Validation of simple indices to assess insulin sensitivity based on the oral glucose tolerance test in the Japanese population

Diabetes Research and Clinical Practice 55 (2002) 229– 235 www.elsevier.com/locate/diabres

Validation of simple indices to assess insulin sensitivity based on the oral glucose tolerance test in the Japanese population Masao Kanauchi *, Nobuhiro Tsujimoto, Toshio Hashimoto First Department of Internal Medicine, Nara Medical Uni6ersity 840, Shijo-cho, Kashihara, Nara, Japan Received 15 May 2001; received in revised form 27 June 2001; accepted 20 July 2001

Abstract The homeostasis model assessment (HOMA) represents a simple index for evaluating insulin sensitivity, but clinical use is limited. Insulin sensitivity indices calculated from plasma glucose and plasma insulin concentrations after an oral glucose tolerant test (OGTT) have been proposed, but have not been validated in the Japanese population. We compared these indices with the euglycemic hyperinsulinemic clamp technique to evaluate the predicting insulin sensitivities in 77 Japanese subjects with varying degrees of glucose tolerance (normal glucose tolerance, n = 40; impaired glucose tolerance, n=22; and type 2 diabetes mellitus, n = 15). Insulin sensitivity was measured by the euglycemic hyperinsulinemic glucose clamp technique using an artificial pancreas, and expressed as the M-value. Weak inverse correlations existed between the HOMA index and M-values (r= − 0.227, P = 0.0468). An alternative index calculated by Matsuda’s formula correlated with the M-value (r= 0.450, P = 0.0001). A second index calculated by Stumvoll’s formula also correlated with the M-value (r= 0.641, P= 0.0001). Finally, a third index calculated by Gutt’s formula also significantly correlated with the M-value (r=0.526, P= 0.0001). All three indices are applicable for clinical use. The second index is the most sensitive measure of insulin sensitivity in the Japanese population. © 2002 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Insulin sensitivity; Oral glucose tolerance test; Glucose clamp; Homeostasis model assessment

1. Introduction In clinical investigation, the measurement of insulin sensitivity is often of interest because of its key role in diabetes, hypertension, and cardiovascular diseases. The homeostasis model assessment (HOMA) represents the most simple index for * Corresponding author. Tel.: + 81-744-22-3051; fax: + 81744-28-1880.

evaluating insulin sensitivity, particularly in epidemiologic studies [1], although some limitations have been reported, restricting its clinical use [2]. Recently, Matsuda and DeFronzo [3], Stumvoll et al. [4], and Gutt et al. [5] have described new insulin sensitivity indices calculated from plasma glucose and plasma insulin concentrations after an oral glucose tolerant test (OGTT). These values correlate closely with the glucose clearance rate estimated via the euglycemic hyperinsuline-

0168-8227/02/$ - see front matter © 2002 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0 1 6 8 - 8 2 2 7 ( 0 1 ) 0 0 3 1 3 - 8

230

M. Kanauchi et al. / Diabetes Research and Clinical Practice 55 (2002) 229–235

mic clamp method. These relationships, obtained from mainly Caucasian subjects, have not been confirmed in the Japanese population, as these metabolic abnormalities may differ among ethnic groups [6] especially because the Japanese population is relatively lean. Moreover, in the Japanese population, the capacity for insulin secretion is characterized by a decrease in the early-phase response to a glucose load [7]. This suggests that impaired insulin secretion, rather than impaired insulin sensitivity, is the major cause of glucose intolerance [8]. Therefore, validation of these insulin sensitivity indices in the Japanese population has not been demonstrated. We compared these insulin sensitivity indices calculated from an OGTT with the euglycemic hyperinsulinemic clamp method in order to evaluate the ability of each method to predict insulin sensitivity in the Japanese population.

gories based on the criteria currently proposed by the American Diabetes Association [9]: normal glucose tolerance (NGT), impaired glucose tolerance (IGT), and diabetes mellitus (DM).

2.3. E6aluation of glucose response and insulin secretion To assess glucose response and total insulin secretion, areas under the plasma response curve to glucose and insulin (AUCglu, AUCins) were calculated from the fasting, 30, 60, 120, and 180 min plasma glucose and insulin concentrations using the trapezoid rule [10]. Insulinogenic index, a widely used index of early-phase insulin response, was defined as the ratio of the increment of immunoreactive insulin to that of plasma glucose at 30 min after glucose loading.

2.4. Insulin sensiti6ity index 2. Methods

2.1. Patients Seventy-seven Japanese individuals, 56 men and 21 women aged 35–77 years (mean, 60.3), participated in this study after informed consent was obtained. Patients were excluded if they had a previous history of type 1 or type 2 diabetes mellitus with prior treatment by either insulin or an oral antidiabetic drugs. Patients with signs of congestive heart failure, chronic infectious diseases, renal failure, serious liver disease, or cancer were excluded. No patient received drugs affecting plasma glucose concentrations or insulin sensitivity.

2.2. Oral glucose tolerance test A standard 75-g OGTT was performed with plasma samples obtained at 0, 30, 60, 120, and 180 min after the glucose load. Plasma glucose was determined with an autoanalyzer using a glucose oxidase method, and immunoreactive insulin was measured by enzyme immunoassay (Entym Insulin test, Roche, Basel, Switzerland). Glucose tolerance was classified into three cate-

The HOMA index was calculated as the product of fasting plasma glucose and fasting plasma insulin divided by 22.5 [1]. Also, insulin sensitivity was assessed by the insulin sensitivity index (ISI) calculated using the OGTT values from three previously proposed formulas. The index, proposed by Matsuda and DeFronzo [3], was calculated as follows: ISI− M = 100 000/{[PG(0 min)× IRI(0 min)] × [PG(mean) × IRI(mean)]}1/2 where the ‘means’ were calculated from concentrations obtained throughout the OGTT. The second formula used was proposed by Stumvoll et al. [4]: ISI− S= 0.226− (0.0032×BMI) −[0.0000645×IRI(120 min)] −[0.00375×PG(90 min)] The third formula used was proposed by Gutt et al. [5]:

M. Kanauchi et al. / Diabetes Research and Clinical Practice 55 (2002) 229–235

ISI − G= m/{[PG(0 min)+ PG(120 min)] ×0.5/log[IRI(0 min) + IRI(120 min) × 0.5]} where m is the glucose uptake rate in peripheral tissues, calculated as: m ={75 000 mg+[PG(0 min)−PG(120 min)] ×0.19×BW}/120 min

2.5. Euglycemic hyperinsulinemic glucose clamp technique Insulin sensitivity was examined by the euglycemic hyperinsulinemic glucose clamp technique using the method of DeFronzo et al. [11] with an artificial pancreas (Model STG-22, Nikkiso, Tokyo, Japan). After an overnight fast, the euglycemic hyperinsulinemic clamp protocol was begun. A priming dose of 3.56 mU/kg per min of short-acting human insulin (Humulin R, Shionogi, Osaka, Japan) was administered during the initial 15 min. This was followed by a continuous insulin infusion at a rate of 1.12 mU/kg per min for 120 min to achieve steady-state hyperinsulinemia. Blood glucose levels were determined every 5 min during the clamp study. The glucose clamp level was set at 95 mg/dl during the 2-h clamp study. These procedures were performed automatically by glucose and insulin algorithms controlled by a computerized system. Plasma glucose concentrations were maintained at similar level in three groups. No difference in plasma insulin concentrations during clamp study were seen among three groups (71.49 8.9 mU/l, 73.6 9 9.9, and 69.999.3, respectively). Insulin sensitivity was estimated as the mean glucose infusion rate during the last 30 min of the clamp study (M-value).

2.6. Statistical analysis Values are presented as the mean9S.D. Statistical analysis was performed using ANOVA with either Scheffe’s test or Spearman’s correlation coefficient, as required. Differences were considered significant when P B0.05.

231

3. Results

3.1. Patient profiles Clinical and metabolic descriptions of study subjects are shown in Table 1. No significant differences were seen between study groups with respect to mean age, systolic and diastolic blood pressures, or serum concentrations of total cholesterol or triglycerides. Fasting plasma glucose were significantly higher in the IGT and DM groups than in NGT group, and were significantly higher in the DM than in IGT group. The same pattern prevailed for AUCglu. These two parameters increased significantly as glucose tolerance worsened (r= 0.674, PB0.0001; r= 0.895, P B0.0001; respectively). Fasting plasma insulin concentrations were significantly higher in the DM group than in the NGT and IGT groups. Plasma insulin concentrations after the 2 h glucose load were significantly higher in the IGT and DM groups than in the NGT group, and were significantly higher in the DM group than in the IGT group. These values increased significantly as glucose tolerance worsened (r= 0.533, PB 0.0001). AUCins was significantly higher in the DM group than in the NGT group. Insulinogenic index were significantly lower in the DM group than in the NGT. This parameters decreased significantly as glucose tolerance worsened (r= −0.324, P= 0.0047). The HOMA index were significantly higher in the DM group than in the NGT. This parameters increased significantly as glucose tolerance worsened (r= 0.365, P=0.0017). The M-value decreased significantly as glucose tolerance worsened (r= − 0.563, PB0.0001).

3.2. Plasma glucose and insulin concentrations during a 75 g OGTT Plasma glucose concentrations during a 75 g OGTT in subjects with NGT, IGT, and type 2 diabetes are presented in Fig. 1A. Fasting plasma insulin concentrations were significantly higher in the DM group than in the NGT and IGT groups (Fig. 2B). Conversely, plasma insulin concentrations at 30 min were significantly lower in the DM

232

M. Kanauchi et al. / Diabetes Research and Clinical Practice 55 (2002) 229–235

group than in the NGT group. Plasma insulin concentrations at 120 min were significantly higher in the IGT group than in the NGTgroup, and plasma insulin concentrations at 120 and 180 min were also significantly higher in the DM group than in the NGT and IGT groups.

3.3. Insulin sensiti6ity indices and M-6alue There was a weak inverse correlation between the HOMA index and M-value (r = −0.227, P = 0.0468)(Fig. 2). By Matsuda’s formula, ISI-M correlated significantly with the M-value (r= 0.450, P B 0.0001) (Fig. 3). By Stumvoll’s formula, ISI-S correlated strongly with the M-value (r= 0.641, P B0.0001) (Fig. 4). By Gutt’s formula, ISI-G also correlated significantly with the M-value (r= 0.526, P B0.0001) (Fig. 5).

4. Discussion The euglycemic hyperinsulinemic clamp method is thought to be the best way to measure insulin sensitivity, but its invasiveness and high cost have limited its use in clinical practice. Other alternative methods for measuring insulin sensitivity, such as the intravenous glucose tolerance test with minimal model analysis [12] necessitating frequent blood sampling, also are invasive and time consuming. These methods are not practical for clinical use. The HOMA is the simplest index for evaluating insulin sensitivity and is particularly useful in epidemiologic studies [1]. Bonora et al. [13] have reported an excellent correlation between the HOMA and the glucose clamp method in 115 subjects with various degree of glucose tolerance (half of whom had type 2 diabetes).

Table 1 Clinical characteristics of subjects

N Age (years) BMI (kg/m2) Systolic blood pressure (mmHg) Diastolic blood pressure (mmHg) Total cholesterol (mg/dl) Triglycerides (mg/dl) Fasting plasma glucose (mg/dl) 2-h plasma glucose (mg/dl) AUCglucose (mg/ml per h) Fasting plasma insulin (mU/ml) 2-h plasma insulin (mU/ml) AUCinsulin (mU ml−1 h) Insulinogenic index HOMA index M-value (mg/kg per min)

Normal glucose tolerance

Impaired glucose tolerance

Diabetes mellitus

40 57.6 9 10.5 23.39 13.4 124.59119.2

22 65.0 9 9.8a 23.492.2 125.9 918.4

15 60.6 9 7.7 26.6 9 4.7b,d 130.8 9 22.4

0.185 0.302 0.115

0.1063 0.0076 0.3215

72.79 15.3

71.89 12.3

73.6 912.4

0.013

0.9120

1839 29 1809 39 91.69 17.7 112.59 20.3 3769 38 7.559 3.88 41.89 24.6 1319 76 0.7439 0.546 1.699 0.84 6.26 9 1.95

185 9 34 185 934 97.0 98.6a 155.4 913.6b 470 9 56b 7.86 96.33 62.8 931.3a 142 9 57 0.536 90.401 1.89 9 1.50 5.30 91.62a

193 9 26 194 957 115.0 911.9b,d 239.2 9 32.9b,d 635 951b,d 11.13 9 4.45a,c 92.0 9 43.0b,d 179 967a 0.251 9 0.141 3.19 91.43 3.13 9 0.81b,d

0.119 0.118 0.674 0.897 0.895 0.249 0.533 0.245 −0.324 0.365 −0.563

0.3408 0.3474 0.0001 0.0001 0.0001 0.0290 0.0001 0.0315 0.0047 0.0014 0.0001

Data are expressed as the mean 9 S.D. a PB0.05 versus normal glucose tolerance. b PB0.01 versus normal glucose tolerance. c PB0.05 versus impaired glucose tolerance. d PB0.01 versus impaired glucose tolerance.

r

P

M. Kanauchi et al. / Diabetes Research and Clinical Practice 55 (2002) 229–235

233

Fig. 1. Plasma glucose (A) and insulin (B) concentrations during a 75 g oral glucose tolerant test in subjects with normal (“) IGT (), and type 2 diabetes ( ). Data are means 9 S.D. c , PB 0.05 versus NGT; c c, PB0.01 versus NGT; $, P B0.05 versus IGT; $$, P B 0.01 versus IGT.

Emoto et al. [14] also reported that the HOMA index correlates strongly with the insulin resistance index determined by the clamp method in 80 type 2 diabetic patients treated with sulfonylureas or diet alone. However, our study did not confirm the existence of a close relationship between HOMA and M-values. It is one of study limitations that we focused on only 15 diabetic patients who were either newly diagnosed, or had relatively mild type 2 diabetes. When statistical analyses for these diabetics were separately performed, the correlation coefficient between HOMA and M-value increased (but not statistically significant, r = −0.402, P = 0.137). Diabetic

group in our study was too small to produce reliable correlation coefficients, because we excluded overt diabetes treated with oral antidiabetic drugs or insulin. These conditions are not appropriate for administration of a 75 g OGTT from an ethical point of view. The OGTT has been used widely to evaluate glucose intolerance and insulin secretion. The validity of insulin sensitivity indices based on this test has recently been confirmed, by three teams of investigators [3–5]. Matsuda and DeFronzo [3] investigated 153 subjects (studied at University of Texas Health Science Center; although the ethnic distribution of the subjects was not described)

234

M. Kanauchi et al. / Diabetes Research and Clinical Practice 55 (2002) 229–235

Fig. 2. Correlation between the HOMA index and M-value, as measured using the euglycemic glucose clamp method.

with varying degrees of glucose tolerance. They proposed a simple index of insulin sensitivity derived from the OGTT, and found that their proposed measurement (ISI-M) correlated closely with the euglycemic insulin clamp study. However, these correlations were stronger in subjects with NGT or IGT than in type 2 diabetic individuals. This most likely reflects the decline in insulin secretion in type 2 diabetic patients. Stumvoll et al. [4] assessed 104 nondiabetic Caucasian individuals (65 NGT and 39 IGT patients) to determine whether plasma glucose and insulin levels during an OGTT predict insulin sensitivity. They reported that the parameter ISI-S correlates better with the glucose clamp technique than HOMA.

Fig. 3. Correlation between the insulin sensitivity index as proposed by Matsuda and DeFronzo (ISI-M), and M-value, as measured using the euglycemic glucose clamp method.

Fig. 4. Correlation between the insulin sensitivity index, as proposed by Stumvoll, et al. (ISI-S), and M-value, as measured using the euglycemic glucose clamp method.

Moreover, Gutt et al. [5] recently investigated 135 individuals (53 black and 82 white patients) with varying degrees of glucose tolerance (NGT, IGT, and diabetic subjects). They proposed a simple sensitive measure of insulin sensitivity (ISI-G) which correlates well with the euglycemic insulin clamp study. We note, however, these data were obtained from mostly Caucasians and ethnic groups other than Japanese. In regard to insulin sensitivity, ethnic differences must be considered [15]. Caucasians, Pima Indians, and MexicanAmericans are relatively obese. Conversely, most Japanese subjects have a lower BMI than in these ethnic groups. Therefore, in Japanese subjects, alternative assessment tools for measuring insulin

Fig. 5. Correlation between the insulin sensitivity index, as proposed by Gutt, et al. (ISI-G), and M-value, as measured using the euglycemic glucose clamp method.

M. Kanauchi et al. / Diabetes Research and Clinical Practice 55 (2002) 229–235

sensitivity remain a matter of controversy. We have presented that the insulin sensitivity index, as proposed by Stumvoll et al., ISI-S is the most sensitive measure of insulin sensitivity, even in the Japanese population. In summary, our study demonstrates that these alternative indices are adaptable for use in a clinical setting. Of these indices, ISI-S is the most sensitive measure of insulin sensitivity in the Japanese population, as validated against the clamp method.

[7]

[8]

[9]

[10]

References [11] [1] D.R. Matthews, J.P. Hosker, A.S. Rudenski, B.A. Naylor, D.F. Treacher, R.C. Turner, Homeostasis model assessment, Diabetologia 28 (1985) 412 –419. [2] S.D. Prato, P. Pozzilli, FIRI: fasting or false insulin resistance index?, Lancet 347 (1996) 132. [3] M. Matsuda, R.A. DeFronzo, Insulin sensitivity indices obtained from oral glucose tolerance testing, Diabetes Care 22 (1999) 1462 –1470. [4] M. Stumvoll, A. Mitrakou, W. Pimenta, T. Jenssen, H. Yki-Jarvinen, T. Van Haeften, W. Renn, J. Gerich, Use of oral glucose tolerance test to assess insulin release and insulin sensitivity, Diabetes Care 23 (2000) 295 –301. [5] M. Gutt, C.L. Davis, S.B. Spitzer, M.M. Llabre, M. Kumar, E.M. Czarnecki, N. Schneiderman, J.S. Skyler, J.B. Marks, Validation of the insulin sensitivity index comparison with other measures, Diabetes Res. Clin. Pract. 47 (2000) 177 –184. [6] M.F. Saad, S. Lillioja, B.L. Nyomba, C. Castillo, R. Ferraro, M. DeGregorio, E. Ravussin, W.C. Knowler, P.H. Bennett, B.V. Howard, Racial differences in the

[12]

[13]

[14]

[15]

235

relation between blood pressure and insulin resistance, New Engl. J. Med. 324 (1991) 733 – 739. K. Kosaka, T. Kuzuya, R. Hagura, H. Yoshinaga, Insulin response to oral glucose load in consistently decreased in established non-insulin-dependent diabetes mellitus, Diabetic Med. 13 (1996) S109 – S119. H. Yoneda, H. Ikegami, Y. Yamamoto, et al., Analysis of early-phase insulin responses in nonobese subjects with mild glucose intolerance, Diabetes Care 15 (1992) 1517 – 1521. The Expert Committee on the Diagnosis and Classification of Diabetes Mellitus, Report of the expert committee on the diagnosis and classification of diabetes mellitus, Diabetes Care 22 (Suppl. 1) (1999) S5 – S19. M.W. Schwartz, E.J. Boyko, S.E. Kahn, E. Ravussin, C. Bogardus, Reduced insulin secretion, J. Clin. Endocrinol. Metab. 80 (1995) 1571 – 1576. R.A. DeFronzo, J.A. Tobin, R. Andres, Glucose clamp technique; a method for quantifying insulin secretion and resistance, Am. J. Physiol. 237 (1979) E214 – 223. R.N. Bergman, R. Prager, A. Volund, J.M. Olefsky, Equivalence of insulin sensitivity index in man derived by the minimal model method and the euglycemic glucose clamp, J. Clin. Invest. 79 (1987) 790 – 800. E. Bonora, S. Kiechel, J. Willeit, F. Oberhallenzer, G. Egger, G. Targher, M. Alberiche, R.C. Bonadonna, M. Muggeo, Prevalence of insulin resistance in metabolic disorders: the Bruneck Study, Diabetes 47 (1998) 1643 – 1649. M. Emoto, Y. Nishizawa, K. Maekawa, Y. Hiura, H. Kanda, T. Kawagishi, T. Shoji, Y. Okuno, H. Moni, Homeostasis model assessment as a clinical index of insulin resistance in type 2 diabetic patients treated with sulfonylureas, Diabetes Care 22 (1999) 818 – 822. K.C. Chiu, P. Cohan, N.P. Lee, L.M. Chuang, Insulin sensitivity differs among ethnic groups with a compensatory response in beta-cell function, Diabetes Care 23 (2000) 1353 – 1358.