I N A L A R T I C L E

Plasma Oxidizability inSubjects With NormalGlucose Tolerance,Impaired GlucosefTolerance, and NIDDMSTEVEN M. HAFFNER, MDAHMAD AGIL, PHDLEENA MYKKANEN, MD

MICHAEL P. STERN, MDISHWARLALJIALAL, MD, FACN, FACB

OBJECTIVE — Several lines of evidence support an atherogenic role for oxidizedlow-density lipoprotein (LDL). Studies on LDL oxidation in diabetes to date haveexamined LDL isolated from plasma, but have failed to evaluate the other pro- andantioxidant factors present in vivo, the balance of which could be crucial in determin-ing the susceptibility of LDL to lipid peroxidation.

RESEARCH DESIGN AND METHODS— We examined the oxidizability ofplasma from Mexican-Americans in the San Antonio Heart Study. The oxidizability ofplasma in 75 subjects with normal glucose tolerance (NGT), impaired glucose toler-ance (IGT), and non-insulin-dependent diabetes mellitus (NIDDM) was studied afterco-incubation with a free radical initiator, 2,2'-azobis-2-amidinopropane hydrochloride(AAPH). Lipid peroxide (LPO) levels were measured by a modified fluorimetric assay.

RESULTS — Baseline LPO levels (jumol/1; means ± SE) were similar in the threeglucose tolerance categories (NGT, 1.99 ± 0.07; IGT, 1.88 ± 0.07; NIDDM, 1.97 ±0.07; P = 0.521). However, after incubation with AAPH (NGT, 4.30 ±0.20; IGT, 4.45± 0.20; NIDDM, 5.35 ± 0.20; P = 0.003), the diabetic plasma had significantlygreater amounts of LPOs compared with the other two groups. There was no significantdifference in LPOs between the NGT and IGT groups. The statistical significance ofincreased oxidizability of the diabetic plasma persisted after exclusion of patients whosmoked cigarettes (n = 15) or who had vascular disease (n = 4).

CONCLUSIONS — In conclusion, this study shows that the plasma of Mexican-American subjects with NIDDM is more prone to lipid peroxidation than that ofnon-Hispanic whites.

From the Division of Clinical Epidemiology (S.M.H., L.M., M.P.S.), Department of Medicine, Univer-sity of Texas Health Science Center at San Antonio, San Antonio, Texas; and the Center for HumanNutrition and Laboratory of Molecular Pathology (A.A., I.J.), Department of Pathology and InternalMedicine, University of Texas Southwestern Medical Center at Dallas, Dallas, Texas.

Address correspondence and reprint requests to Steven M. Haffner, MD, University of Texas HealthSciences Center, 7703 Floyd Curl Drive, San Antonio, TX 78284-7873.

Received for publication 11 April 1994 and accepted in revised form 2 January 1995.AAPH, 2,2'-azobis-2-amidinopropane hydrochloride; ALX, alloxan; BMI, body mass index; CHD,

coronary heart disease; ECG, electrocardiogram; HDL, high-density lipoprotein; IDDM, insulin-dependent diabetes mellitus; IGT, impaired glucose tolerance; LDL, low-density lipoprotein; LPO,lipid peroxide; NGT, normal glucose tolerance; NIDDM, non-insulin dependent diabetes mellitus;WHR, waist-to-hip ratio.

The risk of cardiovascular disease isincreased two- to fourfold in pa-tients with non-insulin-dependent

diabetes mellitus (NIDDM) (1-3). Fur-thermore, this excess risk is explainedonly partially by increased levels of stan-dard risk factors (1-3). Compared withnondiabetic individuals, NIDDM patientshave increased triglyceride concentra-tions and decreased high-density lipopro-tein (HDL) cholesterol concentrations butsimilar absolute concentrations of low-density lipoprotein (LDL) cholesterol(4,5).

Recent data have suggested thepossibility that oxidized LDL is moreatherogenic than ordinary LDL (6). Oxi-dized LDL occurs in atherosclerotic le-sions in experimental animals (7), and an-tibodies to oxidized LDL have been foundto correlate with the progression of ca-rotid atherosclerosis (8). Antioxidantsupplementation inhibits the progressionof atherosclerosis in experimental ani-mals (9), and high intake of vitamin E isassociated with a lower risk of coronaryheart disease (CHD) in both men andwomen (10,11).

A number of recent reviews haveexamined the possible role of oxidizedLDL in diabetes (12,13). Several studieshave examined whether lipid peroxide(LPO) levels or conjugated dienes are el-evated in diabetes (14-22), but the inter-pretation of the data showing elevatedLPO levels has been complicated by thehigh rate of vascular disease in clinicallybased studies (14,15,17,18). In contrast,other studies did not find an increase inperoxide levels in diabetic subjects with-out vascular disease (20,21), suggestingthat the earlier reports (14,15) of in-creased LPO levels in diabetic subjectsmight be due to a disposition to vasculardisease rather than to the diabetes per se.Indeed, Sato et al. (14) found higher LPOlevels in diabetic subjects with vasculardisease than in diabetic subjects withoutvascular disease. In addition, severalstudies (14,15) did not specify whether

646 DIABETES CARE, VOLUME 18, NUMBER 5, MAY 1995

Haffner and Associates

subjects had insulin-dependent diabetesmellitus (IDDM) or NIDDM.

Some authors have speculatedthat insulin resistance may be in part re-sponsible for the increased risk of CHD inNIDDM patients (23). Yet no previousstudy has examined lipid peroxidation insubjects with impaired glucose tolerance(IGT) whose rates of CHD approach thatobserved in NIDDM patients (24) andwho are markedly insulin-resistant (25),in spite of having only mild elevations ofblood glucose. In this study, we examineplasma LPO concentrations among sub-jects with NIDDM, IGT, and normal glu-cose tolerance (NGT) from the San Anto-nio Heart Study, a population-basedstudy of diabetes and cardiovascular dis-ease in which the rate of vascular diseaseis low compared with clinic-based stud-ies. We also examine the oxidizability ofplasma using a free radical initiator, 2,2'-azobis-2-amidinopropane hydrochloride(AAPH), without a transition metal.

RESEARCH DESIGN ANDMETHODS — The San Antonio HeartStudy is a population-based study of dia-betes and cardiovascular disease in Mexi-can-Americans and non-Hispanic whitesthat was designed to separate the effects ofethnicity and socioeconomic status. From1979 through 1982 (phase I) and from1984 through 1988 (phase II), we ran-domly selected households from SanAntonio, TX, census tracts including low-income (barrio), middle-income (tran-sitional), and high-income (suburban)census tracts (26,27). All men and non-pregnant women 25 through 64 years ofage who resided in the randomly selectedhouseholds were eligible to participate.Mexican-Americans were defined as indi-viduals whose ancestry and cultural tradi-tions derived from a Mexican national or-igin (28). Detailed descriptions of the1979 to 1982 survey (phase I) (26) andthe 1984 through 1988 survey (phase II)(27) have been published previously.This study was approved by the institu-tional review board of the University of

Texas Health Science Center at San Anto-nio. All subjects gave informed consent.

In October 1987, we began an8-year follow-up of the phase I cohort todetermine the incidence of NIDDM andcardiovascular disease (29). This surveywas completed in November 1990. Be-ginning in October 1991, we began a sim-ilar 7-year follow-up study of participantsin phase II. The present report is based onsubjects from the first census tract of thephase II follow-up (a middle-income,transitional neighborhood), 210 of whomwere Mexican-Americans and 156 ofwhom were non-Hispanic whites (30).

The ratio of waist-to-hip circum-ferences (WHR) was used as a measure ofbody fat distribution. Body mass index(BMI) was calculated as weight (in kilo-grams) divided by height (in meters)squared. Smoking was assessed by self-report, and subjects were classified assmokers or nonsmokers. The presence ofvascular disease (previous myocardial in-farction, cerebrovascular disease, periph-eral vascular disease, coronary bypass, orangioplasty or positive angiographicstudies) was also assessed by self-report.Subjects with definite or possible myocar-dial infarction on an electrocardiogram(ECG) by Minnesota Code Criteria (31)were also considered to have vascular dis-ease. Three of the four subjects with self-reported vascular disease also had ECGevidence of myocardial infarction.

At the follow-up examination,blood specimens were obtained after a12- to 14-h fast, and a second specimenwas obtained 2 h after the administrationof a 75-g glucose equivalent load (Or-angedex, Custom, Baltimore, MD). Lip-ids, lipoproteins, glucose, and insulinwere measured using methods previouslydescribed (26,27). These measurementswere made in the laboratory of the Divi-sion of Clinical Epidemiology in San An-tonio.

Diabetes (fasting plasma glucoselevel ^7.8 mmol/1 and/or 2-h postglu-cose load plasma glucose level >11.1mmol/1) and IGT (fasting plasma glucoselevel <7.8 mmol/1 and 2-h postglucose

load plasma glucose level between 7.8and 11.1 mmol/1) were diagnosed accord-ing to World Health Organization(WHO) criteria (32). Subjects who didnot meet WHO plasma glucose criteriabut who were under treatment with oralantidiabetic agents or insulin were alsoconsidered to have diabetes. Diabeticsubjects who were treated with diet ororal antihyperglycemic agents were con-sidered to have NIDDM. Diabetic subjectstreated with insulin who had an age ofonset of diabetes ^40 years and a BMI^30 kg/m2 were also considered to haveNIDDM. Insulin-taking subjects whowere considered to have IDDM or to haveunclassifiable diabetes were excludedfrom this report. We randomly selected25 subjects with NGT, 25 subjects withIGT, and 25 subjects with NIDDM fromthe 210 Mexican-American subjects inthe first target area of the follow-up of thephase II cohort. (There were too few non-Hispanic white subjects with NIDDM andwe therefore elected to study a more ho-mogeneous sample.) The original distri-bution of the 210 Mexican-American pa-tients by glucose tolerance status was:NIDDM, 31%; IGT, 14%; and NGT, 55%.

Fasting plasma specimens (col-lected with EDTA as an anticoagulant)were stored at — 70°C for ~ 1 year in SanAntonio. They were shipped on dry ice tothe laboratory of I. Jialal, MD, in Dallas,TX, for measurement of plasma LPO lev-els. These samples had not been thaweduntil the determinations of plasma LPOlevels were performed. Subjects withNIDDM (11.9 months), IGT (12.0months), or NGT (11.9 months) werematched for duration of sample storage.The plasma LPO levels were measured induplicate both in the basal state and afterco-incubation of plasma in the presenceof an aqueous-free radical initiator, 50mmol/1 AAPH, which thermally decom-poses to produce peroxyl radicals at aconstant rate (33). The assay of Yagi (34)was modified to improve specificity forLPOs in the plasma and prevent interfer-ence from bilirubin and sialic acid. Themethod of Yagi has been shown to corre-

DiABtnus CARE, VOLUME 18, NUMBER 5, MAY 1995 647

Lipid peroxidation in NIDDM

Table 1—Clinical and metabolic characteristics of the study sample

nAge (years)Men (%)Smokers (n)Vascular disease (n)BM1 (kg/m2)WHRFasting glucose (mmol/1)2-h glucose (mmol/1)Total cholesterol (mmol/1)LDL cholesterol (mmol/1)HDL cholesterol (mmol/1)Triglyceride (mmol/1)Plasma LPO (/xmol/1)

Unadjusted valuesBaselineAAPH

Adjusted valuesBaselineAAPH

NGT

2546.6 ± 1.9

5272

29.8 ± 1.10.90 ± 0.025.12 ±0.315.90 ± 0.526.12 ±0.254.18 ±0.231.07 ±0.171.95 ± 0.28

1.99 ± 0.074.30 ± 0.20

0.26 ± 0.020.56 ± 0.04

1GT

2547.4 ± 1.9

2040

31.9 ± 1.10.93 ± 0.025.32 ± 0.349.19 ±0.335.68 ± 0.253.66 ± 0.231.13 ±0.062.05 ± 0.28

1.88 ± 0.074.45 ± 0.20

0.26 ± 0.020.60 ± 0.02

NIDDM

2551.5 ± 1.9

4042

32.5 ± 1.10.95 ± 0.028.53 ± 0.7215.5 ± 0.515.68 ± 0.253.55 ± 0.231.02 ± 0.062.66 ± 0.28

1.97 ± 0.075.35 ± 0.20

0.26 ± 0.020.69 ± 0.02

P value

0.0720.0610.471

0.1980.142

<0.001<0.001

0.3450.1220.4230.169

0.5240.003

0.9650.008

Data are means ± SE. For adjusted values, baseline and AAPH values were corrected by mmol/1 of totalcholesterol + triglyceride. P values were calculated by ANOVA.

late strongly with high-pressure liquidchromatography methods for determina-tion of LPO levels (r = 0.85) (35). Fur-thermore, glucose addition to the oxida-tion system did not increase oxidizability.

Statistical analysisStatistical analyses were performed usingSAS statistical software. The analyses in-clude analysis of variance (ANOVA) (Ta-bles 1 and 3), Pearson correlation coeffi-cients (Tables 2), and multiple linearregression (Tables 4 and 5). Unstandard-ized regression coefficients are shown inTables 4 and 5. Because of the number ofcomparisons made in the table with Pear-son correlations (Table 2), caution shouldbe used in interpretation of significant Pvalues. Triglyceride and glucose concen-trations were log-transformed to improvenormality for statistical testing and back-transformed for presentation in tables.Since diabetic subjects may have in-creased cholesterol and triglyceride levelsrelative to subjects with NGT (4,5) andseveral studies have indicated that LPO

levels are correlated with the total amountof lipid in plasma (15,22), we computedLPO levels both per unit volume (micro-moles per liter) and per millimoles perliter of total cholesterol and triglyceride(standardized LPO). P values were calcu-lated by ANOVA.

RESULTS— Table 1 shows the clini-cal, anthropometric, and metabolic char-acteristics of the study sample by glucosetolerance status. Subjects did not differsignificantly by age, BM1, WHR, percent-age with vascular disease, cigarette smok-ing, total, LDL, or HDL cholesterol, ortriglycerides. Basal plasma LPO levelswere not significantly different by glucosetolerance category. However, LPO levelsafter incubation of plasma with AAPHwere significantly higher in diabetic sub-jects than in subjects with NGT or IGT.These differences persisted even when theLPO values were lipid-standardized.

Table 2 shows Pearson coeffi-cients reflecting the correlations betweenplasma LPO levels and clinical, anthropo-metric, and metabolic characteristics. Thebaseline unadjusted plasma LPO levelwas significantly (P < 0.05) inversely cor-related with WHR and positively withLDL cholesterol. The positive correlationof unadjusted AAPH-stimulated plasmaLPO level with female sex and fasting glu-cose and 2-h glucose concentrations washighly significant. The AAPH-stimulatedplasma LPO level was significantly posi-tively correlated with the baseline LPOlevel (r = 0.34, P < 0.01). Importantly,the AAPH-stimulated (but not baseline)LPO level remained significantly corre-

Table 2—Correlations between clinical and metabolic characteristics

AgeBMIWHRFasting glucose2-h glucoseSex (M/F)Total cholesterolLDL cholesterolHDL cholesterolTriglyceridePlasma LPO baseline

LPO

Baseline

-0.03-0.17-0.23*-0.13-0.02

0.150.170.21*0.03

-0.01—

AAPH

0.170.23*

-0.030.30T0.40?

-0.30T0.09

-0.060.200.150.34T

Lipid-standardized LPO

Baseline

-0.01-0.20-0.04-0.15-0.10-0.11

—————

AAPH

0.150.100.150.28*0.30T

-0.25*—————

The values under LPO and Plasma LPO baseline are unadjusted, and the values under the Lipid-standardizedLPO are corrected by mmol/1 of total cholesterol + triglyceride. n = 75. *P < 0.05; TP < 0.01; fP < 0.001.

648 DIABETES CARE, VOLUME 18, NUMBER 5, MAY 1995

Haffner and Associates

Table 3—ANOVA for plasma LPO levels by diabetic status and possible confoundingvariables

SmokingNIDDM

SmokerNonsmoker

NondiabeticSmokerNonsmoker

P valueSmokingNIDDM

SexNIDDM

MenWomen

NondiabeticMenWomen

P valueSexNIDDM

Vascular diseaseNIDDM

YesNo

NondiabeticYesNo

P valueVascular diseaseNIDDM

n

421

1139

1015

7832

223

248

Plasma LPO(/Am

Baseline

2.021.96

1.841.96

0.7950.472

1.912.01

1.881.96

0.3320.698

2.201.95

1.811.94

0.7580.320

ol/l)

AAPH

4.765.46

4.054.46

0.0460.042

5.035.57

3.984.60

0.013<0.001

4.805.40

3.604.40

0.9050.034

Lipid-standardizedplasma LPO

(jumol/mmol lipid)

Baseline

0.300.25

0.250.26

0.4070.607

0.260.25

0.250.27

0.5110.971

0.340.25

0.220.26

0.5640.237

AAPH

0.720.69

0.590.56

0.7900.029

0.660.71

0.520.62

0.0540.018

0.760.67

0.450.59

0.7860.040

Baseline and AAPH for plasma LPO are unadjusted values, and for Lipid-standardized plasma LPO they arecorrected by mmol/1 of total cholesterol + triglyceride.

lated with fasting and 2-h glucose con-centrations when lipid-standardized.

Since the subjects with NGT andIGT had similar levels of plasma LPO, wepooled these subjects in subsequent anal-yses to form a nondiabetic group forgreater statistical power. Table 3 showsthe level of LPOs by diabetic status strat-ified by smoking status, sex, and vasculardisease. Four subjects had vascular dis-ease, and 15 subjects smoked cigarettes.There was no significant difference in

baseline LPO levels between diabetic andnondiabetic subjects. However, diabeticsubjects had higher stimulated LPO levelsby the AAPH method than nondiabeticsubjects even after stratification for sex,smoking, or vascular disease. Since onlyfour subjects had vascular disease, thepower of the analysis is too low to allowus to conclude whether or not LPO wasrelated to the presence of vascular dis-ease.

Table 4 shows the results of mul-

tiple linear regression analyses withplasma LPO level as the dependent vari-able and diabetic status, sex, cigarettesmoking, and vascular disease as inde-pendent variables. Baseline LPO levelswere not significantly related to diabeticstatus. However, subjects with NIDDMhad significantly higher LPO values afterstimulation by AAPH than nondiabeticsubjects. Women had significantly higherstimulated LPO levels than men. The re-sults for self-reported cigarette smokingwere inconsistent, and vascular diseasedid not appear to be related to LPO leveleither at baseline or after stimulation.

Table 5 shows the results of mul-tiple linear regression analysis in the 56nonsmoking subjects who did not havevascular disease. In this analysis, again,subjects with NIDDM had significantlyhigher stimulated LPO values than non-diabetic subjects. However, there was nodifference in LPO values at baseline be-tween diabetic and nondiabetic subjects.

CONCLUSIONS— In the presentstudy, we have compared plasma LPOlevels among subjects with NIDDM, IGT,and NGT, both before and after imposi-tion of an oxidative challenge. In our sub-jects, the prevalence of vascular diseasewas low, and we were able to avoid theconfounding effects of vascular diseasewhile examining the influence of diabeteson plasma LPO levels. However, we ac-knowledge that we may have underesti-mated the prevalence of vascular diseaseby omitting tests such as stress tests forcardiac disease and ankle:arm ratios forperipheral vascular disease. We found nosignificant differences in basal LPO levelsamong the groups, a result consistentwith other studies in which the diabeticand nondiabetic groups were matched forvascular disease (17,18,20,21). Certainother earlier studies that reported in-creased LPO levels in diabetic patients(17-19) could have been confounded bythe greater occurrence of vascular diseasein the diabetic than in the control group.Indeed, Mooradian (22) found that whilediabetic patients with vascular complica-

DIABETES CARE, VOLUME 18, NUMBER 5, MAY 1995 649

Lipid peroxidation in N1DDM

Table 4—Multiple linear regression coefficients with LDL oxidation as a dependentvariable

Plasma LPOLipid-standardized

plasma LPO

Baseline AAPH Baseline AAPH

N1DDMSexSmokingVascular disease

-0.09

0.97t-0.52*-0.35*

-0.02

0.04

0.11*-0.09*

For N1DDM, smoking, and vascular disease, "yes" is 1 and "no" is 0; for Sex, "men" is 1 and "women" is 0.Baseline and AAPH for Plasma LPO are unadjusted values and for Lipid-standardized Plasma LPO they arecorrected by mmol/1 of total cholesterol + triglyceride. *P < 0.05; tP < 0.001. t, did not meet P value toenter (P = 0.50).

tions had significantly higher conjugateddiene levels than healthy control subjects,conjugated diene levels in diabetic sub-jects without complications were inter-mediate between the two groups. On theother hand, increased LPO levels havebeen reported in subjects with IDDM rel-ative to normoglycemic control subjects,regardless of whether the subjects hadvascular disease (21). The area of lipidoxidation and vascular disease has beenreviewed by Stringer et al. (21).

Several reports have indicated arelationship between fasting blood glu-cose and LPO levels. Altamore et al. (19)found higher malondialdehyde concen-trations in subjects with poorly controlledNIDDM compared with normoglycemicsubjects; however, malondialdehyde con-centrations were similar in subjects withwell-controlled diabetes and normoglyce-mic subjects. In addition, malondialde-hyde levels declined after the glycemicconcentrations of subjects with poorlycontrolled diabetes were improved. Like-wise Sato et al. (14) found higher LPOvalues in subjects with poorly controlledthan in subjects with well-controlled dia-betes, but there was no correlation be-tween the glucose tolerance test glucosevalues and LPO levels. Mooradian (22)found that serum glucose level was posi-tively correlated with both conjugateddiene (r = 0.51, P < 0.01) and serumtriglyceride (r = 0.36, P < 0.05) levels.

Thus, the previously reported associationbetween LPO and blood glucose levels indiabetic subjects may merely reflect ele-vated triglyceride levels in the subjectswith poorly controlled diabetes. The re-sults of the present investigation are con-sistent with that possibility. Although ourdata suggested associations between fast-ing glucose and baseline LPO levels (r =0.33, P = 0.10) in diabetic subjects alone,the strength of these associations declinedmarkedly after adjustment for total lipidlevels (r = 0.10, P = 0.630). More workwill be needed to clarify this point.

Few data exist on the susceptibil-ity to oxidation of plasma from diabeticand normal subjects. In contrast to thelack of association between diabetic sta-tus and basal LPO levels, we found thatplasma of NIDDM subjects was signifi-cantly more prone to lipid peroxidationafter stimulation with AAPH than plasma

of subjects with either NGT or IGT. Thegroup of subjects with IGT is likely tohave a degree of insulin resistance similarto that in the NIDDM group, but the for-mation of LPOs in IGT subjects was sim-ilar to that in normoglycemic individuals.Therefore, the increased susceptibility tooxidation in patients with NIDDM ap-pears to be more closely related to ele-vated blood glucose concentration thanto insulin resistance.

We also found that LPO levelswere, surprisingly, higher in women thanin men. There are few data on this area inthe literature. A recent paper has sug-gested that basal LPO levels are higher inwomen with IDDM than in normoglyce-mic women; however, no difference wasseen in LPO levels between normoglyce-mic men and men with IDDM (36). LPOlevels were significantly higher in womenthan in men with IDDM, but no sex dif-ferences were observed in normoglyce-mic subjects (36). The finding of higherLPO levels in women than in men woulddisagree from an ecological standpointwith oxidized LDL playing a major role inatherosclerosis since women have lowrates of CHD relative to men. However,the number of subjects in our study aswell as in the study of Evans and Orchard(36) were small. The majority of subjectsin our report had IGT or NIDDM, and60% of the women were postmenopausal.Further work needs to be done in thisarea.

One limitation of epidemiologicalstudies is that they do not reveal mecha-nisms. In the current study, we have

Table 5—Multiple linear regression for plasma LPO as a dependent variable innondiabetic nonsmoking subjects without vascular diseases

NIDDMSex

Plasma LPO

Baseline

-0.20-0.46*

AAPH

1.081-0.48

Lipid-standardizedplasma LPO

Baseline AAPH

-0.02 0.10*-0.05* -0 .13*

For plasma LPO, Baseline and AAPH are unadjusted values, and for Lipid-standardized plasma LPO they arecorrected by mmol/1 of total cholesterol + triglyceride. n = 56. *P < 0.05; fP < 0.001.

650 DIABETES CARE, VOLUME 18, NUMBER 5, MAY 1995

Haffner and Associates

shown increased susceptibility of plasmato oxidation in NIDDM patients, and thiscould be due to any of a number of bal-ancing factors that promote lipid peroxi-dation or protect against it. The balanceamong these factors should in turn deter-mine the effect of LDL oxidation onatherosclerosis. In both clinical and ex-perimental diabetes, antioxidant levels in-cluding ascorbate, glutathione, and su-peroxide dismutase in plasma and cellsdecrease, suggesting depletion of thesedefense systems through increased de-mand (37-39). Diminished resistance tolipid peroxidation has been found inerythrocytes from patients with IDDM orNIDDM, alloxan (ALX)-induced diabeticrats, and genetically diabetic rats. Insulinadministration reversed the defect inALX-induced diabetic rats (37). Further-more, the levels of glutathione peroxi-dase, superoxide dismutase, and catalasewere decreased in aortic endotheliumfrom ALX-induced diabetic rabbits, andthese deficiencies were reversed by insu-lin treatment (38). A recent study demon-strated reduced total free radical trappingcapacity in plasma and increased suscep-tibility of LDL to oxidation in patientswith poorly controlled IDDM (39); thestudy did not identify any specific defi-ciency. Impaired antioxidant defensescould underlie our findings; alternative-ly, hyperglycemia may increase thesusceptibility of LDL to oxidation (12).Glycosylation of LDL is correlated withglycosylated hemoglobin even in normo-lipidemic IDDM subjects (40). In recentyears, interest has increasingly focusedon combined glycation and oxidation(glycoxidation), which may result in in-creased oxidative modifications of pro-teins (12,13). It has recently been shownthat diabetic LDL was more glycosylatedand also more susceptible to oxidationthan nondiabetic LDL (41). While peroxi-dized LDL is rapidly removed from thecirculation via scavenger receptors, nei-ther minimally oxidized plasma lipopro-teins nor glycated lipoproteins are recog-nized by this receptor and thus remain inthe circulation (42-46). It has been sug-

gested that minimally oxidized LDL maybe increased in diabetic subjects (47). In-creased generation of oxyradicals is athird potential mechanism by which lipidperoxidation could be increased in diabe-tes. Release of superoxide radical by poly-morphonuclear leukocytes and release bymonocytes are both increased in diabeticpatients (48-50).

In conclusion, we have shownthat basal LPO levels are similar insubjects with NGT, IGT, and NIDDM.However, after stimulation by AAPH,NIDDM subjects without vascular di-sease had increased LPO levels relativeto subjects with IGT or NGT. The greaterstimulation of higher LPO levels in sub-jects with NIDDM than with IGT sug-gests that it is hyperglycemia rather thaninsulin resistance that causes increasedsusceptibility to oxidation. Further popu-lation-based studies of diabetic subjectsthat include measurement of specificantioxidants as well as enzymes in-volved in lipid peroxidation would bevaluable in advancing our understand-ing of the relationship between diabetesand increased risk of cardiovascular dis-ease.

Acknowledgments— This work was sup-ported by grants R01-HL24799 and R37-HL36820 from the National Heart, Lung andBlood Institute and American Heart Associa-tion (Texas Affiliate).

References

1. Kannel WB, McGee DL: Diabetes and car-diovascular risk factors: the FraminghamStudy. Circulation 59:8-13, 1979

2. Assmann G, Schulte H: The ProspectiveCardiovascular Munster (PROCAM)Study: prevalence of hyperlipidemia inpersons with hypertension and/or diabe-tes mellitus and the relationship to coro-nary heart disease. Am Heart] 116:1713—1724, 1988

3. Garcia MJ, McNamara PM, Gordon T,Kannel WB: Morbidity and mortality indiabetics in the Framingham population:

sixteen-year follow-up study. Diabetes 23:105-111,1974

4. Falko JM, Parr JH, Simpson RN, Wynn V:Lipoprotein analyses in varying degrees ofglucose tolerance: comparison betweennon-insulin-dependent diabetic, im-paired glucose tolerant, and control pop-ulations. AmJ Med 83:641-647, 1987

5. Laakso M, Barrett-Connor E: Asymptom-atic hyperglycemia is associated with lipidand lipoprotein changes favoring athero-sclerosis. Arteriosclerosis 9:665-672,1989

6. Witztum JL, Steinberg D: Role of oxidizedlow density lipoprotein in atherogenesis. JClin Invest 88:1785-1792, 1991

7. Yla-Herttuala S, Palinski W, RosenfeldME, Parthasarathy S, Carew TE, WitztumJ, Steinberg D: Evidence for the presenceof oxidatively modified LDL in athero-sclerotic lesions of rabbit and man. J ClinInvest 84:1086-1095, 1989

8. SalonenJT, Yla-Herttuala S, Yamamoto R,Butler S, Korpela H, Salonen R, NyssonenK, Palinski W, Witztum J: Autoantibodiesagainst oxidized low density lipoproteinand progression of carotid atherosclero-sis. Lancet 339:883-887, 1992

9. Kita T, Nagano Y, Yokode M, Ishii K,Kume N, Oashima A, Yoshida H, KawaiC: Probucol prevents the progressionof atherosclerosis in Watanabe herita-ble hyperlipidemic rabbit, on animalmodel for familial hypercholesterolemia.Proc Natl Acad Sci USA 84:5928-5931,1987

10. Stampfer MJ, Hennekens CH, MansonJE,Colditz GA, Rosner B, Willett WC: Vita-min E consumption and the risk of coro-nary disease in women. N EnglJ Med 328:1444-1449, 1993

11. Rimm EB, Stampfer MJ, Ascherio A, Gio-vannucci E, Colditz GA, Willett WC: Vi-tamin E consumption and the risk of cor-onary heart disease in men. N EnglJ Med328:1450-1456, 1993

12. Chisolm GM, Irwin KC, Penn MS: Li-poprotein oxidation and lipoprotein-induced cell injury in diabetes. Diabetes41 (Suppl. 2):61-66, 1992

13. Lyons TJ: Glycation and oxidation: a rolein the pathogenesis of atherosclerosis. AmJCardiol 7L26B-31B, 1993

14. Sato Y, Hotta N, Sakamoto N, MatsuokaS, Ohishi N, Yagi K: Lipid peroxide level

DIABETES CARE, VOLUME 18, NUMBER 5, MAY 1995 651

Lipid peroxidation in N1DDM

in plasma of diabetic patients. BiochemMed 21:104-107, 1979

15. Nishigaki I, Hagihara M, Tsunekawa H,Maseki M, Yagi K: Lipid peroxide levels ofserum lipoprotein fractions of diabe-tic patients. Biochem Med 25:373-378,1981

16. Kaji H, Kurasaki M, Ito K, Saito T, Saito K,Niioku T, Koijima Y, Ohsaki Y, Ide H,Tsuji M, Kondo T, Kawakami X: In-creased lipoperoxide value and glutathi-one peroxidase activity in blood plasma oftype 2 (non-insulin-dependent) diabeticwomen. Klin Wochenschr 63:765-768,1985

17. Jennings PE, Jones AF, Florkowski CM,Lunec J, Barnett AH: Increased dieneconjugates in diabetic subjects with mi-croangiopathy. Diabetic Med 4:452-456,1987

18. MacRury SM, Gordon D, Wilson R, Brad-ley H, Gemmell CG, Patterson JR, RumleyAG, MacCuish AC: A comparison of dif-ferent methods of assessing free radicalactivity in type 2 diabetes and peripheralvascular disease. Diabetic Med 10:331—335,1993

19. Altamore E, Vendemaille G, Chicco D,Procacci V, Cirelli F: Increased lipid per-oxide in type 2 poorly controlled diabeticpatients. Diabete Metab 18:264-271,1992

20. Velazquez E, Winocour PH, Kesteven P,Alberti KG, Laker MF: Relation of lipidperoxides to macrovascular disease intype 2 diabetes. Diabetic Med 8:752-758,1991

21. Stringer MD.GorogPG, Freeman A, Kak-kar W : Lipid peroxides and atherosclero-sis. BrMedJ 298:281-284, 1989

22. Mooradian AD: Increased serum conju-gated dienes in elderly diabetic patients. JAm Geriatr Soc 39:571-574, 1991

23. Reaven GM: Banting Lecture 1988: role ofinsulin resistance in human disease. Dia-betes 37:1595-1607, 1988

24. Eschwege E, Ducimetiere P, Papoz L,Claude JR, Richard JL: Blood glucose andcoronary heart disease. Lancet ii:472-473,1980

25. DeFronzo RA: Lilly Lecture: the triumvi-rate: /3-cell, muscle, liver: a collusion re-sponsible for NIDDM. Diabetes 37:667-687,1988

26. Stern MP, Rosenthal M, Haffner SM,Hazuda HP, Franco LJ: Sex difference inthe effects of sociocultural status on dia-betes and cardiovascular risk factors inMexican Americans: the San AntonioHeart Study. Am J Epidemiol 120:834-851,1984

27. Haffner SM, Stern MP, Hazuda HP, PughJA, Patterson JK: Hyperinsulinemia in apopulation at high risk for non-insulindependent diabetes mellitus. N EnglJ Med315:220-224, 1986

28. Hazuda HP, Comeaux PJ, Stern MP,Haffner SM, Eifler CW, Rosenthal M,Franco LJ: A comparison of three indica-tors for identifying Mexican Americans inepidemiological research: methodolog-ical findings from the San AntonioHeart Study. Am] Epidemiol 123:96-112,1986

29. Haffner SM, Hazuda HP, Mitchell BD,Patterson JK, Stern MP: Increased inci-dence of type II diabetes mellitus in Mex-ican Americans. Diabetes Care 14:102-108,1991

30. Haffner SM, Mykkanen L, Valdez R, PaidiM, Stern MP, Howard BV: LDL sizeand subclass pattern in a biethnic popu-lation. Arterioscler Thromb 13:1623-1630,1993

31. Rose GA, Blackburn H, Gillum RF, Prin-eas RF: Cardiovascular Survey Methods.

2nd ed. Geneva, World Health Org.,1982,p.l25-143

32. World Health Organization: Diabetes Mel-litus: Report of a WHO Study Group. Ge-

neva, World Health Org., 1985 (Tech.Rep. Ser.,no. 727)

33. Frei BL, England A, Ames BN: Ascorbateis an antioxidant in human blood plasma.Proc Natl Acad Sci USA 86:6377-6381,1989

34. Yagi K: Lipid peroxides and human dis-eases. Chem Phys Lipids 45:337-351,1987

35. Therasse J, Lemonnier F: Determinationof plasma lipopoxides by high-perfor-mance liquid chromatography. J Chro-matogr 413:237-241, 1987

36. Evans RW, Orchard TJ: Oxidized lipids ininsulin dependent diabetes mellitus: asex-diabetes interaction? Metabolism 43:1196-1200, 1994

37. Gupta BL, Ansari MA, Singh JN, Bacquer

NZ: Effect of insulin and thyroxine oncatalase, glutathione-S-transferase, GSHand GSSG in alloxan diabetic rat red cells.Biochem Int 27:793-802, 1992

38. Tagami S, Kando T, Yoshida K, HirokawaJ, Ohstuda Y, Kawakami Y: Effect of in-sulin on impaired antioxidant activitiesin aortic endothelial cells from diabe-tic rabbits. Metabolism 41:1053-1058,1992

39. Tsai EC, Hirsch IB, Brunzell JD, ChaitA: Reduced plasma peroxyl radical trap-ping capacity and increased suscepti-bility of LDL to oxidation in poorly con-trolled IDDM. Diabetes 43:1010-1014,1994

40. Lyons TJ, Baynes JW, Patrick JS, ColwellJA, Lopes-Virella MF: Glycation of lowdensity lipoprotein in patients with type I(insulin-dependent) diabetes: correla-tions with other parameters of glyce-mic control. Diabetologia 29:685-689,1986

41. Bowie A, Owens D, Collins P, JohnsonA, Tomkin GH: Glycosylated low den-sity lipoprotein is more sensitive tooxidation: implications for the diabe-tic patients? Atherosclerosis 102:63-67,1993

42. Liao F, Berliner JA, Mehrabian M, NavabM, Demer LL, Lusis AJ, Fogelman AM:Minimally modified low density lipo-protein is biologically active in vivoin mice. J Clin Invest 87:2253-2257,1991

43. Steinbrecher UP, Witztum JL, Partha-sarathy S, Steinberg D: Decrease in reac-tive amino groups during oxidation or en-dothelial cell modification of LDL:correlation with changes in receptor-mediated catabolism. Arterioscler Thromb7:135-143, 1987

44. NagelkerkeJF, Havekes L, Van HinsberghVWM, Van Berkel TJC: In vivo catabolismof biologically modified LDL. ArteriosclerThromb 4:256-264, 1984

45. Curtiss LK, Witztum JL: Plasma apoli-poproteins AI, All, B, CI, and E are glyco-sylated in hyperglycemic diabetic sub-jects. Diabetes 34:453-461, 1985

46. Berliner JA, Territo M, Navab M, AndalibiA, Parhami F, Liao F, Kim J, Estworthy S,Lusis AJ, Fogelman AM: Minimally mod-ified lipoproteins in diabetes. Diabetes 41

652 DIABETES CARE, VOLUME 18, NUMBER 5, MAY 1995

Haffner and Associates

(Suppl. 2):74-76, 199247. Morel DW, Chisolm GM: Antioxidant

treatment of diabetic rats inhibits lipopro-tein oxidation and cytotoxicity. J Lipid Res30:1827-1834,1989

48. Freedman SF, Hatchell DL: Enhanced su-peroxide radical production by stimu-

lated polymorphonuclear leukocytes in acat model of diabetes. Exp Eye Res 55:767-773, 1992

49. Nath N, Chari SN, Rathi AB: Superoxidedismutase in diabetic polymorphonu-clear leukocytes. Diabetes 33:586-589,1984

50. Pronai L, Hiramatsu K, Saigusa Y, Naka-zawa H: Low superoxide scavengingactivity associated with enhanced super-oxide generation by monocytes fromhypertriglyceridemia with and with-out diabetes. Atherosclerosis 90:39-47,1991

DIABETES CARE, VOLUME 18, NUMBER 5, MAY 1995 653