OBJECTIVE

To examine the ethnic differences in insulin sensitivity (SI) as measured by the minimal model approach (SI-MM) and the reference method, the euglycemic-hyperinsulinemic clamp (EHC).

RESEARCH DESIGN AND METHODS

In a prospective study design, thirty Black Americans (BA) were age, sex, and BMI matched with non-Hispanic Whites (NHW). Participants underwent frequently sampled intravenous tolerance test (FSIVGTT) and EHC on 2 separate days during a single visit.

RESULTS

SI-MM values were significantly lower in BA when compared with NHW (0.035 ± 0.025 vs. 0.058 ± 0.036 [dL/min]/[μU/mL]; P = 0.003). However, there were no ethnic differences in SI measured by EHC (0.028 ± 0.012 vs. 0.035 ± 0.019 [dL/min]/[μU/mL]; P = 0.18).

CONCLUSIONS

SI-MM systematically underestimates SI in BA when compared with NHW. These findings suggest that studies inferring lower SI in BA based on FSIVGTT and SI-MM should be interpreted cautiously.

The higher death rate and clinical severity among Black Americans (BA) during the ongoing coronavirus disease 2019 (COVID-19) pandemic have highlighted the increased prevalence of type 2 diabetes in this population. It is widely accepted that the lower insulin sensitivity (SI) in BA accentuates their risk for diabetes compared with non-Hispanic White (NHW) Americans (1,2). Understanding ethnic phenotypic variability of SI is vital in ensuring robust outcomes in preventing, diagnosing, and treating metabolic disorders. Therefore, it is crucial to obtain accurately quantified SI measures, especially within high-risk populations.

The reference test for the measurement of SI is the euglycemic-hyperinsulinemic clamp (EHC) technique. Deemed more feasible because of clinical accessibility and lower costs, minimal model analysis of a frequently sampled intravenous glucose tolerance test (FSIVGTT) is often used to infer SI (SI-MM). Widely cited studies, primarily using FSIVGTT, have reported lower SI in BA (1,2). Ethnicity affects the predictive ability of some surrogate indices of insulin resistance that rely on ambient glucose and insulin concentrations (3). Whether ethnicity similarly affects the reliability of SI-MM to accurately predict SI as determined by EHC is unknown. In this study, we examined the ethnic differences in the ability of SI-MM to predict SI as measured by EHC.

This study protocol was approved by the Institutional Review Board of the National Institute of Diabetes and Digestive and Kidney Diseases and was conducted at the Clinical Research Center, National Institutes of Health, in Bethesda, Maryland. Thirty BA and 30 NHW, matched for age, sex, and BMI, were prospectively enrolled in the Study of the Phenotype of Overweight and Obese Adults (ClinicalTrials.gov identifier NCT00428987). Written informed consent was attained from all participants. Participants were admitted for a 3-night visit to the National Institutes of Health Metabolic Research Unit. After an overnight fast, SI was evaluated by the EHC (glucose disposal rate [GDR]) and insulin-modified FSIVGTT (SI-MM) on different days in random order as previously described (4). The rate of glucose disposal (M), a measure of SI was defined as the average of the glucose infusion rate during the steady state (GDR in milligrams per minute) corrected for estimated metabolic body size. The steady-state period of the clamp was defined as a ≥20-min period, 90 min after the initiation of the clamp, where the coefficient of variation for plasma glucose and glucose infusion rate was <5%. Prior to EHC, basal hepatic glucose production (HGP) and hepatic insulin resistance index were measured by using a stable isotope tracer (4). Because of the interruption in pharmacy compounding services and access to tracers, we could conduct tracer studies in only 43 individuals (BA n = 21; NHW n = 22). In addition, we measured circulating IGF binding protein 1 (IGFBP-1), a marker for hepatic SI. Minimal model analysis of the FSIVGTT was used to estimate glucose effectiveness (Sg) and SI-MM values as previously described using MINMOD software (version 6.02) (MinMOD Millenium, Los Angeles, CA) (4). Precision of parameter estimates from SI-MM was assessed by fractional SD. Mean fractional SD of parameter estimates were <10%. Measures of SI from EHC (Sc-Clamp) and SI-MM (Sc-MM), specifically, change in glucose clearance per change in plasma insulin concentration, were expressed in the same units as originally described (5). A model-independent index of SI (calculated SI) was calculated and is related to KG/AUCD (6). KG is the rate of glucose disappearance (slope of log glucose), and AUCD is defined as the dynamic area under the insulin curve in FSIVGTT (0–50 min).

Statistical Analyses

Variables are expressed as mean ± SD. Comparisons between groups were assessed by the independent unpaired t test or Wilcoxon-Mann-Whitney test. A P value <0.05 was considered statistically significant. Data were analyzed with JMP software (version 13.0) (SAS Institute, Cary, NC) and GraphPad Prism 7 software (GraphPad Software, Inc, La Jolla, CA).

Percentage body fat, total body fat, fat-free mass, and fasting plasma glucose and insulin concentrations were similar between the groups (Table 1). BA had a significantly greater A1C than NHW. Six BA and three NHW had impaired fasting glucose (P = 0.27). Fifteen BA and four NHW had prediabetes based on A1C levels (P = 0.002). QUICKI, a surrogate measure of SI based on fasting glucose and insulin concentrations, was not significantly different between the groups (P = 0.07) (Table 1). Direct measurement of SI by EHC was similar between NHW and BA (Table 1). However, SI-MM values were significantly lower in BA when compared with NHW (Table 1). Similarly, when these parameters were expressed in the same units, Sc-Clamp was not significantly different between the groups, but Sc-MM was lower in BA (Table 1). Simple linear regression analyses revealed a modest but significant relationship between log-transformed SI-MM and GDR values in BA (r = 0.44; P = 0.04) and NHW (r = 0.62; P = 0.003). Indeed, Deming regression analysis, which assumes measurement error in Sc-MM and Sc-Clamp, showed a fixed bias (y-intercept) between ethnic groups (P = 0.002). Indeed, a factor of 1.65 applied to Sc-MM in BA (Sc-MM ∗ 1.65) corrects the bias between the ethnic groups. Like SI-MM and Sc-MM values, calculated SI was lower in BA (BA 0.90 ± 0.58; NHW 1.41 ± 0.95 10−4 ⋅ [μU/mL]−1 ⋅ min−1; P = 0.04). We did not observe any significant ethnic differences in estimated whole-body glucose effectiveness by clamp (BA 0.037 ± 0.016; NHW 0.033 ± 0.020 dL ⋅ min−1 ⋅ kg−1; P = 0.40) or IVGTT (BA 0.019 ± 0.008; NHW 0.015 ± 0.006 min−1; P = 0.12). In a subset of our cohort, hepatic insulin resistance index was not different between the groups (BA 6.81 ± 2.77; NHW 6.82 ± 2.91 [mg/kg/min] ⋅ [ng/mL]; P = 0.94). Concentrations of circulating IGFBP-1 were similar (BA 11.7 ± 9.4; NHW 12.3 ± 6.7 ng/mL; P = 0.26).

Table 1

Clinical and metabolic characteristics in NHW and BA

NHW (n = 30)BA (n = 30)P*
Age (years) 38 ± 10 36 ± 11 0.49 
Sex (% female) 47 47 — 
BMI (kg/m229.2 ± 6.3 29.3 ± 6.8 0.98 
Total body fat (%) 33.6 ± 11.5 30.4 ± 11.7 0.29 
Fat-free mass (kg) 56.9 ± 10.1 61.4 ± 12.7 0.13 
Fasting plasma glucose (mg/dL) 89.5 ± 6.1 90.9 ± 8.2 0.42 
Fasting plasma insulin (μU/mL) 9.4 ± 6.8 12.1 ± 6.9 0.06 
Fasting C-peptide (ng/mL) 2.8 ± 1.3 2.8 ± 1.2 0.82 
Hemoglobin A1C (%) 5.3 ± 0.4 5.7 ± 0.4 0.001 
Hemoglobin A1C (mmol/mol) 34.9 ± 3.8 38.3 ± 3.8 0.001 
Total cholesterol (mg/dL) 179 ± 37 172 ± 28 0.63 
LDL cholesterol (mg/dL) 96 ± 38 98 ± 22 0.74 
HDL cholesterol (mg/dL) 59 ± 19 57 ± 10 0.69 
Triglycerides (mg/dL) 122 ± 77 80 ± 31 0.03 
QUICKI 0.35 ± 0.03 0.34 ± 0.04 0.07 
Acute insulin response to glucose (μU ⋅ mL−1 ⋅ min−1524 ± 618 1,127 ± 825 0.0004 
SI-MM (10−4 ⋅ [μU/mL]−1 min−13.88 ± 2.45 2.31 ± 1.54 0.01 
GDR, M (mg/kg fat-free mass + 17.7/min) 12.8 ± 4.7 12.6 ± 3.2 0.54 
Sc-MM ([dL/min]/[μU/mL]) 0.058 ± 0.036 0.035 ± 0.025 0.003 
Sc-Clamp ([dL/min]/[μU/mL]) 0.035 ± 0.019 0.028 ± 0.012 0.18 
NHW (n = 30)BA (n = 30)P*
Age (years) 38 ± 10 36 ± 11 0.49 
Sex (% female) 47 47 — 
BMI (kg/m229.2 ± 6.3 29.3 ± 6.8 0.98 
Total body fat (%) 33.6 ± 11.5 30.4 ± 11.7 0.29 
Fat-free mass (kg) 56.9 ± 10.1 61.4 ± 12.7 0.13 
Fasting plasma glucose (mg/dL) 89.5 ± 6.1 90.9 ± 8.2 0.42 
Fasting plasma insulin (μU/mL) 9.4 ± 6.8 12.1 ± 6.9 0.06 
Fasting C-peptide (ng/mL) 2.8 ± 1.3 2.8 ± 1.2 0.82 
Hemoglobin A1C (%) 5.3 ± 0.4 5.7 ± 0.4 0.001 
Hemoglobin A1C (mmol/mol) 34.9 ± 3.8 38.3 ± 3.8 0.001 
Total cholesterol (mg/dL) 179 ± 37 172 ± 28 0.63 
LDL cholesterol (mg/dL) 96 ± 38 98 ± 22 0.74 
HDL cholesterol (mg/dL) 59 ± 19 57 ± 10 0.69 
Triglycerides (mg/dL) 122 ± 77 80 ± 31 0.03 
QUICKI 0.35 ± 0.03 0.34 ± 0.04 0.07 
Acute insulin response to glucose (μU ⋅ mL−1 ⋅ min−1524 ± 618 1,127 ± 825 0.0004 
SI-MM (10−4 ⋅ [μU/mL]−1 min−13.88 ± 2.45 2.31 ± 1.54 0.01 
GDR, M (mg/kg fat-free mass + 17.7/min) 12.8 ± 4.7 12.6 ± 3.2 0.54 
Sc-MM ([dL/min]/[μU/mL]) 0.058 ± 0.036 0.035 ± 0.025 0.003 
Sc-Clamp ([dL/min]/[μU/mL]) 0.035 ± 0.019 0.028 ± 0.012 0.18 

Data are presented as arithmetic mean ± SD.

*

An unpaired, two-tailed Student t test (or Mann-Whitney U test for values that were not normally distributed) was used to test differences between ethnic groups. P values indicate significance for comparisons between ethnic groups.

Sc-MM is obtained by multiplying SI-MM by VD, where VD is the apparent volume of distribution of glucose and is equal to the ratio of the glucose dose to the increment in plasma glucose during FSIVGTT.

Sc-Clamp was defined as GDR/(G × ΔI), where G is steady-state blood glucose concentration (mg/dL), and ΔI is the difference between basal and steady-state plasma insulin concentrations (μU/mL).

In this study, despite similar M and Sc-Clamp in both groups, SI determined by SI-MM was ∼40% lower in BA. Sc-MM and Sc-Clamp are comparable, but not equivalent. SI is a measure of insulin-mediated glucose uptake and inhibition of HGP. In normal individuals, only ∼17% of SI is due to insulin inhibition of HGP, while the rest is due to insulin-stimulated glucose disposal (5). Nevertheless, measures of hepatic SI were similar in both groups and thus do not explicate the lower SI from SI-MM in BA. GDR from the clamp represents peripheral insulin- and glucose-dependent glucose disposal (glucose effectiveness). We did not observe any significant difference in estimated whole-body glucose effectiveness measures using the clamp or SI-MM. These results suggest that there is no ethnic bias in the measures of glucose disposal during EHC.

In SI-MM, SI is mathematically represented as the partial derivative of glucose disappearance on glucose insulin. Because of the inverse relationship between SI and insulin concentrations, the model likely underestimates SI in individuals who display higher insulin response, especially first-phase insulin secretion (AIR) (Table 1). The robust AIR (approximately 2-fold) and impaired clearance of insulin (337 ± 90 vs. 432 ± 208 mL ⋅ m−2 ⋅ min−1; P = 0.01) in BA may thus play a role in affecting the lumped parameters in the model and estimation of SI (7). Indeed, we recently reported that in simulated FSIVGTT, SI-MM underestimated SI because of its inverse relationship with AIR. This underestimation was context dependent and observed only when high AIR was the result of an increased size of the rapidly releasable pool of insulin (7). Consistent with our results, other studies using EHC have not demonstrated differences in SI between BA and NHW (3,810). In a systematic review of SI-MM studies, BA were more likely to demonstrate lower SI (2). However, presented here is the first prospectively designed study to demonstrate lower SI in BA by FSIVGT, but not EHC, in age-, sex-, and BMI-matched ethnic cohorts. Pisprasert et al. (3) showed that BA were more insulin resistant when assessed by the Matsuda index, HOMA for insulin resistance, and fasting insulin level, despite similar SI levels by EHC. These studies together question the reliability of surrogate measures in assessing SI in BA. Strengths of our study include the prospective study design, BMI matching, and use of the gold-standard EHC technique to compare SI. Limitations include self-reporting of ethnicity and small sample size. Nevertheless, a priori sample size calculation suggested that a sample size of n = 30 was sufficient to detect a 20% difference in SI (by EHC) between groups at a power of 80% and a type I error of 5%.

In conclusion, our results suggest ethnic differences exist in the predictive ability of SI-MM, and studies inferring lower SI in BA without diabetes based on FSIVGTT and minimal modeling should be interpreted cautiously.

Funding. This work was supported in part by the Intramural Research Program of the National Institute of Diabetes and Digestive and Kidney Diseases, Bethesda, MD.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. A.F. researched the data, created the figures, and co-wrote the manuscript. S.Y. participated in the study design, performed the study, researched the data, and edited the manuscript. C.S., S.G., and H.F. researched the data and reviewed and edited the manuscript. R.M. designed the study, obtained funding, performed the study, and co-wrote the manuscript. R.M. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

1.
Haffner
SM
,
Howard
G
,
Mayer
E
, et al
.
Insulin sensitivity and acute insulin response in African-Americans, non-Hispanic whites, and Hispanics with NIDDM: the Insulin Resistance Atherosclerosis Study
.
Diabetes
1997
;
46
:
63
69
2.
Kodama
K
,
Tojjar
D
,
Yamada
S
,
Toda
K
,
Patel
CJ
,
Butte
AJ
.
Ethnic differences in the relationship between insulin sensitivity and insulin response: a systematic review and meta-analysis
.
Diabetes Care
2013
;
36
:
1789
1796
3.
Pisprasert
V
,
Ingram
KH
,
Lopez-Davila
MF
,
Munoz
AJ
,
Garvey
WT
.
Limitations in the use of indices using glucose and insulin levels to predict insulin sensitivity: impact of race and gender and superiority of the indices derived from oral glucose tolerance test in African Americans
.
Diabetes Care
2013
;
36
:
845
853
4.
Armiyaw
L
,
Sarcone
C
,
Fosam
A
,
Muniyappa
R
.
Increased β-cell responsivity independent of insulin sensitivity in healthy African American adults
.
J Clin Endocrinol Metab
2020
;
105
:
e2429
e2438
5.
Bergman
RN
,
Prager
R
,
Volund
A
,
Olefsky
JM
.
Equivalence of the insulin sensitivity index in man derived by the minimal model method and the euglycemic glucose clamp
.
J Clin Invest
1987
;
79
:
790
800
6.
Tura
A
,
Sbrignadello
S
,
Succurro
E
,
Groop
L
,
Sesti
G
,
Pacini
G
.
An empirical index of insulin sensitivity from short IVGTT: validation against the minimal model and glucose clamp indices in patients with different clinical characteristics
.
Diabetologia
2010
;
53
:
144
152
7.
Ha
J
,
Muniyappa
R
,
Sherman
AS
,
Quon
MJ
.
When MINMOD artifactually interprets strong insulin secretion as weak insulin action
.
Front Physiol
2021
;
12
:
601894
8.
Stefan
N
,
Stumvoll
M
,
Weyer
C
,
Bogardus
C
,
Tataranni
PA
,
Pratley
RE
.
Exaggerated insulin secretion in Pima Indians and African-Americans but higher insulin resistance in Pima Indians compared to African-Americans and Caucasians
.
Diabet Med
2004
;
21
:
1090
1095
9.
Ebenibo
S
,
Edeoga
C
,
Wan
J
,
Dagogo-Jack
S
.
Glucoregulatory function among African Americans and European Americans with normal or pre-diabetic hemoglobin A1c levels
.
Metabolism
2014
;
63
:
767
772
10.
Bello
O
,
Mohandas
C
,
Shojee-Moradie
F
, et al
.
Black African men with early type 2 diabetes have similar muscle, liver and adipose tissue insulin sensitivity to white European men despite lower visceral fat
.
Diabetologia
2019
;
62
:
835
844
Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at https://www.diabetesjournals.org/content/license.