Familial Aggregation of Insulin Resistance and Cardiovascular Risk Factors in Hypertension

Abstract

The authors assessed the familial aggregation of cardiometabolic abnormalities (elevated homeostasis model assessment [HOMA], triglycerides [TG], and low‐density lipoprotein [LDL] and reduced high‐density lipoprotein [HDL]) among hypertensive siblings (N=287 from 138 families). Evidence for familial aggregation required sibling‐pair concordance of outcome variables dichotomized according to predefined values (concordance for highest‐quartile HOMA [>3.3], TG [>170 mg/dL], and LDL [>138 mg/dL] and lowest‐quartile HOMA for HDL [<32 mg/dL]). Hypertensive individuals with insulin resistance (high‐quartile HOMA) had higher TG and lower HDL and LDL levels compared with insulinsensitive hypertensives. High‐quartile HOMA, TG, and LDL aggregated in hypertensive families, and TG plus HOMA coaggregated. HDL did not show aggregation. In a multivariate logistic regression, the only significant predictor of an individual's HOMA status was a sibling's HOMA status in a model including age, sex race, and body mass index (odds ratio=9.12; 95% confidence interval, 3.64–23.14; P<.001). Cardiometabolic variables demonstrate heritability in hypertensive families. Further exploration of common genetic susceptibility loci in hypertension involving these factors is warranted.

Clustering of cardiovascular risk factors and metabolic abnormalities is commonly seen in patients with hypertension, ¹ including dyslipidemia, obesity, diabetes, insulin resistance, and inflammatory markers. Characterizing the heritability of metabolic abnormalities in hypertension would facilitate investigation of common underlying genetic mechanisms. Previous studies have examined the genetic susceptibility for metabolic abnormalities associated with increased cardiovascular risk (insulin resistance and dyslipidemia) ² , ³ in diabetics, normotensives, and patients with polycystic ovarian syndrome. ⁴ , ⁵ The heritability of metabolic abnormalities in hypertension, however, is not well studied.

We identified, a priori, metabolic abnormalities associated with hypertension and cardiovascular risk. Previous studies have documented an association between dyslipidemia and insulin resistance (mainly high triglyceride [TG] and low high‐density lipoprotein [HDL] cholesterol levels). ⁶ Increased cardiovascular morbidity and mortality in diabetics with hypertension improves with low‐density lipoprotein (LDL) cholesterol‐lowering therapy. ⁷ Hence, LDL cholesterol is also a significant metabolic abnormality in hypertension. Twin and sibling studies in insulin‐resistant states like polycystic ovarian syndrome have shown LDL cholesterol levels to be heritable, along with other metabolic variables including insulin resistance and TG levels. ⁴

With this as background, we investigated the heritability of metabolic factors associated with cardiovascular risk in hypertension, namely insulin resistance, using the homeostasis model assessment (HOMA), HDL and LDL cholesterol, and TG levels by analyzing their aggregation in multiple‐sibling families.

Subjects and Methods

Data were analyzed from an existing dataset—the HyperPath (Hypertensive Pathotype) Study. ⁸ The HyperPath cohort was initiated in 1994 for the purpose of deciphering the genetic underpinnings of essential hypertension through the use of intermediate phenotyping. A detailed description of this population and the protocol used has been provided previously. ⁸ All studies were performed in General Clinical Research Centers (GCRC) located in Brigham and Women's Hospital (BWH) (Boston), the University of Utah (Salt Lake City), Vanderbilt Medical Center (Nashville), and Broussais Hospital (Paris) to enable strict control of environment and diet. The institutional review boards of each participating institution approved the study protocol, and written informed consent was obtained from each participant. Of the 670 hypertensive patients who had participated in this project as of November 2005, data related to our analysis were available for 402 individuals, which included 287 siblings from 138 families.

Study Participant Selection

Screening history and physical and laboratory data were obtained from each participant before enrollment. Race and ethnicity were self‐defined. Participants were recruited from the general community. To qualify for enrollment, patients were required to have a history of hypertension, with a diastolic blood pressure (BP) ≥100 mm Hg on no medications, diastolic BP ≥90 mm Hg on 1 antihypertensive agent, or the use of 2 or more antihypertensive medications at the time of screening. There were no systolic BP criteria used in the protocol design.

Patients were excluded from participating in the study if they had diastolic BP >110 mm Hg while on 2 or more medications, presence of hypokalemia (serum potassium <3.5 mmol/L), and/or active cardiac or vascular disease. All patients on angiotensin‐converting enzyme inhibitors, angiotensin receptor blockers, or aldosterone antagonists discontinued these medications 3 months before the study because of their known effects on modulation of renin‐angiotensin‐aldosterone system function in some hypertensive patients. ⁹ , ¹⁰ , ¹¹ People on β‐blockers had their medications discontinued for 3–4 weeks. If needed, patients were placed on either a dihydropyridine calcium channel blocker or thiazide diuretic or both to control BP during the washout period. All medications were discontinued 2–4 weeks before the study. No patient with diabetes mellitus, obesity (body mass index [BMI] >34 kg/m²), renal insufficiency, or significant medical history was enrolled.

Protocol

All subjects participated in identical protocols regardless of study site. Subjects were admitted to a GCRC for 1 night and 1 day, during which hemodynamic and laboratory assessments were performed. Study participants remained fasting and supine for 10 hours before the study measurements, which occurred between 8 am and 10 am and included laboratory assessment of TG, HDL and LDL cholesterol, insulin, and glucose levels. Supine BP was recorded by an automatic recording device at 5‐minute intervals and averaged for analysis.

Laboratory Procedures

All blood samples were collected on ice and centrifuged immediately, with plasma separated and frozen until the time of assay. Specimens from all study sites were analyzed at a central laboratory at BWH. HOMA was used to characterize insulin resistance (fasting glucose in mmol × fasting insulin in μU/mL ? 22.5). Plasma glucose was assayed by the glucose oxidase method using a glucose reflectometer. Plasma insulin levels were determined by radioimmunoassay. Lipid levels were measured at the BWH core research laboratory accredited by the Lipid Standardization Program of the Centers for Disease Control and Prevention.

Statistical Analyses

Data are presented as mean ± SEM. In the multi‐center HyperPath study, homogeneity of parameters was investigated and confirmed across the 4 participating centers before data were pooled and analyzed. All metabolic parameters were selected a priori based on their widely acknowledged association with cardiovascular risk and their availability in the present dataset. Predetermined cutoff values for each primary outcome variable were generated by dichotomization at the highest quartile value for HOMA (>3.3), TG (>170 mg/dL), and LDL cholesterol (>138 mg/dL) and the lowest quartile for HDL cholesterol (<32 mg/dL).

To provide evidence of heritability, we first examined the presence of concordant quartile status results for each outcome variable among siblings by testing whether familial aggregation in the high quartile (HOMA, TG, or LDL) or low quartile (HDL) exceeded the expected frequency. The expected frequency represented the probability that 2 randomly selected, unrelated individuals would have laboratory values residing in the same quartile. Thus, for 2 individuals concordant for a high quartile value, the probability is the product of their individual probabilities (0.25 × 0.25, or 0.0625). The probability that 2 randomly selected individuals would have concordant quartile values for the normal/low quartiles would be 0.75 × 0.75, or 0.5625, and that they would be discordant, would be the sum of these frequencies subtracted from 1, or 0.375. The McNemar test was used to test the significance of the excess aggregation in the sibling families where each family received equal weight. After determining whether individual variables aggregated in hypertensive families, we then investigated whether high‐quartile HOMA and TG coaggregated in hypertensive families. Sample size limitations restricted additional exploration of coaggregation.

We examined the specific heritability of HOMA quartile status (high quartile vs normal/low quartile) by performing a multivariate logistic regression where the bivariate dependent variable was the individual's HOMA quartile status and the test predictors were the individual's BMI, age, sex, race, and their sibling's HOMA quartile status.

A 2‐tailed P value of <.05 described statistical significance. The SPSS statistical software package was used for all analyses (SPSS version 12.0; SPSS Inc, Chicago, IL).

Results

Baseline characteristics are shown in Table I according to HOMA quartile status. Subjects in the highest quartile of HOMA were of similar age and BP when compared with normal/low HOMA quartiles. There were no differences in race or sex distributions between the 2 groups. TG and fasting insulin levels were significantly higher (P<.01), and LDL and HDL cholesterol levels were significantly lower (P=.02 and P=.04, respectively) in the high‐quartile HOMA group.

Table I.

Baseline Characteristics of Subjects According to Homeostasis Model Assessment (HOMA) Quartile Status

		Normal/Low‐Quartile
	High‐Quartile HOMA	HOMA	P
Number	107	295
Age, y	48.40±0.78	47.79±0.47
Body mass index, kg/m2	29±0.4	26.7±0.2	<.001
Women, No.	36	46
Caucasian, %	84	88
Systolic blood pressure, mm Hg	145.5±1.91	147.49±1.18
Diastolic blood pressure, mm Hg	87.02±1.02	87.77±0.68
Fasting glucose, mg/dL	90.49±1.36	91.24±0.67
Fasting insulin, μU/mL	14.71±0.61	7.94±0.31	<.001
Triglycerides, mg/dL	205.15±12.42	144.80±6.00	<.001
High‐density lipoprotein, mg/dL	38.89±1.20	41.84±0.79	.04
Low‐density lipoprotein, mg/dL	115.88±3.37	126.25±2.27	.02
Values are reported as mean ± SD except where indicated.

Since HOMA is a continuous variable and there are no well‐defined cut points for normal vs abnormal HOMA values, we compared higher HOMA values (individuals more likely to be insulin‐resistant [high‐quartile HOMA]) with the other quartiles.

The observed proportion of siblings concordant for high‐quartile HOMA status, concordant for normal/low‐quartiles HOMA status, and discordant for HOMA quartile status was significantly different than the expected frequencies (P<.0001) (Figure). The observed frequency of siblings concordant for high‐quartile HOMA was 2.39 times the expected frequency.

Similar findings were observed for concordance of high‐quartile values for TG (P<.001) and LDL cholesterol (P=.03) but not for HDL cholesterol (P=.24). The observed frequency of siblings concordant for high‐quartile TG was 2.3 times the expected frequency and 1.9 times the expected frequency for high‐quartile LDL cholesterol (Figure). The combination of high‐quartile HOMA plus high‐quartile TG also revealed a significant coheritable relationship where the observed frequency was roughly 6 times the expected frequency (P<.001).

Additional evidence of a strong heritable influence for high‐quartile HOMA was observed in a logistic regression model. Sibling HOMA quartile status predicted an individual's HOMA quartile status even after adjusting for age, sex, BMI, and race (odds ratio=9.12; 95% confidence interval, 3.64–23.14; P<.001) (Table II).

Table II.

Multivariate Logistic Regression to Determine Predictors of Sibling HOMA Status*

Independent Predictor	Odds Ratio (95% CI)	P
Final model
Sibling 1 HOMA status*	9.12 (3.64–23.14)	<.001
Variables removed from final model
Male sex	2.48 (0.88–6.75)	.067
White (vs black) race	1.40 (0.27–7.17)	.529
Age	1.02 (0.97–1.08)	.398
Body mass index	1.02 (0.90–1.15)	.707
*High vs normal/low homeostasis model assessment (HOMA) quartile. CI indicates confidence interval.

To provide further evidence for the heritability of metabolic abnormalities in hypertension, we examined the relationship between a reported family history of hypertension (12 patients with a positive family history vs 15 with a negative family history) and the presence of abnormal glucose metabolism in a separate dataset of normotensive individuals. Participants were of similar age (45±3 years) and BMI (25±1 kg/m²). Fasting glucose was significantly higher in subjects who reported a positive family history of hypertension (86.5±2.1 vs 80.2±1.3 mg/ L; P=.01), which remained a significant predictor of family history status after controlling for age and BMI (odds ratio=1.2; 95% confidence interval, 1.1–1.4; P=.O4). HOMA (2.0±0.2 vs 1.8±0.2; P=nonsignificant) and fasting insulin (9.5±1.0 vs 10.2±0.9 μU/mL; P=nonsignificant) were not different based on family history status.

Discussion

We hypothesized that metabolic abnormalities associated with increased cardiovascular risk would aggregate in families with hypertension. In this report we describe 3 separate analyses to provide support for this hypothesis. First, we found that hypertensive siblings were more than twice as likely to be concordant for an elevated HOMA, TG, and LDL cholesterol status and 9 times more likely to be concordant for the combination of elevated HOMA and TG. Second, a hypertensive sibling's HOMA status can predict an individual's HOMA status even after controlling for BMI, age, sex, and race. Finally, the small number of normotensive individuals with a reported family history of hypertension were more likely to have higher fasting glucose values than normotensives who did not report a family history of hypertension. The demonstration of the heritability of factors associated with cardiovascular risk in essential hypertension suggests that a common genetic susceptibility locus might exist.

Despite the significant obstacles inherent in characterizing the genetic underpinnings of complex medical conditions such as hypertension, substantial evidence supports a heritable component to these conditions. Genome‐wide scans in hypertension have provided several chromosomal regions with evidence of linkage, ¹² , ¹³ , ¹⁴ , ¹⁵ although findings have been inconsistent. ¹⁶ Studies have reported an elevated risk of hypertension ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ , ²¹ and insulin resistance ²² , ²³ , ²⁴ , ²⁵ in association with numerous genetic polymorphisms. Specifically related to the heritability of clustering of metabolic factors in hypertension, Shigematsu and colleagues ²⁶ found a genetic predisposition to hypertension in individuals with an elevated HOMA index.

Heritability of cardiovascular risk factors has been reported in diabetes, obesity, the insulin resistance syndrome, and the polycystic ovarian syndrome. ⁴ , ²⁷ , ²⁸ Genetic influences on obesity, hypertension, blood lipids (including HDL cholesterol and TG), type 2 diabetes, insulin levels, and insulin resistance has also been demonstrated in the literature. ²⁹ , ³⁰ , ³¹ , ³² , ³³ , ³⁴ , ³⁵ , ³⁶ , ³⁷ , ³⁸

We found that hypertensive patients with an elevated HOMA index had higher serum TG levels, lower HDL cholesterol and perhaps, surprisingly, lower LDL cholesterol levels when compared with hypertensive patients with a normal/low HOMA index. This may suggest that the ability to cluster metabolic abnormalities is dependent on the baseline population studied, in this case subjects with hypertension.

Our results should be interpreted in the context of the study design, which is a retrospective analysis of an existing hypertensive dataset. It should be noted, however, that the primary Shigematsu and colleagues the genetic underpinnings of essential hypertension. The results suggest, but do not prove, any causal association between combined heritability of metabolic factors associated with cardiovascular risk and hypertension. They suggest that an unobserved underlying characteristic unique to family membership, such as heritability, is in operation. Alternatively, the higher‐than‐expected frequency of concordance could be due to the close or common environmental influence of a family relationship.

Our sample size limited our ability to explore the heritability of combined clustering of these metabolic variables in hypertension, although we did find that TG and HOMA coaggregated in families with hypertension. This might imply that an unknown common genetic defect could be responsible for multiple metabolic derangements present in hypertension. Detailed examination of candidate genes and function are required to better characterize this relationship.

Our intention for presenting the normotensive findings was to demonstrate that altered glucose metabolism is detectable in normotensives when categorized based on family history of hypertension. This supports the hypothesis that hypertensive families carry metabolic factors that confer increased cardiovascular risk. The extrapolation from a small sample size was a limitation in interpretation, which was not robust enough to demonstrate significant differences in HOMA or fasting insulin levels. Generalization of these findings to all hypertensives is limited by the subject selection. Only patients with mild‐to‐moderate hypertension were recruited, although this represents approximately 90% of the general hypertensive population.

We attempted to minimize confounding by utilizing a common protocol and core laboratory, instituting strict environmental control with use of a GCRC setting, including control of dietary intake and prolonged medication washout.

Conclusions

Hypertensive patients with insulin resistance (high HOMA values) have comparatively higher TG levels and lower HDL and LDL cholesterol than hypertensive patients who have lower HOMA values. HOMA, TG, and LDL cholesterol levels aggregate in families with hypertension. HOMA and TG levels coaggregate in hypertensive families. Evidence of heritability of these metabolic factors associated with significant cardiovascular risk warrants investigation of common underlying genetic susceptibility loci in essential hypertension.

Acknowlegement: Dr Annaswamy Raji and Dr Jonathan S. Williams had equal contribution to this paper.

Disclosure: This work was supported by National Institutes of Health grants HL47651, HL59424, HL77234, and DK63214; the Specialized Center of Research in Hypertension (HL55000); and the National Center for Research Resources (General Clinical Research Centers) in Boston, MA (MO1 RR 02635) and Salt Lake City, UT (MO1RR0064).