Is Insulin an Independent Risk Factor for Hypertension? The Paris ...

International Journal of Epidemiology © International Epidemiological Association 1997

Vol. 26, No. 3 Printed in Great Britain

Is Insulin an Independent Risk Factor for Hypertension? The Paris Prospective Study ANNE FAGOT-CAMPAGNA,* BEVERLEY BALKAU,* DOMINIQUE SIMON,*,** PIERRE DUCIMETIÈRE† AND EVELINE ESCHWÈGE* Fagot-Campagna A (INSERM Unit 21, Faculty of Medicine Paris-Sud, 16 avenue Paul Vaillant Couturier, 94807 Villejuif Cedex, France), Balkau B, Simon D, Ducimetière P and Eschwège E. Is insulin an independent risk factor for hypertension? The Paris Prospective Study. International Journal of Epidemiology 1997; 26: 542–550. Background. The link between hyperinsulinaemia and hypertension has been examined in few prospective studies and often diminished after adjustment for obesity, central adiposity and baseline blood pressures. Methods. The incidence of hypertension was studied as a function of baseline insulin and glucose in 4149 Caucasian, non-hypertensive, non-diabetic middle-aged men from the Paris Prospective Study. Blood pressures were measured over the 3 years of follow-up; hypertension incidence was defined as systolic blood pressure >160 mmHg or diastolic blood pressure >95 mmHg or drug treatment for hypertension. Results. Fasting and 2-hour glucose and insulin were predictive of hypertension, after controlling for the known risk factors: age, excessive alcohol consumption and family history of hypertension (FHH). However, after further controlling for body mass index and central adiposity (the iliac circumference), insulin was no longer predictive in men without an FHH. When weight variation was also taken into account, and further adjustment made for baseline blood pressure and heart rate, fasting insulin, only, was predictive when the subject had a weight increase, independently of FHH. Fasting glucose was predictive of hypertension except in the case of no change or weight decrease and a negative FHH; 2-hour glucose was predictive in the presence of a positive FHH. Conclusions. Insulin and glucose levels were both risk factors for hypertension, and this risk was enhanced in the case of a positive FHH. However, obesity, especially central obesity, confounded these relationships and might be an intermediary factor in the relationship between insulin and hypertension. Keywords: essential hypertension, glucose, insulin, obesity, men

Reaven8 described the insulin resistance syndrome as the association of a number of abnormalities: insulin resistance, hyperinsulinaemia, glucose intolerance, increase of very low density triglycerides, decrease of high density lipoprotein cholesterol and hypertension. Björntorp9,10 and more recently Reaven 11 added obesity, emphasizing an abdominal fat distribution. Insulin resistance is thought to be the underlying abnormality of this syndrome12 and the fasting insulin concentration is its surrogate in non-diabetic subjects.13,14 Many crosssectional studies have found associations between the different abnormalities of the insulin resistance syndrome, in particular between hypertension and insulin,15–23 but results differ according to ethnicity and sex: the relation seems to hold in Caucasians and in men. In many studies, the link between these two disorders diminishes after adjustment for obesity and abdominal fat. Few prospective studies have examined the link between insulin and hypertension incidence24–31 and results have not been consistent.

While insulin has been shown to be a risk factor of cardiovascular disease in five prospective epidemiological studies,1–5 other studies have not confirmed these results.6 In contrast, hypertension is an accepted risk factor for cardiovascular disease, but treatment for hypertension with diuretics or beta-blockers does not decrease the risk of coronary heart disease as much as might be expected from epidemiological studies.7 A hypothesis which might explain this apparent discrepancy is that another abnormality, associated with hypertension and not corrected by anti-hypertensive treatment, may be involved in cardiovascular diseases. * INSERM Unit 21, Faculty of Medicine Paris-Sud, 16 avenue Paul Vaillant Couturier, 94807 Villejuif Cedex, France. ** Hôpital Henri Mondor, 51 avenue du Maréchal de Lattre de Tassigny, 94000 Créteil, France. † INSERM Unit 258, Hôpital Broussais, 96 rue Didot, 75674 Paris Cedex 14, France. Reprint requests to: Anne Fagot-Campagna, National Institute of Diabetes and Digestive Kidney Diseases, 1550 East Indian School Road, Phoenix, AZ 85014, USA.

542

543

INSULIN AND HYPERTENSION

We examined the relationship between insulin and glucose levels and the risk of developing hypertension in 4149 active, middle-aged Caucasian men from the Paris Prospective Study, taking into consideration the possible confounding effects of obesity and abdominal fat distribution.

MATERIALS AND METHODS Research Design The Paris Prospective Study included, between 1967 and 1972, 7991 Paris policemen. They were aged 43–54 years at the time of the first screening which consisted of a questionnaire about health and habits and anthropometric measurements including height (to the nearest cm) with the subject wearing no shoes, and iliac circumference (at the iliac crest level, to the nearest mm). In the second year, the baseline year for this analysis, a blood sample was taken at fasting and following a 75 g oral glucose tolerance test to measure plasma glucose (G0, G2), and insulin (I0, I2) by a radioimmunological assay.32 In the second year and during the subsequent three years of follow-up, weight was measured to the nearest 0.5 kg with the subject wearing light clothing and blood pressures were taken once at each exam by the same trained team. Subjects were seated for 15 min and the first and fourth Korotkoff sounds were measured on the right arm to the nearest 10 mmHg (a diastolic blood pressure (DBP) of 95 was recorded as 100 mmHg but a DBP of 94 was recorded as 90 mmHg). The heart rate was also recorded at the same time. Hypertension was defined according to the WHO criteria33 but at only one examination: DBP >95 mmHg (recorded as >100 mmHg in our study) and/or systolic blood pressure (SBP) >160 mmHg and/or current use of antihypertensive medications. A subject who had readings which were elevated at the first follow-up visit but not at the next one was declared hypertensive at the first visit. Body mass index (BMI) was calculated as weight (kg) divided by the square of height (m 2). A family history of hypertension (FHH) included hypertension in any parent, brother or sister. Excessive alcohol consumption was defined by a clinical examination typical of either cirrhosis or clinical alcoholism or by a mean corpuscular volume .98 µ3, a marker linearly related with alcohol intake.34 For this analysis, we selected the 4149 men born in France, attending the second examination, not diabetic or hypertensive at baseline, with no missing values for baseline blood pressures, fasting and 2-hour insulin and glucose, BMI, iliac circumference, excessive alcohol consumption and FHH.

TABLE 1 Characteristics of 4149 subjects at baseline. Median, 10th and 90 percentiles 10th percentile Age (years) 45.8 21.8 BMI (kg/m2) Iliac circumference (mm) 790 –0.40 Weight variation (kg/m2/year) Systolic blood pressure (mmHg) 120 Diastolic blood pressure (mmHg) 70 56 Heart rate (min–1) Fasting insulin (mM) 29 2-hour insulin (mM) 79 Fasting glucose (mM) 4.8 2-hour glucose (mM) 3.6

Median

90th percentile

49.3 25.3 900 0 130 80 64 72 222 5.6 5.3

51.3 29.3 1007 +0.53 150 90 76 136 581 6.2 7.5

Statistical Analysis Analysis of covariance and χ2 tests were used for comparing means and prevalences between subgroups. Insulin levels were log-transformed. The significance of the difference in hypertension rates between groups was determined by a χ2 test for stratified incidence data.35 The relationship between baseline variables and hypertension incidence was assessed by Cox’s proportional hazards regression analysis. 36 Systolic blood pressure was used as a categorical variable because precise measures were not available. For each continuous variable, the hazard ratio of developing hypertension was calculated comparing the 90th and 10th percentiles. Adjustment was made for factors known to be predictive of hypertension: age, BMI, central adiposity (iliac circumference), body weight variation, FHH, excessive alcohol consumption, baseline SBP and heart rate, a marker of stress.37 The interaction between variables was evaluated by a likelihood ratio test.38 The validity of the proportional hazards assumption for the covariates was verified as suggested by Kalbfleish and Prentice.39 For most of the analyses, age and baseline SBP violated the proportionality assumption. Proportional hazards models, therefore, were stratified by SBP at baseline (100–120 mmHg; 130 mmHg; 140 mmHg; 150 mmHg) and two age groups divided by the median (49.3 years) in order to control for these variables. All statistical analyses used the SAS statistical package. A statistically significant level was defined as P , 0.05.

RESULTS The length of follow-up among the 4149 nonhypertensive men at baseline was 3.16 ± 1.35 years

544

INTERNATIONAL JOURNAL OF EPIDEMIOLOGY

FIGURE 1 Incidence rate of hypertension by fasting insulin at baseline divided by quartiles, and by status for family history of hypertension (FHH) and weight variation. Three-year follow-up of 4033 men participating in the Paris Prospective Study

FIGURE 2 Incidence rate of hypertension by fasting insulin at baseline divided by quartiles, and by status for obesity and abdominal fat distribution. Three-year follow-up of 4149 men participating in the Paris Prospective Study

(mean ± SD). Table 1 shows the characteristics of the subjects at baseline. Prevalences of excessive alcohol consumption and FHH were 29% and 44% respectively. When adjusted for age, the 738 (18%) men lost to follow-up before the occurrence of hypertension and the last examination had a higher prevalence of excessive alcohol consumption (P , 0.001), lower mean insulin (P = 0.01 for fasting and 2-hour insulin) and higher heart rate (P = 0.02) than subjects who were followed up. Mean blood pressures were similar. The 47 subjects (1%) who died during the 3 years of followup before the occurrence of hypertension had a higher prevalence of excessive alcohol consumption (P , 0.001) and higher 2-hour insulin (P = 0.004), 2-hour glucose (P = 0.002) and heart rate (P = 0.006) but similar blood pressures to other subjects. There were 116 men with missing information for weight during the 3 years of follow-up, but there were no significant differences between the groups with and without missing values for body weight variation (data not shown). The incidence rate of hypertension was 78.9 per 1000 person-years. Incidence rates of hypertension according to fasting insulin divided by the quartiles are shown in Figure 1 by strata of FHH and weight variation divided by the median. Men in the higher two quartiles for fasting insulin had higher incidence rates of hypertension if the FHH was positive, and among those with similar FHH, men with a weight increase had higher rates (for the first and the second highest insulin quartiles, respectively, P for trend ,0.001 and P = 0.04). This

difference was not observed in men in the lower two insulin quartiles (P for trend = 0.60). To emphasize the influence of obesity, especially central adiposity, on the relationship between fasting insulin and hypertension, the incidence rates of hypertension are shown according to fasting insulin, and by strata of BMI and iliac circumference divided by the median (Figure 2). At all insulin levels, men with high iliac circumference had highest incidence rates of hypertension, and among those with similar iliac circumference, men with high BMI had higher incidence rates (P for trend ,0.001 at all insulin levels), but this effect was stronger in men with high fasting insulin level. In proportional hazards models, a positive FHH but not an excessive alcohol consumption was a predictive factor for hypertension (Table 2). However, further adjustments took into account an excessive alcohol consumption because of its known predictive value for hypertension in other populations. Body mass index, iliac circumference, blood pressures, heart rate, and weight variation were predictive of hypertension when controlled for age, excessive alcohol consumption and FHH. Because of a positive and significant interaction between a FHH and insulin levels (P = 0.03 with fasting insulin and P = 0.05 with 2-hour insulin), analyses were performed separately in men with a positive and with a negative FHH. Men with a positive FHH had higher age-adjusted means for SBP (P = 0.002) and BMI (P = 0.02) as compared to men with a negative FHH, but mean glucose and insulin concentrations were similar. Fasting and

545


TABLE 2 Hazards ratios (HR) and 95% confidence intervals (CI) for hypertension incidence, controlled for age, excessive alcohol consumption and family history of hypertension. 90th compared with 10th percentile. Three-year follow-up of 4149 men participating in the Paris Prospective Study

Excessive alcohol consumptionb Family history of hypertensionb BMI Iliac circumference Weight variationc Systolic blood pressured Diastolic blood pressure Heart rate Fasting insulin 2-hour insulin Fasting glucose 2-hour glucose

HRa

95% CIa

1.06 1.30 2.40 2.58 1.48 4.64 2.60 1.77 1.83e 1.62e 1.58 1.45

(0.93–1.21) (1.15–1.47) (2.07–2.79) (2.21–3.01) (1.29–1.70) (3.73–5.77) (2.23–3.03) (1.55–2.02) (1.56–2.14)e (1.38–1.90)e (1.44–1.73) (1.30–1.62)

a

1035 incident cases. HR controlled for age only. c Because of missing values, 4033 subjects and 1007 incident cases. d HR is given for 3 years or less of follow-up to respect the proportionality assumption of the Cox model. 696 incident cases. e The HR should be interpreted with caution as it is dependent on the family history of hypertension (P < 0.05). b

2-hour insulin and glucose were significant risk factors for hypertension when controlled for age and excessive alcohol consumption whatever the FHH (Table 3). When controlled for age, excessive alcohol consumption, BMI and iliac circumference, insulin levels were predictive of hypertension only in subjects with a positive FHH. Fasting and 2-hour glucose were significant predictive factors whatever the FHH. A significant and positive interaction term was found between weight variation and both fasting and 2-hour insulin (P , 0.001) in subjects with no missing values for weight variation (n = 4033). Therefore, analyses were divided by the median of weight variation into men with weight increase, and men with no change or weight decrease. Men with weight increase had at baseline lower age-adjusted fasting insulin, fasting and 2-hour glucose, BMI, heart rate and blood pressures (P , 0.001 for all variables) but not significantly lower 2-hour insulin (P = 0.09) than men with no change or weight decrease. When controlled for age, excessive alcohol consumption, BMI, iliac circumference, SBP and heart rate, fasting insulin was a significant risk factor for hypertension whatever the FHH, but only in men with a weight increase: fasting insulin was a significant protective factor in the case of a negative FHH and no change or weight decrease (Table 4). Two-hour insulin was not a risk factor for hypertension. Fasting glucose was a risk factor for hypertension when there was a positive FHH, and, if the FHH was negative, in the case of a weight increase. Two-hour glucose was a significant

TABLE 3 Hazards ratios (HR) and 95% confidence intervals (CI) from proportional hazards models, for hypertension incidence, controlled for age and excessive alcohol consumption, and controlled for age, excessive alcohol consumption, BMI and iliac circumference, by status for family history of hypertension. HR comparing the 90th with 10th percentiles. Three-year follow-up of 4149 men participating in the Paris Prospective Study Negative family history of hypertensiona HR

Positive family history of hypertensionb

95% CI

HR

95% CI

(1.25–1.93) (1.12–1.73) (1.32–1.74) (1.13–1.54)

2.18 1.92 1.63 1.62

(1.74–2.73) (1.52–2.42) (1.45–1.84) (1.38–1.89)

Adjusting for age, excessive alcohol consumption, BMI and iliac circumference Fasting insulin 1.09 (0.86–1.38) 2-hour insulin 1.02 (0.81–1.29) Fasting glucose 1.28 (1.10–1.48) 2-hour glucose 1.17 (1.00–1.37)

1.58 1.49 1.38 1.42

(1.24–2.02) (1.18–1.89) (1.21–1.58) (1.21–1.66)

Adjusting for age and excessive alcohol consumption Fasting insulin 1.55 2-hour insulin 1.39 Fasting glucose 1.52 2-hour glucose 1.31

a b

2341 subjects and 525 incident cases. 1808 subjects and 510 incident cases.

546


TABLE 4 Hazards ratios (HR) and 95% confidence intervals (CI) for hypertension incidence from proportional hazards models controlled for age, excessive alcohol consumption, BMI, iliac circumference, systolic blood pressure and heart rate, by status for family history of hypertension and weight variation. HR comparing the 90th with 10th percentiles. Three-year follow-up of 4033 men participating in the Paris Prospective Study Negative family history of hypertensiona

Positive family history of hypertensionb

HR

95% CI

HR

95% CI

No change or weight decrease Fasting insulin 2-hour insulin Fasting glucose 2-hour glucose

0.65 0.74 1.02 1.01

(0.46–0.90) (0.53–1.04) (0.81–1.29) (0.81–1.26)

1.33 1.34 1.24 1.26

(0.92–1.92) (0.96–1.87) (1.02–1.49) (1.02–1.56)

Weight increase Fasting insulin 2-hour insulin Fasting glucose 2-hour glucose

1.43 1.20 1.29 1.10

(1.01–2.01) (0.87–1.64) (1.02–1.64) (0.85–1.44)

1.58 1.31 1.52 1.33

(1.11–2.26) (0.93–1.84) (1.13–2.05) (1.01–1.74)

a b

Respectively, 1187 and 1088 subjects, and 241 and 271 incident cases for men without or with weight increase. Respectively, 924 and 834 subjects, and 238 and 257 incident cases for men without or with weight increase.

risk factor in the case of a positive FHH whatever the weight variation. No difference was observed for insulin and glucose as risk factors for hypertension between subgroups of men who decreased their weight or remained weight-stable (data not shown). The hazards ratios for insulin and glucose differed between the four FHH and weight variation groups (P , 0.02) except for 2-hour glucose (P = 0.21). In forward stepwise models (P = 0.05 for entry into the model) controlled for age, excessive alcohol consumption, BMI, iliac circumference, SBP and heart rate, the fasting and 2-hour insulin and glucose concentrations that entered into the four different models were: fasting insulin for men with a negative FHH and no change or weight decrease (protective factor, P = 0.01); fasting glucose for men with a negative FHH but weight increase (risk factor, P = 0.04); fasting glucose for men with a positive FHH but no change or weight decrease (risk factor, P = 0.03); and fasting glucose and then fasting insulin for men with a positive FHH and weight increase (risk factors, P = 0.01 and P = 0.02, respectively).

DISCUSSION Essential hypertension is partly a genetic disease and we found that a positive FHH, which may partially account for a genetic risk factor, was a predictive factor for hypertension. In other studies, higher levels of insulin have been found in offspring of subjects with

hypertension, as compared to control subjects. 40–44 While we did not find such differences at baseline, offspring of subjects with hypertension had higher baseline blood pressures and BMI. We found that insulin was a risk factor for hypertension only in the case of a positive FHH after controlling for obesity, and that the predictive value of glucose was greater in this case. Therefore, our results support the hypothesis of a strong genetic component for the relationship between insulin and hypertension, whereas this component was less strong for the relationship between glucose and hypertension. Seven prospective studies have analysed the relationship between insulin and hypertension incidence. When controlled for obesity, insulin was predictive of hypertension in Finnish subjects,24 in Swedish men25 and women,26 in white subjects of the Cardia study30 and only in the lean subjects of the San Antonio27 and the San Luis Valley29 studies. However, this relationship was not controlled for baseline blood pressures in three studies,24,26,27 it was no longer significant when controlled for in three other studies,25,29,31 and remained significant in only one study.30 The peculiarity of the Paris Prospective Study population in comparison to others is that the participants were middle-aged policemen who had a high alcohol consumption45 and probably high levels of stress. We controlled for these two factors but a lack of accuracy in the two surrogate measures used (our definition of an excessive alcohol


consumption was not a significant predictor for hypertension) may still be responsible for some confounding effect. However, insulin was found to be a risk factor for cardiovascular disease in this population,3 which may imply a closer link between insulin and hypertension. Weight gain, BMI and iliac circumference, a measure of central adiposity linked with cardiovascular diseases in this population,46 were all strong predictive factors for hypertension in our study. However, the predictive relationship between insulin and hypertension was dependent on BMI and iliac circumference in subjects with a negative FHH, and insulin was a risk factor only in the case of weight gain when controlled for obesity, blood pressure and heart rate, which were strong predictive factors of hypertension. Nevertheless, residual confounding due to BMI, central adiposity, alcohol consumption, etc., could remain and be partially responsible for the former relationship. The results of the present study suggest that obesity may be an intermediary factor in the relationship between insulin and hypertension and that environmental factors are an important component of this relationship. There are a number of possible explanations for the low predictive power of insulin for hypertension incidence in men with a negative FHH or no change or weight decrease. First, the 3-year incidence of hypertension was high (78.9 per 1000 person-years) as compared to other studies.25–28,47 Although the men studied were at high risk for hypertension because of their age and social characteristics, misclassification, because the definition of hypertension was based on a single measurement and on the first and fourth Korotkoff sound, may have resulted in some overestimation of incidence, even though a single measurement of blood pressure may be accurate in epidemiological studies.48 This may have decreased the likelihood of detecting risk factors. The lack of accuracy in the measurement of blood pressure (to the nearest 10 mmHg) is another limitation of the present study: the methodology we used (Cox models and stratification by class of blood pressure at baseline to respect the proportionality assumption) partially avoided this inaccuracy by using classes for blood pressure. A DBP of 95 mmHg was recorded as 100 mmHg and the subject as hypertensive, whereas a DBP of 94 mmHg was recorded as 90 mmHg and the subject as normotensive in any case. Nevertheless, an SBP record of 156 mmHg would have been recorded as 160 mmHg, and the subject included as hypertensive. Another explanation for the low predictive power of insulin for hypertension incidence is that a prevalenceincidence bias may have occurred in these middle-aged subjects:49 hypertensive subjects were excluded, but they were the most hyperinsulinaemic and would have

547

the greatest exposure if insulin is a real risk factor for hypertension. Moreover, confounding pro-insulin and insulin would also decrease the predictive power of insulin on hypertension, as the assays used in this study measured both insulin and pro-insulin. As shown in the San Antonio Study, pro-insulin rather than insulin itself may be the real risk factor for cardiovascular diseases 50,51 but both SBP and DBP were correlated with fasting and 2-hour pro-insulin and true insulin.52 Finally, we found that subjects who gained weight had lower baseline fasting insulin, as has been shown in the Pima Indians53 and it has been suggested that hyperinsulinaemia may represent physiological adaptations to obesity that limit further fat deposition. Therefore, hyperinsulinaemia may also be responsible for a limiting of hypertension incidence in our study. According to physiopathological hypotheses about the effects of insulin on blood pressure levels, insulin is an anti-natriuretic factor,54–56 but there has been no evidence of insulin leading to hypertension through this action.57 Another mechanism found in experimental studies is the stimulation of the sympathic nervous system,58,59 but a decrease in the peripheral vascular vessel resistance may counterbalance this effect, resulting in no increase in blood pressure.57 Obese subjects may have a reduced inhibition of peripheral vascular vessel resistance that cannot counterbalance the stimulation of the sympathetic system which seems intact.60–63 This may explain the confounding effect of BMI and iliac circumference on the relationship between insulin and hypertension, and the predictive power of insulin on hypertension that we observed only when there was a weight gain, even after adjustment for heart rate, a surrogate measure of stress. In our study, glucose was a stronger predictive factor for hypertension than insulin after controlling for obesity. Results from the cross-sectional analysis of the Paris Prospective Study suggested such a link64 and similar results have also been observed in a Finnish study.65 Glucose may be involved in hypertension incidence through non-enzymatic glycation of proteins and production of free radicals:66 a reduction of antioxidant defence is now an established cardiovascular risk factor and is hypothesized to play a role in the development of hypertension. In fact, in our study, the predictive value of glucose was stronger in the case of a positive FHH, which may account for a different effect of hyperglycaemia-related increased free radical production due to a genetic contribution, independently of obesity. In conclusion, in a population where insulin has been shown to be a risk factor for cardiovascular diseases, after controlling for obesity, a high fasting insulin

548


concentration was predictive of hypertension only in subjects who gained weight over the 3 years of followup. Fasting glucose was a significant predictive factor even after controlling for obesity, except in the case of no change or weight decrease and a negative FHH, and 2-hour glucose was predictive in the presence of a positive FHH. We conclude that both glucose and fasting insulin levels were risk factors for hypertension, particularly in the case of a FHH, which may partially account for a genetic predisposition. However, environmental factors confounded these relationships, and obesity, especially central obesity, might be an intermediary factor in the relationship between insulin and hypertension.

ACKNOWLEDGEMENTS The Paris Prospective Study was conducted by the Groupe d’Etudes sur l’Epidémiologie de l’Athérosclérose (GREA) and is supported by INSERM (Units 21, 55, 169 and 258) and the Direction de l’Action Sociale, de l’Enfance et de la Santé de la Ville de Paris (DASES). Dr A Fagot-Campagna was supported by the Fondation pour la Recherche Médicale, France. The authors would like to thank Dr K M Venkat Narayan for reviewing this manuscript and for his advice, and Dr Robert L Hanson for statistical advice.

REFERENCES 1

Pyörälä K. Relationship of glucose tolerance and plasma insulin to the incidence of coronary heart disease: results from two populations studies in Finland. Diabetes Care 1979; 2: 131–41. 2 Welborn T A, Wearne K. Coronary heart disease incidence and cardiovascular mortality in Busselton with reference to glucose and insulin concentrations. Diabetes Care 1979; 2: 154–60. 3 Eschwège E, Richard J L, Thibult N et al. Coronary heart disease mortality in relation with diabetes, blood glucose and plasma insulin levels. The Paris Prospective Study, 10 years later. Horm Metab Res 1985; 15: 41–46. 4 Yarnell J W G, Sweetnam P M, Marks V, Teale J D, Bolton C H. Insulin in ischaemic heart disease: are associations explained by triglyceride concentrations? The Caerphilly prospective study. Br Heart J 1994; 171: 293–96. 5 Desprès J P, Lamarche B, Mauriège P et al. Hyperinsulinemia as an independent risk factor for ischemic heart disease. N Engl J Med 1996; 334: 952–57. 6 Wingard D L, Barrett-Connor E L, Ferrara A. Is insulin really a heart disease risk factor? Diabetes Care 1995; 18: 1299– 304. 7 Collins R, Peto R, MacMahon S et al. Blood pressure, stroke, and coronary heart disease. Part 2, short term reductions in blood pressure: overview of randomised drug trials in their epidemiological context. Lancet 1990; 335: 827–38.

8

Reaven G M. Role of insulin resistance in human disease. Diabetes 1988; 37: 1595–607. Björntorp P. ‘Portal’ adipose tissue as a generator of risk factors for cardiovascular disease and diabetes. Arteriosclerosis 1990; 10: 493–96. 10 Björntorp P. Metabolic implications of body fat distribution. Diabetes Care 1991; 14: 1132–43. 11 Reaven G M. The Fourth Musketeer—from Alexandre Dumas to Claude Bernard. Diabetologia 1995; 38: 3–13. 12 Ferrannini E, Haffner S M, Mitchell B D, Stern M P. Hyperinsulinemia: the key feature of a cardiovascular and metabolic syndrome. Diabetologia 1991; 34: 416–22. 13 Anderson R L, Hamman R F, Savage P J et al. Exploration of simple insulin sensitivity measures derived from frequently sampled intravenous glucose tolerance (FSIGT) tests. The Insulin Resistance Atherosclerosis Study. Am J Epidemiol 1995; 142: 724–32. 14 Laasko M. How good a marker is insulin level for insulin resistance? Am J Epidemiol 1993; 137: 959–65. 15 Cambien F, Warnet J M, Eschwège E, Jacqueson A, Richard J C, Rosselin G. Body mass, blood pressure, glucose and lipids. Does plasma insulin explain their relationships? Arteriosclerosis 1987; 7: 197–202. 16 Bonora E, Zavaroni I, Alpi O et al. Relationship between blood pressure and plasma insulin in non-obese and obese nondiabetic subjects. Diabetologia 1987; 30: 719–23. 17 Saad M F, Knowler W C, Pettitt D J, Nelson R G, Mott D M, Bennett P H. Insulin and hypertension. Relationship to obesity and glucose intolerance in Pima Indians. Diabetes 1990; 39: 1430–35. 18 Collins V R, Dowse G K, Finch C F, Zimmet P Z. An inconsistent relationship between insulin and blood pressure in three Pacific island populations. J Clin Epidemiol 1990; 43: 1369–78. 19 Ferrannini E, Haffner S M, Stern M P. Insulin sensitivity and hypertension. J Hypertens 1990; 8: 169–73. 20 Raison J, Bonithon-Kopp C, Egloff M, Ducimetière P, Guy-Grand B. Hormonal influences on the relationships between body fatness, body fat distribution, lipids, lipoproteins, glucose and blood pressure in French working women. Atherosclerosis 1990; 85: 185–92. 21 Tappy L, Bovet P, Jéquier E, Shamlaye C, Darioli R, Burnand B. Relationship of fasting serum insulin concentrations with blood pressure in a representative sample of the adult population of the Seychelles. Int J Obes 1991; 15: 669–75. 22 Asch S, Wingard D L, Barrett-Connor E L. Are insulin and hypertension independently related? Ann Epidemiol 1991; 1: 231–44. 23 Muller D C, Elahi D, Pratley R E, Tobin J D, Andres R. An epidemiological test of the hyperinsulinemia-hypertension hypothesis. J Clin Endocrinol Metab 1993; 76: 544–48. 24 Niskanen L K, Uusitupa M I, Pyörälä K. The relationship of hyperinsulinemia to the development of hypertension in type 2 diabetic patients and in non-diabetic subjects. J Hum Hypertens 1991; 5: 155–59. 25 Skarfors E T, Lithell H O, Selinus I. Risk factors for the development of hypertension: a 10-year longitudinal study in middle-aged men. J Hypertens 1991; 9: 217–23. 26 Lissner L, Bengtsson C, Lapidus L, Kristjansson K, Wedel H. Fasting insulin in relation to subsequent blood pressure changes and hypertension in women. Hypertension 1992; 20: 797–801. 9

INSULIN AND HYPERTENSION 27

Haffner S M, Valdez R A, Hazuda H P, Mitchell B D, Morales P A, Stern M P. Prospective analysis of the insulinresistance syndrome (Syndrome X). Diabetes 1992; 41: 715–22. 28 Stern M P, Morales P A, Haffner S M, Valdez R A. Hyperdynamic circulation and the insulin resistance syndrome (‘Syndrome X’). Hypertension 1992; 20: 802–08. 29 Shetterly S M, Rewers M, Hamman R F, Marshall J A. Patterns and predictors of hypertension incidence among Hispanic and non-Hispanic Whites: The San Luis Valley Diabetes Study. J Hypertens 1994; 12: 1095–102. 30 Flack J M, Liu K, Savage P et al. Baseline insulin predicts 2 year blood pressure change in young adults: the CARDIA study. Circulation 1991; 83: 724. 31 Vaccaro O, Imperatore G, Iovino V, Iovine C, Rivellese A A, Riccardi G. Does impaired glucose tolerance predict hypertension? A prospective analysis. Diabetologia 1996; 39: 70–76. 32 Rosselin G E, Assan R, Yalow R S, Berson S A. Separation of antibody bound and unbound peptide hormones labelled with iodine 131 by talcum powder and precipitated silica. Nature 1966; 212–355. 33 Organisation Mondiale de la Santé. L’hypertension artérielle. Série de rapports techniques 1978; 628. 34 Chick J, Kreitman N, Plant M. Mean cell volume and gammaglutamyl-transpeptidase as markers of drinking in working men. Lancet 1981; i: 1249–51. 35 Rothman K J, Boice J C. Follow-up (cohort) studies. In: Epidemiologic Analysis with a Programmable Calculator. Washington, DC: National Institutes of Health (NIH publication 79–1649); 1979, pp. 11–17. 36 Cox D R. Regression models and life-tables (with discussion). J R Statist Soc B 1971; 34: 187–220. 37 Shaper A G, Wannamethee G, Macfarlane P W, Walker M. Heart rate, ischaemic heart disease, and sudden cardiac death in middle-aged British men. Br Heart J 1993; 70: 49–55. 38 Breslow N E, Day N E. Unconditional logistic regression for large strata. In: Statistical Methods in Cancer Research. Vol I. The Analysis of Case-Control Studies. Lyon: International Agency for Research on Cancer, IARC Publication No. 32, 1980, pp. 191–259. 39 Kalbfleisch J D, Prentice R L. Likelihood construction and further results on the proportional hazards model. In: The Statistical Analysis of Failure Time Data. New York: John Wiley & Sons, 1980, pp. 119–42. 40 Beatty O L, Harper R, Sheridan B, Atkinson A B, Bell P M. Insulin resistance in offspring of hypertensive parents. Br Med J 1993; 307: 92–96. 41 Ishibashi F. Higher serum insulin level due to greater total body fat mass in offspring of patients with essential hypertension. Diabetes Res Clin Pract 1993; 20: 63–68. 42 Grunfeld B, Balzareti M, Romo M, Gimenez M, Gutman R. Hyperinsulinemia in normotensive offspring of hypertensive parents. Hypertension 1994; 23 (Suppl. I): 12–15. 43 Ohno Y, Suzuki H, Yamakawa H, Nakamura M, Otsuka K, Saruta T. Impaired insulin sensitivity in young, lean normotensive offspring of essential hypertensives: possible role of disturbed calcium metabolism. J Hypertens 1993; 11: 421–26. 44 Allemann Y, Horber F F, Colombo M et al. Insulin sensitivity and body fat distribution in normotensive offspring of hypertensive parents. Lancet 1993; 341: 327–31.

45

549

Randrianjohany A, Balkau B, Cubeau J, Ducimetière P, Warnet J M, Eschwège E. The relationship between behavioural pattern, overall and central adiposity in a population of healthy French men. Int J Obes 1993; 17: 651–55. 46 Ducimetière P. Poids et mortalité. Données épidémiologiques récentes. 33èmes Journées Annuelles de Nutrition et de Diététique. 22 octobre 1992. 47 Wilking S B, Belanger A, Kannel W B, D’Agostino R B, Steel K. Determinants of isolated hypertension. JAMA 1988; 23: 3451–55. 48 Markovic N, Olomu I N, Bunker C H, Huston S L, Ukoli F A, Kuller L H. Adequacy of a single visit for classification of hypertensive status in a Nigerian civil servant population. Int J Epidemiol 1994; 23: 723–29. 49 Sackett D L. Bias in analytic research. J Chron Dis 1979; 32: 51–63. 50 Haffner S M, Mykkänen L, Stern M P, Valdez R A, Heisserman J A, Bowsher R R. Relationship of proinsulin and insulin to cardiovascular risk factors in nondiabetic subjects. Diabetes 1993; 42: 1297–302. 51 Nagi D K, Hendra T J, Ryle A J et al. The relationship of concentrations of insulin, intact proinsulin and 32–33 split proinsulin with cardiovascular risk factors in type 2 (non-insulin-dependent) diabetic subjects. Diabetologia 1990; 33: 532–37. 52 Haffner S M, Mykkänen L, Valdez R A, Stern M P. Evaluation of two insulin assays in insulin resistance syndrome (Syndrome X). Arterioscler Thromb 1994; 14: 1430–37. 53 Schwartz M W, Boyko E J, Kahn S E, Ravussin E, Bogardus C. Reduced insulin secretion: an independent predictor of body weight gain. J Clin Endocrinol Metab 1995; 80: 1571–76. 54 Atchley D W, Loeb R F, Richards D W, Benedict E M, Driscoll M E. On diabetic acidosis. J Clin Invest 1993; 12: 297–326. 55 Skott P, Hother-Nielsen O, Bruun N E et al. Effects of insulin on kidney function and sodium excretion in healthy subjects. Diabetologia 1989; 32: 694–99. 56 Skott P, Vaag A, Bruun N E et al. Effects of insulin on renal sodium handling in hyperinsulinaemic type 2 (non-insulindependent) diabetic patients with peripheral insulin resistance. Diabetologia 1991; 34: 275–81. 57 Hall J E, Brands M W, Mizelle H L, Gaillard C A, Hildebrandt D A. Chronic intrarenal hyperinsulinemia does not cause hypertension. Am J Physiol 1991; 260: 663–69. 58 Landsberg L. Hyperinsulinemia: possible role in obesityinduced hypertension. Hypertension 1992; 19: 161–66. 59 Anderson E A, Mark A L. The vasodilator action of insulin. Implications for the insulin hypothesis of hypertension. Hypertension 1993; 21: 136–40. 60 Baron A D, Brechtel-Hook G, Johnson A, Hardin D. Skeletal muscle blood flow. A possible link between insulin resistance and blood pressure. Hypertension 1993; 21: 129–35. 61 Laasko M, Edelman S V, Brechtel G, Baron A D. Decreased effect of insulin to stimulate skeletal muscle blood flow in obese man: A novel mechanism of insulin resistance. J Clin Invest 1990; 85: 1844–52. 62 Egan B M, Stepniakowski K. Compensatory hyperinsulinemia and the forearm vasodilator response during an oral glucose-tolerance test in obese hypertensives. J Hypertens 1994; 12: 1061–67.

550 63


Hall J E, Brands M W, Hildebrandt D A, Mizelle H L. Obesityassociated hypertension. Hyperinsulinemia and renal mechanisms. Hypertension 1992; 19: 45–55. 64 Filipovsky J, Ducimetière P, Eschwège E, Richard J L, Rosselin G, Claude J R. The relationship of blood pressure with glucose, insulin, heart rate, free fatty acids and plasma cortisol levels according to degree of obesity in middle-aged men. J Hypertens 1996; 14: 229–35. 65 Salomaa V V, Strandberg T E, Vanhanen H, Naukkarinen V, Sarna S, Miettinen T A. Glucose intolerance and blood

pressure: long term follow up in middle aged men. Br Med J 1991; 302: 493–96. 66 Ceriello A, Quatraro A, Giugliano D. Diabetes mellitus and hypertension: the possible role of hyperglycaemia through oxidative stress. Diabetologia 1993; 36: 265–66.

(Revised version received October 1996)