Pathophysiology/Complications ORIGINAL ARTICLE Homeostasis Model Assessment Is a Reliable Indicator of Insulin Resistance During Follow-up of Patients With Type 2 Diabetes AKIRA KATSUKI, MD RIKA ARAKI-SASAKI, MD sex hormone–binding globulin (SHBG) YASUHIRO SUMIDA, MD YASUKO HORI, MD (2–10). Recently, the usefulness of HOMA- ESTEBAN C. GABAZZA, MD YUTAKA YANO, MD IR as an indicator of insulin resistance in SHUICHI MURASHIMA, MD YUKIHIKO ADACHI, MD diabetic patients has been the focus of MASAHIKO FURUTA, MD much attention (11–13). A significant cor- relation has been reported between the insulin resistance index calculated by HOMA and the hyperinsulinemic-eugly- cemic clamp (clamp IR) (11,12). OBJECTIVE — To investigate the usefulness of the homeostasis model assessment as an index We previously reported that the serum of insulin resistance (HOMA-IR) for evaluating the clinical course of patients with type 2 diabetes. levels of SHBG may be an index of insulin RESEARCH DESIGN AND METHODS — The usefulness of HOMA-IR and its rela- resistance only in the hyperinsulinemic state tionship with insulin resistance assessed by the hyperinsulinemic-euglycemic clamp study (before treatment) (7). The usefulness of (clamp IR) were evaluated in 55 Japanese patients with type 2 diabetes before and after treat- HOMA as an index of insulin resistance ment. The patients were subjected to diet (ϳ1,440–1,720 kcal/day) and exercise therapy (walk- during therapy of diabetes has not been as ing 10,000 steps daily) for 6 weeks during their hospitalization. yet evaluated. To evaluate this, in the pres- ent study, we investigated whether HOMA- RESULTS — Univariate regression analysis disclosed a significant correlation between log- IR is correlated with clamp IR before and Ϫ Ͻ transformed HOMA-IR and log-transformed clamp IR before (r = 0.613, P 0.0001) and after after treatment. (r = Ϫ0.734, P Ͻ 0.0001) treatment. Neither the slopes (Ϫ0.71 ± 0.12 vs. Ϫ0.79 ± 0.09, F = 0.25, P = 0.61) nor the intercepts (y-intercept = 1.67 vs. 1.70, x-intercept = 2.36 vs. 2.15, F = 0.02, P = 0.88) of the regression lines between HOMA-IR and clamp IR were significantly different before and after treatment. There was a significant correlation between the decrease in log-trans- RESEARCH DESIGN AND formed HOMA-IR and the increase in clamp IR during treatment (r = Ϫ0.617, P Ͻ 0.0001). METHODS — A total of 55 patients with type 2 diabetes treated with diet alone CONCLUSIONS — HOMA-IR may constitute a useful method not only for diagnosing and with sulfonylureas were enrolled in the insulin resistance, but also for follow-up during the treatment of patients with type 2 diabetes. present study (Table 1). BMI was estimated by dividing the body weight (in kilograms) Diabetes Care 24:362–365, 2001 by the square of the height (in meters). Diabetes was diagnosed according to the criteria of the American Diabetes Asso- nsulin resistance is currently being mea- To date, several methods for evaluating ciation (14). Subjects with fasting plasma sured by using the glucose clamp tech- insulin resistance have been proposed, glucose levels Ն7.0 mmol/l were provi- Inique (1). Although this method is highly including homeostasis model assessment sionally diagnosed as having diabetes. sensitive and shows high reproducibility, it (HOMA-IR), continuous infusion of glu- Thereafter, the subjects underwent a 75-g is time-consuming and very expensive. A cose with model assessment, calculations of oral glucose tolerance test (OGTT) (Trelan rapid, accurate, and low-cost method for sensitivity indexes using insulin and glu- G 75; Shimizu, Shizuoka, Japan), and those assessing insulin resistance would be very cose in the post–oral glucose load state, with fasting plasma glucose levels Ն7.0 useful in clinical practice. fasting insulin levels, and serum levels of mmol/l or 2-h plasma glucose levels Ն11.1 mmol/l were diagnosed as having diabetes. Type 2 diabetes was defined according to the grade of insulin secretion, the age, the From the Third Department of Internal Medicine (A.K., Y.S., E.C.G., M.F., R.A.-S., Y.H., Y.Y., Y.A.), the Depart- pattern of onset, and the existence of fam- ment of Radiology (S.M.), Mie University School of Medicine, Mie, Japan. ily history of diabetes. Address correspondence and reprint requests to Akira Katsuki, MD, Third Department of Internal Med- icine, Mie University School of Medicine, 2-174 Edobashi, Tsu, Mie 514-8507, Japan. E-mail: katuki- On admission, 50 patients were treated [email protected]. with diet alone and 5 with sulfonylureas Received for publication 24 April 2000 and accepted in revised form 24 October 2000. (glibenclamide 2.5–5.0 mg/day). None of Abbreviations: clamp IR, insulin resistance assessed by the hyperinsulinemic-euglycemic clamp; HOMA- the patients was being treated with insulin IR, homeostasis model assessment of insulin resistance; OGTT, oral glucose tolerance test; SHBG, sex hor- mone–binding globulin. or insulin-sensitizing agents. Fifteen A table elsewhere in this issue shows conventional and Système International (SI) units and conversion patients had peripheral neuropathy and factors for many substances. simple diabetic retinopathy and 10 had 362 DIABETES CARE, VOLUME 24, NUMBER 2, FEBRUARY 2001 Katsuki and Associates Table 1—Clinical characteristics of patients with type 2 diabetes before and after treatment Assuming that normal subjects aged Ͻ35 years with normal weight have an IR of 1, the values for a patient can be calculated Before treatment After treatment from the fasting concentrations of insulin n 55 — and glucose using the following formula: Sex (M/F) 44/11 — fasting serum insulin (µU/ml) ϫ fasting Age (years) 54.7 ± 10.6 — plasma glucose (mmol/l)/22.5. Blood sam- Duration of diabetes (years) 8.3 ± 6.7 — ples for HOMA-IR measurements were BMI (kg/m2) 23.6 ± 3.2 22.3 ± 2.9* drawn from each subject from 8:00 A.M. after Visceral fat area (cm2) 109.2 ± 50.9 96.1 ± 49.5* an overnight bed rest. We examined three Subcutaneous fat area (cm2) 121.9 ± 55.3 111.9 ± 54.4 separate insulin samples taken 15 min apart, Systolic blood pressure (mmHg) 131.5 ± 22.0 128.3 ± 17.1 and the averaged insulin level was used for Diastolic blood pressure (mmHg) 79.4 ± 11.7 78.9 ± 10.7 the HOMA-IR calculation. To estimate the Fasting plasma glucose (mmol/l) 8.7 ± 2.6 6.4 ± 1.6* reproducibility of HOMA-IR, we analyzed a HbA1c (%) 9.5 ± 2.4 8.2 ± 1.5* second HOMA-IR in all patients on another Fasting serum insulin (pmol/l) 39.0 ± 22.2 34.8 ± 23.4 occasion within 5 days of the first HOMA-IR Clamp IR (µmol и kgϪ1 и minϪ1) 35.4 ± 16.9 50.1 ± 13.6* before and after treatment. The coefficient of HOMA-IR 2.4 ± 1.6 1.7 ± 1.2* variation for HOMA-IR before treatment was Data are n or means ± SD. *P Ͻ 0.01 vs. before treatment. 10.2% and 9.8% after treatment. The 75-g OGTT was started from 8:00 A.M. after an overnight bed rest (hunger for microalbuminuria. Macrovascular compli- sured using an immunoradiometric assay 11:00 h). Blood was taken at 0, 30, and cations were not detected. kit (Insulin Riabead II kit; Dainabot, 120 min, and plasma glucose and serum Informed consent was obtained from all Tokyo). This kit included 125I-labeled and insulin levels were evaluated. subjects before the beginning of the study. unlabeled anti–human insulin mouse The body fat area was evaluated as pre- monoclonal antibodies. The intra- and viously described (19). The total cross-sec- Study design interassay coefficients of variation of the tional area, the intra-abdominal visceral fat After admission, the patients were sub- assay were 1.9 and 2.0%, respectively. area, and the subcutaneous fat area were jected to diet and exercise therapy for 6 Clamp IR was evaluated by the hyper- measured by abdominal computed tomog- weeks. The dietary treatment was as fol- insulinemic-euglycemic clamp technique raphy taken at the umbilical level. Any lows: 1,440–1,720 kcal/day with a diet using the artificial pancreas (STG-22; intraperitoneal region having the same den- consisting of 20% (energy) protein, 25% Nikkiso, Tokyo) (1,7,15–18). In brief, at sity as the subcutaneous fat layer was fat, and 55% carbohydrates. The compli- 8:00 A.M., two Teflon-coated cannulae were defined as a visceral fat area. ance of dietary therapy was checked by a inserted; one was inserted into the left ante- Blood pressure was determined three dietitian twice a week. During the exercise cubital vein for infusion of insulin (Humulin times in the supine position after a 5-min therapy, the patients walked about 10,000 R; Eli Lilly, Indianapolis, IN) and 10% glu- rest. steps daily; the number of steps per day cose, and the other was inserted into the was counted using a pedometer, and the right contralateral heated hand vein for arte- Statistical analyses count was checked by a nurse on each day. rialized blood sampling. After baseline blood Data are expressed as means ± SD. Stu- Diet and exercise therapies were not mod- collections for glucose and insulin determi- dent’s t test was performed to compare the ified during the course of the treatment. nations, a priming dose of insulin was means of variables measured before and The blood levels of glucose, HbA1c, administered during the initial 10 min in a after treatment. The relationship of clamp and insulin; the values of clamp-IR, logarithmically decreasing manner to raise IR with several clinical indexes of insulin HOMA-IR, and the simple indexes of serum insulin rapidly to the desired level sensitivity was evaluated by univariate insulin sensitivity (the 30-min and 2-h glu- (1,200 pmol/l); this level of insulin was then regression analysis.