Abnormal metabolism of free fatty acids in hypertriglyceridaemic men: apparent insulin resistance of adipose tissue

Journal of Internal Medicine 1998; 243: 265–274

© 1998 Blackwell Science Ltd 265

Abnormal metabolism of free fatty acids inhypertriglyceridaemic men: apparent insulin resistance ofadipose tissue

J. M. MOSTAZA, G. L . VEGA, P. SNELL & S. M. GRUNDYFrom The Center for Human Nutrition, and the Departments of Clinical Nutrition and Internal Medicine at the University of Texas South WesternMedical Center and the Veterans Affairs Medical Center at Dallas, Texas, USA

Abstract. Mostaza JM, Vega GL, Snell P, Grundy SM(The Center for Human Nutrition, and theDepartments of Clinical Nutrition and InternalMedicine, University of Texas South Western MedicalCenter, and the Veterans Affairs Medical Center atDallas, Texas, USA). Abnormal metabolism of freefatty acids in hypertriglyceridaemic men: apparentinsulin resistance of adipose tissue. J InternMed 1998; 243: 265–74.

Objective. There is growing evidence that endoge-nous hypertriglyceridaemia is frequently accompa-nied by a state of insulin resistance. The presentstudy was performed to determine whether patientswith primary endogenous hypertriglyceridaemiacommonly have abnormalities in plasma concentra-tions and turnover rates of free fatty acids (FFA),which could reflect a state of insulin resistance inadipose tissue and could account for raised plasmatriglycerides.Design. Hypertriglyceridaemic and normotriglyc-eridemic control patients underwent measurementsof plasma concentrations and turnover rates of FFA.Fat weights in both groups were determined byhydrodensitometry, and fat distribution was assessedby skin-folds and measurement of waist and hip cir-cumferences. Other measurements included plasmaglucose, insulin, lipids, and lipoproteins.

Subjects. Fifteen men with normal plasma triglyc-erides and 21 men with primary endogenous hyper-triglyceridaemia were studied. Men in both groupsvaried in body weights and total fat weights, but totalfat weights were entirely overlapping for the twogroups. Waist-to-hip ratios and waist circumferencesalso were similar for the two groups.Results. For any total body fat content or waist cir-cumference, most hypertriglyceridaemia patients hadhigher mean plasma concentrations of FFA andhigher turnover rates (flux) for FFA than did nor-motriglyceridemic patients. Hypertriglyceridaemicpatients also had higher fasting insulin concentra-tions for a given body fat content. In general, bothFFA flux and plasma insulin levels were positivelycorrelated with plasma concentrations of triglycerideand inversely with high density lipoprotein (HDL)cholesterol.Conclusions. These studies indicate that manypatients with primary endogenous hypertriglyceri-daemia have increased flux of FFA and hyperinsu-linemia that cannot be explained either by increasedtotal body fat content or by greater waist circumfer-ences than observed in normotriglyceridemicpatients.

Keywords: fat weight, fatty acids, HDL-cholesterol,insulin resistance, triglycerides, VLDL-apo B.

Introduction

The origins of primary endogenous hypertriglyceri-daemia are not well understood. Some investigators[1, 2] report that the predominant abnormality liesin hepatic overproduction of very low densitylipoprotein-triglyceride (VLDL-TG). Others [3–5] pos-

tulate that most cases result from a defect in lipolysisof VLDL-TG. Still others [6, 4, 5] suggest that mostinstances of endogenous hypertriglyceridaemia aredue to a dual defect in triglyceride metabolism, i.e.overproduction of VLDL-TG combined with defectivelipolysis of VLDL-TG.

JINT298F

J. M. MOSTAZA et al.266

© 1998 Blackwell Science Ltd Journal of Internal Medicine 243: 265–274

The most common cause of hepatic overproduc-tion of VLDL-TG almost certainly is obesity. Obeseindividuals typically have higher secretion rates ofVLDL-TG than do normal-weight persons [5, 6].Presumably, a high level of free fatty acids (FFA),which is characteristic of the obese state [7, 8], pro-vides the substrate for excessive synthesis of VLDL-TG. In addition, Reaven et al [9]. and Tobey et al [10].have postulated that production rates of VLDL-TGare accentuated by the presence of hyperinsuline-mia. The latter is a typical manifestation of theinsulin resistance state, which is common in obesepersons. Thus, high FFA levels combined with insulinresistance and hyperinsulinemia probably accountfor the overproduction of VLDL-TG observed in mostobese patients.

Defective lipolysis of VLDL-TG also has beenreported to contribute to endogenous hypertriglyceri-daemia [3, 11, 12]. A large number of mutations inlipoprotein lipase (LPL) have been identified [13], andthese can cause hypertriglyceridaemia. However, it isdoubtful that most cases of endogenous hypertriglyc-eridaemia can be explained by mutations in LPL,which appear to be relatively rare. Another attractivemechanism for defective lipolysis of VLDL-TG hasrecently been proposed. This is an overproduction ofapolipoprotein (apo) CIII. This apolipoprotein inhibitsthe function of LPL [14]. The hepatic synthesis ofapo CIII apparently is insulin responsive [15]; in thepresence of insulin resistance, apo CIII may be over-expressed, which can inhibit lipolysis of plasmaVLDL-TG and lead to hypertriglyceridaemia.

These considerations thus raise the question ofwhether primary endogenous hypertriglyceridaemiais a common manifestation of the insulin resistancestate. Previous reports [9, 10] indeed have noted arelation between glucose tolerance, insulin resis-tance, and plasma triglyceride levels. This associationis further addressed in the present investigation by afocus on FFA metabolism in patients with primaryendogenous hypertriglyceridaemia; to date, severalreports [16–19] have attempted to describe thequantitative link between plasma FFA metabolismand endogenous hypertriglyceridaemia. The currentstudy was designed to compare FFA metabolism inpatients with primary endogenous hypertriglyceri-daemia with those having normal plasma lipids,including normal triglyceride levels. Subjects in thetwo groups were chosen to vary in body mass index(BMI) and the fraction of total body weight as fat.

The primary question being addressed was whetherfor any given weight of body fat, patients with prima-ry endogenous hypertriglyceridaemia have higherturnover rates and plasma concentrations of FFAthan individuals with normotriglyceridemia. Such afinding would imply that defective FFA metabolismcan underlie endogenous hypertriglyceridaemia. Itwas further asked whether fasting insulin levels arehigher in hypertriglyceridaemic patients for a givenamount of total body fat and type of body fat distrib-ution. If so, this too would suggest that some portionof hypertriglyceridaemic patients have primaryinsulin resistance.

Methods

Patients

Thirty-six middle-aged men were recruited for studyfrom the outpatient lipid clinic at the Veterans AffairsMedical Center of Dallas. Fifteen men had plasmatriglyceride levels less than 2.25 mmol L21

(, 200 mg dL21), and the remaining 21 had triglyc-eride levels in the range of 2.25 mmol L21 to5.65 mmol L21 (200–500 mg dL21). Inclusion criteriawere LDL-cholesterol levels less than 4.14 mmol L21

(160 mg dL21), absence of diabetic response to a glu-cose tolerance test, and a stable weight within amonth prior to recruitment for study. Amongst theparticipants, 16 patients had history of smoking, butwere not smoking at the time of recruitment to thestudy; 12 had hypertension controlled by drugs, butnot with beta-adrenergic blocking agents (beta-blockers), and nine patients had coronary heart dis-ease (CHD). Patients were excluded from the study ifthey had unstable CHD, congestive heart failure,endocrine, renal or liver disorders, or any surgicalintervention 6 months before recruitment into thestudy. Subjects also were excluded if they were takinghypolipidemic or hypoglycemic drugs, beta-blockers,or had an alcohol intake greater than 20 g per day.The study protocol was approved by the InstitutionalReview Board of the Veterans Affairs Medical Centerat Dallas.

Clinical characteristics of the patients participat-ing in the study are presented in Table 1. Body massindexes for the whole group ranged from 17.9 to36.5 kg per square meter. Between the two groupsthere were no significant differences in mean age,body mass index, total fat weight, waist circumfer-

FREE FATTY ACID FLUX AND BODY FAT 267


ences, or waist-to-hip ratios. Truncal skin-fold/peripheral skin-fold ratios were slightly higher inthe normotriglyceridemic group. Both plasma choles-terol and triglyceride levels were higher in the hyper-triglyceridaemic group.

Study design

After recruitment for the study, patients were admit-ted to the metabolic unit for four consecutive days.They were maintained on an ad libitum diet through-out the study. During each of the four days of hospi-talization, patients had measurements of levels ofplasma lipids, lipoprotein cholesterol, andapolipoprotein B. They also had anthropometry andmeasurement of turnover rates of FFA, and plasmalevels of FFA, glucose, and insulin.

Anthropometry

Body density was measured by hydrodensitometryafter 12 h of fasting. Simultaneous measurement ofresidual lung volume was made using a nitrogendilution method [20]. The body density was estimat-ed according to Goldman and Burskik [21], and thepercentage body fat was estimated according to Siri[22]. These measurements were performed at theSport’s Center located at the Saint Paul’s MedicalCenter, Dallas, Texas. Waist circumference was mea-sured at the midpoint between the rib cage and theiliac crest (23). The ratio of waist-to-hip circumfer-ence was calculated.

Chemical measurements

Blood samples were drawn into tubes containing

1 mg mL21 of disodium ethylene diamine tetraacetate(EDTA). Plasma was separated immediately afterblood collection by centrifugation at 3000 r.p.m. at4 °C. Analyses were carried out as previouslydescribed in our laboratory [23]. Total cholesteroland triglycerides were determined in the plasmaenzymatically. HDL cholesterol was measured in theplasma supernatant after precipitation of apo B-con-taining lipoproteins with phosphotungstic acid. Forsamples with plasma triglyceride < 3.39 mmoles L21

(, 300 mg dL21), the HDL-cholesterol levels weremeasured in the plasma 1.0063 g mL21 infranatantobtained by ultracentrifugation. Separately, levels oflipoprotein cholesterol were measured as follows. Analiquot of plasma was first adjusted to a density of1.019 g mL21 with salt mixture of sodium chlorideand sodium bromide. Very low density lipoproteins(VLDL) and intermediate density lipoproteins (IDL)were then isolated together by ultracentrifugation.Total cholesterol was measured in the supernatant(density less than 1.019 g mL21) and infranatant(density greater than 1.019 g mL21). The averagerecoveries ranged from 98% to 102%. LDL-choles-terol was estimated as the difference between totalcholesterol and the sum of VLDL 1 IDL 1 HDL cho-lesterol. Apolipoprotein B (apo B) was chemicallydetermined in VLDL 1 IDL and in LDL as describedpreviously [23].

Turnover rates of free fatty acids

Turnover rates of plasma FFA were estimated after a12-h fast using a modification of the constant infu-sion method of Havel et al [24]. Briefly, an indwellingcatheter was placed in a hand vein and was keptpatent by a slow infusion of 0.45% NaCl(0.5 mL min21). The hand was kept in a constanttemperature box at 70 °C to arterialize blood sam-ples. A second catheter was placed in the contralater-al forearm and kept patent by another 0.45% NaClinfusion (0.5 mL min21). Two hours later, an infusionof 9,10 3H-palmitate (Amersham Life Sciences,Arlington Heights, Ill.) in 5% human serum albuminwas started. Isotope was infused at a rate of 0.5 mCiper min over a period of 70 min. The purity of thetracer was confirmed by thin layer chromatographyand was . 98%. Ten ml of arterialized blood weredrawn at 2 15, 0, 40, 50, 60, and 70 min prior toand during the infusion. Blood was collected intochilled tubes containing EDTA (1 mg mL21). Plasma

Table 1 Clinical Characteristics of Subjects

Normo- Hyper-triglyceridemia triglyceridaemia

(Mean 6 SD)Age ( years) 57 6 9 153 6 11BMI (kg/m2) 27.3 6 5.8 128.9 6 3.8Fat weight (kg) 25.9 6 13.1 128.9 6 7.9Waist circumference (cm) 99.9 6 17.3 104.7 6 10.2Waist-to-hip ratio 10.99 6 0.09 111.02 6 0.05Truncal skin-folds/peripheral skin-folds ratio 12.78 6 0.64 112.43 6 0.39†Total cholesterol (mmol L21) 14.76 6 0.67 115.95 6 0.96§Triglyceride (mmol L21) 11.46 6 0.55 115.75 6 2.05§

† Significantly lower from normotriglyceridemic group;P 5 0.049. § Significantly higher from normotriglyceridemicgroup; P 5 , 0.0001.



was separated after blood collection, and analyzedimmediately for levels of free fatty acids and determi-nation of specific activity.

FFA concentrations were determined enzymatical-ly [25]. Mean concentrations represented the aver-age of 3 samples (2 30, 2 15, and 0 min prior to theinfusion). Specific activity of FFA was determinedusing plasma obtained at 40, 50, 60 and 70 min dur-ing the infusion, as follows. A two mL aliquot of plas-ma was taken from each of the samples collectedduring the turnover. 1-C14 palmitate (New EnglandNuclear, Boston, Mass.) was added to each sample asan internal standard to monitor efficiency of freefatty acid extraction for counting. Lipids wereextracted with an organic solvent mixture consistingof isopropyl alcohol: heptane: sulfuric acid (50: 20:1; v:v:v) [24]. This was followed by addition of 6 mLof heptane and 4 mL of distilled water with vigorousmixing. After separation of organic phases, 4 mL ofthe top layer were aliquoted and 0.5 mL of 0.2 M

NaOH was added, and mixed. The mixture was sepa-rated into two phases by centrifugation; the top layerwas discarded and the bottom layer was counted. FFArecoveries after this extraction procedure rangedbetween 85% to 90%. The mean coefficient of varia-tion for FFA specific activity for each subject was7.1%. Rates of FFA flux were estimated as the ratio ofinfusion rate and specific activity.

Plasma glucose and insulin

Plasma glucose was measured using the glucose oxi-dase method, and insulin levels were measured usinga radio- immunoassay. Insulin radioimmunoassayswere performed with a commercial kit (INCSTARCorp., Stillwater, MN). Insulin concentrations weremeasured in the fasting state at the beginning of theFFA turnover study.

Statistical methods

Results are reported as mean 6 SD. Pearson’s coeffi-cients were used to test relationships between vari-ables. Linear regression analysis were carried outusing the least square method. Unless otherwise stat-ed all correlations represent linear regressions. TheStudent-Newman-Keuls test was used for compari-son of means, followed by Dunnett test. Statisticalsignificance was defined as a probability value, 0.05.

Results

Subjects with a wide range of fat weights wererecruited (Table 1). There were no significant differ-ences between BMIs, fat weights, waist circumfer-ences, or waist-to-hip ratios for the two groups.Ratios of truncal-to-peripheral skin folds were slight-ly higher in the normotriglyceridemic group.Figure 1 plots waist-to-hip ratios and waist circum-ference vs. total fat weight for the two groups. Forboth groups, as fat weight increased, the waist-to-hipratios and waist circumferences rose in proportion toincreasing fat weight. However, there was not a dis-proportionate increase in either measurement inhypertriglyceridaemic patients; of note, hypertriglyc-eridaemic patients showed no tendency towardsunusually high waist-to-hip ratios or waist circum-ferences.

1.20

0.700

Fat weight (kg)

Wai

st/h

ip (

ratio

)

1.00

0.80

1.10

0.90

20 40 6010 30 50

r 2 = 0.63P = 0.0004

a

140

600

Wai

st c

ircum

fere

nce

(cm

)

100

120

80

20 40 6010 30 50

r 2 = 0.880P < 0.0001

b

Fig. 1 Fat weight vs. waist-to-hip ratios (a) and waistcircumferences (b). Closed circles 5 normotriglyceridemic group,and open circles 5 hypertriglyceridaemic group. Best linear fitsquared correlation coefficient (r2) are shown fornormotriglyceridemic patients.



Compared to normotriglyceridemic patients,hypertriglyceridaemic patients had significantlyhigher plasma concentrations of FFA, total FFA flux,FFA flux normalized per kg fat weight, FFA flux nor-malized to kg lean body mass, and FFA flux normal-ized per cm waist circumference (Table 2). The FFAflux (mmole/min) in both groups was positively andsignificantly correlated with total fat weight andweight circumference (Fig. 2). For the normotriglyc-eridemic group, r2 was 0.772 for fat weight vs. FFAflux (P , 0.0001); and for the hypertriglyceridaemicgroups, r2 = 0.531 (P , 0.0002) for the same com-parison. The FFA flux per kg fat weight was signifi-cantly higher in the hypertriglyceridaemic group(Table 2); however, the difference although statisti-cally significant, appeared to be relatively small. Thisdifference however, fails to reflect the true differencebetween the two groups. Hypertriglyceridaemic

Table 2 Plasma Concentrations and Flux Rates of Free Fatty Acids

Normo- Hyper-triglyceridemia triglyceridemia

(Mean 6 SD)Plasma free fatty acids conc. (µmoles/min) 354 6 116 477 6 155*Plasma free fatty acid flux (mmoles/min) 265 6 91 376 6 119†Plasma free fatty acid flux (mmole/min/kg fat weight) 111.19 6 2.56 113.12 6 2.88‡Plasma free fatty acid flux (mmole/min/kg lean body mass) 114.44 + 1.25 115.99 1 1.5 ¶Plasma free fatty acid flux (mµmol/min/cm 112.59 6 0.60 113.56 6 0.93§waist circumference)

* Significantly higher than normotriglyceridemic group;P 5 0.014. † Significantly higher than normotriglyceridemicgroup; P 5 0.005. ‡ Significantly higher thannormotriglyceridemic group; P 5 0.046. ¶ Significantly higherthan normotriglyceridemic group; P 5 0.003. § Significantlyhigher than normotriglyceridemic group; P 5 0.001.

700

00

Free

fatty

aci

d flu

x(m

mol

min

Ð1)

400

200

500

300

20 40 6010 30 50

r 2 = 0.797P < 0.0001

a

600

100

Fat weight (kg)

700

060

Free

fatty

aci

d flu

x(m

mol

min

Ð1)

400

200

500

300

100 14080 120

r 2 = 0.772P < 0.0001

b

600

100

Waist circumference (cm)

Fig. 2 Free fatty acid flux vs. fat weight (a) and waistcircumference (b). Closed circles 5 normotriglyceridemic group,and open circles = hypertriglyceridaemic group. Best linear fitsquared correlation coefficient (r2) are shown fornormotriglycidermic patients.

20.00

5.000

10.00

50 6020 30

r 2 = 0.614P < 0.0005

a

15.00

FFA

flux

(mm

ol m

inÐ1

kgÐ1

fat w

eigh

t)

4010

20.00

5.000

10.00

50 6020 30

r 2 = 0.016P < 0.584

b

15.00

FFA

flux

(mm

ol m

inÐ1

kgÐ1

fat w

eigh

t)

4010Fat weight (kg)

Fig. 3 Free fatty acid flux per kg fat weight vs. fat weight fornormotriglyceridemic subject (Panel a; closed circles) andhypertriglyceridaemic patients (Panel b; open circles). Best linearfit squared correlation coefficient (r2) are shown for each group.



patients as a group had higher absolute FFA flux,particularly at higher fat weights (Fig. 2a). When thechange in FFA flux per kg change in fat weight wasestimated from the slopes of the best linear fits of thetwo groups, hypertriglyceridaemic patients hadmuch higher values than normotriglyceridemicpatients (10.8 mmoles/min vs. 6.2 mmoles/min,respectively). Also, the increment in FFA flux per cmwaist circumference was substantially higher inhypertriglyceridaemic patients (Fig. 2b). Finally, forthe normotriglyceridemic group, normalization ofFFA flux per kg fat weight revealed a significantinverse relation with increasing fat weight (Fig. 3a)and with increasing waist circumference (Fig. 4a);these findings indicate that progressive increments inbody fat resulted in declining increments of FFAflux. A similar inverse relationship was not observedin hypertriglyceridaemic patients (Figs 3b, 4b); theselatter data reveal that increments in FFA release per

20

0

60

5

10

120 14080 100

r 2 = 0.515P < 0.0012

a

15

20

560

Free

fatty

aci

d flu

x (m

mol

min

Ð1 k

gÐ1 of

fat w

eigh

t)

10

120 14080 100

r 2 = 0.004P = 0.570

b

15

Waist circumference (cm)

Fig. 4 Free fatty acid flux per kg fat weight vs. waist circumferencefor normotriglyceridemic patients (Panel a; closed circles) andhypertriglyceridaemic patients (Panel b; open circles). Best linearfit squared correlation coefficient (r2) are shown for each group.

50

00

Insu

lin (

IU m

LÐ1)

30

10

40

20

20 40 6010 30

r 2 = 0.542P = 0.0018

50Fat weight (kg)

Fig. 5 Fasting concentration of plasma insulin vs. total body fatweight. Closed circles are the normotriglyceridemic group, andopen circles are the hypertriglyceridaemic group. Best linear fitsquared correlation coefficient (r2) are shown for each group.

Table 3 Levels of Plasma Glucose and Insulin

Normo- Hyper-triglyceridemia triglyceridemia

Fasting Glucose (mg/dL) 11194 6 11 11197 6 14Glucose tolerance test(area under the curve) 17976 6 3610 17664 6 3258Fasting Insulin (mUnits mL21) 11113.7 6 6.4 11124.2 6 11.0†

† Significantly higher than normotriglyceridemic group;P 5 0.002.

Table 4 Univariate Correlation between Free Fatty Acid (FFA) Flux and Insulin Concentrations (Conc.)

Normotriglyceridemia Hypertriglyceridaemia

Parameter r2 p r2 pPlasma Triglycerides vs. FFA flux 0.188 0.060 0.136 0.099vs. Insulin Conc. 0.220 0.075 0.279 0.017VLDL 1 IDL-apo Bvs. FFA flux 0.277 0.073 0.161 0.040vs. Insulin Conc. 0.279 0.043 0.173 0.068HDL cholesterol vs. FFA flux 0.237 0.038 0.014 0.271vs. Insulin conc. 0.255 0.054 0.360 0.005



kg fat were not diminished with increasing amountsof body fat.

Parameters of plasma glucose and insulin for thetwo groups are given in Table 3. There were no differ-ences between the two groups for fasting serum glu-cose or the area under the curve during the glucosetolerance test. In contrast plasma insulin levels in thehypertriglyceridaemic patients averaged twice thoseof normotriglyceridemic patients. The latter differ-ence is again reflected by a plot of fasting insulinlevels vs. total fat weight (Fig. 5). In normotriglyceri-demic patients, fasting insulin levels rose progressive-ly with increasing weight in a highly correlatedmanner; this correlation was lost in hypertriglyceri-daemic patients, and on the whole, insulin levelswere significantly higher in the latter group.

Table 4 examines the relationships between FFAflux (and fasting insulin concentrations) and threeplasma lipid parameters (triglycerides, VLDL 1 IDL-apo B, and HDL-cholesterol). For both FFA flux andinsulin levels, there were strong trends for positivecorrelations for triglyceride and VLDL 1 IDL apo B,and negative trends for correlations with HDL-cho-lesterol concentrations.

Discussion

The primary aim of this study was to determinewhether patients with primary endogenous hyper-triglyceridaemia commonly have abnormally highconcentrations and abnormally high turnover ratesof FFA. If so, such a finding would support the con-cept that endogenous hypertriglyceridaemia is linkedto abnormalities in FFA metabolism and to a state ofinsulin resistance. Previous studies [9, 10] have sug-gested that hypertriglyceridaemia is frequently asso-ciated with insulin resistance, although severalstudies have examined changes in FFA metabolism[16–19]. These latter studies [16–19] are suggestiveof abnormalities in FFA metabolism in hypertriglyc-eridaemic patients. The relations between FFA,insulin sensitivity, and hypertriglyceridaemia arecomplicated by the fact that many patients who haveelevated serum triglycerides are overweight. Obesityitself is known to be accompanied by raised plasmaFFA concentrations [7, 8], by increased FFA flux [8,26], and by hyperinsulinemia [27, 28]. The majorityof obese patients however, do not manifest categori-cal hypertriglyceridaemia [29]; therefore, the ques-

tion of interest is whether development of definitehypertriglyceridaemia is the result of abnormalitiesin metabolism of FFA and insulin that exceed thoseproduced by obesity alone. To examine this question,it was necessary to determine parameters of FFAmetabolism in normolipidemic and hypertriglyceri-daemic individuals at comparable degrees of obesity.To set the stage for a discussion of changes in FFAmetabolism in hypertriglyceridaemic patients, it maybe useful to first summarize the results in currentpatients with normal plasma triglycerides.

In the normotriglyceridemic patients, increasingamounts of body fat were accompanied by progres-sively higher concentrations and flux rates of FFA.This finding accords with previous reports [7, 8].Increments of turnover rates of FFA however, did notlinearly parallel increases in fat weight. As fatweights increased, there was a decline in incrementsof turnover rates of FFA normalized per kg of totalbody fat (Fig. 3). This decline has been observed pre-viously [30, 31]. Several factors may contribute todiminishing increments in flux of FFA at higher fatweights. For example, as body fat increases, moretriglyceride is packed into oversized adipocytes [32];consequently, less of the stored triglyceride may beavailable for hydrolysis than in less obese persons. Inaddition, a rise in insulin levels accompanyingincreasing obesity may dampen FFA release [31].And finally, the blood flow through a given amountof adipose tissue declines with increasing adiposity[33]; this too could limit the amount of FFA mobi-lized from adipose tissue.

Nonetheless, in normolipidemic patients, secretionrates of FFA rise progressively with increasing bodyfat in spite of a rise in insulin concentrations.Although incremental secretion of FFA may beblunted somewhat by hyperinsulinemia, FFA flux isnot suppressed to normal by high insulin concentra-tions. A progressively higher flux of FFA in the pres-ence of hyperinsulinemia suggests that adipose tissuein obese persons is insulin resistant. Thus, whateverthe mechanism, some degree of resistance to insulinaction as it affects the release of FFA inevitablyresults from increasing adiposity. Rising insulin levelswith increasing obesity may blunt the increments inFFA release [31], but do not eliminate them altogeth-er.

In normotriglyceridemic patients, the increasingsecretion of FFA that resulted from a higher contentof body fat was still associated with trends towards



higher plasma triglyceride levels (Table 4).Nevertheless, in none of the patients were triglyc-eride levels categorically increased. As indicatedbefore, the majority of overweight individuals in thegeneral population do not develop categorical hyper-triglyceridaemia [29], even though they have over-production of VLDL-TG [1, 2]; therefore additionalmetabolic abnormalities beyond the overproductionof VLDL-TG apparently are required for definitehypertriglyceridaemia in obese persons. The currentdata suggest that one such abnormality is anincreased flux of FFA that is out of proportion to thedegree of obesity (Fig. 2). In other words, the adiposetissue of many hypertriglyceridaemic patients seemsto be unusually labile and prone to releasing excessFFA.

One cause of an abnormally high release of FFA inhypertriglyceridaemic patients could be a shift in fatdistribution towards truncal obesity. Previous reports[34, 35] indicate that predominant truncal obesity isaccompanied by increased levels of plasma triglyc-eride. Truncal fat has been reported to be more sus-ceptible to releasing FFA than is peripheral adiposetissue [8]. Although the fat distribution in our cur-rent patients was not measured with highly quanti-tative techniques, such as magnetic resonanceimaging [36], the available measurements gave noindication of a difference in fat distribution betweenthe groups. For example, waist-to-hip ratios andwaist circumferences at any given body fat weightwere similar for the two groups (Fig. 1). Anotherindicator of fat distribution is the truncal-to-periph-eral skinfolds ratio [37]. This ratio was not higher inthe hypertriglyceridaemic group (Table 1). Thus, dif-ferences in body fat distribution between the twogroups apparently cannot account for their differ-ences in FFA turnover rates. Moreover, FFA flux inhypertriglyceridaemic patients per cm of waist cir-cumference was significantly higher than in nor-motriglyceridemic patients; this finding implies thatfor a given degree of abdominal obesity, turnoverrates of FFA were greater in hypertriglyceridaemicpatients.

A theoretical mechanism for increased FFA flux inhypertriglyceridaemic patients could be an increasedrate of lipolysis of plasma VLDL-TG. However, the fol-lowing calculation, based on our previous studies[4], shows that this pathway is not large enough toaccount for the observed greater FFA flux in ourhypertriglyceridaemic patients. According to our pre-

vious investigation [4], 27 normotriglyceridemicpatients (VLDL-TG concentration 5 1.55 6 0.41µmoles L21) had a flux rate of VLDL-TG fatty acid(VLDL-TGFA) of 52 6 22 mmoles/min. In contrast,12 patients with hypertriglyceridaemia (VLDL-TGconcentration = 4.82 6 1.48 µmoles L21) had aVLDL-TGFA flux of 88 6 50 mmoles/min. The differ-ence, 36 mmoles/min, could explain only about athird of the observed difference in plasma FFA fluxbetween the two groups, i.e. 111 mmoles/min.Moreover, previous studies have shown that morethan half of FFA released during lipolysis of TG-richlipoproteins in the post-absorptive state is taken updirectly and esterified by adipose tissue [38]; thus lessthan half is released back into the circulation tobecome plasma FFA. Therefore, it is doubtful that thismechanism can account for more than one-sixth ofthe higher flux of plasma FFA observed in hyper-triglyceridaemic patients.

Our hypertriglyceridaemic patients usually hadelevated insulin concentrations as well as high FFAlevels (Table 3); plasma insulin levels likewise wereelevated out of proportion to the degree of over-weight (Fig. 5). An impaired action of insulin inhypertriglyceridaemic patients has been reported pre-viously [9, 10]. The association between high FFAflux and hyperinsulinemia could have either of twoexplanations. First, if there is a specific defect in theregulation of triglyceride lipolysis in the adipose tis-sue, the resulting increase in FFA flux could suppressglucose uptake in muscle, and hence cause insulinresistance. Alternatively, there could be a more gen-eralized defect in insulin action such that high plas-ma insulin levels fail to suppress release of FFA fromadipose tissue. Thus, multiple mechanisms underly-ing the insulin resistance state and associated highflux rates of FFA are possible, and future studies atthe molecular level will be required to distinguishamongst them.

In summary, the results of this investigation areconsistent with the concept that an increased flux ofFFA, out of proportion to the degree of obesity, con-tributes to primary endogenous hypertriglyceri-daemia. Our data support previous reports thatendogenous hypertriglyceridaemia is a commonmanifestation of the insulin resistance state. It more-over extends previous work by confirming that FFAmetabolism is frequently abnormal in patients withprimary hypertriglyceridaemia. It further suggeststhat this abnormal metabolism of FFA can occur



independently of body fat distribution, although pre-dominant truncal (or visceral) obesity could be acause of hypertriglyceridaemia in some patients. Anincrease in FFA flux in hypertriglyceridaemic patientsfurther is accompanied by high insulin levels, sug-gesting a concomitant state of insulin resistance.Whether increased insulin resistance is due to a spe-cific defect in adipose tissue or results from a general-ized defect in insulin action is uncertain. Regardless,this study supports the concept that primary endoge-nous hypertriglyceridaemia is commonly the result ofa generalized metabolic defect in FFA and insulinmetabolism.

Acknowledgments

This work was supported by the Department ofVeterans Affairs grants HL-29252 and HL-22682;National Institute of Health grant MO-IRR00663,Bethesda, Md.; unrestricted grants from Bristol-MyerSquibb, New Brunswick, NJ, and Merck, Rahway, NJ;the South Western Medical Foundation, Dallas, TX;and the Moss Heart Foundation, Dallas, TX. and agrant from Fondo de Investigaciones Sanitarias de LaSeguridad Social no. 92/5614, Madrid, Spain. Theauthors appreciate the excellent technical assistanceof Biman Pramanik, Han T. Nguyen, Han P. Tran,and Long Nguyen. The assistance of Kathleen Gray,R.N., Terri Shumway, R.N., Regina Strowd, R.N., andthe clinical staff of the Metabolic Unit at the VeteransAffairs Medical Center also is appreciated. BeverlyHuet-Adam, M.S., Programme Analyst of theGeneral Clinical Research Center, assisted in the datamanagement and analysis (General Clinical ResearchCenter NIH grant M01 RR00633).

References1 Nikkila EA, Kekki M. Polymorphism of plasma triglyceride

kinetics in normal human adult patients. Acta Med Scand1971; 190: 49–59.

2 Chait A, Albers JJ, Brunzell JD. Very low density lipoproteinoverproduction in genetic forms of hypertriglyceridaemia. EurJ Clin Invest 1980; 10: 17–22.

3 Havel RJ, Kane JP, Balasse EO, Segel N, Basso LV. Splanchnicmetabolism of free fatty acids and production of triglyceridesof very low density lipoproteins in normotriglyceridemic andhypertriglyceridaemic humans. J Clin Invest 1970; 49:2017–35.

4 Beil FU, Crouse JR, Zech L, Grundy SM, Berman M. Synthesisof cholesterol (CH), bile acids (BA), and very low densitylipoprotein-triglyceride (VLDL-TG) in familial combined hyper-lipidemia (FCHL). Circulation 1979; 60: 11–74.

5 Grundy SM. Metabolism of very low density lipoprotein-triglycerides in man, In Gotto AM, Smith LC, Allen B (eds.):Atherosclerosis V. New York, Springer Verlag 1980, 586–590.

6 Grundy SM, Mok HY, Zech L, Steinberg D, Berman M.Transport of very low density lipoprotein triglycerides in vary-ing degrees of obesity and hypertriglyceridaemia. J Clin Invest1979; 63: 1274–83.

7 Bjorntorp P, Bergman H, Varnauskas E. Plasma free fatty acidturnover rate in obesity. Acta Med Scand 1969; 185: 351–6.

8 Jensen MD, Haymond MW, Rizza RA, Cryer PE, Miles JM.Influence of body fat distribution on free fatty acid metabolismin obesity. J Clin Invest 1989; 83: 1168–73.

9 Reaven G, Lerner R, Stern M, Farquhar J. Role of insulin inendogenous hypertriglyceridaemia. J Clin Invest 1967; 46:1756–67.

10 Tobey TA, Greenfield M, Kraemer F. Relationship betweeninsulin resistance, insulin secretion, very-low-density lipopro-tein kinetics and plasma triglyceride levels in normotriglyceri-demic man. Metabolism 1981; 30: 165–71.

11 Boberg J, Carlson LA, Freyschuss U, Lassers BW, WahlqvistML. Sphlanchic secretion rates of plasma triglycerides andtotal and splanchnic turnover of plasma free fatty acids inmen with normo- and hypertriglyceridaemia. Eur J Clin Invest1972; 9: 454–66.

12 Dunn FL, Grundy SM, Bilheimer DW, Havel RJ, Raskin P.Impaired catabolism of very low-density lipoprotein-triglyc-eride in a family with primary hypertriglyceridaemia.Metabolism 1985; 34: 316–24.

13 Lalouel J-M, Wilson DE, Iverius P-H. Lipoprotein lipase andhepatic triglyceride lipase: molecular and genetic aspects. CurrOpin Lipidol 1992; 3: 86–95.

14 Aalto-Setala K, Fisher EA, Chen X, Chajek-Shaul T, Hayek T,Zechner R, Walsh A, Ramakrishnan R, Ginsberg HN. BreslowJL. Mechanism of hypertriglyceridaemia in humanapolipoprotein (apo) CIII transgenic mice: diminished very lowdensity lipoprotein fractional catabolic rate associated withincreased apo CIII reduced apo E on particles. J Clin Invest1992; 90: 1889–900.

15 Dammerman M, Sandkuijl LA, Halaas JL, Chung W, BreslowJL. An apolipoprotein CIII haplotype protective against hyper-triglyceridaemia is specified by promoter and 3¢ untranslatedregion polymorphisms. Proc Natl Acad Sci USA 1993; 90:4562–6.

16 Bruce R, Godsland I, Walton C, Crook D, Wynn V. Associationsbetween insulin sensitivity and free fatty acid and triglyceridemetabolism independent of uncomplicated obesity.Metabolism. Clin & Exp 1992; 41: 711–6.

17 Kissebah AH, Alfarsi S, Adams PW, Wynn V. Role of insulinresistance in adipose tissue and liver in the pathogenesis ofendogenous hypertriglyceridaemia in man. Diabetologia 1976;12: 563–71.

18 Larsson B, Bjorntorp P, Holm J, Schersten T, Sjostrom L, SmithU. Adipocyte metabolism in endogenous hypertriglyceri-daemia. Metabolism 1975; 24: 1375–89.

19 Yki-Jarvinen H, Taskinen MR. Interrelationships amonginsulin’s antilipolytic and glucoregulatory effects and plasmatriglycerides in nondiabetic and diabetic patients with endoge-nous hypertriglyceridaemia. Diabetes; 1988; 37: 1271–8.

20 Wilmore JH. A simplified method for determination of residuallung volumes. J Appl Physiol 1969; 27: 96–100.

21 Goldman RF, Buskirk ER. Body volume measurement byunderwater weighing: description of a method. In: Brozck J,



Henschel A, eds. Techniques for Measuring Body Composition.Washington D.C., USA: National Academy of Sciences, 1961;78–89.

22 Siri WE. Body composition from fluid spaces and density. In:Brozek J, Hanschel A, eds. Techniques for Measuring BodyComposition. Washington D.C., USA: National Academy ofSciences, 1961; 223–4.

23 Lohman TG, Roche AF, Martorell R. Anthropometric stan-dardization reference manual. Champaign, IL, USA: HumanKinetics Books, (1985)

23 Vega GL, Grundy SM. Two patterns of LDL metabolism in nor-motriglyceridemic patients with hypoalphalipoproteinemia.Arterioscler Thromb 1993; 13: 579–89.

24 Havel RJ, Naimark A, Borchgrevink CF. Turnover rate and oxi-dation of free fatty acids of blood plasma in man during exer-cise: Studies during continuous infusion of palmitate-14-C.J Clin Invest 1963; 42: 1054–9.

25 Shimizu S, Inoue K, Tani Y, Yamada H. Enzymatic microdeter-mination of serum free fatty acids. Anal Biochem 1979; 98 (2):341–5.

26 Coon PJ, Rogus EM, Goldberg AP. Time course of plasma freefatty acid concentration in response to insulin: effect of obesi-ty and physical fitness. Metabolism. Clin & Exp 1992; 41:711–6.

27 Elahi D, Nagulesparan M, Herscopf RJ. Feedback inhibition ofinsulin secretion by insulin: the hyperinsulinemia of obesity.N Engl J Med 1989; 306: 1196.

28 Saad MF, Knowler WC, Pettitt DJ, Nelson RG, Mott DM,Bennett PH. The natural history of impaired glucose tolerancein the Pima Indians. N Engl J Med 1988; 319: 1500–6.

29 Denke MA, Sempos CT, Grundy SM. Excess body weight: anunder-recognized contributor to high blood cholesterol levelsin white American men. Arch Intern Med 1993; 153:1093–103.

30 Lillioja S, Foley J, Bogardus C, Mott D, Howard BV. Free fattyacid metabolism and obesity in man: in vivo in vitro compar-

isons. Metabolism 1986; 35: 505–14.31 Campbell PJ, Carlson MG, Nurjhan N. Fat metabolism in

human obesity. Am J Physiol 1994; 266: E600–E5.32 Sjostrom L. Fat cells and body weight. In: Stunkard AJ, ed.

Obesity. Philadelphia, USA: W.B. Saunders Company 1980,72–100.

33 Nielsen SL, Larsen OA. Relationship of subcutaneous adiposetissue blood flow to thickness of subcutaneous tissue and totalbody fat mass. Scand J Clin Lab Invest 1973; 31: 383–8.

34 Seidell JC, Cigolini M, Charzewska J, Ellsinger BM, Bjorntorp P,Hautvast JG, Szostak W. Fat distribution and gender differ-ences in serum lipids in men and women from four Europeancommunities. Atherosclerosis 1991; 87: 203–10.

35 Freedman DS, Jacobsen SJ, Barboriak JJ, Sobocinski KA,Anderson AJ, Kissebah AH, Sasse EA, Gruchow HW. Body fatdistribution and male/female differences in lipids and lipopro-teins. Circulation 1990; 81: 1498–506.

36 Abate N, Burns D, Peshock R, Garg A, Grundy SM. Estimationof adipose tissue mass by magnetic resonance imaging: valida-tion against dissection in human cadavers. J Lipid Res 1994;35: 1490–6.

37 Abate N, Garg A, Peshock RM, Stray-Gundersen J, Adams-Huet B, Grundy SM. Relationship of generalized and regionaladiposity to insulin sensitivity in men with NIDDM. Diabetes1996; 45: 1684–93.

38 Bergman EN, Havel RJ, Wolfe BM, Bohmer T. Quantitativestudies of the metabolism of chylomicron triglycerides andcholesterol by liver and extra hepatic tissues of shepherd dogs.J Clin Invest 1971; 50: 1831–9.

Received 6 March 1997; accepted 10 October 1997.

Correspondence: Dr Gloria Lena Vega PhD, Center for HumanNutrition, University of Texas South Western Medical Center,5323 Harry Hines Boulevard, Dallas, Texas 75235–9052, USA(fax: 1 214 648–4837).

Documents

Abnormal metabolism of free fatty acids in hypertriglyceridaemic men: apparent insulin resistance of adipose tissue