Cardiovascular and Metabolic Risk ORIGINAL ARTICLE

Disposition Index, Glucose Effectiveness, and Conversion to The Resistance Atherosclerosis Study (IRAS)

1 5 CARLOS LORENZO, MD RICHARD N. BERGMAN, PHD ticipants or targeted persons from a single 2 6 LYNNE E. WAGENKNECHT, DRPH ANTHONY J.G. HANLEY, PHD ethnic group. Since direct methods have 3 7 MARIAN J. REWERS, MD STEVEN M. HAFFNER, MD 4 demanding technical requirements, the ANDREW J. KARTER, PHD product of measures of insulin sensitivity and insulin secretion derived from the oral has attracted OBJECTIVE — Disposition index (DI) and glucose effectiveness (SG) are risk factors for interest. This product has been shown to diabetes. However, the effect of DI and SG on future diabetes has not been examined in large have a modest correlation with minimal epidemiological studies using direct measures. model–derived DI, to decrease as glucose tolerance status deteriorates, and to pre- RESEARCH DESIGN AND METHODS — Insulin sensitivity index (S ), acute insulin I dict the development of diabetes inde- response (AIR), and SG were measured in 826 participants (aged 40–69 years) in the Atherosclerosis Study (IRAS) by the frequently sampled intravenous glucose tolerance pendent of other risk factors including ϫ test. DI was expressed as SI AIR. At the 5-year follow-up examination, 128 individuals (15.5%) fasting and 2-h glucose concentrations had developed diabetes. (7,8). It includes the incretin effect and therefore may not always follow the hy- RESULTS — The area under the receiver operating characteristic curve of a model with SI and perbolic law (7–9). The hyperbolic para- AIR was similar to that of DI (0.767 vs. 0.774, P ϭ 0.543). In a multivariate logistic regression digm of the minimal model–derived DI model that included both DI and SG, conversion to diabetes was predicted by both SG (odds ϫ has also been criticized (10). Further- ratio 1 SD, 0.61 [0.47–0.80]) and DI (0.68 [0.54–0.85]) after adjusting for demographic more, its predictive power has not been variables, fasting and 2-h glucose concentrations, family history of diabetes, and measures of obesity. Age, sex, race/ethnicity, glucose tolerance status, obesity, and family history of diabetes tested for large epidemiological studies. In addition to the insulin-dependent did not have a significant modifying impact on the relation of SG and DI to incident diabetes. component of glucose tolerance (or DI),

CONCLUSIONS — The predictive power of DI is comparable to that of its components, SI the insulin-independent component (glu- and AIR. SG and DI independently predict conversion to diabetes similarly across race/ethnic cose effectiveness [SG]) has already been groups, varying states of glucose tolerance, family history of diabetes, and obesity. explored in mice (11) and humans (12). SG is the capacity of glucose to enhance its Diabetes Care 33:2098–2103, 2010 own cellular uptake and to suppress en- dogenous glucose production. Although oth insulin sensitivity and first- tolerance test (FSIGTT) with minimal reduced in individuals with impaired glu- phase insulin secretion are indepen- model analysis, this relationship is hyper- cose tolerance (IGT) and diabetes (4,13), dent determinants of conversion to bolic (2) and similar across glucose toler- B SG contributes to glucose tolerance even diabetes in different ethnic groups and ance categories (3). Known as the in conditions of significant insulin resis- varying states of glucose tolerance, family disposition index (DI), the product of tance, including diabetes (13). Reduced history of diabetes, and obesity (1). First- measures of insulin sensitivity and first- SG has also been described in healthy in- phase insulin secretion compensates for phase insulin secretion, it has been shown dividuals following the infusion of corti- the worsening of insulin sensitivity (2). In to predict conversion to diabetes (4–6). sol or glucagon, individuals in states of studies using direct methods, such as the The evidence, however, comes from stud- very low caloric intake, women with poly- frequently sampled intravenous glucose ies that have enrolled relatively few par- cystic ovary syndrome, and the elderly ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● (13–15). Contrary to the insulin sensitiv- 1 From the Division of Clinical Epidemiology, University of Texas Health Science Center, San Antonio, ity index (SI), SG may not be influenced by Texas; the 2Division of Public Health Sciences, Wake Forest University School of Medicine, Winston- exercise (16) or weight loss interventions Salem, North Carolina; the 3Barbara Davis Center for Childhood Diabetes and Human Medical (17). In relatively small studies, S has Genetics Program, University of Colorado Health Sciences Center, Aurora, Colorado; the 4Division G of Research, Northern California Region, Kaiser Permanente, Oakland, California; the 5Department been shown to predict future diabetes of Physiology and Biophysics, University of Southern California, Los Angeles, California; the 6De- (4,5,18), but its contribution to the devel- partments of Nutritional Sciences and Medicine, Mount Sinai Hospital and the University of Toronto, opment of diabetes remains largely Toronto, Ontario, Canada; and the 7Department of Medicine, Baylor College of Medicine, Houston, unknown. Texas. Since the hyperbolic paradigm has Corresponding author: Carlos Lorenzo, [email protected]. Received 26 January 2010 and accepted 9 June 2010. not been tested in large epidemiological DOI: 10.2337/dc10-0165 studies, our first objective was to analyze © 2010 by the American Diabetes Association. Readers may use this article as long as the work is properly the risk of future diabetes associated with cited, the use is educational and not for profit, and the work is not altered. See http://creativecommons. minimal model-derived DI relative to its org/licenses/by-nc-nd/3.0/ for details. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby components, SI, and acute insulin re- marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. sponse (AIR). The second objective was to

2098 DIABETES CARE, VOLUME 33, NUMBER 9, SEPTEMBER 2010 care.diabetesjournals.org Lorenzo and Associates assess the relative contribution of the in- sitivity across a broad range of glucose of extreme observations. Given that some sulin-independent component of glucose tolerance (20). Second, the reduced sam- individuals had SI and DI equal to zero, ϩ tolerance, SG, to the development of dia- pling protocol (12 samples) was used be- we used the natural logarithms of SI 1 ϩ betes. To meet these aims, we used data cause of the large number of subjects. SI and DI 1 as the transformation for SI from the Insulin Resistance Atherosclero- and SG were calculated using mathemati- and DI, respectively. Accuracy for the pre- sis Study (IRAS), a multicenter observa- cal modeling methods (MINMOD, ver- diction of diabetes was determined by the tional epidemiological study of different sion 3.0 [1994], Los Angeles, California, area under the receiver operating charac- ethnic groups and varying states of glu- courtesy of Richard Bergman, PhD) (21). teristic curve (AUC). cose tolerance (19). AIR was calculated as the mean of 2- and 4-min insulin concentrations after glu- RESULTS — During the 5-year fol- RESEARCH DESIGN AND cose administration and DI as the product low-up period, 128 (15.5%) of 826 IRAS METHODS — The design and meth- of SI and AIR. participants developed type 2 diabetes: ods of the IRAS have been described else- Race/ethnicity was assessed by self- 44 (7.9%) of 557 participants with nor- where (19). Briefly, the study was report. Family history of diabetes was de- mal glucose tolerance at baseline, and 84 conducted at four clinical centers. Cen- fined as diabetes in parents or siblings. (31.2%) of 269 participants with IGT at ters in Oakland and Los Angeles, Califor- Anthropometric variables and blood baseline. Those participants converting to nia, recruited non-Hispanic whites and pressure were obtained by trained per- type 2 diabetes did not differ from non- African Americans from Kaiser Perma- sonnel. Plasma glucose and insulin con- converters in terms of sex or ethnicity (Ta- nente, a nonprofit HMO. Two other cen- centrations were determined by the ble 1). However, older age and family ters in San Antonio, Texas, and San Luis glucose oxidase and dextran-charcoal ra- history of diabetes were more common Valley, Colorado, recruited non-Hispanic dioimmunoassay methods, respectively. among converters. Converters also had whites and Hispanics from two ongoing This last assay displays a high degree of higher BMI and waist circumference, population–based studies, the San Anto- cross-reactivity with proinsulin. The higher glucose and insulin levels, and nio Heart Study and the San Luis Valley 1999 World Health Organization criteria lower SI, AIR, DI, and SG. Diabetes Study. A total of 1,625 individ- were used to define diabetes (fasting glu- In linear regression models, SG was Ն uals were enrolled (56% women) be- cose concentration 7.0 mmol/l, 2-h directly related to SI (parameter esti- tween October 1992 and April 1994. A plasma glucose concentration Ն11.1 mate ϫ 1 SD, 0.11 Ϯ 0.02, P Ͻ 0.001), follow-up examination was performed 5 mmol/l, or treatment with hypoglycemic AIR (parameter estimate ϫ 1 SD, 0.22 Ϯ years after the baseline examination medications) and IGT (2-h plasma glu- 0.03, P Ͻ 0.001), and DI (parameter es- (range 4.5–6.6 years). The response rate cose level between Ն7.8 and Ͻ11.1 timate ϫ 1 SD, 0.29 Ϯ 0.05, P Ͻ 0.001) was 81%. Protocols were approved by lo- mmol/l). after accounting for the effects of age, sex, cal institutional review committees. All race/ethnicity, and research center. There the participants gave written informed Statistical analyses was no interaction effect of family history ϭ consent. The analysis was carried out using the SAS of diabetes on the relation of SG to SI (P The present report includes informa- statistical software (version 9.1, SAS Insti- 0.534), AIR (P ϭ 0.139), and DI (P ϭ tion on 826 (79.2%) participants (332 tute, Cary, NC). One-way ANCOVA or 0.990). non-Hispanic whites, 206 African Amer- logistic regression analysis was used to We used the AUC to quantify the abil- icans, and 288 Hispanics) after excluding compare the differences for continuous or ity of DI to predict conversion to diabetes 153 individuals who failed to return to dichotomous variables between groups, relative to that of a model with both SI and the follow-up visit and 64 individuals respectively. Linear regression analyses AIR (found in the supplemental figure, with missing information on relevant were used to examine the relation of SG to available in an online appendix at http:// variables. Baseline characteristics were SI, AIR, and DI. In separate models, we care.diabetesjournals.org/cgi/content/ ϫ similar in the participants included in this introduced the interaction term SG full/dc10-0165/DC1). The AUC of the analysis and those who were excluded family history of diabetes. Results were model with SI and AIR was similar to that (e.g., age, ethnicity, sex, glucose tolerance adjusted for demographic variables. Risk of DI (0.767 vs. 0.774, P ϭ 0.543). The status, BMI, waist circumference, and SI of future diabetes associated with SI, AIR, 5-year incidence of diabetes by baseline Ն [all comparisons, P 0.2]) except for AIR DI, and SG was assessed by logistic regres- tertiles of SG and each one of the other ϭ (higher in eligible participants, P sion analysis. Odds ratios (ORs) were ex- measures (SI, AIR, or DI) are presented in 0.005). Protocols were identical at the pressed for binary traits or per SD increase Fig. 1. baseline and follow-up examinations. for continuous traits. Demographic vari- SG predicted future development of Each examination required two visits, one ables, family history of diabetes, measures diabetes after adjusting for age, sex, race/ week apart, of ϳ4 h each. During the first of obesity, and plasma glucose concentra- ethnicity, and research center (OR ϫ 1 visit, a 75-g oral glucose tolerance test was tions were also included as covariates. We SD, 0.50 [0.40–0.64]), as did DI (OR ϫ 1 administered to assess glucose tolerance assessed the impact of these covariates on SD, 0.47 [0.40–0.56]). In a multivariate status. the relation of DI and SG to conversion to logistic regression model that included SG During the second visit, insulin sen- diabetes by including appropriate inter- and DI as independent variables, the odds sitivity and first-phase insulin secretion action terms in separate logistic regres- remained statistically significant for both ϫ were measured by FSIGTT with two mod- sion models. In both linear and logistic SG (OR 1 SD, 0.61 [0.47–0.80]) and DI ifications to the original protocol. First, an regression models, log-transformed val- (OR ϫ 1 SD, 0.68 [0.54–0.85]) (Table 2). injection of regular insulin was used to ues of SI, AIR, and DI were used to im- Age, sex, race/ethnicity, research center, ensure adequate plasma insulin levels for prove discrimination and calibration of family history of diabetes, fasting glucose, the accurate computation of insulin sen- the models and to minimize the influence 2-h glucose, BMI, and waist circumfer- care.diabetesjournals.org DIABETES CARE, VOLUME 33, NUMBER 9, SEPTEMBER 2010 2099 Disposition index and glucose effectiveness

Table 1—Baseline characteristics by diabetes status at follow-up dictors of conversion to diabetes similarly across race/ethnic groups, varying states Nonconvertors Convertors P of glucose tolerance, family history of di- abetes, and obesity. The predictive power n 698 128 of DI is comparable to that of the combi- Age (years)* 54.4 Ϯ 0.3 56.5 Ϯ 0.7 0.009 nation of SI and AIR. Female (%)* 55.4 (51.7–59.1) 59.4 (50.7–67.5) 0.410 Expanding a previous analysis by Ethnicity (%)* 0.647 Martin et al. (4), Goldfine et al. (5) exam- African Americans 25.4 (22.3–28.7) 22.7 (16.2–30.7) ined the risk of conversion to diabetes as- Mexican Americans 34.2 (30.8–37.8) 38.3 (30.3–47.0) sociated with directly measured DI in 155 Non-Hispanic whites 40.4 (36.8–44.1) 39.1 (31.0–47.8) offspring of parents who both had type 2 Family history of diabetes (%) 37.6 (34.0–41.4) 50.6 (41.7–59.6) Ͻ0.001 diabetes and 181 subjects with normal BMI (kg/m2) 27.8 Ϯ 0.2 30.8 Ϯ 0.5 Ͻ0.001 glucose tolerance and without family his- Waist circumference (cm) 89.2 Ϯ 0.4 95.5 Ϯ 1.0 Ͻ0.001 tory of diabetes. DI predicted the devel- Fasting glucose (mmol/l) 5.35 Ϯ 0.02 5.83 Ϯ 0.05 Ͻ0.001 opment of type 2 diabetes only in the 2-h glucose (mmol/l) 6.61 Ϯ 0.06 8.37 Ϯ 0.15 Ͻ0.001 former group. Goldfine et al. concluded Impaired glucose tolerance (%) 25.6 (22.6–29.2) 64.3 (55.3–72.4) Ͻ0.001 that DI and particularly insulin resistance Fasting insulin (pmol/l)† 70.8 Ϯ 1.4 106.7 Ϯ 5.5 Ͻ0.001 was not sufficient for the development of Insulin sensitivity index (SI) Ϫ Ϫ Ϫ Ϫ diabetes in individuals without family (ϫ 10 4 min 1 ␮U 1 ml 1)† 1.92 Ϯ 0.06 1.03 Ϯ 0.13 Ͻ0.001 history of the disease. These individuals, Acute insulin response (AIR) (␮U/ml)† 52.5 Ϯ 1.6 38.9 Ϯ 2.8 Ͻ0.001 however, had a very low incidence of di- ϫ Ϯ Ϯ Ͻ Disposition index (SI AIR)† 96.5 2.9 41.2 3.4 0.001 Ϫ abetes (6 cases during a mean follow-up Glucose effectiveness (ϫ 100 min 1) 2.03 Ϯ 0.03 1.58 Ϯ 0.07 Ͻ0.001 period of 25 years). In another study of Data are n (%) with 95% CI or means Ϯ SD. Results are adjusted for age, sex, race/ethnicity, and research Pima Indians, incident diabetes was pre- center. *Unadjusted results; †log-transformed variables. These variables were then back-transformed to their dicted by DI independent of the effect of units for presentation in the table. demographic variables and percentage of body fat (7). Our results validate previous ence were also included as covariates. In a and SG to conversion to diabetes differed studies and extend the findings to men different model that included SI, AIR, and little between categories of age, sex, race/ and women, all three race/ethnic groups, SG, these three measures were all inde- ethnicity, glucose tolerance status, BMI, and varying states of glucose tolerance, pendent predictors of diabetes. and family history of diabetes (Fig. 2A–F). family history of diabetes, and adiposity. In separate logistic regression models, Additionally, the interaction term DI ϫ Ferrannini and Mari (10) have ques- Ն we examined the impact of age, sex, race/ SG was not statistically significant (P tioned the hyperbolic paradigm of the re- ethnicity, research center, BMI, glucose 0.71). DI and SG were independent risk lationship between insulin sensitivity and tolerance status, family history of diabetes factors across different degrees of insulin first-phase insulin secretion. In a recent on the relationship between SG (or DI), sensitivity and insulin secretion (Fig. 2G cross-sectional study (22), however, these and incident diabetes. None of these vari- and H). authors confirmed the paradigm while ar- ables had a significant impact on the rela- guing against the compensation for insu- tion of SG and DI to conversion to diabetes CONCLUSIONS — In cross-sectional lin resistance as the sole mechanism Ն (P 0.16 for all potential interactions) analyses, SG is directly related to insulin responsible for hyperglycemia. They also except for age on the relationship between sensitivity and first-phase insulin secre- indicated that an intrinsic ␤-cell function ϭ DI and conversion to diabetes (P tion. Further in prospective analyses, SG defect is the most important mechanism 0.038). The strength of the relation of DI and DI were found to be independent pre- for hyperglycemia. In agreement with the

Figure 1—Five-year incidence of diabetes by tertiles of SI, AIR, DI, and SG. Results were adjusted for age, sex, race/ethnicity, research center, IGT, ϫ Ϫ4 Ϫ1 ␮ Ϫ1 Ϫ1 Յ Ն family history of diabetes, and BMI. Cut points for tertiles of SI ( 10 min U ml ) were 1.16 (lower), 1.17–2.38 (middle), and 2.39 (upper). Corresponding cut points for tertiles of AIR (␮U/ml) were Յ37.5, 38.0–75.0, and Ն75.5; DI (ϫ 10Ϫ4 minϪ1 ␮UϪ1 mlϪ1 per ␮U mlϪ1) Յ Ն ϫ Ϫ2 Ϫ1 Յ Ն 55.8, 55.9–120.0, and 120.1; SG ( 10 ml ), 1.58, 1.59–2.22, and 2.23.

2100 DIABETES CARE, VOLUME 33, NUMBER 9, SEPTEMBER 2010 care.diabetesjournals.org Lorenzo and Associates

Table 2—Predictors of conversion to type 2 diabetes by multiple logistic regression analysis hyperbolic paradigm, our results suggest that the predictive power of DI is compa- OR (95% CI) P rable with that of its components. DI, however, is not the only independent risk Model 1* factor for the development of diabetes. ϫ Age ( 1 SD) 1.09 (0.87–1.38) 0.449 Since fasting and 2-h plasma glucose val- Female vs. male 0.93 (0.52–1.69) 0.824 ues remain strong predictors of diabetes, Family history of diabetes (yes vs. no) 1.25 (0.79–1.96) 0.340 we cannot disregard other determinants ϫ BMI ( 1 SD) 1.33 (0.87–2.05) 0.187 of glucose homeostasis, such as additional Waist circumference (ϫ 1 SD) 0.74 (0.47–1.18) 0.206 ϫ Ͻ aspects of secretory response to glucose Fasting glucose concentration ( 1 SD) 1.60 (1.24–2.06) 0.001 stimulation (e.g., ␤-cell glucose sensitiv- 2-h glucose concentration (ϫ 1 SD) 1.95 (1.48–2.57) Ͻ0.001 ity and potentiation), from the disease Disposition index (ϫ 1 SD)† 0.68 (0.54–0.85) 0.001 process. Glucose effectiveness (ϫ 1 SD) 0.61 (0.47–0.80) Ͻ0.001 Lopez et al. (23) recently reported Model 2‡ Age (ϫ 1 SD) 1.10 (0.87–1.39) 0.422 that SG is related to SI in individuals with Female vs. male 0.91 (0.50–1.65) 0.754 family history of diabetes but not in those Family history of diabetes (yes vs. no) 1.25 (0.80–1.96) 0.328 without such a history. Our results, how- BMI (ϫ 1 SD) 1.39 (0.90–2.14) 0.137 ever, indicate that family history of diabe- Waist circumference (ϫ 1 SD) 0.72 (0.45–1.17) 0.190 tes has little influence on the relationship Fasting glucose concentration (ϫ 1 SD) 1.54 (1.18–2.00) 0.001 between SG and SI. This is not surprising. 2-h glucose concentration (ϫ 1 SD) 1.91 (1.44–2.53) Ͻ0.001 SG is independent of the dynamic insulin Insulin sensitivity index (ϫ 1 SD)† 0.61 (0.44–0.86) 0.004 response but is influenced by the basal Acute insulin response (ϫ 1 SD)† 0.73 (0.55–0.95) 0.022 insulin level (13). In other words, SG is Glucose effectiveness (ϫ 1 SD) 0.67 (0.50–0.88) 0.005 influenced by the effect of insulin on glu- ORs expressed for binary traits or per 1 SD unit change for continuous traits. *Results in Model 1 adjusted cose uptake by insulin-dependent tissues also for race/ethnicity (P ϭ 0.494) and research center (P ϭ 0.006); †log-transformed variables; ‡results in (basal insulin–dependent component) Model 2 also adjusted for race/ethnicity (P ϭ 0.446) and research center (P ϭ 0.006). (12). Similarly, in longitudinal studies by Martin et al. (4) and Goldfine et al. (5), SG predicted conversion to diabetes inde-

Figure 2—Risk of developing diabetes associated with DI and SG by ethnicity, sex, glucose tolerance status, BMI and age categories, family history of diabetes, and tertiles of SI and AIR. Estimates expressed fora1SDunit change. Age, sex, race/ethnicity, research center, BMI, IGT, family history of diabetes, DI, and SG were all included as independent variables in all eight models. Log-transformed values of DI were used to improve discrimination and calibration of the models and to minimize the influence of extreme observations. care.diabetesjournals.org DIABETES CARE, VOLUME 33, NUMBER 9, SEPTEMBER 2010 2101 Disposition index and glucose effectiveness pendent of DI but only among offspring of No potential conflicts of interest relevant to JK. Evaluation of proposed oral disposi- type 2 diabetic parents. Individuals with- this article were reported. tion index measures in relation to the ac- out family history of diabetes had a very C.L. wrote the manuscript, contributed to tual disposition index. Diabet Med 2009; low risk of future diabetes and neither S the discussion, and reviewed/edited the 26:1198–1203 G manuscript. L.E.W. researched the data, con- 9. Hu¨ cking K, Watanabe RM, Stefanovski D, nor DI predicted diabetes. Osei et al. (18) tributed to the discussion, and reviewed/ Bergman RN. OGTT-derived measures of examined the risk of progression to IGT edited the manuscript. M.J.R. reviewed/edited insulin sensitivity are confounded by fac- or diabetes associated with SG in 81 first- the manuscript. A.J.K. researched the data, tors other than insulin sensitivity itself. degree relatives of African Americans with contributed to the discussion, and reviewed/ Obesity 2008;16:1938–1945 type 2 diabetes. SG as well as directly mea- edited the manuscript. R.N.B. contributed to 10. Ferrannini E, Mari A. function sured insulin resistance and secretion the discussion and reviewed/edited the manu- and its relation to insulin action in hu- were independent predictors of worsen- script. A.J.G.H. contributed to the discussion mans: a critical appraisal. Diabetologia ing glucose tolerance status. Neverthe- and reviewed/edited the manuscript. S.M.H. 2004;47:943–956 researched the data, wrote the manuscript, 11. Pacini G, Thomaseth K, Ahre´n B. Contribu- less, none of these studies adjusted their contributed to the discussion, and reviewed/ tion to glucose tolerance of insulin-inde- results for glucose tolerance status and edited the manuscript. pendent vs. insulin-dependent mechanisms adiposity. In the IRAS, SG is an indepen- Parts of this study were presented in ab- in mice. Am J Physiol Endocrinol Metab dent risk factor for future diabetes in in- stract form at the 70th Scientific Sessions of the 2001;281:E693–E703 dividuals with family history of diabetes American Diabetes Association, Orlando, 12. Kahn SE, Prigeon RL, McCulloch DK, and similar results are demonstrated in all Florida, 25–27 June 2010. Boyko EJ, Bergman RN, Schwartz MW, other categories regardless of age, sex, race/ Neifing JL, Ward WK, Beard JC, Palmer ethnicity, glucose tolerance, and adiposity. JP. The contribution of insulin-dependent and insulin-independent glucose uptake S is directly related to first-phase insu- References G 1. Lorenzo C, Wagenknecht LE, D’Agostino to intravenous glucose tolerance in lin secretion. Experiments in canines have RB Jr, Rewers MJ, Karter AJ, Haffner SM. healthy human subjects. Diabetes 1994; suggested that this relationship is an artifact Insulin resistance, ␤-cell dysfunction, and 43:587–592 of the minimal model method (24). In mice, conversion to type 2 diabetes in a multi- 13. Best JD, Kahn SE, Ader M, Watanabe RM, however, the effect of first-phase insulin se- ethnic population: the Insulin Resistance Ni TC, Bergman RN. Role of glucose cretion (in physiological conditions) on SG Atherosclerosis Study. Diabetes Care effectiveness in the determination of is minimal (20). Furthermore, FSIGTT with 2010;33:67–72 glucose tolerance. Diabetes Care 1996; the minimal model analysis may overesti- 2. Kahn SE, Prigeon RL, McCulloch DK, 19:1018–1030 mate S . In human subjects, the difference Boyko EJ, Bergman RN, Schwartz MW, 14. Tonelli J, Kishore P, Lee DE, Hawkins M. G Neifing JL, Ward WK, Beard JC, Palmer The regulation of glucose effectiveness: between minimal model- and clamp- JP. Quantification of the relationship be- how glucose modulates its own produc- derived SG appears to be related to the as- tween insulin sensitivity and beta-cell tion. Curr Opin Clin Nutr Metab Care sumption of mono-compartmental function in human subjects: evidence for 2005;8:450–456 distribution of glucose by the minimal a hyperbolic function. Diabetes 1993;42: 15. Burattini R, Di Nardo F, Boemi M, Fumelli model analysis; however, both measures are 1663–1672 P. Deterioration of insulin sensitivity and highly correlated (25). Therefore, minimal 3. Festa A, Williams K, D’Agostino R Jr, glucose effectiveness with age and hyper- Wagenknecht LE, Haffner SM. The natu- tension. Am J Hypertens 2006;19:98–102 model-derived SG is considered an ade- quate measure of nonsteady-state glucose ral course of beta-cell function in nondia- 16. Kahn SE, Larson VG, Beard JC, Cain KC, betic and diabetic individuals: the Insulin Fellingham GW, Schwartz RS, Veith RC, kinetics and a dependable index. Resistance Atherosclerosis Study. Diabe- Stratton JR, Cerqueira MD, Abrass IB. Ef- In conclusion, the predictive discrim- tes 2006;55:1114–1120 fect of exercise on insulin action, glucose ination of DI for future diabetes is compa- 4. Martin BC, Warram JH, Krolewski AS, tolerance, and insulin secretion in aging. rable with that of the combination of SI Bergman RN, Soeldner JS, Kahn CR. Role Am J Physiol Endocrinol Metab 1990; and AIR; therefore, our results support of glucose and insulin resistance in devel- 258:E937–E943 the validity of the hyperbolic paradigm. opment of type 2 diabetes mellitus: results 17. Escalante-Pulido M, Escalante-Herrera A, of a 25-year follow-up study. Lancet Milke-Najar ME, Alpizar-Salazar M. Ef- DI and SG are important, independent de- terminants of diabetes in different ethnic 1992;340:925–929 fects of weight loss on insulin secretion 5. Goldfine AB, Bouche C, Parker RA, Kim and in vivo insulin sensitivity in obese di- groups and varying states of glucose tol- C, Kerivan A, Soeldner JS, Martin BC, abetic and non-diabetic subjects. Diabetes erance, family history of diabetes, and Warram JH, Kahn CR. Insulin resistance Nutr Metab 2003;16:277–283 obesity. SG has a direct relationship with is a poor predictor of type 2 diabetes in 18. Osei K, Rhinesmith S, Gaillard T, Schus- SI and AIR. Prospective studies are individuals with no family history of dis- ter D. Impaired insulin sensitivity, insulin needed to examine the natural course of ease. Proc Natl Acad Sci U S A 2003;100: secretion, and glucose effectiveness pre- 2724–2729 dict future development of impaired glu- SG relative to that of insulin resistance and ␤-cell function. 6. Stumvoll M, Tataranni PA, Stefan N, cose tolerance and type 2 diabetes in pre- Vozarova B, Bogardus C. Glucose allosta- diabetic African Americans: implications sis. Diabetes 2003;52:903–909 for primary diabetes prevention. Diabetes 7. Utzschneider KM, Prigeon RL, Faulen- Care 2004;27:1439–1446 Acknowledgments— This study was sup- bach MV, Tong J, Carr DB, Boyko EJ, Le- 19. Wagenknecht LE, Mayer EJ, Rewers M, ported by the National Heart, Lung, and Blood onetti DL, McNeely MJ, Fujimoto WY, Haffner S, Selby J, Borok GM, Henkin L, Institute grants HL-47887, HL-47889, HL- Kahn SE. Oral disposition index predicts Howard G, Savage PJ, Saad MF. The In- 47890, HL-47892, HL-47902 and the General the development of future diabetes above sulin Resistance Atherosclerosis Study Clinical Research Centers Program (National and beyond fasting and 2-h glucose levels. objectives, design, and recruitment re- Center for Research Resources, M01 RR431, Diabetes Care 2009;32:335–341 sults. Ann Epidemiol 1995;5:464–472 M01 RR01346). 8. Retnakaran R, Qi Y, Goran MI, Hamilton 20. Bergman RN, Finegood DT, Ader M. As-

2102 DIABETES CARE, VOLUME 33, NUMBER 9, SEPTEMBER 2010 care.diabetesjournals.org Lorenzo and Associates

sessment of insulin sensitivity in vivo. En- paired beta cell glucose sensitivity rather 24. Finegood DT, Tzur D. Reduced glucose docr Rev 1985;6:45–86 than inadequate compensation for insulin effectiveness associated with reduced in- 21. Welch S, Gebhart SS, Bergman RN, Phil- resistance is the dominant defect in glu- sulin release: an artifact of the minimal lips LS. Minimal model analysis of intra- cose intolerance. Diabetologia 2010;53: model method. Am J Physiol Endocrinol venous glucose tolerance test-derived 749–756 Metab 1996;271:E485–E495 insulin sensitivity in diabetic subjects. 23. Lopez X, Bouche´ C, Tatro E, Goldfine AB. 25. Ni TC, Ader M, Bergman RN. Reassess- J Clin Endocrinol Metab 1990;71:1508– Family history of diabetes impacts on in- ment of glucose effectiveness and insulin 1518 teractions between minimal model esti- sensitivity from minimal model analysis: a 22. Mari A, Tura A, Natali A, Laville M, mates of insulin sensitivity and glucose theoretical evaluation of the single-com- Laakso M, Gabriel R, Beck-Nielsen H, effectiveness. Diabetes Obes Metab 2009; partment glucose distribution assump- Ferrannini E, RISC Investigators. Im- 11:123–130 tion. Diabetes 1997;46:1813–1821

care.diabetesjournals.org DIABETES CARE, VOLUME 33, NUMBER 9, SEPTEMBER 2010 2103