Blood pressure and insulin resistance: role for microvascular function? [Cardiovasc Res 2002;53:271–276]

In a recent issue of Cardiovascular Research, Irving et al. [1] reported on the relationships among microvascular function, blood pressure, and insulin resistance in healthy men. The main objective was to test the hypothesis that defects in microvascular function could explain the relationship between elevated blood pressure and insulin resistance. They report that higher systolic blood pressure was associated with insulin resistance, lower dermal capillary density at rest and during venous occlusion, and impaired maximum dermal blood flow after heating, but not with dilator responses to acetylcholine. Insulin resistance, as assessed by HOMA, did not correlate with indices of microvascular function or structure. It is concluded that the association between hypertension and insulin resistance is unlikely to be explained by altered microvascular function and structure. In contrast, we have demonstrated that higher systolic blood pressure was associated with insulin resistance, capillary recruitment during post-occlusive reactive hyperaemia (PRH), and dilator responses to acetylcholine in both normotensive [2] and hypertensive subjects [3]. Peripheral insulin resistance, as assessed with the euglycaemic, hyperinsulinaemic clamp technique was also associated with these measures of microvascular function [2,3]. Moreover, the association between hypertension and peripheral insulin resistance of glucose uptake could be explained, at least statistically, by altered microvascular structure and function [2,3]. In their article, Irving et al. suggest that their estimate of the relationship between microvascular function and insulin sensitivity is more accurate due to the greater statistical power compared to our studies given the number of studied subjects. However, the probability to detect a significant relationship between microvascular function and insulin sensitivity also depends on the accuracy of the technique to assess insulin sensitivity, the reproducibility of the measurement and the methods used to recruit subjects. Therefore, we have to disagree with Irving et al.'s conclusion for several reasons. Firstly, Irving et al. used an inaccurate measure of insulin sensitivity. The hypothesis investigated states that defects in microvascular function, by affecting insulin-mediated changes in muscle perfusion, modulate insulin-mediated glucose uptake in peripheral tissues [2,3]. Thus, to test this hypothesis it is necessary to measure peripheral insulin-mediated glucose uptake specifically. Homeostatic model assessment (HOMA) relies on the product of fasting plasma glucose and fasting plasma insulin and essentially explores hepatic insulin sensitivity [4,5]. In contrast, the euglycemic, hyperinsulinaemic clamp technique measures insulin-mediated glucose disposal in the peripheral tissues, mainly muscle [5]. Although reasonable correlations can be observed between HOMA and clamp-derived insulin sensitivity (correlations ranged from 0.35 to 0.79 in normal subjects; explained variance of clamp-derived insulin sensitivity 12–62%) [4,6,7], it should be realised that from a pathophysiological standpoint, in any given individual, HOMA (hepatic insulin resistance) and clamp-derived insulin sensitivity (peripheral insulin resistance) provide different information [4,5]. Secondly, the reproducibility (expressed as coefficient of variation) of insulin sensitivity measured with the clamp technique is much better than that measured with HOMA (9 vs. 24%, respectively) [6]. Thirdly, Irving et al. randomly selected 105 subjects, and although the variation in insulin sensitivity is not mentioned in their article, random selection will result in approximately a three- to fourfold variation in insulin sensitivity. In contrast, in our study 18 subjects were recruited to show substantial variability in their clamp-derived insulin sensitivity, which resulted in a ninefold variation in insulin sensitivity. Altogether, this means that the method we have used to measure insulin sensitivity is more valid, more accurate and more reproducible. Moreover, it is likely that our population showed a larger variability in the variable of interest, i.e. insulin sensitivity, making it more suitable to test the hypothesis in question. In addition, Irving et al. suggest that our results may be confounded by age and/or sex, however, as mentioned in our article [2], statistical adjustment for age and sex did not change our findings. Irving et al. express uncertainty whether functional differences in the microcirculation are well represented by recruitment of capillaries during PRH, mainly because of a recently published paper that reported a decrease in capillary density during PRH [8]. We have addressed this issue by demonstrating that methodological differences in determining capillary density in the resting state could explain this discrepancy [9]. Moreover, the relevance of capillary recruitment during PRH is underlined by the fact that it is positively associated with the increase in dermal capillary density during hyperinsulinaemia (r = +0.71; P<0.05) [10]. Insulin-mediated capillary recruitment has been demonstrated, in humans, to be able to modulate glucose uptake [11,12]. In conclusion, the results reported by Irving et al. do not contradict the hypothesis that microvascular function plays an important role in the association between blood pressure and insulin resistance.