Introduction

Hypertension and type 2 diabetes are strongly interrelated and might share common environmental risk factors such as sedentary behaviours, visceral adiposity and insulin resistance [1]. Elevated BP confers an increased risk for incident type 2 diabetes in the general population [2] and hypertension treatments might also modulate the risk of diabetes [3]. Fasting blood glucose, hyperinsulinaemia and HbA1c have been associated with the development of hypertension in non-diabetic individuals [4–8]. However, the pathophysiology underlying the association between elevated glycaemia and incident hypertension has not been well characterised. In particular, the impact on incident hypertension of a parental history of diabetes or genetic predisposition to type 2 diabetes, which is mainly associated with beta cell dysfunction [9], is not known. Furthermore, although the epidemiological link between insulin resistance and hypertension is recognised, the relationship between insulin secretion and the incidence of hypertension and elevated BP has not been investigated in the general population.

The aim of our study was to assess, in the prospective Data from an Epidemiological Study on the Insulin Resistance Syndrome (DESIR) cohort, the impact of a genetic predisposition to type 2 diabetes on the development of hypertension. We analysed the impact of the TCF7L2 rs7903146 polymorphism on incident hypertension, because this polymorphism has the largest effect on susceptibility to diabetes among the predisposing genes discovered to date [10]. As we observed an association of incident hypertension with the TCF7L2 at-risk allele, we tested whether defects in insulin secretion predispose to 3 year incident hypertension in 1,047 non-hypertensive, non-diabetic participants from the Relationship between Insulin Sensitivity and Cardiovascular Disease (RISC) cohort. All of the RISC participants had an accurate evaluation at baseline of both insulin sensitivity by the hyperinsulinaemic–euglycaemic clamp and insulin secretion from an extended OGTT [11].

Methods

The DESIR cohort

We studied men and women aged 30–65 years, who participated in the 9 year follow-up study, DESIR. Participants were recruited from volunteers who were offered periodic health examinations, free of charge, by French social security in ten health examinations centres in western France [12, 13]. All participants signed an informed consent form and the protocol was approved by an ethics committee. Cases of hypertension were defined by treatment for hypertension or a resting systolic BP ≥140 mmHg or diastolic BP ≥90 mmHg at one of the four 3-yearly examinations. After the exclusion of individuals with hypertension at baseline, we studied 2,391 participants with genotype data for TCF7L2 rs7903146 (745 had incident hypertension during the follow-up).

Clinical assessment

Two measures of BP, using a mercury sphygmomanometer, were taken in a supine position after 5 min rest; mean values were used. Weight and height were measured in lightly clad participants, and BMI was calculated. The examining physician noted the parental history of diabetes in a clinical questionnaire; treatment for diabetes and hypertension were recorded at each of the 3-yearly examinations. Smoking habits and alcohol intake were recorded in a patient-completed questionnaire

Biochemical measurements

All biochemical measurements were from one of four health centre laboratories located in France at Blois, Chartres, La Riche and Orléans. The interlaboratory variability for normal and pathological values was assessed monthly. Fasting plasma glucose, measured by the glucose-oxidase method, was applied to fluoro-oxalated plasma using a Technicon RA100 analyser (Bayer Diagnostics, Puteaux, France) or a Specific or Delta device (Konelab, Evry, France). HbA1c was determined by HPLC (L9100 ion-exchange analyser; Hitachi/Merck-VWR, Fontenay-sous-Bois, France) or immunoassay (DCA 2000; Bayer Diagnostics). Insulin was quantified by microparticle enzyme immunoassay with an automated analyser (IMX; Abbott, Rungis, France). Glucose, HbA1c and insulin have been standardised across laboratories and over the years of the study. Indices of insulin resistance (HOMA-IR) [14] and beta cell secretion (HOMA2%B) were computed using software downloaded from www.dtu.ox.ac.uk [15].

Genotyping

TCF7L2 single nucleotide polymorphism rs7903146 genotyping was performed with the SNPlex Genotyping System (Applied Biosystems, Foster City, CA, USA) based on the oligonucleotide ligation assay combined with multiplex PCR target amplification [12].

The RISC cohort

RISC is a prospective observational cohort study whose rationale and methodology, as well as the characteristics of the individuals recruited, have been previously published [11, 16]. Clinically healthy men and women, aged 30–60 years, were recruited from the local populations of 19 centres in 14 European countries. Initial exclusion criteria were: treatment for obesity, hypertension, lipid disorders or diabetes, pregnancy, cardiovascular or chronic lung disease, weight change ≥5 kg in the last 6 months, cancer (in the last 5 years) and renal failure. Exclusion criteria after screening were: arterial BP ≥140/90 mmHg, fasting plasma glucose ≥7.0 mmol/l, 2 h plasma glucose (following a 75 g OGTT) ≥11.0 mmol/l, total serum cholesterol ≥7.8 mmol/l, serum triacylglycerol ≥4.6 mmol/l and ECG abnormalities. In the current investigation, we studied 1,047 healthy individuals (579 women and 468 men) who had an evaluation of both insulin sensitivity and insulin secretion at baseline and who had complete data at the 3 year follow-up. Volunteers were given detailed written information on the study as well as an oral explanation, and all signed consent forms. The protocol was approved by the ethics committee of each recruiting institution.

Height and body weight were measured and BMI was calculated. Alcohol and tobacco consumption were assessed using a standardised semi-quantitative questionnaire [17]. Information on physical activity was collected with the 7 day International Physical Activity Questionnaire, a previously validated assessment tool for international studies that provides a comprehensive evaluation of daily physical activity habits [18].

Blood pressure

BP was measured in triplicate by trained study nurses using an OMRON 705CP BP monitor (Omron Healthcare GmbH, Hamburg, Germany) after 5 min of rest with participants sitting, according to a standard protocol; the median of these readings was used in this analysis for both baseline and follow-up examinations. Hypertension was defined as median systolic BP ≥140 mmHg, median diastolic BP ≥90 mmHg or treatment for hypertension in routine care at follow-up.

OGTT

Blood samples were taken at fasting and 30, 60, 90 and 120 min into the OGTT, for central analysis of routine blood chemistry. Blood collected during the studies was separated into plasma and serum, aliquoted and stored at –20°C for central assays of glucose and insulin.

Glucose concentrations were measured with the glucose-oxidase technique. Plasma insulin and C-peptide were measured by a two-site time-resolved fluoroimmunoassay (AutoDELFIA insulin kit; Wallac Oy, Turku, Finland) using monoclonal antibodies, with the following assay characteristics (for insulin and C-peptide, respectively): sensitivity 1–2 and 5 pmol/l; within-assay variation 5% and 5%; and between-assay variation 5% and 3.5%.

Insulin sensitivity

On a separate day within 1 month of the OGTT, participants underwent a hyperinsulinaemic–euglycaemic clamp. Exogenous insulin was administered as a primed continuous infusion at a rate of 240 pmol m–2 min–1 with a variable 20% dextrose infusion adjusted every 5–10 min to maintain the plasma glucose level within 0.8 mmol/l (±15%) of the target glucose level (4.5–5.5 mmol/l). The clamp procedure was standardised across centres [11]; the data from each clamp study were transferred to the coordinating centre, where they underwent quality-control scrutiny according to predefined criteria.

Insulin sensitivity is expressed as the ratio of the M value during the final 40 min of the 2 h clamp to the mean plasma insulin concentration measured during the same interval (M/I), normalised to fat-free mass and expressed in units of μmol min–1 (kg fat-free mass)–1 (nmol/l)–1.

Insulin secretion

Beta cell function was assessed from the OGTT using a model describing the relationship between insulin secretion (calculated from C-peptide with the method of van Cauter et al [19]) and glucose concentration, previously described in detail [20, 21]. From the model-estimated beta cell dose–response, relating insulin secretion (in pmol m−2 min−1) to glucose concentration, insulin secretion at 5 mmol/l glucose (the average basal glucose in participants with normal glucose tolerance) was estimated. This variable represents insulin secretion in basal conditions, if basal glucose were 5 mmol/l in each participant. Total insulin secretion was also determined using the model (integral during the OGTT, in nmol m–2) [21]. We considered, for all statistical analyses, that the product of either total insulin secretion or insulin secretion at 5 mmol/l glucose with the M/I value from the hyperinsulinaemic–euglycaemic clamp to express the rate of insulin secretion, in relation to the concomitant degree of insulin resistance.

Statistical analysis

Data are expressed as mean ± SD or median (interquartile range) for variables with a skewed distribution, while categorical data are presented as percentages. Variables that were not symmetrically distributed were log10 transformed before analysis. Baseline characteristics, means and percentages, were compared using Student’s t and χ 2 tests, respectively, according to incident hypertension.

For the DESIR cohort, as there was no significant interaction between sex and either parental history of diabetes or TCF7L2 rs7903146 polymorphism on the risk of hypertension, we analysed men and women together. The relationships between both parental diabetes and TCF7L2 rs7903146 genotype with incident hypertension were assessed by logistic regression analysis, with adjustment for sex, age and waist circumference. Further adjustments for fasting insulinaemia, fasting glycaemia and mean BP were made. We also tested for an interaction between HbA1c >5.7% and TCF7L2 variants on incident hypertension.

In the RISC cohort, BP levels at year 3 were compared according to quartiles of insulin secretion rate at baseline by ANOVA; we excluded individuals treated with antihypertensive medications at year 3 for this analysis.

A logistic regression analysis was used to test the association between the insulin secretion rate at baseline, assessed both as a continuous variable and stratified into quartiles, and incident hypertension at year 3 with adjustment for age, sex, recruitment centre, physical activity, waist circumference, fasting glycaemia and systolic and diastolic BP levels at baseline.

Statistical analyses used StatView (version 5.0, SAS Institute, Cary, NC, USA) and SAS version 9.2.

Results

The DESIR cohort

In the DESIR cohort, those who developed incident hypertension had higher blood pressure levels, fasting glucose, HbA1c, BMI, waist circumference and insulin concentrations at baseline, as compared with those who remained normotensive (Table 1).

Table 1 Baseline characteristics of participants in the DESIR cohort according to incident hypertension (HTA) over the 9 year follow-up
Full size table

Fasting glycaemia but not HbA1c levels at baseline remained associated with incident hypertension in a logistic regression analysis, after adjustment for sex and baseline age, waist circumference and mean BP. Neither the HOMA-IR nor the HOMA2%B indices were significantly related to the risk of incident hypertension in this multivariable model.

Parental history of diabetes

A parental history of diabetes was associated with a higher incidence of hypertension at 9 years (36% vs 30%, p = 0.02). This association persisted after controlling for potential confounders such as waist circumference and BMI (Table 2). Further adjustment for baseline fasting glycaemia, fasting insulinaemia, HOMA-IR and HOMA2%B did not alter the significant relationship (Table 2).

Table 2 Association between parental history of diabetes and incident hypertension in the DESIR cohort
Full size table

TCF7L2 rs7903146 polymorphism

The TCF7L2 rs7903146 variant conformed to Hardy–Weinberg equilibrium. BP levels at baseline did not differ according to the TCF7L2 rs7903146 genotype. However, TCF7L2 at-risk variants (CT+TT) were significantly associated with both higher systolic and diastolic BP at year 9 in univariate analysis (CT+TT vs CC genotype: 137 ± 19 vs 135 ± 19 mmHg, respectively, p = 0.02 for systolic BP; 81 ± 10 vs 80 ± 10 mmHg, respectively, p = 0.01 for diastolic BP).

The presence of the T allele was associated with an increased risk of incident hypertension at year 9, after controlling for sex and age, and the association persisted after further adjustment for waist circumference, BMI, mean BP, fasting glycaemia, insulinaemia, HOMA-IR and HOMA2%B at baseline (Table 3).

Table 3 Association between TCF7L2 at-risk variants and incident hypertension (CT+TT vs CC as reference) in the DESIR cohort
Full size table

We further assessed whether baseline BMI or glycaemic status modified the effect of the TCF7L2 at-risk allele on the risk of incident hypertension. There was no significant interaction between BMI (≥ or <27 kg/m2) and TCF7L2 on incident hypertension (p interaction = 0.25). However, there was a significant interaction between baseline HbA1c above the median value (≥5.7%) and the effect of TCF7L2 rs7903146 polymorphism on incident hypertension (p interaction = 0.03). After controlling for sex, age and waist circumference, the risk of hypertension was greater for individuals who carried the T allele and had HbA1c ≥5.7% at baseline (OR 1.82; 95% CI 1.21, 2.74; p = 0.004) than for those with the T allele but HbA1c <5.7% (OR 1.11; 95% CI 0.90, 1.37; p = 0.3).

The RISC cohort

Baseline glycaemia and hypertension at year 3

In univariate analysis, both fasting and 2 h glycaemia at baseline were significantly associated with incident hypertension at year 3 (Table 4). However, in a multivariable analysis, the association between both fasting glycaemia (OR 1.34; 95% CI 0.96, 1.86; p = 0.09] and 2 h glycaemia (OR 1.03; 95% CI 0.92, 1.16; p = 0.63) with the risk of hypertension at year 3 was no longer significant after controlling for sex, age, recruitment centre and waist circumference.

Table 4 Baseline characteristics in the RISC cohort according to incident hypertension (HTA) over the 3 year follow-up
Full size table

Baseline insulin secretion and hypertension at year 3

Individuals who developed hypertension at year 3 had, at baseline, reduced total insulin secretion during OGTT and also reduced basal beta cell function at 5 mmol/l glucose, as compared with those who remained normotensive (Table 4). This suggests that individuals with subtle alterations in insulin secretion, after accounting for the concomitant degree of insulin resistance, were at higher risk of developing hypertension at follow-up.

A lower insulin secretion rate at baseline (first vs fourth quartile of total insulin secretion during OGTT × M/I) was associated with higher BP at year 3 (systolic BP 123 ± 14 vs 118 ± 15 mmHg, p = 0.0001; diastolic BP 78 ± 9 vs 74 ± 9 mmHg, p < 0.0001). A similar inverse association was observed for basal insulin secretion at a fixed normal glucose level of 5 mmol/l from the beta cell dose–response (first vs fourth quartile: systolic BP 125 ± 13 vs 116 ± 15 mmHg, p < 0.0001; diastolic BP: 79 ± 8 vs 73 ± 9 mmHg, p < 0.0001).

The incidence of hypertension at year 3 decreased progressively across the quartiles of both total insulin secretion and insulin secretion at 5 mmol/l at baseline (Fig. 1). Individuals with the lowest total insulin secretion rate during OGTT (quartile 1) had a higher incidence of hypertension as compared with those with the highest insulin secretion (quartile 4), after adjusting for other baseline risk factors (centre, age, sex, physical activity, waist circumference, smoking, alcohol intake, fasting glucose, systolic and diastolic BPs) (OR 2.02; 95% CI 1.06, 3.86; p = 0.03).

Fig. 1
figure 1

Incidence of hypertension at year 3 according to the quartiles of (a) total insulin secretion rate × M/I during the OGTT (p for trend = 0.02) and (b) basal insulin secretion rate at 5 mmol glucose × M/I at inclusion in the RISC cohort (p for trend = 0.003). Error bars represent SEM

Full size image

Similarly, the inverse association between total insulin secretion rate during the OGTT, expressed as a continuous value, and the risk of incident hypertension at 3 years remained significant in a multivariable model after controlling for centre, age, sex, physical activity, waist circumference, smoking, alcohol intake, and systolic and diastolic BP levels at inclusion (OR per 1 SD log total insulin secretion 0.75; 95% CI 0.61, 0.93; p = 0.007). The addition of fasting glucose and 2 h glucose levels into the model did not alter the significant relationship (OR 0.75; 95% CI 0.61, 0.92; p = 0.007 and OR 0.75; 95% CI 0.61, 0.93; p = 0.007, respectively).

Discussion

The main finding of this study is that a genetic predisposition to type 2 diabetes, and in particular the TCF7L2 rs7903146 at-risk allele, was significantly associated with the incidence of hypertension after a 9 year follow-up in the general population; TCF7L2 rs7903146 polymorphism conferred an increased risk of type 2 diabetes through alterations in insulin secretion capacity. Furthermore, we provide evidence, for the first time, that a lower insulin secretion rate during OGTT is a risk factor for the development of elevated BP in a non-diabetic cohort, even after taking into account insulin resistance, as assessed by the gold-standard method of the hyperinsulinaemic–euglycaemic clamp.

Previous evidence has shown that the offspring of individuals with type 2 diabetes have an increased risk of developing type 2 diabetes, with defects in early beta cell secretion [22–25]. It has also been suggested that these offspring have early autonomic dysfunction, even in the absence of glucose intolerance [26], and more often display metabolic abnormalities with high insulin levels and arterial hypertension [27, 28]. In a previous community-based cohort, parental diabetes appeared to be an independent predictor of longitudinal changes in both systolic and diastolic BPs in the offspring, regardless of race and sex [29]. Taken together, these observations and our results suggest that heritable factors related to type 2 diabetes increase the risk of various disorders, including hypertension.

Genetic susceptibility to type 2 diabetes seems to be more related to early beta cell dysfunction rather than to insulin resistance [30, 31]. We selected the TCF7L2 rs7903146 polymorphism because this variant has the largest effect on type 2 diabetes susceptibility among the predisposing genes discovered to date [10, 32]. To our knowledge, this is the first report to study the longitudinal relationship of the TCF7L2 genotype on incident hypertension in the general population. It has been demonstrated that the at-risk TCF7L2 rs7903146 genotype is associated with impaired insulin secretion [33] and a reduced sensitivity of the beta cell to incretins [34]. The fact that the effect of the TCF7L2 rs7903146 polymorphism on BP was independent of waist circumference, glycaemia or the insulin-resistance HOMA2-IR index underscores the potential role of insulin secretion defects on the evolution of BP over time. This is highlighted by our data on the relationship between both total and basal insulin secretion and 3 year incident hypertension in the RISC cohort. However, we suggest that the association of the TCF7L2 genotype with incident hypertension might be related, at least in part, to reduced glucagon-like peptide-1 secretion in these individuals, as recent evidence has suggested that incretins might modulate BP [35]. The TCF7L2 rs7903146 variant might then primarily affect incretin levels and, as a consequence, have independent effects on BP and on insulin secretion.

Our longitudinal results are in favour of a positive interaction between subtle elevations in glycaemia and genetic predisposition leading to defects in insulin secretion in the development of hypertension. Our results are consistent with recent genetic data showing an association between the TCF7L2 rs7903146 variant and increased systolic BP in an endogamous ethnic group from India [36]. However, it should be noted that a recent large genome-wide association study (GWAS) did not report a significant association between the TCF7L2 variant rs7903146 and BP, although a trend was observed for systolic BP (p = 0.06) and pulse pressure (p = 0.04) [37]. A regional plot around TCF7L2 locus from the data publicly available of the ICBP GWAS study suggest that a signal near rs7903146 may modulate BP, although this variant does not appear to be a glycaemic signal (data not shown). The differences between our present study and the GWAS report might be related to the longitudinal follow-up used in our study, as opposed to the case–control design used in the GWAS. Indeed, we did not observe a significant association between the TCF7L2 rs7903146 variant and the presence of hypertension in a cross-sectional analysis of the entire DESIR cohort at baseline (data not shown). Furthermore, in our study, we found that a large part of the association of the TCF7L2 variant with hypertension was observed in people who already had baseline HbA1c above the median value (≥5.7%), which might explain why the TCF7L2 locus did not appear as being associated with hypertension in large cross-sectional GWAS that were not stratified on glucose. Our results need to be confirmed in other large prospective cohorts with long follow-ups.

The causal role of insulin in the development of hypertension has been much debated. It has been proposed that insulin resistance and/or hyperinsulinaemia promote the development of elevated BP over time [5, 6, 38, 39] with a positive association between insulin levels and incident hypertension [39, 40]. In these studies, the insulin secretion rate was not specifically assessed. Our findings confirm an association between fasting insulinaemia and BP but suggest a greater role for reduced insulin secretion in the pathogenesis of hypertension. Fasting insulin is a marker of insulin resistance which could be confounded by the presence of subtle defects in insulin secretion.

To our knowledge, the effects of insulin secretion on BP evolution have not been previously investigated, in contrast to the role of insulin resistance. In the present study, we found that a lower total insulin secretion rate was significantly associated with the incidence of hypertension in people without diabetes, and that this effect was independent of insulin sensitivity. This observation was confirmed when we assessed basal insulin secretion at a fixed normal glucose level of 5 mmol/l glucose, from the beta cell dose–response. Interestingly, in the present study, the inverse association between insulin secretion and incident hypertension was not related to weight gain over the follow-up (data not shown) and was also independent of insulin resistance, as assessed by the gold-standard method of the hyperinsulinaemic–euglycaemic clamp, suggesting alternative mechanisms.

A potential mechanism that might explain the increased hypertension risk with a reduced insulin secretion rate is the protective action exerted by insulin on the blood vessel. Haemodynamic actions of insulin have been suggested [41] and a small BP-lowering effect of insulin has been described in non-diabetic individuals [42]. Furthermore, a partial or complete deficiency in the insulin receptor and/or insulin-receptor substrate-1 in endothelial cells is associated with endothelial dysfunction and increased BP in mice [43, 44]. Together, these findings suggest that reduced activation of the insulin pathway through diminished insulin secretion capacity, particularly in the presence of insulin resistance, might lead to endothelial dysfunction that favours the development of hypertension. Hyperinsulinaemia has to be interpreted as a compensation for enhanced insulin resistance. If this compensation is insufficient, the net effect at the cellular level, including probably the endothelium, is a down-tuned insulin signal.

Furthermore, recent evidence suggests that improving insulin secretion via the modulation of incretin concentrations can lower BP, even in non-diabetic hypertensive individuals [45, 46].

In parallel, subtle defects in beta cell function might facilitate the development of hypertension through possible repeated postprandial elevations in glucose levels over time. Glycaemic variability might contribute to increased BP through chronic induction of inflammation and enhanced oxidative stress [47, 48]. Oxidative stress has been shown to correlate with both glycaemic fluctuations and BP variability, suggesting possible alternative mechanisms to explain the increased incidence of hypertension in relationship to subtle defects in beta cell function [49]. In addition, the STOP-NIDDM trial found that acarbose treatment, which reduces postprandial hyperglycaemia, was associated with a significant reduction in the development of incident hypertension in those with impaired glucose tolerance, suggesting an impact of postprandial excursions on BP [50]. This point needs to be specifically investigated in another prospective study with continuous ambulatory assessment of glucose values.

Limitations of the present study include the absence of accurate measurements of both insulin resistance and beta cell function over time in the DESIR cohort, which precludes us from demonstrating that the association between parental history, the TCF7L2 genotype and incident hypertension is directly related to defects in beta cell function. It is possible that the association observed between parental history and hypertension might be related, at least in part, to enhanced insulin resistance in the parents. The number of individuals with the at-risk TCF7L2 rs7903146 genotype was limited in comparison with large genetic consortium studies and our results, albeit significant, need to be replicated in other populations. Furthermore, the absence of postprandial glucose levels did not allow us to investigate the exact role of this genotype in the development of elevated BP. The strengths of the present study are the use of two complementary cohorts specifically dedicated to the study of glucose metabolism, with a large number of normotensive participants at inclusion, aligned laboratory assays and continuous quality control of data. The RISC cohort is the largest cohort available with both systematic evaluation of insulin secretion and gold-standard measurements of insulin sensitivity by the glucose clamp technique, which provided the opportunity to clarify the role of alterations of insulin secretion in hypertension. For this important question to be addressed, a large number of individuals with high-quality measurements of glucose metabolism are required.

In conclusion, our study shows that a parental history of type 2 diabetes and at-risk variants in the TCF7L2 rs7903146 genotype predispose to incident hypertension. The novel finding of the association between reduced insulin secretion and the development of elevated BP deserves further investigation through the study of the impact of treatments targeting insulin secretion and postprandial hyperglycaemia on BP. Our results suggest that defects in insulin secretion might be a new independent risk factor for the development of hypertension.