 30 min) × (glucose [mmol/l] at 30 min–3.89 mmol/l).

^g Higher score indicates higher depressive symptoms.

* t-test was used to compare the means of continuous variables and chi-squared test was used to compare categorical variables.

Main and independent effects of rs10830963 and depressive symptoms on glycaemic traits

Across adjustment models including all covariates, the addition of each copy of the minor G allele of the rs10830963 was associated with higher fasting glucose, glucose response (AUC), higher insulin resistance (ISI) and lower insulin secretion (CIR and DI) (Table 2). The association with ISI did not survive the FDR correction for multiple testing. Figure 1 displays the associations surviving the FDR correction according to the genotype.

Figure 1. Means (circles), 95% confidence intervals (95% CI) of the means (error bars) and mean differences (MD) in glycaemic traits (panels A to D) according to the MTNR1B rs10830963 genotype (only glycaemic traits showing statistically significant associations surviving the false discovery rate correction for multiple testing in the genetic additive models are displayed).

Table 2. Associations between MTNR1B rs10830963 and glycaemic traits rs10830963 (CC/CG/GG).

Outcome (standard deviation units)	β (95% confidence interval)^a	R^2	P^1	P^2	P^3	P^4	P^5
Glucose
Fasting	0.174 (0.142; 0.206)	0.050	<.001	<.001	<.001	<.001	<.001
Area Under the Curve	0.162 (0.126; 0.198)	0.119	<.001	<.001	<.001	<.001	<.001
Insulin
Fasting	0.011 (–0.030; 0.051)	0.007	.603	.236	.740	.887	.850
Area Under the Curve	–0.020 (–0.065; 0.024)	0.015	.368	.480	.658	.724	.744
Insulin resistance (IR)
Homeostasis Model Assessment-IR	0.048 (0.007; 0.089)	0.012	.023	.002	.058	.097	.088
Homeostasis Model Assessment-2-IR	0.021 (–0.027; 0.068)	0.012	.390	.257	.711	.707	.676
Insulin Sensitivity Index	–0.047 (–0.090; –0.004)	0.027	.031	.006	.037	.048	.044
Insulin secretion
Disposition Index	–0.203 (–0.242; –0.164)	0.094	<.001	<.001	<.001	<.001	<.001
Corrected Insulin Response	–0.198 (–0.240; –0.156)	0.050	<.001	<.001	<.001	<.001	<.001

P¹ – linear regression model 1 is adjusted for sex and age.

P² – linear regression model 2 is adjusted for sex, age and body mass index.

P³ – linear regression model 3 is adjusted for sex, age, body mass index, education, current smoking status, alcohol consumption and physical activity.

P⁴ – linear regression model 4 is adjusted for sex, age, body mass index, education, current smoking status, alcohol consumption, physical activity and season.

P⁵ – linear regression model 5 is adjusted for sex, age, body mass index, education, current smoking status, alcohol consumption, physical activity, season and depressive symptoms.

^a – β coefficients and R² are from the model 1 and show change in standard deviation units in outcome per each addition of rare allele (G) of the MTNR1B rs10830963 variant.

Also, across adjustment models for all covariates, higher depressive symptoms were associated with higher glucose response (AUC), higher fasting insulin and insulin response (AUC), higher insulin resistance (HOMA-IR, HOMA2-IR and ISI) and lower insulin secretion (DI) (Table 3). The association with DI did not survive the FDR correction for multiple testing. Figure 2 displays the associations surviving the FDR correction when depressive symptoms were categorized at the clinical cut-off.

Figure 2. Means (circles), 95% confidence intervals (95% CI) of the means (error bars) and mean differences (MD) in glycaemic traits (panels A to F) according to the clinically relevant depressive symptoms cut-off (60 of the non-reverse scored sumscore) (only glycaemic traits showing statistically significant associations surviving the false discovery rate correction for multiple testing are displayed).

Table 3. Depressive symptoms main and independent effect on glycaemic traits depressive symptoms (standard deviation units).

Outcome (standard deviation units)	β (95% confidence interval)^a	R^2	P^1	P^2	P^3	P^4	P^5
Glucose
Fasting	0.004 (–0.017; 0.025)	0.025	.720	.870	.720	.697	.599
Area Under the Curve	0.033 (0.009; 0.057)	0.104	.007	.020	.013	.012	.008
Insulin
Fasting	0.051 (0.024; 0.078)	0.009	<.001	.001	.002	.002	.002
Area Under the Curve	0.046 (0.017; 0.076)	0.017	.002	.012	.004	.005	.005
Insulin resistance (IR)
Homeostasis Model Assessment-IR	0.051 (0.023; 0.078)	0.013	<.001	.002	.003	.003	.003
Homeostasis Model Assessment-2-IR	0.050 (0.019; 0.081)	0.014	.002	.012	.005	.005	.005
Insulin Sensitivity Index	–0.058 (–0.087; –0.030)	0.029	<.001	.001	<.001	<.001	<.001
Insulin secretion
Disposition Index	–0.033 (–0.059; –0.007)	0.072	.013	.037	.033	.034	.026
Corrected Insulin Response	0.012 (–0.016; 0.040)	0.031	.395	.610	.594	.600	.645

P¹ – linear regression model 1 is adjusted for sex and age.

P² – linear regression model 2 is adjusted for sex, age and body mass index.

P³ – linear regression model 3 is adjusted for sex, age, body mass index, education, current smoking status, alcohol consumption and physical activity.

P⁴ – linear regression model 4 is adjusted for sex, age, body mass index, education, current smoking status, alcohol consumption, physical activity and season.

P⁵ – linear regression model 5 is adjusted for sex, age, body mass index, education, current smoking status, alcohol consumption, physical activity, season and MTNR1B rs10830963.

^a – β coefficients and R² are from model 1 and show change in the outcome in standard deviation units per standard deviation unit change in depressive symptoms.

Tables 2 and 3 (p⁵ in furthest right columns) show that the effects of rs10830963 and depressive symptoms on glycaemic traits were independent of each other.

The effect of rs10830963 on the associations between depressive symptoms and glycaemic traits

The interactions between rs10830963 and depressive symptoms were not significantly associated with glycaemic traits (Supplemental Table 1).

Discussion

We show here that against to what we expected the common MTNR1B diabetes risk variant (rs10830963) does not influence the associations between depressive symptoms and glycaemic traits. Instead, the interesting finding of our study relates to the main effects of the rs10830963 MTNR1B and depressive symptoms on glycaemic traits. Consistent with previous reports [19–21,33], the addition of each copy of the rare G allele of the rs10830963, which confers risk for T2D, was associated with poorer glycaemic profile as indicated by higher fasting glucose concentrations and glucose response to OGTT, higher insulin resistance and lower insulin secretion [23]. Also in consistent with the previous reports [1], higher depressive symptoms were associated with poorer glycaemic profile as indicated by higher glucose response to OGTT, higher fasting insulin and insulin response to OGTT, higher insulin resistance and lower insulin secretion. The association of rs10830963 with insulin resistance measured as ISI and the association of depressive symptoms with insulin secretion measured as DI did not, however, survive the FDR correction for multiple testing. The associations of MTNR1B rs10830963 and depressive symptoms with glycaemic traits were independent of each other, which is not as surprising as the rs10830963 and depressive symptoms were unrelated to each other in our study. The mean differences in the glycaemic traits between the risk allele carriers and non-carriers of the rs10830963 and between those with and without clinically relevant depressive symptoms were mainly small or at the lower medium level in effect size, however. These findings thus show that the common MTNR1B genetic T2D risk variant does not contribute to the common comorbidity between depressive symptoms and T2D, but show that they exert effects on glycaemic traits independently of each other and independently of the covariates associated with T2D and/or depressive symptoms, namely sex, age, season, BMI, education and lifestyle.

While these findings do not provide support for a role of the MTNR1B diabetes risk variant in contributing to the comorbidity of T2D and depression [7], molecular studies have shown increased expression of MTNR1B in β-cells in pancreatic islets of non-diabetic and diabetic individuals carrying the MTNR1B risk variant [21]. It has been suggested that the pathogenic effects are likely mediated through inhibitory effects on β-cell function, which results in decreased insulin secretion [15], increased glucose levels and worsening of insulin resistance [23]. We have recently demonstrated that in the MTNR1B risk genotype carriers, the administration of 4 mg/night of melatonin for three months was associated with impaired insulin secretion [34].

Additionally, bright light therapy [17] and agomelatine [18], which target resetting of the circadian melatonin rhythm, have shown benefits in the treatment of depression. However, it remains unknown whether these therapies also have beneficial effects on glycaemic traits and whether any of these effects are modulated by the MTNR1B T2D risk genotype.

A limitation of the study is that we did not measure melatonin or its circadian variation. A recent study has shown that in MTNR1B risk variant carriers, the duration of melatonin production is prolonged into the morning; risk variant carriers have later melatonin offset and a longer duration of elevated melatonin levels [35]. Melatonin is also known to have pleiotropic effects, including the effects on inflammation [14]. However, it is unknown whether these physiological effects play a role in the associations between MTNR1B or melatonin, or depressive symptoms and glycaemic traits.

We have measures of sleep quantity, quality or timing only for a small subset [36] of current study cohort. While one recent study showed that an earlier, but not later, sleep timing in the risk genotype carriers of the MTNR1B was associated with increased risk of diabetes [35], other studies testing interactions between MTNR1B variants and sleep quantity or quality on glycaemic traits have been mostly negative [37–40].

While the variant rs10830963 in the MTNR1B is also common in other ethnic groups, these associations may be exaggerated in Finns living at latitudes with large seasonal variations in light and darkness. In a recent study on adults of European ancestry living in Northern Sweden, the effect of circadian rhythm-regulating loci (CRY1, CRY2 and MTNR1B) on glucose homeostasis was modified by seasonal variations. Unexpectedly, these findings showed that rs10830963 G allele was associated with lower 2 h glucose during dark season but not at the light season at baseline visit and during both dark and light seasons at a follow-up visit a decade later [41]. This study did not analyse interactions with depression so it remains unknown whether depressive symptoms could have influenced the results.

In conclusion, our study did not unravel the biological underpinnings of the common comorbidity between depressive symptoms and T2D as far as the MTNR1B rs10830963 genetic T2D risk variant is concerned. Future studies in larger sample sizes than ours are needed to either confirm or refute the lack of interaction of rs10830963 and depressive symptoms in the analysis of glycaemic traits. Our study encourages further studies focusing on other genetic variants, regardless of the null genetic correlation between depression and T2D, as our recent study demonstrated that a number of pleiotropic SNPs may underpin depression and T2D [42].

Acknowledgements

The skilful assistance of the Botnia Study Group is gratefully acknowledged.

Disclosure statement

The authors report no biomedical financial interests or potential conflicts of interest.

Additional information

Funding

This work was supported by the Academy of Finland, the ERC Advanced Scientist Grant, the Sigrid Jusélius Foundation and the Doctoral Programme of Psychology, Learning and Communication (University of Helsinki). The PPP-Botnia study has been financially supported by grants from the Sigrid Jusélius Foundation, the Folkhälsan Research Foundation, the Nordic Center of Excellence in Disease Genetics, EU (EXGENESIS, EUFP7-MOSAIC), the Signe and Ane Gyllenberg Foundation, the Swedish Cultural Foundation in Finland, the Finnish Diabetes Research Foundation, the Foundation for Life and Health in Finland, the Finnish Medical Society, the Paavo Nurmi Foundation, the Helsinki University Central Hospital Research Foundation, the Perklén Foundation, the Ollqvist Foundation, the Närpes Health Care Foundation and the Ahokas Foundation. The study has also been supported by the Ministry of Education in Finland, Municipal Heath Care Center and Hospital in Jakobstad and Health Care Centers in Vaasa, Närpes and Korsholm.