Treatment With Dietary trans10cis12 Conjugated Linoleic Acid Causes Isomer-Specific Insulin Resistance in Obese Men With the Metabolic Syndrome

Sixty Caucasian men (35–65 years old) with signs of the metabolic syndrome (abdominal obesity, insulin resistance, dyslipidemia, and hypertension) (15) were recruited from the local community through media advertisements. Because subjects with combined abdominal obesity and dyslipidemia are nearly always insulin resistant (8,16), both of these disorders were included in the inclusion criteria: waist girth >102 cm, waist-to-hip ratio >0.95, BMI 27–39 kg/m², triglycerides (TG) >1.7 mmol/l, and/or HDL cholesterol <0.9 mmol/l and stable body weight for the preceding 3 months. No one had diagnosed heart, liver, or renal disease or diabetes. Six subjects had fasting plasma glucose ≥7.0 but ≤7.2 mmol/l, indicating mild diabetes; all men had HbA_1c ≤5% (Table 1). All subjects gave written consent. The protocol was approved by the Ethics Commitee of the Medical Faculty, Uppsala University.

The study was performed after a 4-week run-in screening period. Outcome measures included insulin sensitivity; fasting levels of glucose, lipids, and leptin; and body composition. Subjects were randomly assigned to 3.4 g/day of CLA (isomer mixture), purified t10c12 CLA, or placebo. Six capsules were taken daily with morning and evening meals. The isomer content of the CLA preparation (80% free fatty acids [FFAs]) was 35.9% t10c12 CLA, 35.4% c9t11 CLA, 13.1% 18:1c9, 7.2% 16:0, 2.4% 18:0, and 1.3% 18:2c9c12; that of the t10c12 CLA preparation (75% FFAs) was 76.5% t10c12 CLA, 11.4% 18:1c9, 3.4% 16:0, 2.9% c9t11 CLA, 2.0% 18:2c9c12, and 0.8% 18:0. All capsules (identical in appearance) were prepared by Natural Lipids (Hovebygda, Norway), which also assessed isomer separation using gas chromatography (17) (possible isomerization products from the t10c12 CLA preparation were very low as identified on the chromatogram) and generated the randomization numbers for patients and incorporation into double-blind labeling. The solution code was broken after the study was completed.

All measurements were obtained in the morning, after instructions to fast (12 h) and refrain from smoking, taking snuff (nicotine-containing moist), or engaging in physical activity in the morning and to avoid alcohol and exercise the day before visits. Subjects completed a questionnaire concerning diabetes heredity (having at least one first-degree relative with type 2 diabetes) and the use of medication or dietary supplements. All men were encouraged to maintain their usual diet and exercise habits during the study. To assess possible changes in dietary intake during the study, a 3-day weighed-food record was completed during weeks 1 and 8.

The sagittal abdominal diameter (SAD) was measured at L4–5 level, and waist and hip girth were measured as previously described in detail (7). Anthropometrics were measured by a single investigator. Bioelectrical impedance analysis (BIA) was done with a multifrequency analyzer (Xitron Technologies, San Diego, CA). From the estimation of body water, body fat content was calculated based on the assumption that fat-free mass contains 73.2% water (18). Lean body mass was calculated with the formula provided by the manufacturer.

A 120-min hyperinsulinemic clamp was conducted to determine insulin sensitivity, as described by Defronzo et al. (19) with slight modification previously described in detail (20). Insulin (Actrapid Human; Novo, Copenhagen, Denmark) was infused (336 pmol/l · m⁻² · min⁻¹) resulting in a mean steady-state insulin level of 624 pmol/l, a level shown to almost completely suppress hepatic glucose production in insulin-resistant and type 2 diabetic subjects (21). Venous blood sampling was obtained in the left hand, which was kept warmed to provide arterialized blood. Insulin and glucose were infused in the right antecubital vein. Plasma glucose was assayed in duplicate in a Beckman Glucose Analyzer II (Beckman Instruments, Fullerton, CA), using an enzymatic method. Glucose disposal (M) was calculated as the glucose infusion rate (mg · kg⁻¹ · min⁻¹) during the last 60 min of the clamp.

Venous blood was drawn into vacuum tubes, coagulated, and centrifuged at room temperature and then frozen at −20°C. Lipoproteins were isolated from fresh serum by a combination of preparative ultracentrifugation (22) and precipitation with a sodium phosphotungstate and magnesium chloride solution (23). Serum lipoproteins and TG were assayed by enzymatic techniques using a Monarch 2000 centrifugal analyzer (Instrumentation Laboratories, Lexington, MA). Serum samples were stored at −70°C. Serum free glycerol and FFAs were measured using an enzymatic colorimetric method (Boehringer Mannheim and Wako Chemical, respectively) in the Monarch centrifuge. Samples from each subject were analyzed within the same run. Plasma insulin was measured by an enzyme immunosorbent assay (ELISA) kit (Mercodia AB, Uppsala, Sweden) in a Bio-Rad Coda automated EIA analyzer (Bio-Rad Laboratories, Hercules, CA). Serum leptin was measured using a human leptin radioimmunoassay kit (Linco Research, St. Charles, MO).

Values are expressed as means ± SD. Variables with skewed distributions were logarithmically transformed before analysis. A nonparametric test was used if data were not normally distributed after logarithmic transformation. Paired t test was used for within-group effects from baseline. Differences between groups from baseline to 12 weeks were assessed using overall test (ANOVA or nonparametric test). In case of a significant overall test result, ANOVA or Mann Whitney’s nonparametric test was used for differences between two groups. Pearson’s correlation coefficient was determined. It was estimated that n = 20/group would be needed to detect a 20% change in cholesterol levels with a power of 0.80 at a significance level of 0.05. All tests were two-tailed. P < 0.05 was regarded as significant. JMP software (SAS Institute, Cary, NC) was used.

Baseline characteristics and diabetes heredity were similar in all groups (Table 1). Of 60 men, 58 completed the study. Reasons for withdrawal were gastrointestinal symptoms (CLA group) and weight gain (t10c12 CLA group). One patient (placebo) was hypertensive and treated (candesartan) during the trial. This patient completed the trial, but his data were excluded from statistics; results were not influenced if his data were included. Data are based on 57 subjects with complete follow-up data. For M, n = 56 because in one subject only insulin values were available. Supplements were generally well tolerated, with only minor transient gastrointestinal problems reported. No adverse events occurred, and no changes in liver enzymes occurred (data not shown). Compliance (capsule count) did not significantly differ between groups (89.5%, n = 57). No significant changes in dietary intake occurred during the study (Table 4).

There were no significant differences between groups in body weight, BMI, body fat, lean body mass, or waist girth at 12 weeks (Table 2). CLA preparations did not decrease weight or change body fat, BMI, or waist girth compared with placebo, although these variables significantly decreased within the t10c12 CLA group, whereas SAD and body fat decreased within the CLA group (paired t test). SAD tended to decrease more with both CLA treatments than with placebo (P = 0.07) (Table 2).

Insulin sensitivity (M) decreased significantly in the t10c12 CLA group compared with placebo but not compared with CLA (Table 3). Also, when correcting M for the mean plasma insulin level during the clamp (M/I = M/pmol/l × 100) (19), insulin sensitivity decreased (−19%; P < 0.01) after t10c12 CLA treatment (data not shown). Fasting insulin increased in the t10c12 CLA group (P < 0.05) but not differently from placebo (Table 3). The significant reduction in M after t10c12 CLA treatment was not affected by adjustment for age or changes in glucose levels, body fat, BMI, or abdominal fat. Baseline adjustments for HbA_1c, diabetes, or diabetes heredity did not affect significance, nor did exclusion of six subjects with plasma glucose ≥7.0 mmol/l. Of all variables, only adjustment for changes in VLDL TG abolished the significant change in M after t10c12 CLA treatment. Fasting glucose increased after t10c12 CLA compared with placebo (P = 0.0009) (Table 3), and HbA_1c decreased after t10c12 CLA treatment versus baseline (P = 0.01) but was not significantly different from placebo.

The changes in serum lipids or leptin did not differ, except for HDL cholesterol, which decreased with both CLA and t10c12 CLA treatment (P = 0.03 and P = 0.006, respectively) versus placebo (Table 3). VLDL TG tended to increase with t10c12 CLA versus placebo and CLA (P = 0.06).

The changes in M in the t10c12 CLA group correlated with the changes of VLDL TG (r = −0.74, P = 0.0003) but not with other variables. The decrease in HDL cholesterol in both CLA groups was correlated only to a change in leptin (r = 31, P = 0.02), which remained significant when including all subjects.

This randomized placebo-controlled trial has revealed unexpected metabolic actions by conjugated fatty acids in humans—actions that seem isomer-specific. The t10c12 CLA isomer, but not a CLA mixture, significantly increased insulin resistance, fasting glucose, and dyslipdemia in abdominally obese men. Such men are prone to develop type 2 diabetes and are considered a possible target group for the putative beneficial effect of CLA. Commercial CLA had no metabolic benefit, which is consistent with previous data (7,9,12). Instead, t10c12 CLA might be diabetogenic in the metabolic syndrome.

Previous human studies have used CLA mixtures, which do not provide separate information regarding t10c12 CLA, although in a study on healthy women, insulin levels tended to increase after treatment with a CLA mixture containing 23% t10c12 CLA (12), indicating impaired insulin action. However, unlike other studies, diet and physical activity were controlled using metabolic suites in that study, and the relative amount of c11t13 CLA was higher and c9t11 CLA was lower, which should be taken into account when comparing that study with our data.

The current results are dissimilar from the antidiabetic effects of CLA in male ZDF rats (4,5) but are in agreement with the effects in female C57BL/6J mice that became severely insulin resistant and lipodystrophic after receiving a CLA mixture containing 36% t10c12 CLA (6). Similarly, male AKR/J mice showed signs of insulin resistance after CLA treatment (24). Whether insulin resistance in CLA-fed mice shares a common mechanism with the t10c12 CLA-induced insulin resistance is unclear. Thus, the mechanism of current insulin resistance is unknown, but some animal data are worth considering. An intriguing speculation is that CLA induces adipocyte apoptosis as shown in mice (6) and in vitro (6,25,26), and by the t10c12 isomer in particular (26). t10c12 CLA might also inhibit the formation of new, small, and insulin-sensitive fat cells (27), possibly via downregulated peroxisome proliferator-activated receptor (PPAR)-γ (6). Other mechanisms for a t10c12 CLA-induced insulin resistance might involve impaired cell membrane function (28), possibly via increasing intramuscular fat content (29). Insulin resistance appeared to occur mainly in the peripheral tissues rather than in the liver, as serum IGF binding protein (IGFBP)-1, a marker of hepatic insulin sensitivity (30), did not change (unpublished data). To elucidate the mechanism, it is probably necessary to investigate insulin-sensitive tissue directly, which is more complicated in humans than in mice. Our primary goal, however, was to investigate the clinical effects of CLA in an insulin-resistant phenotype.

Neither CLA preparation decreased weight or changed body composition compared with placebo. SAD tended to decrease after CLA treatment, but the difference was not significant versus placebo. The standard deviations for changes in anthropometrics were large, and the measurement errors for SAD, BIA, and waist girth have to be considered when interpreting the data.

Adjustment for several variables did not affect the t10c12 CLA-induced insulin resistance, with one exception. Adjusting for VLDL TG abolished statistical significance, reflecting the well-known tight relation between these two variables (31). It is likely that impaired insulin action after t10c12 CLA treatment preceded dyslipidemia (31), but this remains to be proven.

HDL cholesterol decreased after treatment with both CLA forms, although the effect was more pronounced with t10c12 CLA. This was also observed in overweight, normolipidemic subjects using a CLA mixture (10), but not in normal-weight subjects (9,32). Because low HDL cholesterol is an independent cardiovascular risk factor (33), the current reduction of 0.04 mmol/l (−4%) with t10c12 CLA is of clinical concern, as a change of 0.026 mmol/l has been inversely related to coronary risk (33). Interestingly, the current changes in HDL cholesterol were not related to insulin resistance, but were positively related to leptin levels.

Further, the t10c12 CLA-induced insulin resistance is impressive. Whereas the concurrent increase in fasting glucose was clinically irrelevant, the 19% reduction in insulin sensitivity was large enough to have clinical relevance (34), especially considering the prediabetic high-risk group studied. Estimated from dose-response clamp studies in obese and lean control subjects, the relative decrease in insulin sensitivity after t10c12 CLA treatment was equivalent to having an excess body weight of ∼15 kg at a current mean weight of ∼100 kg (35), indicating a rather powerful effect of this isomer. The amounts of t10c12 CLA in the diet (where the major isomer is c9t11 CLA) are very low, but commercial CLA mixtures do contain ∼20–45% t10c12 CLA, indicating that a long-term use might be of concern. Considering that nonprescription “weight loss agents” such as CLA are used by a significant number of people (36), the present results should be of interest for health care professionals treating obese patients. However, more clinical trials are needed to make firm conclusions regarding the clinical safety of CLA isomers and mixtures in obese patients.

In summary, t10c12 CLA increased insulin resistance and dyslipidemia in men with the metabolic syndrome, in agreement with data in CLA-fed mice (6,24) but unlike data in ZDF rats (4), suggesting important isomer-specific metabolic effects of CLA in humans.

This study was supported by the Swedish Medical Research Council (no. 27X-13083, 03X-1034-35C, 04224), Swedish Society for Nutrition Research, Swedish National Fund for Industrial and Technical Development, Swedish National Association against Heart and Lung Disease, and Swedish Diabetes Foundation. We thank Natural Lipids Ltd. AS, Norway, for supplying the CLA preparations.