Disclosure: The authors declared no conflict of interest.
Funding agencies: E. C. Eringa is supported by the Netherlands Organisation for Scientific Research (grant 916.76.179). E.H. Serné is supported by a fellowship from The Netherlands Heart Foundation (grant no. 2010T041).
The mechanisms underlying obesity-related hypertension are incompletely understood. Microvascular dysfunction might play a role by increasing peripheral vascular resistance (PVR). Metabolic and microvascular effects of insulin are impaired in obesity, but how these impairments contribute to disturbed blood pressure homeostasis is unclear. Specifically, it is unknown whether local microvascular vasoactive effects of insulin play a role in determining systemic vascular resistance. The aim of this study was to investigate the association between PVR and local microvascular effects of insulin.
Design and Methods
Thirty-seven healthy, overweight subjects (age 25-55 years, BMI 25-30 kg/m2) were cross-sectionally studied. Local insulin-mediated vasodilation was measured using skin laser Doppler fluxmetry combined with transcutaneous iontophoresis of insulin. For comparison, local vasodilatory effects of acetylcholine and sodium nitroprusside were measured. PVR was calculated from mean arterial pressure and cardiac output, assessed by pulse-dye densitometry.
PVR was inversely correlated with insulin-mediated vasodilation (r = −0.50; P < 0.01). This finding was maintained after adjustment for age, sex, blood pressure, and smoking. PVR was not associated with local microvascular effects of acetylcholine.
Our study in overweight subjects suggests that insulin's role in the microvasculature may contribute to blood pressure control.
The prevalence of obesity has been increasing rapidly worldwide, reaching epidemic proportions in the industrialized world [1, 2]. Obesity is an important risk factor for cardiovascular disease. One of the major sequelae of obesity is hypertension [3, 4], but the pathophysiological pathways linking obesity to disturbed blood pressure homeostasis are incompletely understood.
Blood pressure is the product of cardiac output (CO) and peripheral vascular resistance (PVR). CO may be increased in obesity as a result of sodium retention and increased circulating volume . Although PVR would be expected to decrease in obesity as a result of the increase in capillary cross-sectional area in adipose tissue, it may be inappropriately high because of, for example, hyperactivity of the sympathetic nervous system (SNS), renin-angiotensin system activation, and impaired vasodilator function . PVR is mainly determined by microvascular structure and function. In accordance, some of the pro-hypertensive effects of, for example, SNS activity, pro-inflammatory cytokines, and endocrine factors may be explained by their effects on the microcirculation .
Over the past two decades, it has been increasingly recognized that insulin regulates microvascular function and, consequently, may influence blood pressure regulation. Microvascular endothelial cells respond to insulin by increasing phosphorylation of endothelial nitric oxide synthase (eNOS) via a PI-3 kinase pathway, increasing Ca2+-independent nitric oxide synthase activity leading to vasodilation [7, 8]. Interestingly, insulin also activates the mitogen-activated protein kinase pathway in endothelial cells, which enhances the generation of the vasoconstrictor endothelin-1 . This can lead to insulin-stimulated vasoconstriction or impaired vasodilation if signaling from the insulin receptor to eNOS is inhibited pharmacologically or downregulated by insulin resistance, as seen in conditions of obesity and hypertension in animal studies [9, 10]. In obese human subjects, a blunted (micro-)vascular responsiveness to insulin has been demonstrated . Thus, endothelial insulin resistance may contribute to the development of hypertension in obesity and account in part for the epidemiological relationship between hypertension and insulin resistance [6, 8, 14].
Up till now, the contribution of direct vasoactive effects of insulin to blood pressure homeostasis in humans has been unclear. The systemic hyperinsulinemia often used in physiological test protocols precludes separation of local vasomotor and systemic effects of insulin, such as SNS and renin-angiotensin system activation, which have a considerable impact on both CO and PVR [15, 16].
We hypothesize that, in susceptible patients such as those with overweight, the direct local microvascular vasodilator response to insulin is inversely associated with PVR. As a model for this microvascular responsiveness to insulin we measured local vasomotor effects of insulin using skin laser Doppler fluxmetry combined with transcutaneous iontophoresis of insulin. Our study suggests that the direct microvascular vasodilator response to insulin is an important correlate of PVR, and thus may play a role in blood pressure homeostasis.
We cross-sectionally studied 37 healthy subjects aged between 25 and 55 years who were overweight (BMI between 25 and 30 kg/m2). Subjects were recruited via advertisements in local media. A telephone questionnaire and a screening visit were used to ascertain suitability for inclusion. The following exclusion criteria applied:  history of diabetes mellitus, cardiovascular, pulmonary, neurological, gastrointestinal, or other chronic disease,  use of any medication affecting cardiovascular function,  waist circumference of < 80 cm (women) or < 90 cm (men),  office systolic blood pressure >160 mm Hg and/or diastolic blood pressure >95 mm Hg,  abnormal laboratory results at screening: anemia (hemoglobin level <7 mmol/L), renal insufficiency (serum creatinine >130 μmol/L), diabetes mellitus (fasting plasma glucose >6.9 mmol/L), or  pregnancy. The study protocol was approved by the local medical ethics committee. All participants gave written informed consent.
All measurements were performed in the morning in a quiet, temperature-controlled room (22-24°C) after an overnight fast. No alcohol or caffeine containing beverages were allowed during the previous 24 hours. In addition, smoking was prohibited during the study day. Subjects were in a supine position during the measurements.A venous catheter was positioned in the antecubital vein for blood sampling and for measurement of CO (see ‘hemodynamic measurements’).
Measurements were started after a 30-min acclimatisation period.
Skin temperature was above 28°C at the start of each microvascular measurement. Skin blood flow was measured in conventional perfusion units (PU) by means of a laser Doppler system (Periflux 4000, Perimed, Stockholm, Sweden). Insulin-mediated vasodilation of finger skin microcirculation was evaluated with laser Doppler measurements in combination with iontophoresis of insulin. For comparison, specific endothelium-dependent and endothelium-independent vasodilations were tested using iontophoresis of acetylcholine (ACh) and sodium nitroprusside (SNP), respectively. Iontophoresis measurements were performed on the dorsal skin of the middle phalanx of the second (ACh), third (SNP), and fourth (insulin) finger of the nondominant hand. ACh (1%, Miochol, IOLAB, Bournonville Pharma, The Netherlands) was delivered using an anodal current; seven doses of 0.1 mA for 20 s were administered, with a 60-s interval. Insulin (0.20 mL Actrapid 100 IU/mL; Novo Nordisk, Denmark) and SNP (0.01%, Nipride, Roche, The Netherlands) were delivered with a cathodal current with a 90-s interval between each dose (0.20 mA for 20 s). Seven doses of ACh, nine doses of SNP, and 12 doses of insulin were delivered, which have previously been shown to result in an incremental time - response curve of skin blood flow . Baseline skin blood flow was recorded for 1 min before the start of the iontophoresis protocols. Peak skin blood flow was defined as the mean skin blood flow reached during the final two iontophoresis deliveries. Absolute and percentage increase in skin blood flow was defined as, respectively, the absolute and percentage change from baseline to peak skin blood flow.
Subjects had been in supine position for at least 2 h at the time of the hemodynamic measurements. CO was measured using a dye-dilution method. Pulse dye densitometry (PDD) is a validated, indirect, relatively low-invasive technique to measure CO. Details on the PDD method are provided elsewhere . To supplement published validation studies of this method, we have performed an in-house validation of this method by comparing it to pulmonary artery catheter thermodilution, showing excellent validity and reproducibility .
The mean of three sequential CO readings was taken. Parallel to the CO measurement, three sequential blood pressure measurements were performed and the mean was calculated. Blood pressure measurements were performed on the nondominant arm using an automated oscillometric device (Press-Mate BP-8800, Colin Co, San Antonio, USA, normal cuff size). Subjects were in the supine position during hemodynamic measurements. Thus, assuming that central venous pressure is negligible, PVR was calculated from mean CO and mean arterial blood pressure, using the following equation: Peripheral vascular resistance (mm Hg min/L) = mean arterial pressure (mm Hg) / cardiac output (L/min)
Continuous variables are presented as mean ± standard deviation (SD) or, in case of a nonparametric distribution, as median (interquartile range). Because microvascular data showed a skewed distribution, log-transformed variables were used for statistical analyses where appropriate. Pearson's and Spearman correlation analyses were performed to examine the relationship between hemodynamic and microvascular measurements, using the Pearson correlation coefficient (r) and Spearman's rank correlation coefficient (rs), respectively. Multiple regression analysis was used to investigate whether age, sex, smoking, blood pressure, ACh-mediated vasodilation, and SNP-mediated vasodilation were significant confounders and/or effect-modifiers in the relationships between PVR and microvascular measurements. Analyses were performed with and without inclusion of smokers. A two-tailed P-value of < 0.05 was regarded as significant. All analyses were performed on a personal computer using the statistical software package SPSS version 15.0 (SPSS, Chicago, IL, USA).
Baseline characteristics of the study population are shown in Table 1.
Table 1. Baseline characteristics
Continuous variables are expressed as mean ± standard deviation or median (interquartile range), depending on the normality of the distribution. BMI, body mass index; WHR, waist-to-hip-ratio; SBP, systolic blood pressure; DBP, diastolic blood pressure; HOMA-IR, homeostasis model assessment of insulin resistance [fasting insulin (mU/L) × fasting glucose (mmol/L) / 22,5)]; LDL, low-density lipoprotein; HDL, high-density lipoprotein.
45.7 ± 8.2
84.0 ± 10.5
28.0 ± 1.5
103.7 ± 4.1
90.6 ± 8.4
SBP (mm Hg)
118 ± 9.5
DBP (mm Hg)
81 ± 8.3
Fasting glucose (mmol/L)
5.0 ± 0.48
Fasting insulin (pmol/L)
Fasting cholesterol (mmol/L)
5.0 ± 0.9
2.9 ± 1.0
1.5 ± 0.5
Fasting triglycerides (mmol/L)
Hemodynamic and microvascular data are shown in Table 2. Mean CO was 5.1 L/min, which is in the range expected for healthy subjects. As expected, all iontophoresed substances induced a significant increase in blood flow (P < 0.001).
Table 2. Haemodynamic and microvascular measurements
Continuous variables are expressed as mean ± standard deviation or median (interquartile range), depending on the normality of the distribution. MAP, mean arterial pressure; PVR, peripheral vascular resistance; PU, arbitrary perfusion units; SNP, sodium nitroprusside; ACh, acetylcholine.
Cardiac output (min−1)
5.1 ± 1.1
MAP (mm Hg)
91.0 ± 7.4
Heart rate (beats per min)
59.0 ± 9.8
PVR (mm Hg min/L)
18.7 ± 4.0
Baseline perfusion (PU)
Peak perfusion (PU)
Absolute increase (PU)
Percentage increase (%)
Baseline perfusion (PU)
Peak perfusion (PU)
Absolute increase (PU)
Percentage increase (%)
Baseline perfusion (PU)
Peak perfusion (PU)
Absolute increase (PU)
Percentage increase (%)
As demonstrated in Table 3, insulin-mediated vasodilation was inversely correlated with PVR (absolute increase rs = −0.50, percent increase rs = 0.39). The association is visualized in Figure 1, which shows a scatter plot of PVR and log-transformed insulin-mediated vasodilation. This correlation remained significant after adjustment for age, sex, blood pressure, and smoking. Similar results were found after exclusion of smokers. Associations between insulin-mediated vasodilation and blood pressure were nonsignificant either for SBP (rs = 0.26; P = 0.12), DBP (rs = 0.25; P = 0.13), or MAP (rs = 0.25; P = 0.14). Insulin-mediated vasodilation was significantly associated with CO (rs = 0.49; P = 0.002).
Table 3. Spearman correlation analyses of peripheral vascular resistance and microvascular parameters
Microvascular measures were not significantly correlated to BMI within the range of the studied subjects (data not shown). To adjust for noninsulin-specific endothelial responses and for endothelium-independent vasoreactivity, we adjusted the correlation between PVR and insulin-mediated vasodilation for the responses to ACh and SNP. Entering these covariates into a linear regression model with PVR as the dependent variable and insulin-mediated vasodilation as the primary independent variable did not change the strength of the association.
We have shown that PVR is inversely correlated to the local vasodilatory effects of insulin in the microcirculation of overweight individuals.
As outlined in the Introduction section, microvascular effects of insulin may play a role in blood pressure regulation by causing variation in peripheral vascular resistance. We used iontophoresis to administer insulin locally, rather than the hyperinsulinemic clamp technique used in many other studies, thereby avoiding the hemodynamic effects of systemic hyperinsulinemia.
In the context of vascular resistance, the site of interest is the microcirculation, consisting of small arteries, arterioles, capillaries, and venules. It is at the level of the microcirculation that a substantial drop in hydrostatic pressure occurs [20, 21]. In patients with hypertension, microvascular structure is altered and microvascular function is impaired [20, 22]. We and others have hypothesized that microvascular abnormalities may not only be a consequence of hypertension, but may in fact precede - and contribute to - the elevation of blood pressure, hence creating a vicious circle [6, 23]. Indeed, structural and functional microvascular abnormalities have been reported in normotensive subjects at risk for hypertension [14, 24]. Our understanding of the role of obesity-associated microvascular abnormalities in the development of hypertension has been enhanced by studies in the spontaneously hypertensive rat. Defects in vascular responses to insulin can be detected in spontaneously hypertensive rats before the onset of hypertension, suggesting that elevated blood pressure per se does not determine insulin resistance in this model . Consistent with the findings of this study, spontaneously hypertensive rat show impaired NO-depentent vasodilator responses to insulin, whereas endothelial function with respect to ACh appears normal.
On a biological level, the variable local response to insulin may be explained by variation of signaling effects of insulin in the microvascular endothelium. Mechanistically, the net effect of insulin on vessel diameter is the result of a balance between NO-dependent vasodilator effects and endothelin-mediated vasoconstrictor effects, normally resulting in vasodilation in vivo. In obesity, however, a disturbance of this balance in relative favor of endothelin seems to exist, leading to impaired insulin-mediated vasodilation or even insulin-mediated vasoconstriction. Elevated levels of circulating free fatty acids and AngII, and increased local concentrations of pro-inflammatory cytokines (such as TNFα) are all candidates for induction of impaired insulin-mediated vasodilation in overweight and obesity . In man, increased ET-1 activity has been observed in type 2 diabetes and hypertension [29, 30]. Taken together, these data suggest that vascular effects of insulin contribute to increased ET-1 activity in hypertension and type 2 diabetes.
In this study, insulin-mediated vasodilation was associated with PVR, whereas ACh- and SNP-mediated vasodilation were not. All three vasoactive substances have distinct signaling pathways. SNP acts as an NO-donor, thus bypassing the endothelium. ACh and insulin both act via the endothelium. Insulin's actions on vascular tone occur via effects on the balance between nitric oxide and endothelin production [9, 31]. Iontophoretically applied ACh is generally agreed to induce an endothelium-dependent vasodilation, but the relative contribution of the different mediators is still a matter of debate. Studies suggests an important role for other substances than NO, such as prostaglandins, which only play a minor role in determining basal vascular tone . Although we cannot pinpoint if and how exactly differences in vasoactive mediators explain the fact that insulin, but not ACh, correlates with vascular resistance, it is important to stress that “endothelium-dependent vasodilation” in fact represents different mechanisms. Also, the physiological interpretation of effects of insulin and ACh is quite different in that insulin is, as opposed to ACh, normally present in the circulation, appearing in variable concentrations relating to obesity, insulin resistance, etc., which are all conditions interacting with blood pressure homeostasis.
Our finding of an inverse, continuous relationship between insulin-mediated vasodilation and PVR in mildly overweight subjects supports the hypothesis that diminished insulin-mediated vasodilation plays a role in regulating PVR in obesity. In line with this concept, several previous studies revealed the ability of insulin to reduce PVR in healthy, lean individuals [35, 36]. Animal studies suggest that this ability may be blunted in obesity [37, 38].
It should be stressed at this point that a lower vascular insulin sensitivity alone may not lead to a sustained increment of blood pressure as long as adaptive responses of CO to increased vascular resistance (e.g., renal pressure-natriuresis) remain intact . This may explain why in this study, a significant association was observed between insulin-iontophoresis and PVR, but not blood pressure. Only few studies investigated the relationship between vascular effects of insulin and blood pressure. Baron et al. demonstrated a continuous inverse relationship between basal mean arterial pressure and insulin-induced skeletal muscle blood flow in healthy individuals . Another study showed a blood pressure increase in obese subjects compared to lean subjects in response to a euglycemic hyperinsulinemic clamp . Unfortunately, it is uncertain whether these findings can be attributed to direct microvascular effects of insulin, since systemic hyperinsulinemia was applied in these studies.
A few limitations of our study merit discussion. As a result of the cross-sectional design this study cannot distinguish between cause and effect, nor can it fully exclude the presence of another unmeasured variable explaining the association between PVR and insulin-mediated vasodilation. However, a potential common determinant such as SNS activity would equally affect vascular responses to insulin, ACh and SNP. Nonetheless, future studies should attempt to address direct causality and mechanisms underlying the associations we observed. Another limitation is the small number of participants. Particularly the number of men is small, precluding conclusions on sex-specific effects. It is also important to realize that the study population is overweight but not obese, and relatively insulin-sensitive, as judged from the HOMA model assessment. This prohibits extrapolation of the results to obese and metabolically more unhealthy, insulin-resistant individuals.
In addition, transcutaneous laser doppler fluxmetry predominantly measures flow in precapillary arterioles, whereas PVR is mainly determined by resistance arteries located just upstream. Theoretically, there may be differences in insulin responses between these microvascular structures. A limitation of the method of iontophoresis of insulin is the fact that the exact local insulin concentration reached by iontophoresis of insulin is unknown. Finally, no vehicle response was measured in this study to adjust for potential nonspecific hyperemic responses to cathodal iontophoresis of saline alone. However, this is, at least partially, overcome by adjusting for vasodilator responses to SNP, which is, as insulin, delivered by cathodal iontophoresis.
In conclusion, our results suggest that insulin's role in the microvasculature may contribute to blood pressure control. Further studies should attempt to repeat and extend our findings, and should address causative links and mechanisms. Such studies should also involve patients with marked obesity, diabetes, and/or hypertension.