Multi-Year Reproducibility of Hyperinsulinemic Euglycemic Clamp-Derived Insulin Sensitivity in Free-Living Adults: Association with Incident Prediabetes in the POP-ABC Study

Abstract

OBJECTIVE:

The hyperinsulinemic euglycemic clamp (HEC) is the “gold standard” for measuring insulin sensitivity (Si-clamp). Here, we determined the reproducibility of serial HEC data in healthy subjects.

RESEARCH DESIGN AND METHODS:

The Pathobiology of Prediabetes in A Biracial Cohort study assessed incident prediabetes in healthy African Americans (AA) and European Americans (EA) with parental type 2 diabetes mellitus during 5.5 years of follow-up. Assessments included anthropometry, OGTT, and HEC. Ninety subjects (44 AA, 46 EA) who underwent Year-1 HEC consented to Year-3 HEC. We calculated coefficients of variation (CVs), 95% limits of agreement, and repeatability coefficients for Year-1 and Year-3 data, and assessed the association of change in Si-clamp with incident prediabetes.

RESULTS:

The mean (SD) baseline age was 47.5 ± 8.13y, body mass index was 30.4 ± 9.16 kg/m², fasting plasma glucose was 93.7 ± 7.82 mg/dl and 2-hrPG was 126 ± 26.8 mg/dl. Si-clamp (umol/kg/min.pmol/L-1) was 0.071 ± 0.04 in Year 1 and 0.067± 0.04 in Year 3 (P =0.22). Year 1 and Year 3 values were strongly correlated (r=0.81, P<0.0001); the CV was 13.6% and repeatability coefficient was ± 0.025. Intrasubject differences in serial Si-clamp were less than the repeatability coefficients and within the 95% limits of agreement. After 5.5 years of followup, 40 subjects progressed to prediabetes and 50 were nonprogressors. The change in Si-clamp was greater in progressors than nonprogressors (−10% vs. −2.5%, P=0.02).

CONCLUSIONS:

The HEC is reproducible over ~2 years in free-living individuals, with a temporal decline in Si-clamp that predicts prediabetes risk.

1.1. INTRODUCTION

Insulin resistance (or decreased insulin sensitivity) is a characteristic of diverse medical conditions, including overweight/obesity, physical inactivity, hypertension, dyslipidemia, metabolic syndrome, polycystic ovary disease and type 2 diabetes mellitus (T2DM) (1-4). Methods for estimating insulin sensitivity range from anthropometry (5,6), calculation of formulaic indices based on the ratio of insulin and glucose levels in the fasting state (7,8) or during oral glucose tolerance test (OGTT) (9), to more complex approaches, such as minimal modeling (10) and the hyperinsulinemic euglycemic clamp (HEC) (11). Of the latter, the HEC is regarded as the “gold standard” methodology that has been used widely for the direct assessment of insulin sensitivity (11-13).

During HEC, a constant intravenous infusion of insulin is administered to raise plasma insulin levels to ~10-fold above the fasting physiological range. To prevent hypoglycemia and maintain euglycemia, dextrose (20%) is concurrently infused at variable rate that is adjusted based on frequently sampled (every 5-10 min) bedside plasma glucose levels. A typical HEC lasts 120-180 minutes; additional time usually is required for post-clamp stabilization (11, 13). Owing to the methodological complexity, reports of multiple HEC procedures in the same subjects are scant (14-17). Potential sources of variability that could affect the reproducibility of clamp-derived insulin sensitivity values include technical factors, such as operator-dependent decision on the magnitude of adjustment of glucose infusion rate in response to a given plasma glucose level. Being an informed guess (based on operator’s experience), over-shooting or undershooting of the euglycemic target could occur with varying frequencies during the 180 minutes. In addition, changes in lifestyle habits, body weight, health status, and other environmental exposures could alter insulin sensitivity values obtained measured at different times.

Given the many potential sources of variability, we were interested in determining the temporal stability and reproducibility of data derived from repeated HEC studies in otherwise healthy, free-living individuals. In previous reports based on seven (14) or ten (15) healthy subjects, who underwent repeated HEC studies separated by intervals of 2 days to 14 days, the intraindividual coefficient of variation (CV) for insulin-stimulated glucose disposal was reported as 5.8% (14) and 10.3% (15), respectively. Bokemark et al., who studied 32 older men (mean age 58 years) with consecutive HEC, separated by 2 weeks, reported a CV of 15% for insulin-mediated glucose disposal (16). In another report, the intraindividual CV in insulin-mediated glucose disposal was 21% among of 8 subjects with non-alcoholic fatty liver disease, who underwent two HEC studies, performed 2 months apart (17).

In the present study, we aimed to assess the reproducibility of HEC measurements in a larger, diverse cohort studied more than two years apart in the longitudinal Pathobiology of Prediabetes in A Biracial Cohort (POP-ABC) study (18-20). The POP-ABC study enrolled initially normoglycemic African Americans (AA) and European Americans (EA) with parental T2DM and followed them for the occurrence of incident prediabetes over 5 years. Our assessment utilized a rigorous approach that included calculation of intraindividual CV, repeatability coefficients, and the 95% limits of repeatability for HEC procedures performed in 90 POP-ABC participants during Year 1 and Year 3 of the study. These were the participants who consented to undergo repeated HEC studies. We further analyzed the relationship between longitudinal change in Si-clamp and the risk of progression from normoglycemia to prediabetes.

2.1. RESEARCH DESIGN AND METHODS

2.1.1. Participants

Study subjects were participants in the POP-ABC study (18-20). Eligible for enrollment in POP- ABC study were individuals with a history of T2DM in one or both biological parents, whose screening OGTT showed normal fasting plasma glucose (FPG) (<100 mg/dL {5.6 mmol/L}) and normal glucose tolerance (NGT) (2-hour plasma glucose {2hrPG} <140 mg/dL {7.8 mmol/L}) (18-20). Persons with diabetes or those taking antidiabetic medications or other pharmacological agents known to affect glucose metabolism or body weight were excluded. Additional exclusion criteria were enrollment in a weight loss program, a history of bariatric surgery (including liposuction), and recent hospitalization within 6 weeks of the planned screening visit (18-20). The POP-ABC study protocol was approved by the University of Tennessee Institutional Review Board. All participants gave written informed consent prior to initiation of the study, which was conducted within the University of Tennessee Health Science Center General Clinical Research Center (GCRC).

2.1.2. Assessments

After an approximately 8-12-hour overnight fast, participants arrived at the GCRC for initial procedures, which included a medical interview, clinical examination, anthropometric measurements (height, weight, waist circumference), and standard 75-g oral glucose tolerance test (OGTT). The body mass index (BMI) was calculated as weight in kilograms divided by the height in meters squared.

2.1.3. Hyperinsulinemic euglycemic clamp

Insulin sensitivity was measuring with the HEC technique, as described by DeFronzo, Tobin and Andres (11). Briefly, after an 8-12 hour overnight fast, participants presented to the GCRC for the HEC procedure. Study subjects were given written instructions to maintain their usual habits and to avoid strenuous physical exercise or major changes in food intake 3 days before study visit. Participants were allowed to drink water in the morning of the study, but refrained from smoking.

At ~0800 on the day of the study, intravenous catheters were inserted in both arms for separate infusions of insulin and glucose, and for blood sampling. Regular human insulin was infused at a continuous rate of 2 mU.kg⁻¹.min⁻¹ (12 pmol.kg⁻¹.min⁻¹) for 180 min along with a variable-rate dextrose (20%) infusion to maintain a target plasma glucose level of ~100 mg/dl (5.6 mmol/L). Blood sampling for determination of bedside glucose levels was performed every 5-10 minutes. To ensure “arterialization” of the blood specimens, the sampling forearm was placed in a thermoregulated chamber (~65-70° C). The mean steady-state plasma glucose and insulin levels during the final hour of the clamp procedures were calculated by averaging plasma levels obtained from 120-180 minutes. Insulin-stimulated glucose disposal (M) was derived from the glucose infusion rate (GIR) during the last 60 minutes of steady state (11). The M value was corrected for ambient steady-state plasma insulin levels to obtain a final measure of insulin sensitivity (Si-clamp), as previously described (11).

Following initial assessment during Year 1, the HEC was repeated in Year 3 in POP-ABC study participants who consented to undergo repeated testing. Different members of the research team performed the HEC procedure, using a standard protocol, during Year 1 and Year 3. It was not feasible to assign participants to a dedicated clamp operator during the study visits.

2.1.4. Determination of glycemic progression

The primary outcome of the main POP-ABC study was progression from normoglycemia to incident prediabetes (18-21). Prediabetes was diagnosed as impaired fasting glucose (IFG), indicated by FPG values of 100-125 mg/dL (5.5-6.9 mmol/L) or impaired glucose tolerance (IGT), indicated by 2hrPG values of 140-199 mg/dL (7.8 – 11.0 mmol/L) (22, 23). All endpoints were confirmed using the standard 75-g OGTT and independently adjudicated by the Institutional Data and Safety Officer (Murray Heimberg, MD, PhD).

2.1.5. Laboratory analyses

Plasma glucose levels were measured using the glucose oxidase method (Yellow Spring Instruments Co., Inc., Yellow Spring, OH). Plasma insulin was measured with a chemiluminescent assay, using commercial kits (Immulite, Siemens Ltd., Llanberis, Gwynedd, UK). The sensitivity of the insulin assay was 2 μIU/ml and the within-run and between-run coefficients of variation were 4.7% and 8%, respectively.

2.1.6. Statistical analysis

Data are reported as means ± standard deviations (SD); significance level was set as P< 0.05. Differences in continuous variables between defined groups were analyzed using descriptive statistics. Means and SD of the within-subject differences between Year-1 and Year-3 HEC measurements were calculated. The intraindividual coefficient of variation (CV) was calculated as the individual SD for the values obtained during Year-1 and Year-3 HEC study days divided by the mean of the two values and expressed as percent. Repeatability was analyzed according to Bland and Altman (24, 25). The 95% limits of agreement were defined as the mean difference ± 1.96 SD of the within-subject differences in Year-1 and Year-3 HEC measurements (24,25). The repeatability coefficient was calculated as 2.77Sw (where Sw is the within-subject SD for the repeated HEC measurements) (24, 25).

Linear regression models were used to analyze the relationships between changes in insulin sensitivity vs. putative predictive variables. All analyses were performed using StatView statistical software (SAS Institute Inc., Cary, NC).

3.1. RESULTS

3.1.1. Characteristics of study subjects

A total of 90 adults (44 AA, 46 EA; 62 women, 28 men) underwent HEC studies during Year 1 and Year 3 of the POP-ABC study. The mean interval between the two clamp procedures was 2.30 ± 0.24 years (median 2.28; range 2.00 – 3.53 years). The mean age of the participants at enrollment was 47.5 ± 8.13 years (median 48 years; range 26-65 years), and their body mass index was 30.4 ± 9.16 kg/m². The mean baseline FPG was 93.7 ± 7.82 mg/dl and mean 2-hrPG was 126 ± 26.8 mg/dl. Apart from a significant increase in age, the mean baseline characteristics remained stable upon re-testing in Year 3 (Table 1).

Table 1.

Characteristics of Study Participants

	Year-1	Year-3	P value
Number	90	90
AA/EA	44/46	44/46
Female/Male	62/28	62/28
Age (yr)	47.5 ± 8.20	49.8 ± 8.14	<0.0001
Weight (kg)	84.1± 18.1	84.8 ± 18.8	NS
FPG (mg/dL)	93.7 ± 7.82	94.8 ± 7.92	NS
2hrPG (mg/dL)	126 ± 26.8	127 ± 27.7	NS

AA, African American; EA, European American; FPG, fasting plasma glucose; 2hrPG, two-hour plasma glucose; NS, not significant. To convert glucose from mg/dL to mmol/L, divide by 18.

3.1.2. Clamp measures

Figure 1 shows plasma glucose and insulin levels during Year-1 and Year-3 clamp procedures. The steady-state plasma glucose level attained during the final 60 minutes of the clamp procedure was similar in Year 1 and. Year 3, and the 2-mg/dL numerical difference was within the limits of agreement and the repeatability coefficient (Table 2). The CV of the steady-state plasma glucose value between the two HEC studies was 2.55%. The “control deviation” (i.e., the mean absolute deviation of plasma glucose from the target of 100 mg/dl) was 2.7 mg/dl and 3. 4 mg/dl during Year 1 and Year 3 clamp, respectively.

Figure 1.

Plasma glucose (A) and insulin (B) levels during Year 1 (blue) and Year 3 (red) hyperinsulinemic euglycemic clamps. Insulin was infused continuously at a rate of 2 mU.kg⁻¹.min⁻¹ (14.4 pmol.kg⁻¹.min⁻¹) along with a variable-rate infusion of dextrose (20%) during the clamp procedures.

Table 2.

Comparison of Year-1 and Year-3 Hyperinsulinemic Euglycemic Clamp Studies in 90 Healthy Subjects

	Year-1 Study	Year-3 Study	95% Limits of agreement	Repeatability coefficient	Coefficient of variation (%)
SSPG (mg/dL)	102 ± 5.30	104 ±6.00	−12.6 to 14.7	± 7.76	2.55
SSPI (pmol/L)	718 ± 265	807 ± 360	−357 to 232	± 269	8.22
GIR (mg/kg/min)	9.20 ± 3.50	9.80 ± 3.40	−5.87 to 4.67	± 3.10	11.7
Si-clamp (μmol/kg/min.pmol/L−1)	0.085 ± 0.046	0.081 ± 0.048	−0.065 to 0.068	± 0.033	13.6

The difference was calculated as Year 3 – Year 1 clamp values. SSPG, steady-state (120-180 min) plasma glucose; SSPI, steady-state (120-180 min) plasma insulin; GIR, glucose infusion rate; Si-clamp, insulin sensitivity (GIR in micromoles adjusted for SSPI in pmol/L) expressed as micromol/kg/min.pmol/L⁻¹.

The steady-state plasma insulin level was 718 ± 265 pmol/L during Year-1 and 807 ±360 pmol/L during Year-3 HEC studies (P<0.001), with 95% limits of agreement of −357 to 232 pmol/L, repeatability coefficient of ± 269 and CV of 8.22% (Table 2). The intrasubject differences in steady-state plasma insulin levels were contained within the 95% limits of agreement and the repeatability coefficient (Table 2). The GIR, a measure of insulin-mediated glucose disposal, was higher during Year-3 (9.80 ± 3.40 mg/kg/min) than Year-1 (9.20 ± 3.50 mg/kg/min) HEC study (P<0.001), consistent with the higher steady-state plasma insulin levels during Year-3 procedure (Table 2). The Si-clamp (i.e. GIR adjusted for ambient insulin levels) decreased insignificantly during Year-1 vs. Year-3 studies (0.071 ± 0.04 vs. 0.067± 0.04, P=0.22). The within-subject differences in GIR and Si-clamp values were contained within the respective 95% limits of agreement and repeatability coefficients (Table 2). The CV was 11.7% and 13.6% for GIR and Si-clamp values, respectively (Table 2).

Figure 2 shows the scatter diagram of individual Si-clamp values determined during Year 1 and Year 3. Both values were highly correlated (r=0.81, P<0.0001). Despite the high reproducibility, there were modest temporal changes in Si-clamp values, which were significantly correlated with changes in weight (r= −0.35, P=0.009). The mean Si-clamp was ~15% lower in AA than EA participants in Year-1 (0.065 ± 0.04 vs. 0.076 ± 0.04) and Year-3(0.063 ± 0.044 vs. 0.074 ± 0.039), although the differences were not statistically significant (P=0.21). There were no significant differences in insulin sensitivity by sex.

Figure 2.

Correlation of Year 1 (x-axis) and Year 3 (y-axis) insulin sensitivity (Si-clamp) values in African-American (closed symbol) and European-American (open symbol) study subjects.

3.1.3. Incident Prediabetes

During a follow-up period of 5.5 years (mean 2.62 years), 40 subjects progressed to prediabetes and 50 were nonprogressors. As previously reported in the main results from the POP-ABC study (21), participants who subsequently progressed to prediabetes (progressors) had lower insulin sensitivity at baseline compared with participants who maintained normoglycemia (nonprogressors). In the present subgroup of POP-ABC participants who underwent serial HEC studies, the lower baseline insulin sensitivity among progressors was evident (0.075 ± 0.045 vs. 0.093 ± 0.049 μmol/kg/min.pmol/L⁻¹, P=0.04). Data from the follow-up HEC measurement indicated that insulin sensitivity remained stable of among nonprogressors (Year 1 vs. Year 3 Si-clamp: 0.093 ± 0.049 vs. 0.091 ± 0.047 μmol/kg/min.pmol/L⁻¹) but declined significantly by ~10% decrease among progressors to prediabetes (Year 1 vs. Year 3 Si-clamp: 0.075 ± 0.045 vs. 068 ± 0.047, P=0.025). Thus, the Year-3 Si-clamp data indicate that the lower insulin sensitivity observed at baseline in progressors persisted and worsened during follow up, compared with non nonprogressors (Figure 3).

Figure 3.

Comparison of Year 1 and Year 3 insulin sensitivity (Si-clamp) values in participants who developed incident prediabetes (progressors, open bars) and those who maintained normoglycemia (nonprogressors, striped bars) during 5 years of follow-up. * P= 0.04, ** P= 0.02 for comparison between progressors and nonprogressors. Repeated measures ANOVA P=0.025 for change in Si-clamp from Year 1 to Year 3 among progressors.

4.1. DISCUSSION

The hyperinsulinemic euglycemic clamp is the “gold standard” method for assessment of insulin sensitivity (11-17). However, data on the technical performance parameters derived from HEC during repeated measurements are scant. Here, we have shown that insulin sensitivity measures derived from repeated HEC studies performed during a mean interval of 2.3 years in a large group of healthy, free-living people are reproducible, with a correlation coefficient of 0.81. We found that the intraindividual CV of repeated measurements of insulin-mediated glucose disposal (GIR) was 11.7% and that of the related Si-clamp was 13.6%. These CVs were within the range of 5.8% - 21% previously reported from studies that assessed the reproducibility of HEC in fewer subjects over much shorter intervals of 2 days to 2 months (14-17). Besides the smaller sample sizes, the previous studies lacked age, gender and ethnic diversity (14-17). Thus, the findings from our much larger sample of men and women, with nearly equal representation of black and white subjects, confirm and extend previous reports of the reproducibility of the HEC procedure in a diverse population with a broad age range (26-65 years).

The CV (ratio of the SD to the mean) indicates the extent of agreement between the means of repeated measurements. More robust measures of reproducibility proposed by Bland and Altman include the 95% limits of agreement and the repeatability coefficient (24,25). The 95% limits of agreement encompass the mean difference ± 1.96 SD of the within-subject differences for repeated measurement. (24,25). Derived from the within-subject SD for repeated measurements, the repeatability coefficient is a measure of the variation of replicated measurements obtained by the same method under similar conditions. A test is considered reproducible if intraindividual differences lie within the 95% limits of agreement and are smaller than the repeatability coefficient (24,25). Using these metrics, outputs from HEC studies (namely, steady-state plasma glucose and insulin levels, GIR and Si-clamp), performed 2.3 years apart, are quite reproducible, with commensurate repeatability coefficients and 95% limits of agreement.

Although not the main focus of the present study, our longitudinal cohort enabled us to evaluate the association between serial changes in insulin sensitivity and the risk of incident prediabetes. As reported in the main results from the POP-ABC study, a total of 101 (52 AA, 49 EA) out of 343 enrolled subjects (192 AA, 151 EA) experienced glycemic progression from normoglycemia to prediabetes during 5 years of follow-up (mean 2.62 years) (21). Among the predictors of incident prediabetes was lower baseline insulin sensitivity (21), a finding that was consistent in lean as well as obese partcipants (26). Consistent with our previous findings, the present study (based on 90 POP-ABC participants who underwent HEC in Year 1 and Year 3) showed lower Year-1 Si-clamp values in progressors to prediabetes compared with nonprogressors. Additionally, analysis of Year-3 data showed a markedly greater decline in insulin sensitivity among progressors compared with nonprogressors. Specifically, progressors to prediabetes showed approximately 4% per year decline in insulin sensitivity, compared with ~1% per year in nonprogressors.

There are several limitations of the present study. First, the main POP-ABC study enrolled only subjects with parental history of T2DM, which may limit the generalizability of our findings. Second, we did not assign a dedicated HEC operator for each subject but used different operators to perform the clamp procedures, which poses a methodological risk of inter-operator variability. Third, we had no influence on individual behaviors (including dietary and physical activity practices) that could affect insulin sensitivity among our study participants between the two clamp measurements. Indeed, physical activity is an important longitudinal predictor of insulin sensitivity (27). Furthermore, we utilized an insulin infusion dose of 2 mU/kg/min in the present study, whereas others have used a lower dose (1 mU/kg/min or 40 mU/m²/min). In a study comparing progressors from NGT to T2DM versus nonprogressors, insulin sensitivity did not differ significantly between the groups when assessed with low dose infusion (40 mU/m²/min); however, maximal insulin-stimulated glucose disposal (assessed using an insulin infusion dose of 400 mU/m²/min) decreased by 31% in progressors, but remained unchanged in nonprogressors (28). Although direct comparison is difficult due to possible variations due to differences in study population, insulin clearance, and research and analytical methodologies, the study by Weyer et al. (28) showed that the higher dose of insulin infusion uncovered differences in insulin sensitivity between progressors and non-progressors. Our study, which utilized an insulin infusion dose that was 2-fold higher than the low dose but much lower than the high dose employed by Weyer et al. (28), demonstrated a significant decrease in insulin sensitivity in progressors to prediabetes compared with nonprogressors. It is unknown whether use of a lower dose of insulin infusion in our study would have produced different results.

Despite these limitations, it is remarkable that we found such a robust correlation (r=0.81) between Si-clamp values measured more than two years apart. Arguably, we might have seen an even closer correlation between Year-1 and Year-3 clamp values, had we controlled for operator variability and lifestyle habits. As the POP-ABC is a natural history study, no pharmacological or lifestyle interventions were offered to study participants. Thus, our present findings demonstrate the reproducibility and relative stability of hyperinsulinemic euglycemic clamp-derived measures of glucose disposal and insulin sensitivity among free-living white and black adults.

Highlights.

1. Insulin-mediated glucose disposal, assessed with the hyperinsulinemic euglycemic clamp, is reproducible in healthy African Americans and European Americans.

2. The correlation coefficient between consecutive clamp measurements performed after an interval of more than 2 years was 0.81 (P<0.0001) and the coefficient of variation was 13.6%.

3. Intraindividual differences in outputs from serial clamp studies (including steady-state plasma glucose and insulin levels and glucose infusion rates) were well within the 95% limits of agreement for repeated measures.

4. During 5.5 years of following, the change in insulin sensitivity was associated with risk of incident prediabetes among the initially normoglycemic study cohort.

List of Abbreviations

AA

African Americans

BMI

Body mass index

EA

European Americans

FPG

Fasting plasma glucose

GCRC

General Clinical Research Center

GIR

Glucose infusion rate

HEC

Hyperinsulinemic euglycemic clamp

HOMA-B

Homeostasis model assessment of beta-cell function

HOMA-IR

Homeostatic model assessment of insulin resistance

NGT

Normal glucose tolerance

OGTT

Oral glucose tolerance test

POP-ABC

Pathobiology of Prediabetes in a Biracial Cohort

Si-clamp

Insulin sensitivity measured using hyperinsulinemic euglycemic clamp

SSPG

Steady state plasma glucose

SSPI

Steady state plasma insulin

2hrPG

Two-hour post-load plasma glucose during OGTT

T2DM

Type 2 diabetes mellitus