EvidenceofReducedBetaCell Function inAsian IndiansWithMild DysglycemiaLISA R. STAIMEZ, MPH

1

MARY BETH WEBER, PHD, MPH2

HARISH RANJANI, PHD, CDE3

MOHAMMED K. ALI, MBCHB, MSC1,2

JUSTIN B. ECHOUFFO-TCHEUGUI, MD, PHD2

LAWRENCE S. PHILLIPS, MD4,5

VISWANATHAN MOHAN, MD, FRCP, PHD, DSC3

K.M. VENKAT NARAYAN, MD, MSC, MBA1,2,6

OBJECTIVEdTo examine b-cell function across a spectrum of glycemia among AsianIndians, a population experiencing type 2 diabetes development at young ages despite low BMI.

RESEARCH DESIGN AND METHODSdOne-thousand two-hundred sixty-four indi-viduals without known diabetes in the Diabetes Community Lifestyle Improvement Program inChennai, India, had a 75-g oral glucose tolerance test, with glucose and insulinmeasured at 0, 30,and 120min. Type 2 diabetes, isolated impaired fasting glucose (iIFG), isolated impaired glucosetolerance (iIGT), combined impaired fasting glucose and impaired glucose tolerance, and normalglucose tolerance (NGT) were defined by American Diabetes Association guidelines. Measuresincluded insulin resistance and sensitivity (homeostasis model assessment of insulin resistance[HOMA-IR], modified Matsuda Index, 1/fasting insulin) and b-cell function (oral dispositionindex = [Dinsulin0–30/Dglucose0–30] 3 [1/fasting insulin]).

RESULTSdMean age was 44.2 years (SD, 9.3) and BMI 27.4 kg/m2 (SD, 3.8); 341 individualshad NGT, 672 had iIFG, IGT, or IFG plus IGT, and 251 had diabetes. Patterns of insulin re-sistance or sensitivity were similar across glycemic categories. With mild dysglycemia, the ab-solute differences in age- and sex-adjusted oral disposition index (NGT vs. iIFG, 38%; NGT vs.iIGT, 32%) were greater than the differences in HOMA-IR (NGT vs. iIFG, 25%; NGT vs. iIGT,23%; each P , 0.0001). Compared with NGT and adjusted for age, sex, BMI, waist circumfer-ence, and family history, the odds of mild dysglycemia were more significant per SD of oraldisposition index (iIFG: odds ratio [OR], 0.36; 95% CI, 0.23–0.55; iIGT: OR, 0.37; 95% CI,0.24–0.56) than per SD of HOMA-IR (iIFG: OR, 1.69; 95% CI, 1.23–2.33; iIGT: OR, 1.53; 95%CI, 1.11–2.11).

CONCLUSIONSdAsian Indians with mild dysglycemia have reduced b-cell function, re-gardless of age, adiposity, insulin sensitivity, or family history. Strategies in diabetes preventionshould minimize loss of b-cell function.

Type 2 diabetes mellitus is a globalproblem, with 80% of all casesworldwide occurring in low- and

middle-income countries (1). However,despite the increasing prevalence of type2 diabetes, the etiology of the disease re-mains incompletely understood. Previ-ously invoked as the driving feature ofdiabetes, increased insulin resistance can

trigger increased insulin production tomaintain normoglycemia and, over time,can strain b cells to the point at whichinsulin production is no longer adequate(2–4), i.e.,b-cell “exhaustion.”Character-istics associated with insulin resistance,particularly older age, obesity, and phys-ical inactivity, are strong risk factors fordiabetes (5).

Yet, poor b-cell function also mayhave more of a primary role in diabetesdevelopment. The inadequate b-cell re-sponse to physiologic needs for insulinnot only may be an acquired feature (e.g.,as a result of insulin resistance) but also, atleast in some individuals, may be an inher-ent feature. b-cell dysfunction has been de-tected early in the pathogenesis of thedisease (6), with recent cross-sectionaland longitudinal studies detecting dysfunc-tion in people with prediabetes or evennormoglycemia (7–10). Supported by re-cent genetic discoveries (11), these studiessuggest that some individuals have an un-derlying susceptibility to poor b-cell func-tion (12) and that b-cell dysfunction maybe an early driving metabolic feature of di-abetes development.

Most studies of diabetes pathogenesishave been conducted in populations ofEuropean descent; however, more peoplehave diabetes in other populations world-wide. Asian Indians, in particular, expe-rience high rates of type 2 diabetes (13) atyounger ages and lower BMI values (14)compared with other populations. Theyhave high basal insulin levels (15) thatare not entirely explained by obesity oradverse fat distribution (16), which arecommonly cited factors related to insulinresistance. Considering these characteris-tics, Asian Indians may be an ideal popu-lation to utilize for developing a betterunderstanding of the relative roles ofb-cell function and insulin resistance inthe pathogenesis of type 2 diabetes. Pre-vious studies that have examined the eti-ology of diabetes in Asian Indians haveproduced conflicting findings. Alteredb-cell function has been associated withimpaired glucose tolerance (IGT) (17),has not been associated with IGT(18,19), and has been associated with im-paired fasting glucose (IFG) but not IGT(20). Furthermore, b-cell function hasnot always been evaluated rigorously inAsian Indians (i.e., expressed relative tothe insulin resistance of each individual)(21). We investigated the associations be-tween the pathophysiologic mechanismsof insulin resistance and b-cell functionwith glycemic status in a large cohort(n = 1,264) of Asian Indians in Chennai,India.

c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c

From the 1Nutrition and Health Sciences Program, Graduate Division of Biomedical and Biological Sciences,Laney Graduate School, Emory University, Atlanta, Georgia; the 2Hubert Department of Global Health,Rollins School of Public Health, Emory University, Atlanta, Georgia; the 3Madras Diabetes Research Foundation,Global Diabetes Research Center, WHO Collaborating Centre for Noncommunicable Diseases Prevention andControl and International Diabetes Federation Centre of Education, Chennai, India; the 4Division of Endocri-nology, Department of Medicine, School of Medicine, Emory University, Atlanta, Georgia; the 5Veterans AffairsMedical Center, Decatur, Georgia; and the 6Department of Medicine, School of Medicine, Emory University,Atlanta, Georgia.

Corresponding author: Lisa R. Staimez, lisa.staimez@emory.edu.Received 6 November 2012 and accepted 25 February 2013.DOI: 10.2337/dc12-2290. Clinical trial reg. no. NCT01283308, www.clinicaltrials.gov.© 2013 by the American Diabetes Association. Readers may use this article as long as the work is properly

cited, the use is educational and not for profit, and thework is not altered. See http://creativecommons.org/licenses/by-nc-nd/3.0/ for details.

care.diabetesjournals.org DIABETES CARE 1

P a t h o p h y s i o l o g y / C o m p l i c a t i o n sO R I G I N A L A R T I C L E

Diabetes Care Publish Ahead of Print, published online April 17, 2013

RESEARCH DESIGN ANDMETHODS

Study populationStudy subjects were individuals in theDiabetes Community Lifestyle Improve-ment Program, a primary prevention trialin Chennai (formerly Madras) testing theeffects of a stepwise model of diabetesprevention, including a culturally tailoredand intensive lifestyle intervention plusmetforminwhen needed (22). Community-wide recruitment targeted men andwomen at large-scale community events,housing or apartment complexes, localbusinesses, places of worship, and educa-tional institutions, through clinic recordsat the study site, and through direct re-ferral by health care providers at theclinic. Community-based screening (n =19,377) included a short survey, anthro-pometric measurements, and randomcapillary blood glucose test using a glu-cose meter (Lifescan; Johnson & Johnson,Milpitas, CA). Screened volunteers whowere 20–65 years old with a random cap-illary blood glucose of $6.1 mmol/L(110 mg/dL) and without known type 2diabetes were eligible for clinic-basedscreening (Diabetes Community LifestyleImprovement Program baseline testing),which included a 75-g oral glucose toler-ance test (OGTT) performed after an over-night fast.

Individuals who were pregnant,breastfeeding, with a history or evidenceof heart disease, or with any other seriousillness were excluded from the study. Allsubjects provided informed consent andparticipated in clinic-based screening be-tween 2008 and 2011. Among the 1,285individuals tested in clinic-based screen-ing, 14 participants were excluded formissing glucose or insulinmeasures of theOGTT at any time point (i.e., 0, 30, or 120min). An additional seven individualswere excluded for having negative orzero values of the insulinogenic index(IGI), a measure of the early insulin re-sponse in the OGTT (23), calculated asthe ratio of change in insulin to the changein glucose from 0 to 30 min (i.e., DI0–30 /DG0–30). The final number of participantsincluded in the present analyses was1,264. The study was approved by theEmory University Institutional ReviewBoard and the Madras Diabetes ResearchFoundation Ethics Committee.

Study proceduresDiabetes Community Lifestyle Improve-ment Program baseline testing included

the collection of demographic, anthropo-metric, and glucose tolerance data (22).After fasting overnight for at least 8 h,subjects participated in a standard 75-goral OGTT (24), with plasma glucoseand insulin sampled at 0, 30, and120 min. Other collected data includeddemographics, body weight, height, waistcircumference, and family history of dia-betes (defined as having one or more first-degree relatives with type 2 diabetes). Forbody weight assessment, subjects wereasked to wear light clothing and weightwas recorded after shoes and heavy jew-elry were removed. Height was measuredwith a stadiometer to the nearest centime-ter with subjects standing upright with-out shoes. BMI was calculated as mass(kg) divided by height squared (m2).Waist circumference was measured twiceat the smallest horizontal girth betweenthe costal margins and the iliac crestswere measured at minimal respirationusing a nonelastic measuring tape and av-eraged. OGTT samples were collected inEDTA, separated, and stored at 2808C.Plasma glucose (hexokinase method)was measured on a Hitachi 912 Autoan-alyzer (Hitachi, Mannheim, Germany) us-ing kits supplied by Roche Diagnostics(Mannheim, Germany). Insulin concen-trations were estimated using an electro-chemiluminescence method (COBAS E411; Roche Diagnostics, Mannheim, Ger-many). The intra-assay and interassay co-efficients of variation for the biochemicalassays ranged between 3.1 and 7.6%.Samples were processed in a laboratoryaccredited nationally by the National Ac-creditation Board for Testing and Calibra-tion Laboratories and internationally bythe College of American Pathologists.

Key variablesThe glycemic status outcomes for this studywere defined by the following AmericanDiabetes Association criteria (25): diabetesas fasting plasma glucose $7.0 mmol/L(126 mg/dL) or 2-h postload glucose$11.1 mmol/L (200 mg/dL), or both; iso-lated IFG (iIFG) as fasting plasma glucose5.6–6.9 mmol/L (100–125 mg/dL) and2-h postload glucose ,7.8 mmol/L(140 mg/dL); isolated IGT (iIGT) as 2-hpostload glucose 7.8–11.0 mmol/L(140–199 mg/dL) and fasting plasmaglucose ,5.6 mmol/L (100 mg/dL); NGTas fasting plasma glucose ,5.6 mmol/L(100 mg/dL) and 2-h postload glucose,7.8 mmol/L (140 mg/dL); and com-bined IFG and IGT (IFG plus IGT) as fast-ing plasma glucose 5.6–6.9 mmol/L

(100–125 mg/dL) and 2-h postload glu-cose 7.8–11.0 mmol/L (140–199 mg/dL).Prediabetes was defined as iIFG, iIGT, orIFG and IGT. Mild dysglycemia was de-fined as iIFG or iIGT.

The primary measure for b-cell func-tion was the oral disposition index, deno-ted as DIo, calculated as follows as IGIadjusted for insulin sensitivity: DIo =([DI0–30/DG0–30] 3 [1/fasting insulin])(23). Insulin resistance was estimated us-ing homeostasis model assessment of in-sulin resistance (HOMA-IR; [fastinginsulin 3 fasting glucose]/22.5) (26).The modified Matsuda Index for whole-body insulin sensitivity was calculated asfollows: (10,000/square root of [fastingglucose 3 fasting insulin] 3 [mean glu-cose3mean insulin]), withmean glucoseand mean insulin each calculated fromvalues at 0, 30, and 120 min of theOGTT (27,28). Total area under the curve(AUC) for insulin and AUC for glucosewere calculated using the trapezoidalrule, and a ratio of the two was created(AUCins/glu). Secondary measures ofb-cell function included the following:([DI0–30/DG0–30] 3 [1/HOMA-IR]);([DI0–30 /DG0–30] 3 [modified MatsudaIndex]); AUCinsulin/glucose 3 (1/fasting in-sulin); AUCinsulin/glucose 3 (1/HOMA-IR);and AUCinsulin/glucose 3 (modified Mat-suda Index). Other covariates includedBMI and waist circumference. Categoriesof overweight/obesity were those definedby the World Health Organization (29)both generically (25.0#BMI,30.0 kg/m2 for overweight; BMI $30.0 kg/m2

for obesity) and specifically for Asian pop-ulations (23.0,BMI#27.4 kg/m2foroverweight; BMI $27.5 kg/m2 for obe-sity). Waist circumference was categorizeddichotomously, using waist $90 cm formen and $80 cm for women (30), andby using tertiles.

Statistical analysisAnalyses were performed using SAS 9.3(SAS Institute, Cary, NC). ANOVA al-lowed comparison of DIo across glycemicstatus categories, including the Tukey testfor multiple comparisons. Non-normallydistributed variables were log-transformedas required to meet assumptions of re-gression. A Score test was conducted toevaluate the possible use of ordinal logis-tic regression; the proportional oddscould not be assumed as required (i.e.,the null hypothesis that the model wasconstrained by the proportional odds as-sumption was rejected [x2 = 98.59; de-grees of freedom = 3; P , 0.0001]).

2 DIABETES CARE care.diabetesjournals.org

Reduced beta cell function and dysglycemia

Therefore, polytomous logistic regression,rather than ordinal logistic regression, wasdetermined appropriate, and we assessedassociations of glycemic status with eachcontinuous independent variable (per SDchange).

Interaction effects of DIo and covariates(sex, age, BMI, and waist circumference)on DIo were tested by fitting interactionterms in the models (including all othercovariates) and using hierarchical backwardelimination to remove interaction termsthat were insignificant at the P, 0.05 level.Age, BMI, and waist circumference werefitted as continuous and categorical varia-bles. The hyperbolic relationship of insulinsecretion and insulin sensitivity (21) wasassessed using linear regression to estimateln(DI0–30 /DG0–30) or ln(AUCinsulin/glucose)as a function of ln(1/fasting insulin) or ln(1/HOMA-IR) or ln(modified MatsudaIndex). A hyperbolic relationship wasconfirmed if the slope was approxi-mately equal to 21 and 95% CI ex-cluded 0 (23,31). Data were expressedas mean and SD.

RESULTSdTable 1 provides baselinecharacteristics for the participants across gly-cemic status groups. Among all participants,the mean age was 44.2 years (9.3), and 37%were female. According to specific WorldHealth Organization cut-offs for Asians,47% were overweight and 45% were obese.According to generic BMI cut-points, 52%were overweight and 21% were obese.Mean waist circumference among men andwomen was 97.1 cm (SD, 8.5) and 88.6 cm(SD, 8.5), respectively. One-fifth had newlydiagnosed diabetes (19.9%), more than half(53.2%) had prediabetes (15.8% with iIFG;16.0% with iIGT; and 21.4% with IFG plusIGT), and 31.9% had mild dysglycemia(15.8% iIFG and 16.0% iIGT).

Using our primary measure for b-cellfunction, DIo, the unadjusted mean(L/mmol) was highest in NGT and lowestamong those with more severe disease(IFG plus IGT and diabetes): NGT, 2.49;iIFG, 1.56; iIGT, 1.69; IFG plus IGT,0.99; and diabetes, 0.51. HOMA-IR, fast-ing insulin, and 1/modified Matsuda in-dex were all lowest in NGT and all highestin diabetes. DIo was significantly different(P , 0.05) for every pair-wise compari-son of glycemic status groups except iIFGand iIGT. There was a steep difference inmean DIo between NGT and iIFG(238%) and NGT and iIGT (232%; bothP, 0.0001). Like DIo, HOMA-IR was sig-nificantly different between every pair-wisecomparison of glycemic status groups

Table

1dCharacteristics

ofparticipants

inthe

Diabetes

Com

munity

Lifestyle

Improvem

entProgram

subdividedby

glycemic

status

Characteristic

Total

N=1,264

NGT

n=341

iIFGn=200

iIGT

n=202

IFGplus

IGT

n=270

Diabetes

n=251

Male,n

(%)

803(63.5)

209(61.3)

118(59.0)

133(65.8)

168(62.2)

175(69.7)

Age,years

44.219.27

41.999.03

44.699.33

43.139.22

45.309.27

46.548.89

BMI,kg/m

227.41

3.7626.77

3.6227.44

3.7927.43

3.3528.26

4.1627.34

3.65Waist

circumference,cm

93.989.40

92.219.30

93.099.63

94.258.88

95.759.34

94.989.39

Fastingglucose,m

mol/L

5.800.95

5.070.29

5.870.30

5.130.27

6.030.36

7.031.22

30-min

glucose,mmol/L

9.811.88

8.411.21

9.881.31

9.051.15

10.121.31

11.951.96

2-hglucose,m

mol/L

8.813.05

6.240.91

6.610.89

8.740.79

9.270.97

13.642.70

Fastinginsulin,pm

ol/L77.44

45.8065.17

45.2070.26

35.0479.35

38.9685.48

54.0889.66

44.7730-m

ininsulin,pm

ol/L504.83

343.78594.59

363.96532.53

354.34636.40

407.96433.96

268.41331.14

210.552-h

insulin,pmol/L

784.44570.28

615.40447.49

615.07426.08

1,145.99725.91

871.88538.24

764.02560.86

1/fastinginsulin,L/pm

ol0.017

0.0090.020

0.0100.017

0.0080.016

0.0090.015

0.0090.014

0.009Insulinogenic

index,IG

I(pmolin

s /mmolglu )

124.23156.43

188.20252.44

127.32108.48

148.35104.13

89.9164.52

52.3743.37

HOMA-IR

[mmolglu

3pm

olins /(L

2)]20.22

12.8414.72

10.2218.36

9.2618.17

9.1322.88

14.0527.96

15.07Modified

Matsuda

[(L2)/m

molglu

3pm

olins ]

10.526.20

13.597.16

11.275.68

9.395.16

9.114.99

8.175.35

Oraldisposition

index

(L/mmolglu )*

1.302.35

2.492.05

1.561.90

1.691.75

0.991.86

0.511.86

IGI3

(1/HOMA-IR

)(L/m

molglu ) 2*

5.112.60

11.082.07

5.971.94

7.431.77

3.701.89

1.652.02

IGI3

mod

Matsud

a(L/m

molglu ) 2*

804.922.35

1,702.501.94

1,000.631.80

988.641.60

594.211.72

287.341.77

AUCins/glu

31/fasting

insulin*0.77

1.761.10

1.500.81

1.501.04

1.520.67

1.460.40

1.66AUCins/glu

31/H

OMA-IR

*3.00

1.974.90

1.533.12

1.544.57

1.552.52

1.491.29

1.88AUCins/glu

3modified

Matsud

a*473.50

1.69752.12

1.30522.73

1.29608.33

1.28404.96

1.28225.70

1.55

Data

aremean

andSD

unlessoth

erwise

stated.*Geom

etricmean

andgeom

etricSD

providedfor

DIoand

secondary

measures

ofbeta-cellfunction.

care.diabetesjournals.org DIABETES CARE 3

Staimez and Associates

except iIFG and iIGT. Compared withNGT, HOMA-IR was 25% greater in iIFGand 23% greater in iIGT (both P ,0.0001). However, unlike HOMA-IR, themodified Matsuda Index was significantlydifferent between every pair-wise compar-ison of glycemic status groups except be-tween iIGT and IFG plus IGT, iIGT anddiabetes, and IFG plus IGT and diabetes.Differences between glycemic status levelsremained after adjustment for age and sex(Fig. 1). Furthermore, differences in meanDIo persisted after adjustment for age, sex,BMI,waist circumference, and family historyas follows: DIo was 2.48 L/mmol in NGT vs.1.55, 1.69, 1.00, and 0.50 L/mmol in iIFG,iIGT, IFG plus IGT, and diabetes, respec-tively (all P, 0.0001 vs. NGT). Across thesame glycemic categories, HOMA-IR(mmolglu 3 pmolins/L

2) was 15.29, 18.52,18.02, 21.98, and 28.18 (all P , 0.05 vs.NGT) and the modified Matsuda Index[(L2)/mmolglu 3 pmolins] was 13.31,11.14, 9.49, 9.55, and 8.09 (all P ,0.0001 vs. NGT), and all were adjusted forage, sex, BMI, waist circumference, andfamily history.

A hyperbolic relationship was foundbetween insulin sensitivity and insulinsecretion. Using linear regression to esti-mate ln(IGI) as a function of ln(1/fastinginsulin), the slope and its 95% CI foreach glycemic category did not equal zeroand were negative. For example, amongindividuals with diabetes, the slope ofregression was 20.8 (95% CI, 20.9 to20.6). Secondary measures of b-cellfunction also were evaluated for hyper-bolic relationships between components.The hyperbolic relationship of IGI and1/HOMA-IR was poorer than IGI and1/fasting insulin; among those with dia-betes, the slope of regression between IGIand 1/HOMA-IR was 20.6 (95% CI,20.7 to 20.4). In contrast, the relation-ship improved slightly using the modifiedMatsuda Index instead of 1/fasting insulin(20.9; 95% CI, 21.1 to 20.8). The hy-perbolic relationships between AUCins/glu

and the insulin sensitivity measures weresimilar to or better than those between IGIand the various insulin sensitivity mea-sures. The weakest relationship withAUCins/glu was found with 1/HOMA-IR.

No interactions were found betweenDIo and sex, age, BMI, or waist circumfer-ence. Polytomous logistic regression wasused to determine the odds of each hyper-glycemic status category compared withNGT for incremental changes in DIo,HOMA-IR, and other covariates. Regres-sion results are shown in Table 2. DIo andHOMA-IR were each independently asso-ciated with glycemic status, as shown inmodel 1. The odds for each glycemic cat-egory compared with NGT were signifi-cantly lower for every SD increase in DIo,both in prediabetes and in diabetes (eachP, 0.0001). In particular, the odds ratio(OR) of diabetes compared with NGTwasextremely small for each SD increase inDIo. Almost no change in the magnitudeof association between DIo and glycemicstatus was found after adjustment for age,for age, BMI, waist circumference, or forage, BMI, waist circumference, and familyhistory (Table 2,models 2, 3, 4). In contrastto DIo, the odds for any glycemic categorycompared with NGT were greater for everySD increase in HOMA-IR. Adjustment forage, BMI, waist circumference, and familyhistory did not substantially change themagnitude of association between HOMA-IR and glycemic status.

Polytomous regression with second-ary measures of b-cell function yieldedmore pronounced findings, with the rel-ative contributions of b-cell function onglycemic status exceeding that of HOMA-IR. For the measure AUCins/glu 3 modi-fied Matsuda, which exhibited the besthyperbolic relationship between insulinsecretion and insulin sensitivity, b-cellfunction was independently associatedwith glycemic status (iIFG: OR, 0.11;95% CI, 0.07–0.16; iIGT: OR, 0.34;95% CI, 0.25–0.46) and HOMA-IR wasnot associated with glycemic status (iIFG:OR, 0.94; 95% CI, 0.66–1.33; iIGT: OR,1.16; 95% CI, 0.84–1.60).

CONCLUSIONSdThis study under-scores the importance of b-cell dysfunc-tion relative to insulin resistance acrossglycemic status groups in Asian Indians,particularly iIFG and iIGT. Using an in-dex of b-cell function relative to insulinsensitivity, DIo, a highly significant differ-ence in DIo was observed between NGTand iIFG or iIGT. A difference in insulinresistance, as measured by HOMA-IR,also was observed; however, the differ-ence in mean HOMA-IR was greatest be-tween IFG plus IGT and diabetes,whereas the greatest difference in meanDIo was between NGT and iIFG. Results

Figure 1dAge- and sex-adjusted mean oral disposition index and mean HOMA-IR across gly-cemic status. White bars indicate DIo; filled triangles (▲) indicate HOMA-IR. Error bars indicateSE estimates. Geometric means and geometric SE were calculated for DIo (L/mmol). HOMA-IR ispresented as (mmolglu 3 mmolins)/L

2).

4 DIABETES CARE care.diabetesjournals.org

Reduced beta cell function and dysglycemia

from statistical modeling showed thatthe relative contributions of DIo to iIFGand to iIGT were greater than those ofHOMA-IR, even after adjustment forvariables known to impact disease devel-opment, including age, BMI, waist cir-cumference, and family history ofdiabetes. These findings suggest that de-spite conflicting studies (17–20) in AsianIndians, a decrease in b-cell function maybe a primary etiological factor in the de-velopment of type 2 diabetes in this ethnicgroup.

DIo was selected as the primary mea-sure of b-cell function for several reasons.First, the measure agrees with the biolog-ical constructs known for b-cell function,because secretion of insulin is measuredrelative to the prevailing levels of insulinsensitivity in the body (21). Second,mathematical confirmation in studieswith more rigorous designs has de-monstrated a hyperbolic relationship ofthe two components (i.e., insulin secre-tion and insulin sensitivity) (23,32). Third,autocollinearity between the componentsof DIo is unlikely to drive the hyperbolicrelationship, as described elsewhere (31).Fourth, longitudinal studies have indicatedthat DIo is a good measure of disease pro-cesses related to b-cell function, because itpredicts the development of future diabetes(23). In addition to DIo, alternative mea-sures of b-cell function were analyzed.

Like another study, our results suggest that(AUCins/glu3modifiedMatsuda Index) and(AUCins/glu 3 1/fasting insulin) should beexamined further as potentially strongmeasures of DIo (32). Across measures ofb-cell function, similar findings indicatedthat the relative contribution of b-cell func-tion toward glycemic status categories wasgreater than that of insulin resistance.

Studies that have examined b-cellfunction in various ethnic groups (i.e.,b-cell function defined as insulin secre-tion with adjustment for insulin sensitiv-ity within individuals) (12,21) haveshown alternative patterns across glyce-mic phenotypes compared with thosefound in the current study. In a study of1,399 normal-weight Japanese adults,those with iIFG were found to have some-what preserved b-cell function using theratio of change in AUC for plasma insulinto plasma glucose (DAUCPI/DAUCPG)from 0 to 120 min of the OGTT after ad-justment of the Matsuda Index. This mea-sure of the disposition index was reducedin iIFG (mean 6 SEM, 7.1 6 0.5; P ,0.05; with iIFG defined as fasting plasmaglucose 6.1–7.0 mmol/L) compared withthe NGT group (9.7 6 0.2). A more ex-treme and statistically significant differ-ence was detected in the iIGT (mean 6SEM, 2.7 6 0.2) and the combined IFGplus IGT (2.1 6 0.2) groups comparedwith both the NGT and iIFG groups

(P , 0.05) (33). In another study of1,272 Chinese adults, a decline in earlyphase DIo (DAUCPI/DAUCPG at 30 min,adjusted with the Matsuda Index) wasfound in both iIFG (fasting glucose5.6–7.0 mmol/L) and iIGTdboth signif-icantly lower than NGTdbut DIo in iIGTalso was significantly lower than DIo inIFG (both P , 0.05) (34). The currentstudy showed that early-phase DIo wassignificantly reduced in iIFG as in iIGT.These findings are supported by the lon-gitudinal Inter99 study involving Danishadults. Early-phase DIo was measured atboth baseline and 5-year follow-up usingfasting insulin, 30-min insulin, and30-min plasma glucose adjusted for insu-lin sensitivity index (10). The authors re-ported that DIo was lower in those withincident iIGT and IFG plus IGT than inthose with incident iIFG (fasting plasmaglucose 6.1–6.9 mmol/L) after 5 years offollow-up. However, among all individualswith NGT at baseline (n = 3,145), thosewho would later develop iIFG (NGT toiIFG) had lower mean DIo at baseline com-paredwith those whowould eventually de-velop iIGT (NGT to iIGT). These resultssuggest that individuals who develop iIFGalready have significantly lower DIo in nor-moglycemia, and the reduction in DIo dur-ing normoglycemia may be as severe ormore severe in those who eventually de-velop iIFG compared with iIGT. In

Table 2dStandardized polytomous logistic regression estimates for the OR of each glycemic status group

Normal (reference)n =341

iIFGn = 200

iIGTn = 202

IFG plus IGTn = 270

Diabetesn = 251

Model 1DIo 1 0.35 (0.22–0.54)‡ 0.37 (0.24–0.57)‡ 0.02 (0.01–0.04)‡ 0.001 (0.001–0.001)‡xHOMA-IR 1 1.56 (1.17–2.07)† 1.53 (1.15–2.03)† 1.97 (1.50–2.59)‡ 1.91 (1.44–2.55)‡

Model 2DIo 1 0.35 (0.22–0.54)‡ 0.37 (0.24–0.57)‡ 0.02 (0.01–0.04)‡ 0.001 (0.001–0.001)‡xHOMA-IR 1 1.69 (1.26–2.27)† 1.61 (1.20–2.15)† 2.18 (1.65–2.89)‡ 2.14 (1.59–2.88)‡Age 1 1.44 (1.20–1.74)† 1.21 (1.00–1.45)* 1.60 (1.32–1.93)‡ 1.90 (1.52–2.38)‡

Model 3DIo 1 0.35 (0.23–0.55)‡ 0.37 (0.24–0.56)‡ 0.02 (0.01–0.04)‡ 0.001 (0.001–0.001)‡xHOMA-IR 1 1.70 (1.23–2.33)† 1.53 (1.11–2.11)† 2.07 (1.52–2.81)‡ 2.31 (1.66–3.22)‡Age 1 1.46 (1.21–1.76)‡ 1.20 (0.99–1.44) 1.59 (1.32–1.93)‡ 1.85 (1.48–2.32)‡BMI 1 1.11 (0.87–1.42) 0.98 (0.77–1.25) 1.06 (0.84–1.35) 0.71 (0.53–0.96)*Waist 1 0.88 (0.70–1.11) 1.11 (0.88–1.41) 1.08 (0.85–1.37) 1.16 (0.87–1.55)

Model 4DIo 1 0.36 (0.23–0.55)‡ 0.37 (0.24–0.56)‡ 0.02 (0.01–0.05)‡ 0.001 (0.001–0.001)‡xHOMA-IR 1 1.69 (1.23–2.33)† 1.53 (1.11–2.11)† 2.07 (1.52–2.81)‡ 2.31 (1.66–3.22)‡Age 1 1.47 (1.21–1.77)‡ 1.20 (0.99–1.44) 1.62 (1.33–1.96)‡ 1.84 (1.47–2.32)‡BMI 1 1.11 (0.87–1.41) 0.98 (0.77–1.25) 1.06 (0.83–1.35) 0.71 (0.53–0.96)*Waist 1 0.88 (0.70–1.12) 1.11 (0.88–1.41) 1.09 (0.86–1.39) 1.16 (0.86–1.55)Family history NI{ NI{ NI{ NI{ NI{

Data presented as OR and 95% CI. *P,0.05. †P,0.01. ‡P,0.001. xOR and CI are ,0.001. {Standardized ORs for family history are not interpretable.

care.diabetesjournals.org DIABETES CARE 5

Staimez and Associates

populations in which the progression to-ward diabetes is particularly rapid, the levelof DIo in normoglycemia may be an impor-tant factor related to disease risk. Futuremultiethnic studies are needed to deter-mine if and how b-cell decline varies byethnicity and to what degree differenceslie between iIFG and iIGT.

This study has several importantstrengths. It contains a large, well-characterized, community-based samplestemming from the screening of almost20,000 people and, therefore, was notlimited to hospital- or clinic-based sam-ples. All cases of diabetes were newlydiagnosed. A hyperbolic relationship wasfound between insulin sensitivity andinsulin secretion using linear regressionto estimate ln(DI0–30 /DG0–30) as a func-tion of ln(1/fasting insulin), indicating ap-propriate use of the disposition index as ameasure of b-cell function. We used stan-dardized regression to enable comparisonof variables that were measured in differ-ent units and, consequently, to directlycompare the relative contributions ofb-cell function and HOMA-IR to the de-velopment of prediabetes and diabetes.

The limitations of this study includeits cross-sectional design, limiting furtherinquiry regarding temporality, and thedegree of representation of the sample,because results may not be generalizableto other populations (e.g., other racial/ethnic groups). In addition, all subjectshad a random blood glucose level of$6.1 mmol/L before receiving theOGTT and, therefore, the NGT groupmay not be representative of normoglyce-mia. However, if the NGT group had nor-mal random blood glucose levels, evengreater differences between NGT andiIFG or iIGT groups would be expected.Therefore, our results provide conserva-tive estimates of the lower b-cell functionand higher insulin resistance in iIFG andiIGT relative to NGT groups. We ex-cluded seven individuals for having neg-ative or zero values of the insulinogenicindex, values considered biologicallyimplausible; however, this numbercomprises a very small percentage of thetotal sample (0.5%), smaller than repor-ted in another study (2.7%) (23). Onlyone OGTT test was performed for eachparticipant, and thus the classification ofsome individuals may have changed if theOGTT had been performed a secondtime. Another limitation pertains to theuse of an OGTT-derived measure ofb-cell function, rather than estimatesbased on the glucose clamp technique or

the frequently sampled intravenous glu-cose tolerance test as used by Bergmanet al. (21), who showed that insulin secre-tion had a hyperbolic relationship (i.e.,y = constant/x) with the existing state ofinsulin sensitivity. The findings here, us-ing the measure of DIo for b-cell function,are consistent with several other studiesthat used euglycemic clamps and intrave-nous glucose tolerance tests. These stud-ies have shown that poor insulin secretionbegins sometime during normoglycemia(35,36). Also, they have highlighted thatpotential differences in b-cell dysfunctionmay exist between IFG and IGT pheno-types (37–39), particularly that impair-ment of b-cell function may occur iniIFG at least as much as in iIGT.

Although the disposition index wasoriginally based on intravenous samplingtechniques, an OGTT-derived measurehas been shown to be valid (23) andmay include additional benefits in thestudy of b-cell function. The hyperbolicrelationship between insulin sensitivityand insulin secretion has been demon-strated using OGTT data, indicating thatDIo is a valid measure of b-cell functionand is highly predictive of 10-year inci-dence of diabetes (23). Several metabolicfactors, such as glucose disposition, differbetween oral and intravenous glucoseloads as related to different responses inthe liver and in the periphery and, thus,OGTTs may probe important physiolog-ical processes in glucose metabolism thatcannot be studied through intravenoustesting. Finally, OGTTs are easy to per-form in populations, making them moreconvenient for epidemiological studies inwhich recruitment of sufficiently largestudy samples can correct for within-subject variability (23,31,40). The mini-mum sample size needed to detect a 20%change when using the insulinogenic in-dex is 181 (40), a number far exceeded bythe sample size of the current study. Thus,DIo is a valid, informative, and practicalapproach to the study of b-cell function.

Using a robustmeasure ofb-cell func-tion, the current study provides evidencethat those in any category of dysglycemia,including iIFG and iIGT, have lowerb-cell function compared with those inthe NGT category. This study alsoshowed that insulin resistance was greateracross all categories of dysglycemia com-pared with NGT. However, the relativecontribution of insulin resistance wasnot as great as that of b-cell function inany given glycemic status group. As such,the primary prevention and control of

diabetes will require strategies to preserveb-cell function and reduce b-cell decline.

In conclusion, our data demonstratemarkedly reduced b-cell function amongAsian Indians with mild dysglycemia.These abnormalities cannot be attributedto differences in age, adiposity, insulinsensitivity, or family history. Prospectivestudies are needed to further investigatethe relative roles ofb-cell dysfunction andinsulin resistance in the early natural his-tory of diabetes.

AcknowledgmentsdThe project is sup-ported by a BRiDGES grant from the In-ternational Diabetes Federation. BRiDGES, anInternational Diabetes Federation project, issupported by an educational grant from LillyDiabetes. This project also was supported byEmory Global Health Institute, Molecules toHumankind Program sponsored by theBurroughsWellcomeFund (L.R.S), theNationalInstitutes ofHealth T32 grant (5T32DK007298-33 to M.B.W), VA award HSR&D IIR 07-138(L.S.P. and K.M.V.N.), and Cystic FibrosisFoundation award PHILLI12A0 (L.S.P.).No potential conflict of interest relevant to

this article was reported.L.R.S. analyzed data, wrote the manuscript,

drafted tables and figures, and reviewed andrevised the manuscript. M.B.W. and H.R. re-searched data and reviewed and revised themanuscript. M.K.A. and J.B.E.-T. reviewedand revised the manuscript. L.S.P. contributedto analysis, discussion, and reviewed and re-vised the manuscript. V.M. contributed toanalysis, discussion, and reviewed and revisedthe manuscript. K.M.V.N. contributed toconcept, design, and analysis, and reviewedand revised the manuscript. L.R.S. is theguarantor of this work and, as such, had fullaccess to all the data in the study and takesresponsibility for the integrity of the data andthe accuracy of the data analysis.This is the 8th paper from Global Diabetes

Research Centre (GDRC)da collaborationbetween Madras Diabetes Research Founda-tion, Chennai, India, and Emory University,Atlanta, Georgia (GDRC- 8). Parts of this studywere presented at the 71st Scientific Sessionsof the American Diabetes Association, SanDiego, California, 24–28 June 2011.

References1. International Diabetes Federation. The

Diabetes Atlas. 5th ed. Brussels, Belgium,International Diabetes Federation, 2011

2. DeFronzo RA. Glucose intolerance andaging. Diabetes Care 1981;4:493–501

3. Saad MF, Knowler WC, Pettitt DJ, NelsonRG, Charles MA, Bennett PH. A two-stepmodel for development of non-insulin-dependent diabetes. Am J Med 1991;90:229–235

6 DIABETES CARE care.diabetesjournals.org

Reduced beta cell function and dysglycemia

4. Goldstein BJ. Insulin resistance as the coredefect in type 2 diabetes mellitus. AmJ Cardiol 2002;90:3G–10G

5. Amati F, Dubé JJ, Coen PM, Stefanovic-Racic M, Toledo FG, Goodpaster BH.Physical inactivity and obesity underliethe insulin resistance of aging. DiabetesCare 2009;32:1547–1549

6. Gerich JE. Contributions of insulin-resistance and insulin-secretory defects tothe pathogenesis of type 2 diabetes mel-litus. Mayo Clinic Proc 2003;78:447–456

7. Sjaarda LA, Michaliszyn SF, Lee S, TfayliH, Bacha F, Farchoukh L, Arslanian SA.HbA1c diagnostic categories and beta-cellfunction relative to insulin sensitivity inoverweight/obese adolescents. DiabetesCare 2012;35:2559–2563

8. Cnop M, Vidal J, Hull RL, et al. Pro-gressive loss of beta-cell function leads toworsening glucose tolerance in first-degree relatives of subjects with type 2diabetes. Diabetes Care 2007;30:677–682

9. Cali AM,ManCD, Cobelli C, et al. Primarydefects in beta-cell function further exac-erbated by worsening of insulin resistancemark the development of impaired glu-cose tolerance in obese adolescents. Di-abetes Care 2009;32:456–461

10. Faerch K, Vaag A, Holst JJ, Hansen T,Jørgensen T, Borch-Johnsen K. Natural his-tory of insulin sensitivity and insulin secre-tion in the progression from normal glucosetolerance to impaired fasting glycemia andimpaired glucose tolerance: the Inter99study. Diabetes Care 2009;32:439–444

11. Elbein SC. Genetics factors contributingto type 2 diabetes across ethnicities. J Di-abetes Sci Tech 2009;3:685–689

12. Kahn SE, Hull RL, Utzschneider KM.Mechanisms linking obesity to insulinresistance and type 2 diabetes. Nature2006;444:840–846

13. Anjana RM, Pradeepa R, Deepa M, et al.;ICMR–INDIAB Collaborative Study Group.Prevalence of diabetes and prediabetes(impaired fasting glucose and/or impairedglucose tolerance) in urban and rural India:phase I results of the Indian Council ofMedical Research-INdia DIABetes (ICMR-INDIAB) study. Diabetologia 2011;54:3022–3027

14. Chan JC, Malik V, Jia W, et al. Diabetes inAsia: epidemiology, risk factors, and path-ophysiology. JAMA 2009;301:2129–2140

15. Mohan V, Sharp PS, Cloke HR, Burrin JM,Schumer B, Kohner EM. Serum immu-noreactive insulin responses to a glucoseload in Asian Indian and European type 2(non-insulin-dependent) diabetic patientsand control subjects. Diabetologia 1986;29:235–237

16. Dowse GK, Zimmet PZ, Alberti KG, et al.;Mauritius NCD Study Group. Serum in-sulin distributions and reproducibility ofthe relationship between 2-hour insulin

and plasma glucose levels in Asian Indian,Creole, and Chinese Mauritians. Metabo-lism 1993;42:1232–1241

17. Dowse GK, Qin H, Collins VR, ZimmetPZ, Alberti KG, Gareeboo H; The Mau-ritius NCD Study Group. Determinants ofestimated insulin resistance and beta-cellfunction in Indian, Creole and ChineseMauritians. Diabetes Res Clin Pract 1990;10:265–279

18. Snehalatha C, Ramachandran A, SatyavaniK, Latha E, Viswanathan V. Study of ge-netic prediabetic south Indian subjects.Importance of hyperinsulinemia and beta-cell dysfunction. Diabetes Care 1998;21:76–79

19. Snehalatha C, Satyavani K, Sivasankari S,Vijay V, Ramachandran A. Insulin secre-tion and action in different stages of glu-cose tolerance in Asian Indians. DiabeticMed 1999;16:408-414

20. SnehalathaC, RamachandranA, SivasankariS, SatyavaniK, VijayV. Insulin secretion andaction show differences in impaired fastingglucose and in impaired glucose tolerance inAsian Indians. Diabetes Metab Res Rev2003;19:329–332

21. Bergman RN, Phillips LS, Cobelli C.Physiologic evaluation of factors control-ling glucose tolerance in man: measure-ment of insulin sensitivity and beta-cellglucose sensitivity from the response tointravenous glucose. J Clin Invest 1981;68:1456–1467

22. WeberMB, Ranjani H,Meyers GC,MohanV, Narayan KM. A model of translationalresearch for diabetes prevention in lowand middle-income countries: The Di-abetes Community Lifestyle Improve-ment Program (D-CLIP) trial. Prim CareDiabetes 2012;6:3–9

23. Utzschneider KM, Prigeon RL,Faulenbach MV, et al. Oral dispositionindex predicts the development of futurediabetes above and beyond fasting and 2-h glucose levels. Diabetes Care 2009;32:335–341

24. World Health Organization.WHOExpertCommittee on Diabetes Mellitus: secondreport. World Health Organ Tech Rep Ser1980;646:1–80

25. American Diabetes Association. Standardsof medical care in diabetesd2010. Di-abetes Care 2010;33(Suppl 1):S11–S61

26. Matthews DR, Hosker JP, Rudenski AS,Naylor BA, Treacher DF, Turner RC. Ho-meostasis model assessment: insulin re-sistance and beta-cell function fromfasting plasma glucose and insulin con-centrations inman. Diabetologia 1985;28:412–419

27. Matsuda M, DeFronzo RA. Insulin sensi-tivity indices obtained from oral glucosetolerance testing: comparison with theeuglycemic insulin clamp. Diabetes Care1999;22:1462–1470

28. DeFronzo RA, Matsuda M. Reduced timepoints to calculate the composite index.Diabetes Care 2010;33:e93

29. WHO Expert Consultation. Appropriatebody-mass index for Asian populationsand its implications for policy and in-tervention strategies. Lancet 2004;363:157–163

30. International Diabetes Federation. TheIDF Consensus worldwide definition of themetabolic syndrome. Brussels, Belgium,International Diabetes Federation, 2006

31. Retnakaran R, Shen S, Hanley AJ, VuksanV, Hamilton JK, Zinman B. Hyperbolic re-lationship between insulin secretion andsensitivity on oral glucose tolerance test.Obesity (Silver Spring) 2008;16:1901–1907

32. Retnakaran R, Qi Y, Goran MI, HamiltonJK. Evaluation of proposed oral disposi-tion index measures in relation to the ac-tual disposition index. Diabetic Med2009;26:1198-1203

33. Miyazaki Y, Akasaka H, Ohnishi H, SaitohS, DeFronzo RA, Shimamoto K. Differencesin insulin action and secretion, plasma lip-ids and blood pressure levels between im-paired fasting glucose and impaired glucosetolerance in Japanese subjects. HypertensRes 2008;31:1357–1363

34. Bi Y, Zhu D, Jing Y, et al. Decreased betacell function and insulin sensitivity con-tributed to increasing fasting glucose inChinese. Acta Diabetol 2012;49(Suppl 1):S51–S58

35. Szoke E, Shrayyef MZ, Messing S, et al.Effect of aging on glucose homeostasis:accelerated deterioration of beta-cellfunction in individuals with impairedglucose tolerance. Diabetes Care 2008;31:539–543

36. Slentz CA, Tanner CJ, Bateman LA, et al.Effects of exercise training intensity onpancreatic beta-cell function. DiabetesCare 2009;32:1807–1811

37. Kanat M, Mari A, Norton L, et al. Distinctb-cell defects in impaired fasting glucoseand impaired glucose tolerance. Diabetes2012;61:447–453

38. Weyer C, Bogardus C, Pratley RE. Meta-bolic characteristics of individuals withimpaired fasting glucose and/or impairedglucose tolerance. Diabetes 1999;48:2197–2203

39. Hong J, Gui MH, Gu WQ, Zhang YF, XuM, Chi ZN, Zhang Y, Li XY, Wang WQ,Ning G. Differences in insulin resistanceand pancreatic B-cell function in obesesubjects with isolated impaired glucosetolerance and isolated impaired fastingglucose. Diabetic Med 2008;25:73–79

40. Utzschneider KM, Prigeon RL, Tong J,et al. Within-subject variability of meas-ures of beta cell function derived from a 2h OGTT: implications for research stud-ies. Diabetologia 2007;50:2516–2525

care.diabetesjournals.org DIABETES CARE 7

Staimez and Associates