The Journal of Nutrition

Nutrition and Disease

Conjugated Linoleic Acids Reduce Body Fatin Healthy Postmenopausal Women1–3

Marianne Raff,4* Tine Tholstrup,4 Søren Toubro,4 Jens M. Bruun,5 Pia Lund,6 Ellen M. Straarup,6

Robin Christensen,7 Maria B. Sandberg,8 and Susanne Mandrup8

4Department of Human Nutrition, Faculty of Life, University of Copenhagen, Frederiksberg 1958, Denmark; 5Department of

Endocrinology and Metabolism C, Aarhus University Hospital, DK-8000 Aarhus C, Denmark; 6Biochemistry and Nutrition Group,

BioCentrum-DTU, Technical University of Denmark, Lyngby 2800, Denmark; 7Parker Institute, Musculoskeletal Statistics Unit,

Frederiksberg Hospital, Frederiksberg C 2000, Denmark; and 8Department of Biochemistry and Molecular Biology, University of

Southern Denmark, Odense 5000, Denmark

Abstract

Isomers of conjugated linoleic acids (CLA) reduce fat mass (FM) and increase insulin sensitivity in some, but not all, murine

studies. In humans, this effect is still debatable. In this study, we compared the effect of 2 CLA supplements on total and

regional FM assessed by dual energy X-ray absorptiometry, changes in serum insulin and glucose concentrations, and

adipose tissue (AT) gene expression in humans. In a double-blind, parallel, 16-wk intervention, we randomized 81 healthy

postmenopausal women to 1) 5.5 g/d of 40/40% of cis9,trans11-CLA (c9,t11-CLA) and trans10,cis12-CLA (t10,c12-CLA)

(CLA-mix); 2) cis9, trans11-CLA (c9,t11-CLA); or 3) control (olive oil). We assessed all variables before and after the

intervention. The CLA-mix group had less total FM (4%) and lower-body FM (7%) than the control (P = 0.02 and , 0.001,

respectively). Post hoc analyses showed that serum insulin concentrations were greater in the CLA-mix group (34%) than

the control group (P = 0.02) in the highest waist circumference tertile only. AT mRNA expression of glucose transporter 4,

leptin, and lipoprotein lipase was lower, whereas expression of tumor necrosis factor-a was higher in the CLA-mix group

than in the control group (P, 0.04). In conclusion, a 50:50mixture of c9,t11- and t10,c12-CLA isomers resulted in less total

and lower-body FM in postmenopausal women and greater serum insulin concentrations in the highest waist

circumference tertile. Future research is needed to confirm the insulin desensitizing effect of the CLA mixture and the

effect on the mRNA expression of adipocyte-specific genes in humans. J. Nutr. 139: 1347–1352, 2009.

Introduction

Conjugated linoleic acid (CLA)9 is a collective term for a groupof naturally occurring and industrially produced linoleic acid(C18:2) isomers with conjugated double bonds. cis9, trans11-CLA (c9,t11-CLA), found in dairy products and ruminant meat,is the most abundant naturally occurring isomer (1), whereasother isomers only occur naturally in negligible amounts. Themajority of CLA studies in humans and animals have examined

the effect of a mixture containing primarily the trans10, cis12-CLA isomer (t10,c12-CLA) and the c9,t11-isomer in equalamounts. This mixture is promoted and sold worldwide as aweight-loss agent, because supplementation with the CLAmixture and t10,c12-CLA has been shown to reduce fat mass(FM) in mice (2,3). In humans, the results are inconclusive inthat some studies (4,5), but not all (6,7), have confirmed the fat-reducing effect of a CLA mixture. The rodent studies also foundthat adipose tissue (AT) in different regions differed in theirresponsiveness to a CLA mixture (8,9), the retroperitonealregion being the most sensitive. In addition to the fat-reducingeffect, the CLA mixture also increased insulin sensitivity in rats(10), whereas decreased insulin sensitivity has been found inmice (2). In humans, both a CLA mixture and the c9,t11- andt10,c12-isomers separately have been reported to reduce insulinsensitivity (7,11,12) or to have no effect on fasting insulin andglucose concentrations (13,14), whereas only a single studyreported improved insulin sensitivity in young participants aftertreatment with a CLA mixture (15).

The mechanisms underlying the effects of CLA are not yetestablished. However, there are clear indications that theadipogenic program is specifically inhibited by the t10, c12CLA isomer. In vitro studies using murine adipocyte cell lines

1 Supported by the Danish Dairy Research Foundation and the Danish Research

Development Program for Food Technology.2 Author disclosures: M. Raff, T. Tholstrup, S. Toubro, J. M. Bruun, P. Lund,

E. M. Straarup, R. Christensen, M. B. Sandberg, and S. Mandrup, no conflict

of interest.3 Supplemental Table 1 is available with the online posting of the paper at jn.

nutrition.org.

* To whom correspondence should be addressed. E-mail: mrf@life.ku.dk.

9 Abbreviations used: AT, adipose tissue; BP, blood pressure; CE, cholesterol

ester; CLA, conjugated linoleic acid; CLA-mix, mixture of cis9,trans11-CLA and

trans10,cis12-CLA; c9,t11-CLA, cis9,trans11-CLA; CVD, cardiovascular disease;

DEXA, dual energy X-ray absorptiometry; E%, percent of energy; EI, energy

intake; FA, fatty acid; FM, fat mass; HOMA-R, homeostatic model assessment

for insulin resistance; LBM, lean body mass; LPL, lipoprotein lipase; t10,

c12-CLA, trans10,cis12-CLA; TNFa, tumor necrosis factor-a; wt%, relative

weight content.

0022-3166/08 $8.00 ã 2009 American Society for Nutrition.

Manuscript received January 27, 2009. Initial review completed March 13, 2009. Revision accepted May 14, 2009. 1347First published online June 3, 2009; doi:10.3945/jn.109.104471.

at DE

F - D

anish Electronic R

es Consortium

on June 29, 2009 jn.nutrition.org

Dow

nloaded from

(16) as well as primary human adipocytes (17) show that t10,c12-CLA supplementation reduces triacylglycerol accumulationand adipocyte differentiation. These studies are supported by invivo animal studies where t10,c12-CLA has been shown toreduce the transcription of several adipocyte-specific genes suchas adiponectin, glucose transporter 4, leptin, and lipoproteinlipase (LPL) in white AT (16,18,19). The key adipogenictranscription factor PPARg has been suggested to be involvedin these effects of t10,c12-CLA on gene expression, becauseseveral of the adipogenic genes are PPARg target genes.

Because the 50:50 mixture of CLA is the one commerciallyavailable to population groups interested in losing weight, wefound it pertinent to examine the effect of this mixture and tocompare this effect with the c9,t11-isomer and olive oil in a16-wk intervention study of postmenopausal volunteers whowere prone to changes in FM. The c9,t11-CLA isomer has notbeen as well examined as the t10,c12-isomer in humans. Becausethis is the one most abundant in natural foods and is possible toincrease the content of this isomer in bovine milk via feedingstrategies (20), we found it relevant to include this isomer in ourstudy. Our main focus was on the in vivo changes in total andregional FM, as well as measures of glucose metabolism andgene expression of adipogenic genes. The effect of CLA onadipogenic genes in human AT samples has, to our knowledge,been examined in only 1 other study (21), which was publishedafter the initiation of this human intervention trial.

Methods and Participants

This article is based on data from a large CLA supplementationstudy. Here we report effects on total and regional FM and leanbody mass (LBM), insulin sensitivity, and the relative geneexpression of adipose derived proteins. We have presentedresults from cardiovascular disease (CVD) markers, inflamma-tion, and lipid peroxidation elsewhere (22).We performed a16-wk, double-blind, randomized, dietary parallel study withdietary supplements that differed in fatty acid (FA) composition.The participants were stratified according to their bloodpressure (BP), waist circumference, and smoking habits to 1 ofthe 3 supplements. We assessed all outcome variables at the startand end of the intervention and fully explained the protocol andaim of the study to the participants (orally and in writing) beforethey gave their written informed consent.

Participants. All applicable institutional and governmentalregulations concerning the ethical use of human volunteers werefollowed during this research. The Scientific Ethics Committeeof the City of Copenhagen and Frederiksberg Municipality (KF)01–166/03 approved the research protocol.

We recruited 81 postmenopausal women by advertising inlocal newspapers and froma list obtained from theDanish centralnational register. Exclusion criteria were: BMI .35 kg/m2,hypertension, chronic disease, regular use of medication, and,1 y since last menstruation. All women were apparentlyhealthy as indicated by a medical and lifestyle questionnaire.The baseline characteristics did not differ between the groups(Supplemental Table 1). All participants agreed to refrain fromdonating blood 2 mo prior to and during the study and fromtaking dietary supplements and medication that might interferewith study measurements (e.g. acetylsalicylic acid). We assessedphysical activity by a questionnaire and instructed all participants tomaintain the same level of physical activity throughout the study.

The study participants received the dietary supplements (andreturned leftover capsules) at monthly visits to the Department

of Human Nutrition. Within a period 2 wk prior to, and in wk 8of the intervention, the participants completed a 3-d weighedfood record on 2 weekdays and 1 d the following weekend toassess potential difference in dietary intake between the groupsbefore or during the intervention. We gave each participantdetailed instructions on how to fill out the food records. Aclinical dietician coded all records before evaluating them. Shealso calculated the energy intake (EI) and dietary composition byusing a national database (Dankost; National Food Agency).Habitual dietary intake before the intervention did not differbetween the 3 groups (22).

Test fats. During the intervention, we supplemented theparticipants with either a commercial mix of CLA-isomers(CLA-mix) or a synthetic preparation of the isoform found inmilk (c9,t11-CLA). Olive oil, which is considered neutral orslightly beneficial to blood lipids (and possibly to new CVD riskmarkers, as well), was selected as control. The FA compositionof the supplements has been presented elsewhere (22). The testfats were provided as 6 capsules/d and the participants received atotal of 5.5 g test fat/d, which corresponded to 2.3 g of c9,t11-CLA + 2.2 g t10,c12-CLA (mixture of 41.2 and 39.9% wt ofeither isomer) or 4.7 g c9t11-CLA. The control group received5.5 g/d of olive oil, which corresponded to 4.2 g of oleic acid[18:1(n-9)].

Compliance.We provided all participants with surplus capsulesonce per month (exact number known to staff) and leftovercapsules were counted when returned the following month. Aparticipant was considered compliant when .75% of theexpected capsules were taken. We used the relative weightcontent (wt%) of c9,t11-, and t10,c12-CLA in the AT andplasma cholesterol esters (CE) as a biomarker of compliance.

Fatty acid composition. We determined the FA composition inplasma CE and in AT. Total lipids were extracted as describedelsewhere (22). We isolated CE by TLC, solvent system heptane:isopropanol:acetic acid 95:5:1 (v:v), and prepared the lipidfractions FAME by transesterification with potassium hydroxidein methanol at room temperature. GC analyses were performedusing a gas chromatograph (HP 6890 Hewlett-Packard). Webased identification and quantification of FA on standards ofFAME (Nu-Chek Prep). Further details are presented elsewhere(22).

Body weight and composition.We used a regularly calibratedelectronic scale (SCALE; Lindells) to measure the participants’body weight to the nearest 0.1 kg while they were wearingunderwear, with no subtractions for clothes. Height wasmeasured without shoes to the nearest 0.5 cm. Body composi-tion was measured by dual energy X-ray absorptiometry(DEXA; Lunar Radiation) with LUNAR PRODIGY software(version 4.6c; Lunar Radiation); CV ,0.1%. In total, the scantook 45 min/person and provided a maximum radiation of 19.3mRem (0.193 mSv) (both visits included).

Blood sampling and analysis. We collected venous bloodafter a 12-h overnight fast on the first and last day of theintervention. Blood for serum adiponectin and FA analyses wascollected in tubes containing EDTA kept on ice and the sampleswere centrifuged at 3000 3 g; 15 min 48C. We assessed theserum concentration of adiponectin by a specific highly sensitivehuman ELISA. The adiponectin assay (B-Bridge International)had an intra-assay CV of 5.0% (n = 12). We collected blood for

1348 Raff et al.

at DE

F - D

anish Electronic R

es Consortium

on June 29, 2009 jn.nutrition.org

Dow

nloaded from

serum insulin and glucose concentrations into dry tubes andafter coagulation, the samples were centrifuged at 3000 3 g;15 min at 208C. Serum (500 mL) was stored at 2208C. Insulinconcentration (pmol/L) was assessed by solid 2-site fluoroim-munometric assay (Insulin kit B080–101, Wallac 04) on anAutoDELFIA system (1235–514, Wallac), CV = 3.2%. Weassessed glucose concentrations (mmol/L) by a hexokinaseendpoint procedure in serum (Gluco-quant Glucose/HK kit,Roche Diagnostics) using a Cobas Mira Plus analyzer (RocheDiagnostic Systems), CV = 1.2%. The homeostatic modelassessment for insulin resistance (HOMA-R) was calculatedusing the following equation: [fasting serum insulin (pmol/L) 3fasting serum glucose (mmol/L)]/135.

AT biopsies and gene expression analysis. We tooksubcutaneous AT biopsies from the upper outer quadrant of abuttock with a needle (16 or 17 G) attached to a vacuum tube.We left the samples (~35 mg) in the connector and placed bothneedle and connector in a marked tube filled with nitrogen andstored the samples at 2808C (23). Many of our biopsies couldnot be analyzed for the relative mRNA level of the AT-specificand inflammatory genes due to the small size of the adiposebiopsies. In total (before/after): 14/10, 12/12, and 14/11 sampleswere analyzed from the CLA-mix, c9,t11-CLA, and controlgroups, respectively. RNA was isolated from biopsies and usedfor quantitative RT-PCR analysis. We performed RNA isolation,cDNA synthesis, and real time quantitative RT-PCR as describedby Brown et al. (17). The relative expression of the AT-specificgenes was calculated after normalization to glyceraldehyde-3-phosphate dehydrogenase.

Statistics. We used the SAS statistical package (version 9.1; SASInstitute) for all statistical analyses. We applied a mixed-effectsmodel ANCOVA to compare the effects of the 3 supplementsusing the baseline value as a covariate. If we detected asignificant treatment effect (P , 0.05), we included the Tukey-Kramer adjusted test for multiple post hoc pair-wise compari-sons. We transformed data logarithmically when necessary, e.g.to normalize the distribution of residuals and to obtain variancehomogeneity. We performed statistical tests on the transformeddata. Baseline BP (both systolic and diastolic BP), waistcircumference, and smoking status were all tested for influenceon the results. Waist circumferences influenced fasting insulinconcentrations and HOMA-R and were therefore included inthe statistical analysis. We analyzed fasting serum insulinconcentrations post hoc on data that was divided into thefollowing baseline waist circumference tertiles: low (71–87 cm),medium (88–92 cm), and high (93–109 cm). The Pearsoncorrelation coefficient was determined from pairwise correla-tions of D values (end – start values) from the total sample ofwomen. The sparse number of available AT samples for relativemRNA analysis made formal model assumptions of a specificprobability distribution speculative. Accordingly, we employedthe Fisher’s exact test to determine whether the missing datavaried between the 3 groups. Because data were missingcompletely at random (24), the computer-intensive methodbootstrapping was applied, which enabled us to estimate somequantity (25). We calculated the mean median values followingmultiple bootstrap samples (10,000 samples with replacement),with the corresponding SE associated with this quantity. We havesummarized data for baseline variables as mean 6 SD and foroutcome variables as LSMean 6 SEM, unless noted otherwise.

We determined Pearson correlation coefficient from pairwisecorrelations of D values (end – start values) from the total sampleof women.

Results

Participants and compliance. A total of 81 participants wereincluded in our study and 75 completed the 16-wk intervention.Three of the 6 dropouts were caused by withdrawal of consentand 3 by nonsevere side effects. The dropouts were distributed as2, 3, and 1 in the CLA-mix, c9,t11-CLA, and control groups,respectively. The main intervention period was (mean 6 SD)112 6 3 d. Capsule counting showed a mean compliance of98.2 6 22.4% in all groups together and of 98.8 6 1.9%,98.7 6 1.4%, and 98.2 6 2.4% in the CLA-mix, c9,t11-CLA,and control group, respectively. The number of returned capsulesdid not differ between the groups. The wt% of the specific CLAisomers in the plasma CE and AT all reflected the dietarysupplements after the intervention (Table 1). Similar results wereobtained for wt% of the specific CLA isomers in the plasmatriacylglycerol and phospholipids (data not shown).

Dietary intake. The c9,t11-CLA group reported a lower EI atwk 8 than the CLA-mix group (LSMean 6 SEM) (7330 6 300kJ vs. 8420 6 320 kJ; P = 0.01) and lower fat intake comparedwith the control group [31.56 1.0% of energy (E%) vs. 34.2 60.9 E%; P = 0.03].

However, when expressed as g/d, the fat intake did not differbetween any of the 3 groups. Both the EI and the fat intake in thec9,t11-CLA groups at wk 8 were lower compared with theirhabitual intake. The intake of carbohydrate and protein did notdiffer between the 3 groups intake when expressed as E% or g/dat wk 8.

Body weight and DXA scans. Body weight tended to be less inthe CLA-mix group than in the control (P = 0.05) and the c9,t11-CLA group (P = 0.09) (Table 2). Three DEXA scans weremissing due to technical problems, 2 from the CLA-mix groupand 1 from the c9,t11-CLA group. After the intervention, theCLA-mix group had a lower total FM than the control group(P = 0.02) and tended to have lower total FM compared with thec9,t11-CLA group (P = 0.07). Regional measurements showedthat the CLA-mix group had less lower-body fat than both thecontrol group (P = 0.0008) and the c9,t11-CLA group (P =0.009). Lower-body LBM was higher in the CLA-mix groupthan in the control (P = 0.02) and c9,t11-CLA (P = 0.04) groups.None of the other DEXA scan data differed among the 3 groups.

TABLE 1 Fatty acid composition in plasma CE and AT offasting, postmenopausal women who wheresupplemented with CLA-mix, c9,t11-CLA, or oliveoil for 16 wk1

CLA-mix c9,t11-CLA Olive oil

Plasma CE, n 25 24 26

18:1(n-9), %wt 19.81 6 0.26a 20.38 6 0.28b 21.33 6 0.26c

c9,t11-CLA, %wt 0.42 6 0.02a 0.76 6 0.04b 0.11 6 0.01c

t10,c12-CLA, %wt 0.26 6 0.01a 0.11 6 0.01b 0.07 6 0.01c

AT,2 n n = 12 n = 13 n = 11

18:1(n-9) , %wt 46.17 6 0.31 46.55 6 0.38 45.77 6 0.33

c9,t11-CLA, %wt 0.49 6 0.02a 0.68 6 0.03b 0.37 6 0.02c

t10,c12-CLA, %wt 0.08 6 0.03 ND ND

1 Data are LSMean6 SEM. Means in a row with superscripts without a common letter

differ, P , 0.05 (ANCOVA following intervention; baseline values applied as

covariates).2 Due to the small amount of sample collected, it was not possible to analyze all FA

composition in all AT biopsies. ND, Nondetectable.

Conjugated linoleic acids and body fat mass in women 1349

at DE

F - D

anish Electronic R

es Consortium

on June 29, 2009 jn.nutrition.org

Dow

nloaded from

Serum metabolites. We found no treatment effect on serumadiponectin, insulin, glucose concentrations, or the calculatedHOMA-R. However, waist circumference (measured only atbaseline) influenced the fasting insulin concentration after theintervention (P = 0.004). Post hoc analysis in the tertile ofwomen with the highest waist circumference (94–109 cm)showed that thee fasting serum insulin concentration wasgreater in the CLA-mix group than in the control (34%; P =0.02) and the c9,t11-CLA (28%; P = 0.04) groups. There wereno differences between the effects of the supplements in the 2lower tertiles. Similar post hoc results were obtained for theHOMA-R (data not shown). The serum adiponection concen-trations did not differ in the 3 groups.

Relative mRNA expression of adipogenic genes and

tumor necrosis factor-a in AT from the buttocks. In theCLA-mix group, we found a lower relative AT expression ofGLUT 4, LPL, and leptin compared with the control (P = 0.001,0.04, and 0.04, respectively) (Fig. 1). The leptin expression in theCLA-mix group also tended to be lower than in the c9,t11-CLAgroup (P = 0.06). The expression of tumor necrosis factor-a(TNFa) was greater in the CLA-mix group than in the control(P = 0.04). In the c9,t11-CLA group, the expression of TNFatended to be lower than in the control (P = 0.07). Whencomparing the 2 a priori-defined active interventions, theexpression of TNFa was greater in the CLA-mix group than inthe c9,t11-CLA group (P = 0.04).

Correlations. The changes in total and lower-body FM wereinversely correlated to the changes in the relative content of t10,c12-CLA in plasma CE (r =20.29, P = 0.02 and r = 20.43, P ,0.001 for total and lower-body FM, respectively). Correlationanalysis could not be performed for the t10,c12-CLA in AT,because detectable levels were found only in participants fromthe CLA-mix group after the intervention. No other correlationswere detected.

Discussion

This randomized, controlled, double-blind study showed that a40:40 mixture of c9,t11- and t10,c12-CLA isomers significantly

reduced total and lower-body FM compared with olive oil. Thiswas supported by a similar tendency in body weight and, veryinterestingly, also by differences in the pattern of expression ofadipogenic genes. The location-specific effect of CLA on the FMis interesting, as the FM in the lower-body was previously shownto correlate with insulin sensitivity (26).

The total body FM-reducing effect of CLA mixtures issupported by other human studies (4,5) as well as by a recentlypublished meta-analysis (27), but there are also several studiesthat did not find this effect (6,7). The reason for the discrepancyis not clear. A recent study also found a lower-body FMreduction after supplementation with a CLA mixture whenassessed by DEXA (28), whereas several others found nodifference in hip circumference regardless of the effect on totalFM (7,13). We speculate that differences in the measuringmethods may explain the discrepancy. The changes in total andlower-body FMwere significantly and inversely correlated to thechange in the relative content of t10,c12-CLA in plasma CE,which is consistent with the findings of others (29). We found nocorrelation between the change in FM and c9,t11-CLA in CE orAT. Mice and cell studies have found that the t10,c12-isomer isresponsible for the fat-reducing effect of CLA (3,17). Only theCLA-mix supplement contained the t10,c12-isomer; thus, thecorrelation to this isomer is consistent with our finding of FMreduction in this group only.

Notably, we found greater lower-body LBM in the CLA-mixgroup than in the other 2 groups. This is in contrast to mosthuman studies that find no significant difference between theeffect of a CLA supplement and a placebo (5,7,30). However, 2human studies (4,30) and some animal studies (9,31) found asimilar increase in LBM in response to CLA mixtures. The effectof a CLA supplement on LBM may be small and thus moredifficult to detect. This is partly supported by the fact that manyof the studies that examine the effect of CLA on LBM findnonsignificant increases or smaller decreases in LBM than acontrol (5,7,30). However, because we did not assess physicalactivity during the intervention, we cannot rule out that thechange in LBM in our CLA-mix group was caused by change inphysical activity.

We found no differences in the effect of the 3 supplements onfasting serum insulin or glucose concentrations or the calculatedHOMA-R, but because the baseline waist circumference had a

TABLE 2 Weight, body composition, insulin sensitivity, andadiponectin of fasting, postmenopausal women whowhere supplemented with CLA-mix, c9,t11-CLA, orolive oil for 16 wk1

CLA-mix c9,t11-CLA Olive oil

n 25 24 26

Body composition

Weight, kg 70.2 6 0.3 71.0 6 0.3 71.1 6 0.3

Total FM, kg 24.6 6 0.3a 25.5 6 0.3ab 25.6 6 0.3b

Lower-body FM, kg 9.5 6 0.2a 10.1 6 0.2b 10.2 6 0.1b

Upper-body FM, kg 11.1 6 0.1 11.4 6 0.1 11.3 6 0.1

Total LBM, kg 43.3 6 0.2 43.1 6 0.2 43.1 6 0.2

Lower-body LBM, kg 14.7 6 0.1a 14.3 6 0.1b 14.3 6 0.1b

Upper-body LBM, kg 20.9 6 0.2 21.0 6 0.1 21.0 6 0.1

Serum measures

Insulin, pmol/L 41.6 6 2.4 40.9 6 2.4 40.3 6 2.3

Glucose, mmol/L 5.1 6 0.1 5.1 6 0.1 5.1 6 0.1

HOMA-R 1.4 6 0.1 1.4 6 0.1 1.4 6 0.1

Adiponectin, mg/L 11.9 6 0.4 12.6 6 0.5 12.5 6 0.4

1 Data are LSMeans 6 SEM. Means in a row with superscripts without a common

letter differ, P , 0.05 (ANCOVA following intervention; baseline values applied as

covariates).

FIGURE 1 Changes from baseline in AT mRNA expression of

adipogenic genes and TNFa in subcutaneous fat biopsies from fasting,

postmenopausal women who where supplemented with CLA-mix, c9,

t11-CLA, or olive oil for 16 wk. The relative expression of mRNA was

calculated after normalization to glyceraldehyde-3-phosphate dehy-

drogenase. Values are means 6 SE, n = 10–14. Means for a variable

without a common letter differ, P , 0.05 (bootstrapping, 10,000

samples with replacement).

1350 Raff et al.

at DE

F - D

anish Electronic R

es Consortium

on June 29, 2009 jn.nutrition.org

Dow

nloaded from

significant influence on the insulin concentrations, we conducteda post hoc analysis where the women were divided into tertilesaccording to their waist circumference. In the women with thelargest waists, we found a significantly greater serum insulinconcentration in the CLA-mix group than in the 2 other groups,whereas the supplementation groups did not differ in the leanerwomen. This finding was supported by similar results for theHOMA-R. A large waist circumference is a characteristic ofmetabolic syndrome and a risk factor for diabetes type II (32).Thus, our post hoc analysis indicates that participants with alarge waist circumference are more sensitive to a negative effectof a CLA mixture on insulin metabolism. This is particularlyinteresting in light of the previous CLA studies in humans, asmost studies with participants in the risk group (e.g. men withabdominal obesity or diabetes type II or suffering from meta-bolic syndrome) showed that supplementation with a CLAmixture or t10,12-CLA decreased insulin sensitivity (7,11).Abdominal obesity is a promoter of low-grade inflammation(33) and we speculate that participants with large waistcircumferences may have increased potential for inducingproinflammatory signals in response to a CLA mixture, whichin turn mediates the insulin desensitizing effect.

In relation to the inflammatory response, TNFa mRNAexpression was higher in AT in the CLA-mix group than in thecontrol group. The isomer t10, c12 CLA has been shown toactivate expression of proinflammatory genes in human adipo-cytes and our results indicate that t10,c12 CLAmay have similareffects on adipocytes in vivo. The proinflammatory effect ofCLA is supported by the finding of higher serum C-reactiveprotein concentrations, a well-established inflammatory marker,in the CLA-mix group in this (22) and at least 1 other study (30).However, not all studies found such proinflammatory effects(11,13).

Interestingly, when taken together, our results seem tosupport a delipidating effect of t10,c12 CLA by mechanismsreminiscent of those in animals (18) and in vitro studies withhuman adipocytes (17). Thus, like those studies, we foundsignificantly lower relative AT expression of PPARg target genessuch as glucose transporter 4 and LPL and a significant inductionof TNFa in the CLA-mix group compared with the control. Inaddition, there was a tendency for the AT expression of A-FABPand adiponectin, also PPARg target genes, to be less in the CLA-mix group than in the control. The reduction in leptin expressioncontrasts with data from acute in vitro studies (17). Thediscrepancy between in vivo and in vitro studies in this case islikely to be due to the length of CLA exposure. TNFa and otherproinflammatory cytokines have been found to suppress bothactivation and expression of PPARg (34), the key adipocytetranscription factor. Studies using human adipocytes in cultureindicate that repression of PPARg expression and activity isinvolved in the delipidating and dedifferentiating effects of thet10,c12-isomers (17). Although our in vivo results are based onfew samples, they support the relevance of the above-mentionedex vivo results. It is of interest that the relative AT expression ofTNFa tended to be lower in the c9,t11-CLA group than in thecontrol group.

To our knowledge, only 1 other study has examined the effectof CLA on the relative expression of adipogenic genes in humanAT (21). In contrast to our findings, they found greater PPARgmRNA expression and no effect on body composition. Thereason for the different results is not clear, because CLA dose,isomeric composition, participant bodyweight, and interventionlength were similar to ours. Their study participants were ofboth genders and younger than ours, but because the fat-

reducing effect has been found both in men (13) and in youngparticipants (~30 y) (14), this is unlikely to account for thedifferent results.

Adiponectin is expressed by the adipocytes and is inverselyassociated with body weight and insulin resistance (35). Wefound no difference in the effect of a CLA mixture or the c9,t11-isomer on serum adiponectin concentration, which is consistentwith the findings of others (13).

A strength of this study is the high compliance and lowdropout rate, which indicated good tolerance of the CLAsupplements. The EI, examined using 3-d food records, did notdiffer significantly between the CLA-mix and the control group,which suggest that the observed differences between these 2groups were independent of diet. The lower EI reported by thec9,t11-CLA group at wk 8 compared with the CLA-mix groupwould have been expected to reduce the FM in the c9,t11-CLAgroup, but the CLA-mix group had a lower FM than the c9,t11-CLA group; thus, the lower EI in the c9,t11-CLA group is notlikely to have influenced our FM results.

This study included healthy, primarily normal-weight womenfor whom an increase in a risk parameter within the normalrange may be less relevant than a change from normal toincreased risk. However, the participating women had a smallincreased risk of developing CVD and diabetes type II due totheir age and postmenopausal status and any change in a riskparameter that would indicate increased CVD risk may beconsidered unbeneficial.

Another concern may be the power in this study. Theprestudy power calculation was based on differences in bloodlipids and for the fasting serum insulin concentrations, we mayhave had too little power to detect a change caused by the c9,t11-isomer, as found by Riserus et al. (12). This also applies tothe upper-body FM and LBM as well as serum adiponectinconcentrations.

In conclusion, we found that a mixture of t10, c12- and c9,t11-CLA resulted in lower total and lower-body FM in healthypostmenopausal women. The change in FM was correlated withthe content of t10, c12- isomer in CE and not the c9,t11-isomer.Future research is needed to confirm if a CLA mixture, asindicated by our current results, can reduce insulin sensitivity inwomen with large, but not lower, waist circumference and tofurther examine if a CLA mixture containing the t10,c12-CLAisomer reduces mRNA expression of adipocyte-specific genesand increases the mRNA expression of TNFa in AT in humans.

AcknowledgmentsWe thank our technicians Ella Jessen, Kirsten Ebbesen, LeifJacobsen, Martin Kreutzer, and Hanne Lysdal Petersen from theDepartment of Human Nutrition for technical assistance.

Literature Cited

1. Jiang J, Bjorck L, Fonden R. Conjugated linoleic acid in Swedish dairyproducts with special reference to the manufacture of hard cheeses. IntDairy J. 1997;7:863–7.

2. Tsuboyama-Kasaoka N, Takahashi M, Tanemura K, Kim HJ, Tange T,Okuyama H, Kasai M, Ikemoto S, Ezaki O. Conjugated linoleic acidsupplementation reduces adipose tissue by apoptosis and developslipodystrophy in mice. Diabetes. 2000;49:1534–42.

3. Park Y, Storkson JM, Albright KJ, Liu W, Pariza MW. Evidence that thetrans-10,cis-12 isomer of conjugated linoleic acid induces body com-position changes in mice. Lipids. 1999;34:235–41.

4. Gaullier JM, Halse J, Hoye K, Kristiansen K, Fagertun H, Vik H,Gudmundsen O. Conjugated linoleic acid supplementation for 1 yreduces body fat mass in healthy overweight humans. Am J Clin Nutr.2004;79:1118–25.

Conjugated linoleic acids and body fat mass in women 1351

at DE

F - D

anish Electronic R

es Consortium

on June 29, 2009 jn.nutrition.org

Dow

nloaded from

5. Blankson H, Stakkestad JA, Fagertun H, Thom E, Wadstein J,Gudmundsen O. Conjugated linoleic acid reduces body fat mass inoverweight and obese humans. J Nutr. 2000;130:2943–8.

6. Petridou A, Mougios V, Sagredos A. Supplementation with CLA: isomerincorporation into serum lipids and effect on body fat of women. Lipids.2003;38:805–11.

7. Riserus U, Arner P, Brismar K, Vessby B. Treatment with dietarytrans10cis12 conjugated linoleic acid causes isomer-specific insulinresistance in obese men with the metabolic syndrome. Diabetes Care.2002;25:1516–21.

8. Azain MJ, Hausman DB, Sisk MB, Flatt WP, Jewell DE. Dietaryconjugated linoleic acid reduces rat adipose tissue cell size rather thancell number. J Nutr. 2000;130:1548–54.

9. DeLany JP, Blohm F, Truett AA, Scimeca JA, West DB. Conjugatedlinoleic acid rapidly reduces body fat content in mice without affectingenergy intake. Am J Physiol. 1999;276:R1172–9.

10. Houseknecht KL, Vanden Heuvel JP, Moya-Camarena SY, PortocarreroCP, Peck LW, Nickel KP, Belury MA. Dietary conjugated linoleic acidnormalizes impaired glucose tolerance in the Zucker diabetic fatty fa/farat. Biochem Biophys Res Commun. 1998;244:678–82.

11. Moloney F, Yeow TP, Mullen A, Nolan JJ, Roche HM. Conjugatedlinoleic acid supplementation, insulin sensitivity, and lipoproteinmetabolism in patients with type 2 diabetes mellitus. Am J Clin Nutr.2004;80:887–95.

12. Riserus U, Vessby B, Arnlov J, Basu S. Effects of cis-9,trans-11conjugated linoleic acid supplementation on insulin sensitivity, lipidperoxidation, and proinflammatory markers in obese men. Am J ClinNutr. 2004;80:279–83.

13. Taylor JS, Williams SR, Rhys R, James P, Frenneaux MP. Conjugatedlinoleic acid impairs endothelial function. Arterioscler Thromb VascBiol. 2006;26:307–12.

14. Malpuech-Brugere C, Verboeket-van de Venne WP, Mensink RP, ArnalMA, Morio B, Brandolini M, Saebo A, Lassel TS, Chardigny JM, et al.Effects of two conjugated linoleic Acid isomers on body fat mass inoverweight humans. Obes Res. 2004;12:591–8.

15. Eyjolfson V, Spriet LL, Dyck DJ. Conjugated linoleic acid improvesinsulin sensitivity in young, sedentary humans. Med Sci Sports Exerc.2004;36:814–20.

16. Kang K, Liu W, Albright KJ, Park Y, Pariza MW. trans-10,cis-12 CLAinhibits differentiation of 3T3–L1 adipocytes and decreases PPARgamma expression. Biochem Biophys Res Commun. 2003;303:795–9.

17. Brown JM, Boysen MS, Chung S, Fabiyi O, Morrison RF, Mandrup S,McIntosh MK. Conjugated linoleic acid induces human adipocytedelipidation: autocrine/paracrine regulation of MEK/ERK signaling byadipocytokines. J Biol Chem. 2004;279:26735–47.

18. Warren JM, Simon VA, Bartolini G, Erickson KL, Mackey BE, KelleyDS. Trans-10,cis-12 CLA increases liver and decreases adipose tissuelipids in mice: possible roles of specific lipid metabolism genes. Lipids.2003;38:497–504.

19. Poirier H, Shapiro JS, Kim RJ, Lazar MA. Nutritional Supplementationwith trans-10, cis-12-conjugated linoleic acid induces inflammation ofwhite adipose tissue. Diabetes. 2006;55:1634–41.

20. Bauman DE, Barbano DM, Dwyer DA, Griinari JM. Technical note:production of butter with enhanced conjugated linoleic acid for use inbiomedical studies with animal models. J Dairy Sci. 2000;83:2422–5.

21. Nazare JA, de la Perriere AB, Bonnet F, Desage M, Peyrat J,Maitrepierre C, Louche-Pelissier C, Bruzeau J, Goudable J, et al. Dailyintake of conjugated linoleic acid-enriched yoghurts: effects on energymetabolism and adipose tissue gene expression in healthy participants.Br J Nutr. 2007;97:273–80.

22. Tholstrup T, Raff M, Straarup EM, Lund P, Basu S, Bruun JM. An oilmixture with trans-10, cis-12 conjugated linoleic acid increases markersof inflammation and in vivo lipid peroxidation compared with cis-9,trans-11 conjugated linoleic acid in postmenopausal women. J Nutr.2008;138:1445–51.

23. Beynen AC, Katan MB. Rapid sampling and long-term storage ofsubcutaneous adipose-tissue biopsies for determination of fatty acidcomposition. Am J Clin Nutr. 1985;42:317–22.

24. Gadbury GL, Coffey CS, Allison DB. Modern statistical methods forhandling missing repeated measurements in obesity trial data: beyondLOCF. Obes Rev. 2003;4:175–84.

25. Henderson AR. The bootstrap: a technique for data-driven statistics.Using computer-intensive analyses to explore experimental data. ClinChim Acta. 2005;359:1–26.

26. Buemann B, Sorensen TI, Pedersen O, Black E, Holst C, Toubro S,Echwald S, Holst JJ, Rasmussen C, et al. Lower-body fat mass as anindependent marker of insulin sensitivity: the role of adiponectin. Int JObes (Lond). 2005;29:624–31.

27. Whigham LD, Watras AC, Schoeller DA. Efficacy of conjugated linoleicacid for reducing fat mass: a meta-analysis in humans. Am J Clin Nutr.2007;85:1203–11.

28. Gaullier JM, Halse J, Hoivik HO, Hoye K, Syvertsen C, Nurminiemi M,Hassfeld C, Einerhand A, O’Shea M, et al. Six months supplementationwith conjugated linoleic acid induces regional-specific fat mass de-creases in overweight and obese. Br J Nutr. 2007;97:550–60.

29. Belury MA, Mahon A, Banni S. The conjugated linoleic acid (CLA)isomer, t10c12-CLA, is inversely associated with changes in bodyweight and serum leptin in participants with type 2 diabetes mellitus.J Nutr. 2003;133:S257–60.

30. Steck SE, Chalecki AM, Miller P, Conway J, Austin GL, Hardin JW,Albright CD, Thuillier P. Conjugated linoleic acid supplementation fortwelve weeks increases lean body mass in obese humans. J Nutr.2007;137:1188–93.

31. Park Y, Albright KJ, Liu W, Storkson JM, Cook ME, Pariza MW. Effectof conjugated linoleic acid on body composition in mice. Lipids.1997;32:853–8.

32. Rexrode KM, Carey VJ, Hennekens CH, Walters EE, Colditz GA,Stampfer MJ, Willett WC, Manson JE. Abdominal adiposity andcoronary heart disease in women. JAMA. 1998;280:1843–8.

33. Forouhi NG, Sattar N, McKeigue PM. Relation of C-reactive protein tobody fat distribution and features of the metabolic syndrome inEuropeans and South Asians. Int J Obes Relat Metab Disord. 2001;25:1327–31.

34. Xing H, Northrop JP, Grove JR, Kilpatrick KE, Su JL, Ringold GM.TNF alpha-mediated inhibition and reversal of adipocyte differentiationis accompanied by suppressed expression of PPARgamma withouteffects on Pref-1 expression. Endocrinology. 1997;138:2776–83.

35. Pischon T, Girman CJ, Hotamisligil GS, Rifai N, Hu FB, Rimm EB.Plasma adiponectin levels and risk of myocardial infarction in men.JAMA. 2004;291:1730–7.

1352 Raff et al.

at DE

F - D

anish Electronic R

es Consortium

on June 29, 2009 jn.nutrition.org

Dow

nloaded from