Fasting and Post-Glucose Load Measures of Insulin Resistance and Risk of Ischemic Stroke in Older Adults
Background and Purpose—Few studies have assessed post-glucose load measures of insulin resistance and ischemic stroke risk, and data are sparse for older adults. We investigated whether fasting and post-glucose load measures of insulin resistance were related to incident ischemic stroke in nondiabetic, older adults.
Methods—Participants were men and women in the Cardiovascular Health Study, age 65+ years and without prevalent diabetes or stroke at baseline, followed for 17 years for incident ischemic stroke. The Gutt insulin sensitivity index was calculated from baseline body weight and from fasting and 2-hour postload insulin and glucose; a lower Gutt index indicates higher insulin resistance.
Results—Analyses included 3442 participants (42% men) with a mean age of 73 years. Incidence of ischemic stroke was 9.8 strokes per 1000 person-years. The relative risk (RR) for lowest quartile versus highest quartile of Gutt index was 1.64 (95% CI, 1.24–2.16), adjusted for demographics and prevalent cardiovascular and kidney disease. Similarly, the adjusted RR for highest quartile versus lowest quartile of 2-hour glucose was 1.84 (95% CI, 1.39–2.42). In contrast, the adjusted RR for highest quartile versus lowest quartile of fasting insulin was 1.10 (95% CI, 0.84–1.46).
Conclusions—In nondiabetic, older adults, insulin resistance measured by Gutt index or 2-hour glucose, but not by fasting insulin, was associated with risk of incident ischemic stroke.
See related article, page 3333.
Insulin resistance is a precursor and mechanism of type 2 diabetes, and is associated with development of atherosclerosis and hypercoagulability.1 In prospective studies of middle-aged, nondiabetic adults, measures of insulin resistance have been associated with coronary heart disease and stroke.1,2
Because vascular disease is more strongly associated with impaired glucose tolerance than with impaired fasting glucose,3 post-glucose load measures of insulin resistance may be important for assessing the relationship between insulin resistance and cardiovascular risk. Although recent studies of insulin resistance and ischemic stroke have examined fasting insulin,4,5 there has been little study of ischemic stroke risk in relation to post-glucose load measures of insulin resistance. One such measure is the Gutt insulin sensitivity index, which is derived from body weight and from fasting and 2-hour insulin and glucose.6
Data are sparse on the association between insulin resistance and ischemic stroke in older adults because in most previous studies, insulin resistance was measured in young-to-middle-aged adults at baseline. We examined a cohort of nondiabetic participants in whom measures of insulin resistance were obtained at a mean age of 73 years, with 17 years of follow-up for incident stroke. We hypothesized that insulin resistance, reflected by higher fasting insulin and lower Gutt insulin sensitivity index, would be associated with higher risk of ischemic stroke among nondiabetic, older adults.
Setting and Participants
The Cardiovascular Health Study (CHS) is a community-based, prospective study of men and women age 65 years and older.7 The cohort included 5201 participants enrolled in 1989–1990 at field centers in Washington County, MD; Pittsburgh, PA; Forsyth County, NC; and Sacramento County, CA. The study was approved by institutional review boards at the University of Washington and at each field center.
For this analysis, we included participants who at baseline had no history of stroke, were free of diabetes, and had fasting and 2-hour oral glucose tolerance test (OGTT) insulin and glucose measurements. An additional cohort of black participants enrolled later was not included because 2-hour insulin and glucose were not measured at their baseline visit. Diabetes at baseline was defined as use of insulin or oral hypoglycemic drugs, fasting serum glucose ≥7.0 mmol/L (126 mg/dL), random serum glucose ≥11.1 mmol/L (200 mg/dL), or 2-hour serum glucose ≥11.1 mmol/L (200 mg/dL). History of stroke at baseline was ascertained from participant report and was confirmed with medical records.8 We excluded 199 participants who had a history of stroke, 1107 participants who had diabetes, 173 participants for whom data were incomplete to classify diabetes status, 239 participants who were missing at least 1 fasting or 2-hour insulin or glucose measurement, and 41 participants who had incomplete covariate data; this left 3442 participants for analysis.
Insulin and Glucose Measures
Serum samples were obtained after an overnight fast of at least 8 hours, and again 2 hours after a 75-g oral glucose challenge. Insulin was measured with a competitive radioimmunoassay (Diagnostic Products Corporation), and glucose was measured with an enzymatic method.9 The Gutt insulin sensitivity index was calculated as insulin sensitivity=m/(G×I), where m is a measure of glucose uptake during the OGTT calculated from body weight and from fasting and 2-hour glucose, G is the mean of fasting and 2-hour glucose, and I is a log10 transformation of the mean of fasting and 2-hour insulin. Units for the Gutt index are mg×L2/mmol×mU×min. The complete formula and a correction are reported elsewhere.6,10
Age, sex, race, and smoking behavior were determined from participant report. Body weight, waist circumference, and systolic blood pressure were measured with standardized protocols. Plasma triglycerides, high-density lipoprotein cholesterol, and serum creatinine were measured with enzymatic methods.9 Plasma low-density lipoprotein cholesterol was estimated with the Friedewald equation.11 Estimated glomerular filtration rate was estimated with the CKD-EPI equation.12 Physical activity was measured with a questionnaire.13 Antihypertensive medication use was measured with a medication inventory.14 Prevalent coronary heart disease (including history of myocardial infarction, angina, coronary angioplasty, or coronary artery bypass graft), congestive heart failure, and peripheral arterial disease were determined from the baseline examination and medical record review.15 Atrial fibrillation was determined by 12-lead resting electrocardiogram at the baseline examination. Interim development of diabetes was determined from annual medication inventories throughout follow-up and fasting (or random) glucose measurements at 3, 5, 7, 9, and 16 years after baseline.
Ischemic Stroke Outcome
The outcome was definite incident ischemic stroke, adjudicated by the CHS Cerebrovascular Events Committee. Incident strokes were ascertained from participant report, questions at annual visits, telephone contacts every 6 months between annual visits and after annual visits ended, and screens of hospitalizations for key International Classification of Diseases, 9th Revision codes. Strokes were confirmed with information from participant interviews, medical records, test results, and brain images; detailed diagnostic criteria have been described previously.8 Follow-up for each participant extended from the 1989–1990 baseline examination until occurrence of incident ischemic stroke (n=417), death (n=1775), loss to follow-up (n=149), or end of follow-up on June 30, 2007 (n=1101), whichever occurred first.
We assessed correlations among Gutt index components with Pearson correlation coefficients. We used Cox proportional hazards models to assess associations of Gutt index and each of its components with incident ischemic stroke. We compared stroke risks across quartiles of each measure, with the least-insulin-resistant quartile as reference category (highest quartile for Gutt index; lowest quartile for each component), and tested for trend across quartiles. Relative risks (RR) were adjusted with risk-set stratification for baseline age in 1 year categories; indicator variables for male sex, black race, estimated glomerular filtration rate <60 mL/min per 1.73 m2, coronary heart disease, congestive heart failure, atrial fibrillation, peripheral arterial disease, and antihypertensive medication use; and continuous linear variables for systolic blood pressure, triglycerides, high-density lipoprotein cholesterol, and low-density lipoprotein cholesterol. We did not adjust for waist circumference or physical activity because they were not associated with ischemic stroke risk in this study. Adjustment for smoking, which was unrelated to measures of insulin resistance, did not affect our results.
In additional analyses, we calculated sex-specific RRs and tested whether they were different. We also recalculated RRs using follow-up time and strokes only before interim development of diabetes using models with a time-varying diabetes indicator variable and a multiplicative interaction between diabetes and each baseline measure of insulin resistance.
Participants had a mean age of 73 years at baseline; 42% were men; and 4% were black. Participants with lower Gutt insulin sensitivity index (more insulin resistant) tended to have a more adverse cardiovascular risk profile (Table 1). During follow-up, 417 participants experienced ischemic stroke, at a mean age of 82 years. Ischemic stroke incidence was 9.8 per 1000 person–years.
Gutt Insulin Sensitivity Index and Incident Ischemic Stroke
Compared with the highest quartile of Gutt index (least insulin resistant), lower quartiles (more insulin resistant) were associated with higher risk of ischemic stroke (Table 2). RRs adjusted for demographics, and prevalent diseases were similar to unadjusted RRs. Additional adjustment for blood pressure, antihypertensive medication use, and lipids attenuated the RRs.
Gutt Index Components and Incident Ischemic Stroke
The Pearson correlations of Gutt index with each of its components were −0.37 for fasting glucose, −0.36 for fasting insulin, −0.85 for 2-hour glucose, −0.65 for 2-hour insulin, and −0.18 for body weight. For 2-hour glucose, RRs of ischemic stroke comparing higher quartiles (more insulin resistant) with the lowest quartile (least insulin resistant) were similar to RRs observed for Gutt index (Table 2). Two-hour insulin was also positively associated with ischemic stroke; however, fasting glucose, fasting insulin, and body weight were not (Table 2). In a model that included all individual Gutt index components simultaneously, only 2-hour glucose was independently associated with stroke risk (data not shown).
The association between Gutt index and ischemic stroke was stronger in men (lowest quartile versus highest quartile: RR, 2.39; 95% CI, 1.59–3.59) than in women (lowest quartile versus highest quartile: RR, 1.16; 95% CI, 0.79–1.69; P for interaction=0.04), adjusted for demographics and prevalent disease. Also for 2-hour insulin, the RR comparing extreme quartiles was higher for men (RR, 1.93; 95% CI, 1.25–2.97) than for women (RR, 1.11; 95% CI, 0.76–1.62; P for interaction=0.009). For other Gutt index components, RRs comparing extreme quartiles were either similar for men and women, or only slightly higher for men. Associations of baseline measures of insulin resistance with stroke before interim development of diabetes were similar to associations observed when ignoring interim development of diabetes (Supplemental Table I; http://stroke.ahajournals.org).
In this study of nondiabetic, older adults, lower Gutt insulin sensitivity index was associated with higher risk of ischemic stroke. Higher 2-hour glucose was as strongly associated with ischemic stroke risk as was lower Gutt index. In contrast, higher fasting insulin was not associated with ischemic stroke in this study. This study contributes new information concerning older adults, a group in whom insulin resistance is common and stroke risk is high, and for whom data are relatively lacking from previous studies.
One plausible interpretation of our finding that Gutt index and 2-hour glucose were associated with stroke risk, but fasting insulin was not, is that peripheral, rather than hepatic, insulin resistance may play a dominant role in cardiovascular risk in older adults. Post-glucose load measures of insulin resistance reflect whole-body or peripheral insulin resistance. This is because they incorporate information about insulin response to an oral glucose load and glucose uptake during the OGTT by skeletal muscle and fat, in addition to the liver. In contrast, fasting insulin reflects hepatic insulin resistance, because fasting insulin concentrations are determined by the liver and pancreas. However, hepatic and whole-body (or peripheral) insulin resistance are correlated, so making a sharp distinction based on fasting and post-glucose load measures in epidemiological studies is difficult.16
Other studies have found positive associations between fasting measures of insulin resistance and ischemic stroke risk. Among Atherosclerosis Risk in Communities (ARIC) Study participants without cardiovascular disease or diabetes at baseline, fasting insulin was positively associated with ischemic stroke risk (445 incident ischemic strokes), adjusted for age, sex, race, and study site.4 Positive associations between fasting insulin and stroke were also observed in smaller studies.1,5 The discrepancy between our findings and observations of previous studies that fasting insulin was associated with stroke risk, could be that in some previous studies, some participants with high fasting insulin also had undiagnosed diabetes; this may have been detected if 2-hour glucose had been measured. The association of insulin resistance with ischemic stroke in our study was stronger for post-glucose load measures than for fasting measures, suggesting that the association of insulin resistance and stroke may be underestimated in studies that use fasting measures alone.
We observed a stronger association between Gutt index and stroke in men than in women. However, the P value was not adjusted for multiple comparisons, the sex difference was not significant for 2-hour glucose, and the finding was not consistent with an earlier study that observed a stronger association between fasting insulin and stroke in women than in men.4
Insulin resistance may be a mechanism through which characteristics such as age, obesity, and inactivity lead to hypertension, dyslipidemia, and other factors, ultimately increasing stroke risk.1 We considered demographics, obesity, activity levels, prevalent cardiovascular disease, and kidney dysfunction as potential confounders of the association between insulin resistance and stroke. Hypertension and dyslipidemia may also confound the relationship of insulin resistance with stroke, or they may be intermediate variables. Lower Gutt index and higher 2-hour glucose were associated with higher stroke risk after adjustment for all potential confounding and intermediate variables we considered, and after excluding follow-up time and strokes occurring after interim development of diabetes; this suggests that insulin resistance may influence ischemic stroke risk through mechanisms other than those we considered in this article.
A limitation of our study is that the fasting and 2-hour measures of insulin resistance we used are only modestly correlated with more direct measures of insulin resistance. However, indirect measures of insulin resistance are commonly used in place of invasive tests that place a higher burden on research participants. Another limitation is that we used only baseline measures of insulin resistance. This study also has strengths, including the focus on older adults, who have the highest burden of stroke and insulin disorders, the prospective design with high retention of participants during follow-up, assessment of interim development of diabetes, and adjudicated ischemic stroke outcomes.
In summary, among nondiabetic, older adults, lower Gutt insulin sensitivity index, a post-glucose load measure of insulin resistance that mainly reflected higher 2-hour glucose, was associated with a higher risk of ischemic stroke. In contrast, higher fasting insulin was not associated with ischemic stroke risk among nondiabetic, older adults. Additional research is needed to define clinical and public health implications of these findings.
Sources of Funding
The research reported in this article was supported by contracts N01-HC-85239, N01-HC-85079 through N01-HC-85086, N01-HC-35129, N01-HC-15103, N01-HC-55222, N01-HC-75150, N01-HC-45133, and grants HL-080295 and HL-094555 from the National Heart, Lung, and Blood Institute (NHLBI), with additional contribution from the National Institute of Neurological Disorders and Stroke (NINDS). Additional support was provided through AG-023629, AG-15928, AG-20098, and AG-027058 from the National Institute on Aging (NIA). EL Thacker was supported by NHLBI training grant T32-HL007902 (PI: Siscovick). JB Meigs was supported by grant K24-DK080140 from the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK). EJ Boyko was supported by NIDDK grant P30-DK17047 and by the Veterans Affairs Puget Sound Health Care System.
We thank the CHS participants and staff and the CHS Diabetes Working Group. A full list of principal CHS investigators and institutions can be found at http://www.chs-nhlbi.org/pi.htm.
Bruce Ovbiagele, MD, was the Guest Editor for this paper.
The online-only Data Supplement is available at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.111.620773/-/DC1.
- Received March 17, 2011.
- Accepted July 6, 2011.
- © 2011 American Heart Association, Inc.
- Kernan WN,
- Inzucchi SE,
- Viscoli CM,
- Brass LM,
- Bravata DM,
- Horwitz RI
- Sarwar N,
- Sattar N,
- Gudnason V,
- Danesh J
- Rasmussen-Torvik LJ,
- Yatsuya H,
- Selvin E,
- Alonso A,
- Folsom AR
- Longstreth WT Jr.,
- Bernick C,
- Fitzpatrick A,
- Cushman M,
- Knepper L,
- Lima J,
- et al
- Cushman M,
- Cornell ES,
- Howard PR,
- Bovill EG,
- Tracy RP
- Hanley AJ,
- Williams K,
- Gonzalez C,
- D'Agostino RB Jr.,
- Wagenknecht LE,
- Stern MP,
- et al
- Friedewald WT,
- Levy RI,
- Fredrickson DS
- Siscovick DS,
- Fried L,
- Mittelmark M,
- Rutan G,
- Bild D,
- O'Leary DH
- Matsuda M,
- DeFronzo RA