Relations of Insulin Sensitivity to Longitudinal Blood Pressure Tracking
http://www.100md.com
《循环学杂志》
the National Heart, Lung, and Blood Institute’s Framingham Heart Study (J.., M.J.P., B.H.N., C.S.F., D.L., R.B.D., R.S.V.), National Heart, Lung, and Blood Institute, Bethesda, Md (C.S.F., D.L.)
the Cardiology Section (R.S.V.), Preventive Medicine and Epidemiology (D.L., R.S.V.), Department of Medicine, Boston University School of Medicine, Boston, Mass
the Mathematics Department, Boston University, Boston, Mass (M.J.P., B.H.N., R.B.D.)
Massachusetts General Hospital, Harvard Medical School, Boston, Mass (J.B.M.)
the Section of Geriatrics, Department of Public Health and Caring Sciences, Uppsala University, Uppsala, Sweden (J..).
Abstract
Background— The relations of insulin sensitivity (IS) to hypertension incidence may vary according to baseline age, body mass index (BMI), and blood pressure (BP).
Methods and Results— We investigated the relations of IS (insulin sensitivity index, ISI0,120) to 4-year incidence of hypertension and BP progression in 1933 nonhypertensive Framingham Study participants (median age, 51 years; 56% women). Analyses were stratified by age (less than versus greater than or equal to median), BMI (<25 [normal], 25 to <30 [overweight], 30 kg/m2 [obese]), and BP category (systolic BP130 or diastolic BP85, "high normal" per the sixth report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High BP [JNC-VI] versus BP<130/85 mm Hg). On follow-up, 41% of participants had BP progression (1 BP stage increase) and 18% had development of hypertension (systolic BP140 or diastolic BP90 mm Hg or antihypertensive medication use). In younger (<51 years) people with normal BMI and baseline BP<130/85 mm Hg, the second-to-fourth ISI0,120 quartiles were associated with lower multivariable-adjusted odds for hypertension incidence (0.27; 95% CI, 0.09 to 0.83; P<0.05) and BP progression (0.37; 95% CI, 0.18 to 0.77; P<0.01) relative to the lowest (most insulin resistant) quartile. IS was not related to BP progression or hypertension incidence in older individuals, in obese participants, or in people with BP130/85 mm Hg.
Conclusions— In our large community-based sample, reduced IS predicted BP tracking principally in younger people with normal BMI and BP<130/85 mm Hg. Effect modification by age, BMI, and baseline BP may explain variation in the results of prior clinical investigations relating IS to hypertension incidence.
Key Words: blood pressure epidemiology hypertension insulin obesity
Introduction
There is an ongoing debate on whether insulin resistance is a cause or a consequence of hypertension or whether both conditions arise from a common substrate.1–20 Experimental evidence strongly suggests that insulin resistance can result in higher blood pressure (BP) through effects on the vasculature21–24 and the kidneys.25,26 However, clinical investigations have yielded less consistent results. Whereas several investigators have reported that insulin resistance is associated with future risk of hypertension,1–13 others report an attenuation of the association on adjustment for baseline BP or body mass index (BMI).14–19 One potential reason for the inconsistencies in the literature may be that some prior investigations were performed in small referral samples,4–7,10,11,14,19 predominantly male participants1,2,11,16,19, and with variable adjustment for potential confounders in the analyses.6–8,10–13,19,20 An alternative explanation may be that the impact of insulin resistance on the risk of development of hypertension may vary according to factors such as age, presence of obesity, and the level of baseline BP, being stronger in younger people,3,27 lean individuals15,17,28–31, and in those at the lower end of the BP spectrum. Because multiple risk factors for hypertension cluster in older individuals, obese people, and in those at the upper end of the BP spectrum,32 associations of insulin sensitivity with the development of hypertension may be more confounded and less discernible in these subgroups.
See p 1678
Accordingly, we investigated the relations of insulin sensitivity to the development of hypertension and BP tracking in a large community-based sample of middle-aged adults stratifying our analyses by age, baseline BMI, and BP level.
Methods
Study Sample
The design and selection criteria of the Framingham Heart Study have been previously described.33 We evaluated 3386 subjects who attended both the fifth (1991 to 1995) and the sixth (1995 to 1998) examination cycles of the Framingham Offspring Study. We excluded participants at examination cycle 5 for the following reasons: prevalent cardiovascular disease (n=44); diagnosed diabetes (n=127); hypertension (n=1143); and nonavailable oral glucose tolerance test (n=139). After exclusion, 1933 nonhypertensive participants remained eligible. All participants gave written informed consent, and the Institutional Review Board at Boston Medical Center approved the study protocol.
Baseline Clinical Examination and Laboratory Measurements (Examination Cycle 5)
At the baseline examination, all attendees had a physical examination (including medical history, BP examinations, and anthropometry) and laboratory assessment of vascular risk factors.
Height and weight were measured by using a standardized protocol, and BMI was calculated as weight in kilograms divided by square of height in meters. BMI categories were defined as follows: normal weight (BMI<25 kg/m2), overweight (BMI25 to <30 kg/m2), and obesity (BMI30 kg/m2).34 Waist circumference was measured to the nearest 0.25 inch while the participant was standing erect with weight equally distributed on both feet, and with the measuring tape placed at the level of the umbilicus.
Blood pressure was assessed as the average of 2 physician-obtained measurements taken on the left arm of subjects after they were seated for at least 5 minutes. The participants were classified according to their BP category defined by Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High BP (JNC-6):35 optimal (systolic BP<120 and diastolic BP<80 mm Hg), normal (systolic 120 to 129 or diastolic 80 to 84 mm Hg), and high-normal (systolic 130 to 139, or diastolic 85 to 89 mm Hg). Physical activity was assessed by using the Physical Activity Index,36 a measure of metabolic work during a typical day derived from responses to questions that ask participants how many hours per day they typically spend asleep, sitting, and involved in light, moderate, and heavy activities.
Assessment of Insulin Sensitivity
At the baseline examination, a standard 75-g oral glucose tolerance test was performed, with measurements of fasting and 2-hour postload plasma glucose and insulin levels. Plasma glucose was measured in fresh specimens with a hexokinase reagent kit (A-gent glucose test; Abbott). Glucose assays were run in duplicate with an intra-assay coefficient of variation of <3%. Fasting insulin was measured in plasma as total immunoreactive insulin (Coat-A-Count Insulin, Diagnostics Products Corp) and was standardized to serum levels for reporting purposes. The lower limit of sensitivity was 8.0 pmol/L (1.1 μU/mL), and the intra-assay and interassay coefficients of variation ranged from 5.0% to 10.0%.
We used the insulin sensitivity index developed by Gutt and coworkers,37 ISI0,120, which estimates the disposition of plasma glucose given body weight and ambient insulin levels, is highly correlated with insulin sensitivity as measured by the gold standard euglycemic hyperinsulinemic clamp-method, and is the best available surrogate measure of insulin resistance for prediction of incident type 2 diabetes.38 ISI0,120 was defined as (m/MPG)/log MSI, where m=[75 000 mg+(fasting glucose –2-hour glucose)x0.19xbody wt (kg)]/120 minutes, MPG=mean of fasting and 2-hour glucose concentrations (mg/dL), and MSI=mean of fasting and 2-hour serum insulin concentrations (mU/L); ISI0,120 is expressed as mg · L2/mmol · mU · min,37 and a low value indicates greater insulin resistance.
BP Outcomes at Follow-Up (Examination Cycle 6)
At the follow-up examination, approximately 4 years from baseline, participants underwent routine assessment of their BP, using the same standardized protocol as at the baseline examination. We defined the occurrence of 2 BP outcomes: (1) development of hypertension (systolic BP 140 mm Hg or diastolic BP 90 mm Hg, or the use of antihypertensive medication); and (2) BP progression (increase in BP by 1 category as defined by the JNC-6).35
Statistical Analyses
We investigated the relations of baseline ISI0,120 to the incidence of hypertension and BP progression over a 4-year follow-up period. Sex-specific quartiles of ISI0,120 were used as the predictor variables. We incorporated first-order statistical interaction terms into regression models to evaluate variation in the relations of ISI0,120 to BP outcomes according to sex, baseline age (dichotomized at the median of 51 years), BMI (normal weight, overweight, and obesity), and BP ([optimal or normal] versus high normal) categories. Statistically significant effect modification (P<0.05 for interactions) was observed by age, BMI, and BP categories but not by sex. Thus, in the 2 age groups (<51 versus 51 years), we performed sex-pooled analyses, further stratifying our sample by BMI and by BP categories. The optimal and normal BP categories were combined to provide adequate statistical power to evaluate the relation to hypertension incidence, given prior knowledge of hypertension incidence rates in different BP categories in our cohort.39
Additionally, within each age stratum we divided our sample further into 4 groups, based on the cross-classification of BMI and BP categories: (1) BMI <25 kg/m2 and optimal or normal BP, (2) BMI <25 kg/m2 and high normal BP, (3) BMI 25 kg/m2 and optimal or normal BP, and (4) BMI 25 kg/m2 and high normal BP. However, the number of subjects in the subgroups with high normal BP in the 2 age strata was low (n=83 and 31, respectively, in younger and older age strata for group 2; and n=188 and 92, respectively, in younger and older age strata for group 4). Hence, when we evaluated strata defined by both BP and BMI categories, the latter groups were not analyzed due to insufficient statistical power.
We used multiple logistic regression40 to relate ISI0,120 to hypertension incidence and BP progression, constructing models in a hierarchical fashion adjusting for age and sex (model A); age, sex, and BMI (model B); age, sex, BMI, and systolic and diastolic BP (model C); age, sex, BMI, systolic and diastolic BP, and weight change during follow-up (model D)
As secondary analyses, we also evaluated the following additional models: age, sex, waist circumference (instead of BMI), and systolic and diastolic BP (model E); age, sex, BMI, level of physical activity, and systolic and diastolic BP (model F).
We chose a strategy of sequentially adding covariates because other investigators have suggested that adjusting for either baseline BP or BMI may represent "overadjustment" because both these factors are cross-sectionally related to insulin sensitivity and may represent intermediate states along the causal pathway from insulin resistance to hypertension.12,17 All of the above confounders, except for sex, were used as continuous variables in the analyses.
We prespecified 2 kinds of analyses: (1) models comparing the odds of developing BP outcomes in each of quartiles 2 to 4, with the first quartile (the most insulin-resistant quartile) serving as a referent. These models assessed the possibility of a continuous relation between ISI0,120 and BP tracking; (2) models comparing quartile 1 with quartiles 2 through 4 (combined) because quartile 1 indicates the group with lowest insulin sensitivity. These models assessed a potential threshold effect, where the ISI0,120 quartile with the greatest insulin resistance41 (corresponding to lowest insulin sensitivity) is associated with BP tracking.
For analyses of different subgroups, we used the sex-specific quartiles defined for the entire sample. We performed additional analyses in which we assessed the relation between ISI0,120 and changes in systolic and diastolic BP (modeled as continuous variables separately) during follow-up. In these analyses, subjects with hypertension treatment at follow-up were excluded (n=146). Furthermore, we also investigated if fasting plasma glucose by itself was a predictor of hypertension incidence and BP progression.
Odds ratios and their 95% confidence intervals are presented. A 2-tailed probability value of <0.05 was considered statistically significant in all analyses. All analyses were performed with SAS version 8 for Windows (SAS Inc).
Results
The baseline characteristics of our study sample including the sex-specific distribution of ISI0,120 are displayed in Table 1. ISI0,120 was significantly correlated with BMI (Spearman rank r=–0.26, P<0.001), waist circumference (r=–0.24, P<0.001), and fasting glucose (r=–0.40, P<0.001) at baseline.
On follow-up 4 years from baseline, 351 participants (18.2%) had hypertension (168 men) and 786 subjects (40.7%) had an increase in BP by one or more BP categories (367 men). The incidence of hypertension and BP progression decreased with increasing quartiles of ISI0,120, that is, with increasing insulin sensitivity (Table 2).
Insulin Sensitivity and BP Outcomes: Relations in the Whole Sample
In the whole sample, the odds of development of hypertension or BP progression decreased with increasing ISI0,120 quartiles in multivariable models A and B (Table 2). However, adjustment for baseline systolic and diastolic BP attenuated the relation between ISI0,120 and hypertension incidence and BP progression, rendering it statistically nonsignificant (Table 2, model C).
As noted previously, the interaction terms for baseline age, BMI, and baseline BP category with ISI0,120 were significant (P<0.05). Accordingly, we stratified further analyses by age, BMI, and BP categories.
Insulin Sensitivity and BP Outcomes: Relations Within BMI Strata
Among younger individuals with a normal BMI (<51 years and BMI <25 kg/m2), those in quartiles 2 through 4 of ISI0,120 had a statistically significant lower odds for the development of hypertension in models A and B (Table 3, Q2-Q4 versus Q1; P<0.01) and a lower odds of BP progression in all models (P<0.05).
In the younger overweight group (<51 years and BMI >25 to <30 kg/m2), participants in quartiles 2 through 4 of ISI0,120 had a statistically significant lower odds for the development of hypertension in models A and B (Table 3, Q2-Q4 versus Q1; P<0.01) but no relation to BP progression in any of the models.
In the older normal BMI group (>51 years and BMI <25 kg/m2), participants in quartiles 2 through 4 of ISI0,120 had a statistically significant lower odds for the development of hypertension in models A and B (Table 3, Q2-Q4 versus Q1; P<0.01 and P<0.05, respectively). No significant relation was observed between ISI0,120 and BP progression in this subgroup.
No significant relations between ISI0,120 and hypertension incidence or BP progression were observed in older overweight or obese individuals (>51 years and BMI>25 kg/m2, Table 3).
Insulin Sensitivity and BP Outcomes: Relations Within BP Strata
Among younger participants (<51 years) with optimal/normal BP, those in quartiles 2 through 4 of ISI0,120 had a statistically significant lower odds for the development of hypertension in all models (Table 4, Q2-Q4 versus Q1; P<0.01) and the odds of development of hypertension or BP progression was lower in the top quartile of ISI0,120 compared with quartile 1 in all models (Table 4, P<0.05).
No significant relations between the quartiles of ISI0,120 and hypertension incidence or BP progression were observed in individuals with high normal BP regardless of age or in older participants with optimal/normal BP (Table 4).
Insulin Sensitivity and BP Outcomes: Relations Within Combined BMI and BP Strata
In younger participants with BMI <25 kg/m2 and an optimal/normal BP, individuals in quartile 2 to 4 of ISI0,120 had a significantly lower odds ratios for hypertension incidence in all models (Table 5, model A, P<0.01 and models B and C P<0.05) and lower odds of BP progression in all models (P<0.01 for all models).
In younger participants with BMI >25 kg/m2 and an optimal/normal BP, subjects in quartile 2 to 4 of ISI0,120 had a significantly lower odds ratios for hypertension incidence in models A and B (Table 5, P<0.05) that became borderline significant in model C (P<0.06) No significant relations were observed between ISI0,120 and BP progression in this subgroup.
No significant relations between the quartiles of ISI0,120 and hypertension incidence and BP progression were noted in older participants, regardless of BMI status (Table 5).
Secondary Analyses
The results remained essentially unchanged after adjusting for weight change on follow-up, the baseline level of physical activity, and on adjusting for waist circumference instead of BMI in model C (models D, E, and F; data not shown). Fasting plasma glucose did not independently predict either hypertension incidence or BP progression in any subgroup (data not shown).
Models Evaluating BP Change as a Continuous Variable
In the models treating BP change as a continuous variable (and restricted to participants not on treatment on follow-up), the relations between ISI0,120 and increases in systolic BP displayed a pattern consistent with the analyses of BP change as a categorical variable. A 1-SD increment in ISI0,120 was associated with a 1-mm lower systolic BP on follow-up in younger participants (<51 years; P<0.05 in all models, using pooled BMI and BP categories) but not in older participants. We performed exploratory analyses cross-classifying individuals according to BMI and BP categories. In younger participants with a normal BMI and optimal/normal BP at baseline, those in ISI0,120 quartiles 2 to 4 had approximately 4 mm Hg lower systolic BP increase compared with people in quartile 1 independent of age, sex, BMI, and baseline systolic BP (P<0.05 in all models). We did not observe any association of ISI0,120 with BP change in any of the other subgroups defined on the basis of age, BMI, and BP category. ISI0,120 was not associated with changes in diastolic BP (assessed as a continuous variable) in any of the groups evaluated.
Statistical Power
Because we found no independent associations of ISI0,120 with BP outcomes in older individuals, in those with high normal BP and in those with BMI 25 kg/m2, we assessed our statistical power to detect relations in different strata with an -level of 0.05. In the subgroup of old individuals with BMI <25 kg/m2 and optimal/normal BP, we had >80% power to detect an odds ratio of 0.45 or lower for BP progression (quartile 2 to 4 versus quartile 1). In the subgroups of individuals with BMI 25 kg/m2 and optimal/normal BP, we had >80% power to detect an odds ratio of 0.49 (younger participants) and 0.56 (older individuals) for BP progression. In older people with BMI 25 kg/m2 and an optimal/normal BP, we had >80% power to detect an odds ratio of 0.41 for BP progression. As noted earlier, we did not analyze select individuals with high normal BP further stratified by BMI and age because of an inadequate number of participants.
Discussion
Principal Findings
We observed evidence of varying relations of insulin resistance and BP tracking according to age, BMI, and baseline BP. An independent inverse association of ISI0,120 with BP outcomes was evident principally in younger people with a normal BMI and a BP <130/85 mm Hg. Insulin sensitivity was not associated with BP progression or hypertension incidence in the older age group (regardless of baseline BMI or BP), in obese participants (regardless of age or BP), or in people with BP 130/85 mm Hg (regardless of age and BMI). We had adequate statistical power to detect an odds ratio of 0.5 (quartile 2 to 4 versus quartile 1) in all subgroups analyzed (with the exception of the BMI- and BP-stratified groups for people with high normal BP). It is noteworthy that adjustment for BMI or waist circumference did not attenuate the association of insulin sensitivity with BP progression in several of the subgroups. We noted only modest correlations of insulin sensitivity with BMI and waist circumference in our sample that probably contributed to this observation.
Comparison With the Literature
As noted previously, both positive1–13 and null/less positive14–19 associations of insulin resistance have been reported in the literature. Our investigation suggests that effect modification by age, BMI, and BP is an important factor to be considered when relating measures of insulin sensitivity to longitudinal BP tracking. An association of lesser insulin sensitivity with greater BP progression is more likely to be observed in samples that consist of younger, lean people with lower BP levels.
Our finding that insulin sensitivity mainly predicts BP progression in younger participants (<51 years) is consistent with some previous reports3,20,27 and supports the thesis that that the effects of insulin sensitivity are more important early along the natural history of BP progression.
Our observation of an association of insulin sensitivity with BP tracking in those with a normal BMI but not in overweight/obese people is consistent with some prior cross-sectional and prospective reports15,17,28–31 but differs from some other studies.2,13 There are 3 possible explanations for these findings. First, insulin resistance may only contribute to BP progression in lean individuals. The well-established increased risk of BP progression observed in overweight and obese individuals may be explained by factors other than insulin resistance itself, such as hemodynamic adaptations to a greater body mass, increased circulating levels of leptin and free fatty acids, or increased sympathetic activation caused by obstructive sleep apnea.42 Second, insulin resistance may contribute less to BP progression in overweight and obese individuals compared with its contribution in normal-weight people. As several other factors associated with BP progression cluster in overweight and obese individuals,42 factors other than insulin resistance may assume a greater importance for BP progression in these individuals. Third, insulin resistance may contribute equally to BP progression in lean and overweight/obese people. However, the overweight and obese individuals may have been insulin resistant for a longer period of time (compared with people with normal weight). The long-term influences of insulin resistance may result in vascular hypertrophy and remodeling and renal effects; these chronic alterations may mask the contribution of insulin resistance to BP progression in the overweight and obese. Further studies are warranted to understand the mechanisms behind the divergent associations of insulin sensitivity and BP progression in normal-weight versus overweight/obese individuals.
Our finding of a longitudinal association between insulin sensitivity and BP outcomes principally in normal-weight individuals at the lower end of the BP spectrum merits comment. For subjects at the higher end of the baseline BP spectrum, it is possible that factors other than insulin resistance regulate BP progression (such as impaired vascular reactivity, elevated vascular stiffness, renal hemodynamic responses to insulin, and so forth) and that once higher BP develops, the mechanisms affecting BP progression may become insensitive to the systemic effects of insulin resistance. It is also conceivable that a relation between insulin sensitivity and BP tracking is more discernible in individuals at the lower end of the BP distribution because of a lesser degree of confounding by factors associated with both insulin sensitivity and BP in obese individuals and in those with high normal BP.
Mechanisms Underlying Association of Insulin Sensitivity and BP Tracking
There are several mechanisms by which impaired insulin sensitivity may predispose to hypertension. Experimental data suggest that insulin resistance and compensatory hyperinsulinemia can increase BP through effects on the renin-angiotensin-aldosterone system,43 the vascular smooth muscle cell membrane,21–23 smooth muscle proliferation,24 increased sympathetic tone,44–46 and through effects of insulin on sodium retention.25,26 A potential causal link between insulin sensitivity and BP is also suggested by the observation that insulin-sensitizing agents (thiazolidinediones) can lower BP.47
Strengths and Limitations
The strengths of our study include the large community-based sample of nonhypertensive participants, the standardized measurements of BP and BMI, and the multivariable analyses adjusting for factors known to influence insulin sensitivity and BP progression. The use of ISI0,120 is an additional strength of our investigation. Most prior reports evaluating relations of insulin resistance to the development of hypertension used fasting plasma insulin as a measure of insulin sensitivity. Fasting insulin may be limited as a marker of insulin sensitivity because it reflects not only insulin sensitivity but also indicates insulin secretion. Measurements of insulin sensitivity that also include post–glucose load plasma levels of insulin and glucose are closely correlated to insulin sensitivity, as measured by the gold standard euglycemic hyperinsulinemic clamp-method.37,48 Furthermore, it has been shown that ISI0,120 predicts diabetes better than any of the other indirect measurements of insulin sensitivity.38
Our study has several limitations. We did not measure insulin sensitivity by directly using the clamp technique, given the constraints of a large epidemiological study, although ISI0,120 correlates very well with the clamp, as noted above. Our sample was predominantly middle-aged and white, limiting the generalizability of our results to other age groups and ethnicities. Other reports suggest that there may be age- and ethnicity-related differences in the relations between insulin sensitivity and hypertension incidence3,9,15,20; stronger relations have been observed in younger people and in whites. It is important to note that multiple statistical tests were performed in our investigation. However, all analyses were prespecified, and the relative consistency of results across the different models would argue against a type I error as an explanation for our findings.
Conclusions
In our large, community-based sample of nonhypertensive individuals, impaired insulin sensitivity predicted hypertension incidence and BP tracking principally in younger people with normal BMI and optimal/normal BP. These longitudinal associations were independent of baseline BP and BMI. Additional studies are warranted to confirm our findings.
Acknowledgments
This study was supported by a Bergmarks travel grant, Viking Bjrks Hedersledamotstipendium, a Capio travel grant (Dr rnlv), and by research grants (NHLBI contract N01-HC-25195, and 2K24 HL04334 [Dr Vasan]) from the National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, Md. Dr Meigs was supported by a career development award from the American Diabetes Association.
Footnotes
Guest Editor for this article was Robert H. Eckel, MD.
References
Fagot-Campagna A, Balkau B, Simon D, Ducimetiere P, Eschwege E. Is insulin an independent risk factor for hypertension The Paris Prospective Study. Int J Epidemiol. 1997; 26: 542–550.
Salonen JT, Lakka TA, Lakka HM, Valkonen VP, Everson SA, Kaplan GA. Hyperinsulinemia is associated with the incidence of hypertension and dyslipidemia in middle-aged men. Diabetes. 1998; 47: 270–275.
Dyer AR, Liu K, Walsh M, Kiefe C, Jacobs DR Jr, Bild DE. Ten-year incidence of elevated blood pressure and its predictors: the CARDIA study: Coronary Artery Risk Development in (Young) Adults. J Hum Hypertens. 1999; 13: 13–21.
Imazu M, Yamamoto H, Toyofuku M, Sumii K, Okubo M, Egusa G, Yamakido M, Kohno N. Hyperinsulinemia for the development of hypertension: data from the Hawaii-Los Angeles-Hiroshima Study. Hypertens Res. 2001; 24: 531–536.
Tsuruta M, Hashimoto R, Adachi H, Imaizumi T, Nomura G. Hyperinsulinaemia as a predictor of hypertension: an 11-year follow-up study in Japan. J Hypertens. 1996; 14: 483–488.
Niskanen LK, Uusitupa MI, Pyorala K. The relationship of hyperinsulinaemia to the development of hypertension in type 2 diabetic patients and in non-diabetic subjects. J Hum Hypertens. 1991; 5: 155–159.
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.
Haffner SM, Valdez RA, Hazuda HP, Mitchell BD, Morales PA, Stern MP. Prospective analysis of the insulin-resistance syndrome (syndrome X). Diabetes. 1992; 41: 715–722.
Mitchell BD, Haffner SM, Hazuda HP, Valdez R, Stern MP. The relation between serum insulin levels and 8-year changes in lipid, lipoprotein, and blood pressure levels. Am J Epidemiol. 1992; 136: 12–22.
Kashiwabara H, Inaba M, Maruno Y, Morita T, Awata T, Negishi K, Iitaka M, Katayama S. Insulin levels during fasting and the glucose tolerance test and Homa’s index predict subsequent development of hypertension. J Hypertens. 2000; 18: 83–88.
Hirose H, Saito I, Kawabe H, Saruta T. Insulin resistance and hypertension: seven-year follow-up study in middle-aged Japanese men (the KEIO study). Hypertens Res. 2003; 26: 795–800.
Goff DC Jr, Zaccaro DJ, Haffner SM, Saad MF. Insulin sensitivity and the risk of incident hypertension: insights from the Insulin Resistance Atherosclerosis Study. Diabetes Care. 2003; 26: 805–809.
Haffner SM, Miettinen H, Gaskill SP, Stern MP. Metabolic precursors of hypertension: the San Antonio Heart Study. Arch Intern Med. 1996; 156: 1994–2001.
Vaccaro O, Imperatore G, Iovino V, Iovine C, Rivellese AA, Riccardi G. Does impaired glucose tolerance predict hypertension A prospective analysis. Diabetologia. 1996; 39: 70–76.
Shetterly SM, Rewers M, Hamman RF, Marshall JA. Patterns and predictors of hypertension incidence among Hispanics and non-Hispanic whites: the San Luis Valley Diabetes Study. J Hypertens. 1994; 12: 1095–1102.
Skarfors ET, Lithell HO, Selinus I. Risk factors for the development of hypertension: a 10-year longitudinal study in middle-aged men. J Hypertens. 1991; 9: 217–223.
Liese AD, Mayer-Davis EJ, Chambless LE, Folsom AR, Sharrett AR, Brancati FL, Heiss G. Elevated fasting insulin predicts incident hypertension: the ARIC study: Atherosclerosis Risk in Communities Study Investigators. J Hypertens. 1999; 17: 1169–1177.
Nesbitt SD, Amerena JV, Grant E, Jamerson KA, Lu H, Weder A, Julius S. Home blood pressure as a predictor of future blood pressure stability in borderline hypertension: The Tecumseh Study. Am J Hypertens. 1997; 10: 1270–1280.
Mattiasson I, Endre T, Hulthen UL. Insulin sensitivity with respect to glucose metabolism in hypertension-prone men did not predict the blood pressure increase in 5 years. J Hum Hypertens. 2001; 15: 781–785.
Jiang X, Srinivasan SR, Bao W, Berenson GS. Association of fasting insulin with longitudinal changes in blood pressure in children and adolescents: the Bogalusa Heart Study. Am J Hypertens. 1993; 6: 564–569.
Aviv A. The roles of cell Ca2+, protein kinase C and the Na(+)-H+ antiport in the development of hypertension and insulin resistance. J Am Soc Nephrol. 1992; 3: 1049–1063.
Ferrannini E, Taddei S, Santoro D, Natali A, Boni C, Del Chiaro D, Buzzigoli G. Independent stimulation of glucose metabolism and Na+-K+ exchange by insulin in the human forearm. Am J Physiol. 1988; 255: E953–E958.
Standley PR, Rose KA, Sowers JR. Increased basal arterial smooth muscle glucose transport in the Zucker rat. Am J Hypertens. 1995; 8: 48–52.
Stout RW, Bierman EL, Ross R. Effect of insulin on the proliferation of cultured primate arterial smooth muscle cells. Circ Res. 1975; 36: 319–327.
DeFronzo RA. The effect of insulin on renal sodium metabolism: a review with clinical implications. Diabetologia. 1981; 21: 165–171.
Skott P, Hother-Nielsen O, Bruun NE, Giese J, Nielsen MD, Beck-Nielsen H, Parving HH. Effects of insulin on kidney function and sodium excretion in healthy subjects. Diabetologia. 1989; 32: 694–699.
Fossum E, Hoieggen A, Reims HM, Moan A, Rostrup M, Eide I, Kjeldsen SE. High screening blood pressure is related to sympathetic nervous system activity and insulin resistance in healthy young men. Blood Press. 2004; 13: 89–94.
Bonora E, Zavaroni I, Alpi O, Pezzarossa A, Bruschi F, Dall’Aglio E, Guerra L, Coscelli C, Butturini U. Relationship between blood pressure and plasma insulin in non-obese and obese non-diabetic subjects. Diabetologia. 1987; 30: 719–723.
Grandi AM, Gaudio G, Fachinetti A, Zanzi P, Bianchi L, Ceriani L, Guasti L, Venco A. Influence of family history of hypertension on insulin sensitivity in lean and obese hypertensive subjects. Eur J Clin Invest. 1997; 27: 774–779.
Mykkanen L, Haffner SM, Ronnemaa T, Watanabe RM, Laakso M. Relationship of plasma insulin concentration and insulin sensitivity to blood pressure: is it modified by obesity J Hypertens. 1996; 14: 399–405.
Haffner SM, Ferrannini E, Hazuda HP, Stern MP. Clustering of cardiovascular risk factors in confirmed prehypertensive individuals. Hypertension. 1992; 20: 38–45.
Muntner P, He J, Roccella EJ, Whelton PK. The impact of JNC-VI guidelines on treatment recommendations in the US population. Hypertension. 2002; 39: 897–902.
Kannel WB, Feinleib M, McNamara PM, Garrison RJ, Castelli WP. An investigation of coronary heart disease in families: the Framingham offspring study. Am J Epidemiol. 1979; 110: 281–290.
Clinical guidelines on the identification, evaluation, and treatment of overweight and obesity in adults. NIH Publication No. 98–4083. 1998. National Institutes of Health.
Sixth report of the Joint National Committee on prevention, detection, evaluation, and treatment of high blood pressure. Arch Intern Med. 1997; 157: 2413–2446.
Kannel WB, Belanger A, D’Agostino R, Israel I. Physical activity and physical demand on the job and risk of cardiovascular disease and death: the Framingham Study. Am Heart J. 1986; 112: 820–825.
Gutt M, Davis CL, Spitzer SB, Llabre MM, Kumar M, Czarnecki EM, Schneiderman N, Skyler JS, Marks JB. Validation of the insulin sensitivity index (ISI[0,120]): comparison with other measures. Diabetes Res Clin Pract. 2000; 47: 177–184.
Hanley AJ, Williams K, Gonzalez C, D’Agostino RB Jr, Wagenknecht LE, Stern MP, Haffner SM. Prediction of type 2 diabetes using simple measures of insulin resistance: combined results from the San Antonio Heart Study, the Mexico City Diabetes Study, and the Insulin Resistance Atherosclerosis Study. Diabetes. 2003; 52: 463–469.
Vasan RS, Larson MG, Leip EP, Kannel WB, Levy D. Assessment of frequency of progression to hypertension in non-hypertensive participants in the Framingham Heart Study: a cohort study. Lancet. 2001; 358: 1682–1686.
Hosmer DW, Lemeshow S. Applied Logistic Regression. New York, NY: John Wiley & Sons Inc, 1989.
Balkau B, Charles MA. Comment on the provisional report from the WHO consultation. European Group for the Study of Insulin Resistance (EGIR). Diabet Med. 1999; 16: 442–443.
Montani JP, Antic V, Yang Z, Dulloo A. Pathways from obesity to hypertension: from the perspective of a vicious triangle. Int J Obes Relat Metab Disord. 2002; 26 (suppl 2): S28–S38.
Sowers JR, Frohlich ED. Insulin and insulin resistance: impact on blood pressure and cardiovascular disease. Med Clin North Am. 2004; 88: 63–82.
Rowe JW, Young JB, Minaker KL, Stevens AL, Pallotta J, Landsberg L. Effect of insulin and glucose infusions on sympathetic nervous system activity in normal man. Diabetes. 1981; 30: 219–225.
Young JB, Landsberg L. Stimulation of the sympathetic nervous system during sucrose feeding. Nature. 1977; 269: 615–617.
Troisi RJ, Weiss ST, Parker DR, Sparrow D, Young JB, Landsberg L. Relation of obesity and diet to sympathetic nervous system activity. Hypertension. 1991; 17: 669–677.
Gilling L, Suwattee P, DeSouza C, Asnani S, Fonseca V. Effects of the thiazolidinediones on cardiovascular risk factors. Am J Cardiovasc Drugs. 2002; 2: 149–156.
Soonthornpun S, Setasuban W, Thamprasit A, Chayanunnukul W, Rattarasarn C, Geater A. Novel insulin sensitivity index derived from oral glucose tolerance test. J Clin Endocrinol Metab. 2003; 88: 1019–1023.(Johan rnlv, MD, PhD; Mich)
the Cardiology Section (R.S.V.), Preventive Medicine and Epidemiology (D.L., R.S.V.), Department of Medicine, Boston University School of Medicine, Boston, Mass
the Mathematics Department, Boston University, Boston, Mass (M.J.P., B.H.N., R.B.D.)
Massachusetts General Hospital, Harvard Medical School, Boston, Mass (J.B.M.)
the Section of Geriatrics, Department of Public Health and Caring Sciences, Uppsala University, Uppsala, Sweden (J..).
Abstract
Background— The relations of insulin sensitivity (IS) to hypertension incidence may vary according to baseline age, body mass index (BMI), and blood pressure (BP).
Methods and Results— We investigated the relations of IS (insulin sensitivity index, ISI0,120) to 4-year incidence of hypertension and BP progression in 1933 nonhypertensive Framingham Study participants (median age, 51 years; 56% women). Analyses were stratified by age (less than versus greater than or equal to median), BMI (<25 [normal], 25 to <30 [overweight], 30 kg/m2 [obese]), and BP category (systolic BP130 or diastolic BP85, "high normal" per the sixth report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High BP [JNC-VI] versus BP<130/85 mm Hg). On follow-up, 41% of participants had BP progression (1 BP stage increase) and 18% had development of hypertension (systolic BP140 or diastolic BP90 mm Hg or antihypertensive medication use). In younger (<51 years) people with normal BMI and baseline BP<130/85 mm Hg, the second-to-fourth ISI0,120 quartiles were associated with lower multivariable-adjusted odds for hypertension incidence (0.27; 95% CI, 0.09 to 0.83; P<0.05) and BP progression (0.37; 95% CI, 0.18 to 0.77; P<0.01) relative to the lowest (most insulin resistant) quartile. IS was not related to BP progression or hypertension incidence in older individuals, in obese participants, or in people with BP130/85 mm Hg.
Conclusions— In our large community-based sample, reduced IS predicted BP tracking principally in younger people with normal BMI and BP<130/85 mm Hg. Effect modification by age, BMI, and baseline BP may explain variation in the results of prior clinical investigations relating IS to hypertension incidence.
Key Words: blood pressure epidemiology hypertension insulin obesity
Introduction
There is an ongoing debate on whether insulin resistance is a cause or a consequence of hypertension or whether both conditions arise from a common substrate.1–20 Experimental evidence strongly suggests that insulin resistance can result in higher blood pressure (BP) through effects on the vasculature21–24 and the kidneys.25,26 However, clinical investigations have yielded less consistent results. Whereas several investigators have reported that insulin resistance is associated with future risk of hypertension,1–13 others report an attenuation of the association on adjustment for baseline BP or body mass index (BMI).14–19 One potential reason for the inconsistencies in the literature may be that some prior investigations were performed in small referral samples,4–7,10,11,14,19 predominantly male participants1,2,11,16,19, and with variable adjustment for potential confounders in the analyses.6–8,10–13,19,20 An alternative explanation may be that the impact of insulin resistance on the risk of development of hypertension may vary according to factors such as age, presence of obesity, and the level of baseline BP, being stronger in younger people,3,27 lean individuals15,17,28–31, and in those at the lower end of the BP spectrum. Because multiple risk factors for hypertension cluster in older individuals, obese people, and in those at the upper end of the BP spectrum,32 associations of insulin sensitivity with the development of hypertension may be more confounded and less discernible in these subgroups.
See p 1678
Accordingly, we investigated the relations of insulin sensitivity to the development of hypertension and BP tracking in a large community-based sample of middle-aged adults stratifying our analyses by age, baseline BMI, and BP level.
Methods
Study Sample
The design and selection criteria of the Framingham Heart Study have been previously described.33 We evaluated 3386 subjects who attended both the fifth (1991 to 1995) and the sixth (1995 to 1998) examination cycles of the Framingham Offspring Study. We excluded participants at examination cycle 5 for the following reasons: prevalent cardiovascular disease (n=44); diagnosed diabetes (n=127); hypertension (n=1143); and nonavailable oral glucose tolerance test (n=139). After exclusion, 1933 nonhypertensive participants remained eligible. All participants gave written informed consent, and the Institutional Review Board at Boston Medical Center approved the study protocol.
Baseline Clinical Examination and Laboratory Measurements (Examination Cycle 5)
At the baseline examination, all attendees had a physical examination (including medical history, BP examinations, and anthropometry) and laboratory assessment of vascular risk factors.
Height and weight were measured by using a standardized protocol, and BMI was calculated as weight in kilograms divided by square of height in meters. BMI categories were defined as follows: normal weight (BMI<25 kg/m2), overweight (BMI25 to <30 kg/m2), and obesity (BMI30 kg/m2).34 Waist circumference was measured to the nearest 0.25 inch while the participant was standing erect with weight equally distributed on both feet, and with the measuring tape placed at the level of the umbilicus.
Blood pressure was assessed as the average of 2 physician-obtained measurements taken on the left arm of subjects after they were seated for at least 5 minutes. The participants were classified according to their BP category defined by Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High BP (JNC-6):35 optimal (systolic BP<120 and diastolic BP<80 mm Hg), normal (systolic 120 to 129 or diastolic 80 to 84 mm Hg), and high-normal (systolic 130 to 139, or diastolic 85 to 89 mm Hg). Physical activity was assessed by using the Physical Activity Index,36 a measure of metabolic work during a typical day derived from responses to questions that ask participants how many hours per day they typically spend asleep, sitting, and involved in light, moderate, and heavy activities.
Assessment of Insulin Sensitivity
At the baseline examination, a standard 75-g oral glucose tolerance test was performed, with measurements of fasting and 2-hour postload plasma glucose and insulin levels. Plasma glucose was measured in fresh specimens with a hexokinase reagent kit (A-gent glucose test; Abbott). Glucose assays were run in duplicate with an intra-assay coefficient of variation of <3%. Fasting insulin was measured in plasma as total immunoreactive insulin (Coat-A-Count Insulin, Diagnostics Products Corp) and was standardized to serum levels for reporting purposes. The lower limit of sensitivity was 8.0 pmol/L (1.1 μU/mL), and the intra-assay and interassay coefficients of variation ranged from 5.0% to 10.0%.
We used the insulin sensitivity index developed by Gutt and coworkers,37 ISI0,120, which estimates the disposition of plasma glucose given body weight and ambient insulin levels, is highly correlated with insulin sensitivity as measured by the gold standard euglycemic hyperinsulinemic clamp-method, and is the best available surrogate measure of insulin resistance for prediction of incident type 2 diabetes.38 ISI0,120 was defined as (m/MPG)/log MSI, where m=[75 000 mg+(fasting glucose –2-hour glucose)x0.19xbody wt (kg)]/120 minutes, MPG=mean of fasting and 2-hour glucose concentrations (mg/dL), and MSI=mean of fasting and 2-hour serum insulin concentrations (mU/L); ISI0,120 is expressed as mg · L2/mmol · mU · min,37 and a low value indicates greater insulin resistance.
BP Outcomes at Follow-Up (Examination Cycle 6)
At the follow-up examination, approximately 4 years from baseline, participants underwent routine assessment of their BP, using the same standardized protocol as at the baseline examination. We defined the occurrence of 2 BP outcomes: (1) development of hypertension (systolic BP 140 mm Hg or diastolic BP 90 mm Hg, or the use of antihypertensive medication); and (2) BP progression (increase in BP by 1 category as defined by the JNC-6).35
Statistical Analyses
We investigated the relations of baseline ISI0,120 to the incidence of hypertension and BP progression over a 4-year follow-up period. Sex-specific quartiles of ISI0,120 were used as the predictor variables. We incorporated first-order statistical interaction terms into regression models to evaluate variation in the relations of ISI0,120 to BP outcomes according to sex, baseline age (dichotomized at the median of 51 years), BMI (normal weight, overweight, and obesity), and BP ([optimal or normal] versus high normal) categories. Statistically significant effect modification (P<0.05 for interactions) was observed by age, BMI, and BP categories but not by sex. Thus, in the 2 age groups (<51 versus 51 years), we performed sex-pooled analyses, further stratifying our sample by BMI and by BP categories. The optimal and normal BP categories were combined to provide adequate statistical power to evaluate the relation to hypertension incidence, given prior knowledge of hypertension incidence rates in different BP categories in our cohort.39
Additionally, within each age stratum we divided our sample further into 4 groups, based on the cross-classification of BMI and BP categories: (1) BMI <25 kg/m2 and optimal or normal BP, (2) BMI <25 kg/m2 and high normal BP, (3) BMI 25 kg/m2 and optimal or normal BP, and (4) BMI 25 kg/m2 and high normal BP. However, the number of subjects in the subgroups with high normal BP in the 2 age strata was low (n=83 and 31, respectively, in younger and older age strata for group 2; and n=188 and 92, respectively, in younger and older age strata for group 4). Hence, when we evaluated strata defined by both BP and BMI categories, the latter groups were not analyzed due to insufficient statistical power.
We used multiple logistic regression40 to relate ISI0,120 to hypertension incidence and BP progression, constructing models in a hierarchical fashion adjusting for age and sex (model A); age, sex, and BMI (model B); age, sex, BMI, and systolic and diastolic BP (model C); age, sex, BMI, systolic and diastolic BP, and weight change during follow-up (model D)
As secondary analyses, we also evaluated the following additional models: age, sex, waist circumference (instead of BMI), and systolic and diastolic BP (model E); age, sex, BMI, level of physical activity, and systolic and diastolic BP (model F).
We chose a strategy of sequentially adding covariates because other investigators have suggested that adjusting for either baseline BP or BMI may represent "overadjustment" because both these factors are cross-sectionally related to insulin sensitivity and may represent intermediate states along the causal pathway from insulin resistance to hypertension.12,17 All of the above confounders, except for sex, were used as continuous variables in the analyses.
We prespecified 2 kinds of analyses: (1) models comparing the odds of developing BP outcomes in each of quartiles 2 to 4, with the first quartile (the most insulin-resistant quartile) serving as a referent. These models assessed the possibility of a continuous relation between ISI0,120 and BP tracking; (2) models comparing quartile 1 with quartiles 2 through 4 (combined) because quartile 1 indicates the group with lowest insulin sensitivity. These models assessed a potential threshold effect, where the ISI0,120 quartile with the greatest insulin resistance41 (corresponding to lowest insulin sensitivity) is associated with BP tracking.
For analyses of different subgroups, we used the sex-specific quartiles defined for the entire sample. We performed additional analyses in which we assessed the relation between ISI0,120 and changes in systolic and diastolic BP (modeled as continuous variables separately) during follow-up. In these analyses, subjects with hypertension treatment at follow-up were excluded (n=146). Furthermore, we also investigated if fasting plasma glucose by itself was a predictor of hypertension incidence and BP progression.
Odds ratios and their 95% confidence intervals are presented. A 2-tailed probability value of <0.05 was considered statistically significant in all analyses. All analyses were performed with SAS version 8 for Windows (SAS Inc).
Results
The baseline characteristics of our study sample including the sex-specific distribution of ISI0,120 are displayed in Table 1. ISI0,120 was significantly correlated with BMI (Spearman rank r=–0.26, P<0.001), waist circumference (r=–0.24, P<0.001), and fasting glucose (r=–0.40, P<0.001) at baseline.
On follow-up 4 years from baseline, 351 participants (18.2%) had hypertension (168 men) and 786 subjects (40.7%) had an increase in BP by one or more BP categories (367 men). The incidence of hypertension and BP progression decreased with increasing quartiles of ISI0,120, that is, with increasing insulin sensitivity (Table 2).
Insulin Sensitivity and BP Outcomes: Relations in the Whole Sample
In the whole sample, the odds of development of hypertension or BP progression decreased with increasing ISI0,120 quartiles in multivariable models A and B (Table 2). However, adjustment for baseline systolic and diastolic BP attenuated the relation between ISI0,120 and hypertension incidence and BP progression, rendering it statistically nonsignificant (Table 2, model C).
As noted previously, the interaction terms for baseline age, BMI, and baseline BP category with ISI0,120 were significant (P<0.05). Accordingly, we stratified further analyses by age, BMI, and BP categories.
Insulin Sensitivity and BP Outcomes: Relations Within BMI Strata
Among younger individuals with a normal BMI (<51 years and BMI <25 kg/m2), those in quartiles 2 through 4 of ISI0,120 had a statistically significant lower odds for the development of hypertension in models A and B (Table 3, Q2-Q4 versus Q1; P<0.01) and a lower odds of BP progression in all models (P<0.05).
In the younger overweight group (<51 years and BMI >25 to <30 kg/m2), participants in quartiles 2 through 4 of ISI0,120 had a statistically significant lower odds for the development of hypertension in models A and B (Table 3, Q2-Q4 versus Q1; P<0.01) but no relation to BP progression in any of the models.
In the older normal BMI group (>51 years and BMI <25 kg/m2), participants in quartiles 2 through 4 of ISI0,120 had a statistically significant lower odds for the development of hypertension in models A and B (Table 3, Q2-Q4 versus Q1; P<0.01 and P<0.05, respectively). No significant relation was observed between ISI0,120 and BP progression in this subgroup.
No significant relations between ISI0,120 and hypertension incidence or BP progression were observed in older overweight or obese individuals (>51 years and BMI>25 kg/m2, Table 3).
Insulin Sensitivity and BP Outcomes: Relations Within BP Strata
Among younger participants (<51 years) with optimal/normal BP, those in quartiles 2 through 4 of ISI0,120 had a statistically significant lower odds for the development of hypertension in all models (Table 4, Q2-Q4 versus Q1; P<0.01) and the odds of development of hypertension or BP progression was lower in the top quartile of ISI0,120 compared with quartile 1 in all models (Table 4, P<0.05).
No significant relations between the quartiles of ISI0,120 and hypertension incidence or BP progression were observed in individuals with high normal BP regardless of age or in older participants with optimal/normal BP (Table 4).
Insulin Sensitivity and BP Outcomes: Relations Within Combined BMI and BP Strata
In younger participants with BMI <25 kg/m2 and an optimal/normal BP, individuals in quartile 2 to 4 of ISI0,120 had a significantly lower odds ratios for hypertension incidence in all models (Table 5, model A, P<0.01 and models B and C P<0.05) and lower odds of BP progression in all models (P<0.01 for all models).
In younger participants with BMI >25 kg/m2 and an optimal/normal BP, subjects in quartile 2 to 4 of ISI0,120 had a significantly lower odds ratios for hypertension incidence in models A and B (Table 5, P<0.05) that became borderline significant in model C (P<0.06) No significant relations were observed between ISI0,120 and BP progression in this subgroup.
No significant relations between the quartiles of ISI0,120 and hypertension incidence and BP progression were noted in older participants, regardless of BMI status (Table 5).
Secondary Analyses
The results remained essentially unchanged after adjusting for weight change on follow-up, the baseline level of physical activity, and on adjusting for waist circumference instead of BMI in model C (models D, E, and F; data not shown). Fasting plasma glucose did not independently predict either hypertension incidence or BP progression in any subgroup (data not shown).
Models Evaluating BP Change as a Continuous Variable
In the models treating BP change as a continuous variable (and restricted to participants not on treatment on follow-up), the relations between ISI0,120 and increases in systolic BP displayed a pattern consistent with the analyses of BP change as a categorical variable. A 1-SD increment in ISI0,120 was associated with a 1-mm lower systolic BP on follow-up in younger participants (<51 years; P<0.05 in all models, using pooled BMI and BP categories) but not in older participants. We performed exploratory analyses cross-classifying individuals according to BMI and BP categories. In younger participants with a normal BMI and optimal/normal BP at baseline, those in ISI0,120 quartiles 2 to 4 had approximately 4 mm Hg lower systolic BP increase compared with people in quartile 1 independent of age, sex, BMI, and baseline systolic BP (P<0.05 in all models). We did not observe any association of ISI0,120 with BP change in any of the other subgroups defined on the basis of age, BMI, and BP category. ISI0,120 was not associated with changes in diastolic BP (assessed as a continuous variable) in any of the groups evaluated.
Statistical Power
Because we found no independent associations of ISI0,120 with BP outcomes in older individuals, in those with high normal BP and in those with BMI 25 kg/m2, we assessed our statistical power to detect relations in different strata with an -level of 0.05. In the subgroup of old individuals with BMI <25 kg/m2 and optimal/normal BP, we had >80% power to detect an odds ratio of 0.45 or lower for BP progression (quartile 2 to 4 versus quartile 1). In the subgroups of individuals with BMI 25 kg/m2 and optimal/normal BP, we had >80% power to detect an odds ratio of 0.49 (younger participants) and 0.56 (older individuals) for BP progression. In older people with BMI 25 kg/m2 and an optimal/normal BP, we had >80% power to detect an odds ratio of 0.41 for BP progression. As noted earlier, we did not analyze select individuals with high normal BP further stratified by BMI and age because of an inadequate number of participants.
Discussion
Principal Findings
We observed evidence of varying relations of insulin resistance and BP tracking according to age, BMI, and baseline BP. An independent inverse association of ISI0,120 with BP outcomes was evident principally in younger people with a normal BMI and a BP <130/85 mm Hg. Insulin sensitivity was not associated with BP progression or hypertension incidence in the older age group (regardless of baseline BMI or BP), in obese participants (regardless of age or BP), or in people with BP 130/85 mm Hg (regardless of age and BMI). We had adequate statistical power to detect an odds ratio of 0.5 (quartile 2 to 4 versus quartile 1) in all subgroups analyzed (with the exception of the BMI- and BP-stratified groups for people with high normal BP). It is noteworthy that adjustment for BMI or waist circumference did not attenuate the association of insulin sensitivity with BP progression in several of the subgroups. We noted only modest correlations of insulin sensitivity with BMI and waist circumference in our sample that probably contributed to this observation.
Comparison With the Literature
As noted previously, both positive1–13 and null/less positive14–19 associations of insulin resistance have been reported in the literature. Our investigation suggests that effect modification by age, BMI, and BP is an important factor to be considered when relating measures of insulin sensitivity to longitudinal BP tracking. An association of lesser insulin sensitivity with greater BP progression is more likely to be observed in samples that consist of younger, lean people with lower BP levels.
Our finding that insulin sensitivity mainly predicts BP progression in younger participants (<51 years) is consistent with some previous reports3,20,27 and supports the thesis that that the effects of insulin sensitivity are more important early along the natural history of BP progression.
Our observation of an association of insulin sensitivity with BP tracking in those with a normal BMI but not in overweight/obese people is consistent with some prior cross-sectional and prospective reports15,17,28–31 but differs from some other studies.2,13 There are 3 possible explanations for these findings. First, insulin resistance may only contribute to BP progression in lean individuals. The well-established increased risk of BP progression observed in overweight and obese individuals may be explained by factors other than insulin resistance itself, such as hemodynamic adaptations to a greater body mass, increased circulating levels of leptin and free fatty acids, or increased sympathetic activation caused by obstructive sleep apnea.42 Second, insulin resistance may contribute less to BP progression in overweight and obese individuals compared with its contribution in normal-weight people. As several other factors associated with BP progression cluster in overweight and obese individuals,42 factors other than insulin resistance may assume a greater importance for BP progression in these individuals. Third, insulin resistance may contribute equally to BP progression in lean and overweight/obese people. However, the overweight and obese individuals may have been insulin resistant for a longer period of time (compared with people with normal weight). The long-term influences of insulin resistance may result in vascular hypertrophy and remodeling and renal effects; these chronic alterations may mask the contribution of insulin resistance to BP progression in the overweight and obese. Further studies are warranted to understand the mechanisms behind the divergent associations of insulin sensitivity and BP progression in normal-weight versus overweight/obese individuals.
Our finding of a longitudinal association between insulin sensitivity and BP outcomes principally in normal-weight individuals at the lower end of the BP spectrum merits comment. For subjects at the higher end of the baseline BP spectrum, it is possible that factors other than insulin resistance regulate BP progression (such as impaired vascular reactivity, elevated vascular stiffness, renal hemodynamic responses to insulin, and so forth) and that once higher BP develops, the mechanisms affecting BP progression may become insensitive to the systemic effects of insulin resistance. It is also conceivable that a relation between insulin sensitivity and BP tracking is more discernible in individuals at the lower end of the BP distribution because of a lesser degree of confounding by factors associated with both insulin sensitivity and BP in obese individuals and in those with high normal BP.
Mechanisms Underlying Association of Insulin Sensitivity and BP Tracking
There are several mechanisms by which impaired insulin sensitivity may predispose to hypertension. Experimental data suggest that insulin resistance and compensatory hyperinsulinemia can increase BP through effects on the renin-angiotensin-aldosterone system,43 the vascular smooth muscle cell membrane,21–23 smooth muscle proliferation,24 increased sympathetic tone,44–46 and through effects of insulin on sodium retention.25,26 A potential causal link between insulin sensitivity and BP is also suggested by the observation that insulin-sensitizing agents (thiazolidinediones) can lower BP.47
Strengths and Limitations
The strengths of our study include the large community-based sample of nonhypertensive participants, the standardized measurements of BP and BMI, and the multivariable analyses adjusting for factors known to influence insulin sensitivity and BP progression. The use of ISI0,120 is an additional strength of our investigation. Most prior reports evaluating relations of insulin resistance to the development of hypertension used fasting plasma insulin as a measure of insulin sensitivity. Fasting insulin may be limited as a marker of insulin sensitivity because it reflects not only insulin sensitivity but also indicates insulin secretion. Measurements of insulin sensitivity that also include post–glucose load plasma levels of insulin and glucose are closely correlated to insulin sensitivity, as measured by the gold standard euglycemic hyperinsulinemic clamp-method.37,48 Furthermore, it has been shown that ISI0,120 predicts diabetes better than any of the other indirect measurements of insulin sensitivity.38
Our study has several limitations. We did not measure insulin sensitivity by directly using the clamp technique, given the constraints of a large epidemiological study, although ISI0,120 correlates very well with the clamp, as noted above. Our sample was predominantly middle-aged and white, limiting the generalizability of our results to other age groups and ethnicities. Other reports suggest that there may be age- and ethnicity-related differences in the relations between insulin sensitivity and hypertension incidence3,9,15,20; stronger relations have been observed in younger people and in whites. It is important to note that multiple statistical tests were performed in our investigation. However, all analyses were prespecified, and the relative consistency of results across the different models would argue against a type I error as an explanation for our findings.
Conclusions
In our large, community-based sample of nonhypertensive individuals, impaired insulin sensitivity predicted hypertension incidence and BP tracking principally in younger people with normal BMI and optimal/normal BP. These longitudinal associations were independent of baseline BP and BMI. Additional studies are warranted to confirm our findings.
Acknowledgments
This study was supported by a Bergmarks travel grant, Viking Bjrks Hedersledamotstipendium, a Capio travel grant (Dr rnlv), and by research grants (NHLBI contract N01-HC-25195, and 2K24 HL04334 [Dr Vasan]) from the National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, Md. Dr Meigs was supported by a career development award from the American Diabetes Association.
Footnotes
Guest Editor for this article was Robert H. Eckel, MD.
References
Fagot-Campagna A, Balkau B, Simon D, Ducimetiere P, Eschwege E. Is insulin an independent risk factor for hypertension The Paris Prospective Study. Int J Epidemiol. 1997; 26: 542–550.
Salonen JT, Lakka TA, Lakka HM, Valkonen VP, Everson SA, Kaplan GA. Hyperinsulinemia is associated with the incidence of hypertension and dyslipidemia in middle-aged men. Diabetes. 1998; 47: 270–275.
Dyer AR, Liu K, Walsh M, Kiefe C, Jacobs DR Jr, Bild DE. Ten-year incidence of elevated blood pressure and its predictors: the CARDIA study: Coronary Artery Risk Development in (Young) Adults. J Hum Hypertens. 1999; 13: 13–21.
Imazu M, Yamamoto H, Toyofuku M, Sumii K, Okubo M, Egusa G, Yamakido M, Kohno N. Hyperinsulinemia for the development of hypertension: data from the Hawaii-Los Angeles-Hiroshima Study. Hypertens Res. 2001; 24: 531–536.
Tsuruta M, Hashimoto R, Adachi H, Imaizumi T, Nomura G. Hyperinsulinaemia as a predictor of hypertension: an 11-year follow-up study in Japan. J Hypertens. 1996; 14: 483–488.
Niskanen LK, Uusitupa MI, Pyorala K. The relationship of hyperinsulinaemia to the development of hypertension in type 2 diabetic patients and in non-diabetic subjects. J Hum Hypertens. 1991; 5: 155–159.
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.
Haffner SM, Valdez RA, Hazuda HP, Mitchell BD, Morales PA, Stern MP. Prospective analysis of the insulin-resistance syndrome (syndrome X). Diabetes. 1992; 41: 715–722.
Mitchell BD, Haffner SM, Hazuda HP, Valdez R, Stern MP. The relation between serum insulin levels and 8-year changes in lipid, lipoprotein, and blood pressure levels. Am J Epidemiol. 1992; 136: 12–22.
Kashiwabara H, Inaba M, Maruno Y, Morita T, Awata T, Negishi K, Iitaka M, Katayama S. Insulin levels during fasting and the glucose tolerance test and Homa’s index predict subsequent development of hypertension. J Hypertens. 2000; 18: 83–88.
Hirose H, Saito I, Kawabe H, Saruta T. Insulin resistance and hypertension: seven-year follow-up study in middle-aged Japanese men (the KEIO study). Hypertens Res. 2003; 26: 795–800.
Goff DC Jr, Zaccaro DJ, Haffner SM, Saad MF. Insulin sensitivity and the risk of incident hypertension: insights from the Insulin Resistance Atherosclerosis Study. Diabetes Care. 2003; 26: 805–809.
Haffner SM, Miettinen H, Gaskill SP, Stern MP. Metabolic precursors of hypertension: the San Antonio Heart Study. Arch Intern Med. 1996; 156: 1994–2001.
Vaccaro O, Imperatore G, Iovino V, Iovine C, Rivellese AA, Riccardi G. Does impaired glucose tolerance predict hypertension A prospective analysis. Diabetologia. 1996; 39: 70–76.
Shetterly SM, Rewers M, Hamman RF, Marshall JA. Patterns and predictors of hypertension incidence among Hispanics and non-Hispanic whites: the San Luis Valley Diabetes Study. J Hypertens. 1994; 12: 1095–1102.
Skarfors ET, Lithell HO, Selinus I. Risk factors for the development of hypertension: a 10-year longitudinal study in middle-aged men. J Hypertens. 1991; 9: 217–223.
Liese AD, Mayer-Davis EJ, Chambless LE, Folsom AR, Sharrett AR, Brancati FL, Heiss G. Elevated fasting insulin predicts incident hypertension: the ARIC study: Atherosclerosis Risk in Communities Study Investigators. J Hypertens. 1999; 17: 1169–1177.
Nesbitt SD, Amerena JV, Grant E, Jamerson KA, Lu H, Weder A, Julius S. Home blood pressure as a predictor of future blood pressure stability in borderline hypertension: The Tecumseh Study. Am J Hypertens. 1997; 10: 1270–1280.
Mattiasson I, Endre T, Hulthen UL. Insulin sensitivity with respect to glucose metabolism in hypertension-prone men did not predict the blood pressure increase in 5 years. J Hum Hypertens. 2001; 15: 781–785.
Jiang X, Srinivasan SR, Bao W, Berenson GS. Association of fasting insulin with longitudinal changes in blood pressure in children and adolescents: the Bogalusa Heart Study. Am J Hypertens. 1993; 6: 564–569.
Aviv A. The roles of cell Ca2+, protein kinase C and the Na(+)-H+ antiport in the development of hypertension and insulin resistance. J Am Soc Nephrol. 1992; 3: 1049–1063.
Ferrannini E, Taddei S, Santoro D, Natali A, Boni C, Del Chiaro D, Buzzigoli G. Independent stimulation of glucose metabolism and Na+-K+ exchange by insulin in the human forearm. Am J Physiol. 1988; 255: E953–E958.
Standley PR, Rose KA, Sowers JR. Increased basal arterial smooth muscle glucose transport in the Zucker rat. Am J Hypertens. 1995; 8: 48–52.
Stout RW, Bierman EL, Ross R. Effect of insulin on the proliferation of cultured primate arterial smooth muscle cells. Circ Res. 1975; 36: 319–327.
DeFronzo RA. The effect of insulin on renal sodium metabolism: a review with clinical implications. Diabetologia. 1981; 21: 165–171.
Skott P, Hother-Nielsen O, Bruun NE, Giese J, Nielsen MD, Beck-Nielsen H, Parving HH. Effects of insulin on kidney function and sodium excretion in healthy subjects. Diabetologia. 1989; 32: 694–699.
Fossum E, Hoieggen A, Reims HM, Moan A, Rostrup M, Eide I, Kjeldsen SE. High screening blood pressure is related to sympathetic nervous system activity and insulin resistance in healthy young men. Blood Press. 2004; 13: 89–94.
Bonora E, Zavaroni I, Alpi O, Pezzarossa A, Bruschi F, Dall’Aglio E, Guerra L, Coscelli C, Butturini U. Relationship between blood pressure and plasma insulin in non-obese and obese non-diabetic subjects. Diabetologia. 1987; 30: 719–723.
Grandi AM, Gaudio G, Fachinetti A, Zanzi P, Bianchi L, Ceriani L, Guasti L, Venco A. Influence of family history of hypertension on insulin sensitivity in lean and obese hypertensive subjects. Eur J Clin Invest. 1997; 27: 774–779.
Mykkanen L, Haffner SM, Ronnemaa T, Watanabe RM, Laakso M. Relationship of plasma insulin concentration and insulin sensitivity to blood pressure: is it modified by obesity J Hypertens. 1996; 14: 399–405.
Haffner SM, Ferrannini E, Hazuda HP, Stern MP. Clustering of cardiovascular risk factors in confirmed prehypertensive individuals. Hypertension. 1992; 20: 38–45.
Muntner P, He J, Roccella EJ, Whelton PK. The impact of JNC-VI guidelines on treatment recommendations in the US population. Hypertension. 2002; 39: 897–902.
Kannel WB, Feinleib M, McNamara PM, Garrison RJ, Castelli WP. An investigation of coronary heart disease in families: the Framingham offspring study. Am J Epidemiol. 1979; 110: 281–290.
Clinical guidelines on the identification, evaluation, and treatment of overweight and obesity in adults. NIH Publication No. 98–4083. 1998. National Institutes of Health.
Sixth report of the Joint National Committee on prevention, detection, evaluation, and treatment of high blood pressure. Arch Intern Med. 1997; 157: 2413–2446.
Kannel WB, Belanger A, D’Agostino R, Israel I. Physical activity and physical demand on the job and risk of cardiovascular disease and death: the Framingham Study. Am Heart J. 1986; 112: 820–825.
Gutt M, Davis CL, Spitzer SB, Llabre MM, Kumar M, Czarnecki EM, Schneiderman N, Skyler JS, Marks JB. Validation of the insulin sensitivity index (ISI[0,120]): comparison with other measures. Diabetes Res Clin Pract. 2000; 47: 177–184.
Hanley AJ, Williams K, Gonzalez C, D’Agostino RB Jr, Wagenknecht LE, Stern MP, Haffner SM. Prediction of type 2 diabetes using simple measures of insulin resistance: combined results from the San Antonio Heart Study, the Mexico City Diabetes Study, and the Insulin Resistance Atherosclerosis Study. Diabetes. 2003; 52: 463–469.
Vasan RS, Larson MG, Leip EP, Kannel WB, Levy D. Assessment of frequency of progression to hypertension in non-hypertensive participants in the Framingham Heart Study: a cohort study. Lancet. 2001; 358: 1682–1686.
Hosmer DW, Lemeshow S. Applied Logistic Regression. New York, NY: John Wiley & Sons Inc, 1989.
Balkau B, Charles MA. Comment on the provisional report from the WHO consultation. European Group for the Study of Insulin Resistance (EGIR). Diabet Med. 1999; 16: 442–443.
Montani JP, Antic V, Yang Z, Dulloo A. Pathways from obesity to hypertension: from the perspective of a vicious triangle. Int J Obes Relat Metab Disord. 2002; 26 (suppl 2): S28–S38.
Sowers JR, Frohlich ED. Insulin and insulin resistance: impact on blood pressure and cardiovascular disease. Med Clin North Am. 2004; 88: 63–82.
Rowe JW, Young JB, Minaker KL, Stevens AL, Pallotta J, Landsberg L. Effect of insulin and glucose infusions on sympathetic nervous system activity in normal man. Diabetes. 1981; 30: 219–225.
Young JB, Landsberg L. Stimulation of the sympathetic nervous system during sucrose feeding. Nature. 1977; 269: 615–617.
Troisi RJ, Weiss ST, Parker DR, Sparrow D, Young JB, Landsberg L. Relation of obesity and diet to sympathetic nervous system activity. Hypertension. 1991; 17: 669–677.
Gilling L, Suwattee P, DeSouza C, Asnani S, Fonseca V. Effects of the thiazolidinediones on cardiovascular risk factors. Am J Cardiovasc Drugs. 2002; 2: 149–156.
Soonthornpun S, Setasuban W, Thamprasit A, Chayanunnukul W, Rattarasarn C, Geater A. Novel insulin sensitivity index derived from oral glucose tolerance test. J Clin Endocrinol Metab. 2003; 88: 1019–1023.(Johan rnlv, MD, PhD; Mich)