点击显示 收起
1 From INSERM, Nutriomique, U872 (team 7), Paris, France (NN, KC, MG-M, and SWR); AP-H, Hôtel-Dieu Hospital, Departments of Diabetes and Nutrition, Paris, France (MK, GS, VP, GS, KC, and SWR); University Pierre and Marie Curie-Paris 6, Center of Research of Cordeliers, Paris, France (NN, KC, MG-M, and SWR); the Center of Research on Human Nutrition (CRNH-Ile de France), Paris, France (KC, MG-M, and SWR); INSERM, U870-INRA U1235, Faculty of Medicine Rene Laennec, Lyon, France (EM, SR, and HV); and INRA-INA-PG, UMR 914, Paris, France (AQ-B)
2 This study was presented in part in the 64th ADA meeting, held in Orlando, FL, 2004, and at the 24th International Symposium on Diabetes and Nutrition DNSG of the EASD, held in Salerno, Italy, 2006. 3 Supported by grants from the National Institute of Health and Medical Research (INSERM); the Pierre and Marie Curie University, Paris 6; the Association of Young Subjects with Diabetes (AJD), Paris; and the National Agency of Research (ANR Obcat) and Programme of Research in Human Nutrition (PRNA), France. The fish oil and placebo capsules were provided by Pierre Fabre Médicament (Castre, France). 4 Reprints not available. Address correspondence SW Rizkalla, Department of Nutrition, CRNH-Ile de France, INSERM U872, Hôtel-Dieu Hospital, 1 Place du Parvis Notre-Dame, 75004 Paris, France. E-mail: salwa.rizkalla{at}htd.aphp.fr.
ABSTRACT
Background: Information is lacking on the potential effect of n–3 polyunsaturated fatty acids (PUFAs) on the adipose tissue of patients with type 2 diabetes.
Objective: We evaluated whether n–3 PUFAs have additional effects on adiposity, insulin sensitivity, adipose tissue function (production of adipokines and inflammatory and atherogenic factors), and gene expression in type 2 diabetes.
Design: Twenty-seven women with type 2 diabetes without hypertriglyceridemia were randomly allocated in a double-blind parallel design to 2 mo of 3 g/d of either fish oil (1.8 g n–3 PUFAs) or placebo (paraffin oil).
Results: Although body weight and energy intake measured by use of a food diary were unchanged, total fat mass (P < 0.019) and subcutaneous adipocyte diameter (P < 0.0018) were lower in the fish oil group than in the placebo group. Insulin sensitivity was not significantly different between the 2 groups (measured by homeostasis model assessment in all patients and by euglycemic-hyperinsulinemic clamp in a subgroup of 5 patients per group). By contrast, atherogenic risk factors, including plasma triacylglycerol (P < 0.03), the ratio of triacylglycerol to HDL cholesterol (atherogenic index, P < 0.03), and plasma plasminogen activator inhibitor-1 (P < 0.01), were lower in the fish oil group than in the placebo group. In addition, a subset of inflammation-related genes was reduced in subcutaneous adipose tissue after the fish oil, but not the placebo, treatment.
Conclusions: A moderate dose of n–3 PUFAs for 2 mo reduced adiposity and atherogenic markers without deterioration of insulin sensitivity in subjects with type 2 diabetes. Some adipose tissue inflammation-related genes were also reduced. These beneficial effects could be linked to morphologic and inflammatory changes in adipose tissue. This trial was registered at clinicaltrials.gov as NCT0037.
Key Words: Adiposity • fish oil • type 2 diabetes • women • adipocyte size • PAI-1 • atherogenic index • adipose tissue inflammation-related genes
INTRODUCTION
Since the early 1980s, fish oil consumption has been found to exert beneficial effects on plasma lipids and to be associated with decreased coronary artery disease in nondiabetic subjects (1). These promising effects of fish oil consumption on health were dampened by the discovery of deleterious effects of fish oil on glucose control in subjects with diabetes (2, 3). These effects, however, were mainly observed in response to high doses of fish oil (5–18 g/d) or in studies comprising patients who did not comply with their treatments (3). Two meta-analyses of randomized controlled trials with moderate doses of fish oil finally concluded that fish oil supplementation decreases plasma triacylglycerol in subjects with type 2 diabetes without adverse effects on plasma glucose control (4, 5).
In rodents, additional effects of n–3 polyunsaturated fatty acids (PUFAs) on insulin sensitivity and adipose tissue metabolism and gene expression have been reported. Both in rats and in mice, the intake of n–3 PUFAs reduces adipose tissue mass, which is preferentially located in visceral depots (6-9). Moreover, in insulin-resistant rodents, dietary fish oil ameliorates insulin sensitivity (8, 10) and decreases rates of glucose oxidation (6) and lipolysis in isolated adipocytes (7). Gene expression studies showed the up-regulation of genes encoding mitochondrial biogenesis and oxidative metabolism proteins in the epididymal adipose tissue of mice fed n–3 PUFAs (9). Fish oil may also target adipose tissue secretory factors, because circulating concentrations of leptin and adiponectin and adipose tissue gene expression were modulated in rodents fed n–3 PUFAs (8, 11-13).
In humans, only one preliminary intervention study, in 6 healthy adults, reported a beneficial effect of fish oil on body fat mass (14). Data are lacking regarding the potential effect of n–3 PUFAs on adiposity markers, insulin sensitivity, and adipose tissue gene expression in patients with altered insulin sensitivity or type 2 diabetes.
Increasing evidence indicates that metabolic disorders, including type 2 diabetes, are associated with a low-grade inflammatory state. Additionally, adipose tissue has been recognized to contribute to the production of inflammatory factors (15). The expression of a wide panel of inflammation-related factors is increased in the obese state and is reduced after weight loss, a condition associated with marked improvement of insulin sensitivity (16). A series of recent animal and human studies showed that blood-derived macrophages infiltrate the adipose tissue in obesity and therefore contribute to the inflammatory state of adipose tissue (17-20). Experimental and clinical evidence suggest that fish oil exerts immunomodulatory effects in various pathologic situations, including inflammatory joint and bowel diseases (21) and cancer (22). Fish oil may influence monocyte and macrophage functions (23). Given these observations, we hypothesized that n–3 PUFA supplementation may modify plasma inflammatory markers and inflammatory gene expression in adipose tissue.
Therefore, we conducted the present study to evaluate whether the intake of a moderate dose of n–3 PUFAs, which is generally prescribed in France, might be of benefit on 1) insulin sensitivity, 2) adiposity (fat mass and adipocyte diameter), 3) proteins secreted by adipose tissue (plasma concentrations of adipokines, atherogenic factors, and inflammatory factors), and 4) the expression of a subset of inflammation-related genes in adipose tissue in a homogeneous group of women with well-controlled type 2 diabetes. This is a population with a special adipose tissue distribution that differs from that of men. It has been shown that, for a given waist circumference, women tend to have more abdominal subcutaneous and less visceral adipose tissue than do men independent of age (24).
SUBJECTS AND METHODS
Patients and treatments
Postmenopausal women with type 2 diabetes were recruited from the Diabetes Department outpatient clinic at the Hôtel-Dieu hospital in Paris. We excluded from the study patients with abnormal renal, hepatic, or thyroid function or gastrointestinal disorders. Twenty-nine patients who met the following criteria were initially included in the study: fasting plasma glucose between 7.7 and 14.0 mmol/L, glycated hemoglobin (HbA1c) of 7–10.5%, age from 40 to 60 y, body mass index (BMI; in kg/m2) between 27 and 40, and plasma triacylglycerol <2.5 mmol/L. The clinical and biological characteristics of the women who participated in the study are shown in Table 1. This sample size was calculated after fixing the probability of type 1 error at 0.05 and that of type 2 error at 0.10 for changes in fat mass (25). During the follow-up, 3 patients were excluded from the study, 2 because of acute medical events (severe acute pancreatitis and ischemic cerebral vascular accident) and 1 because of noncompliance with the treatment dose. Three of the remaining subjects were treated by dietary regimen alone, whereas the other 23 patients were taking usual oral hypoglycemic treatments: biguanides alone (7 patients) or bi-therapy with sulfonylureas and biguanides (16 patients). None of the patients had been treated with thiazolidinediones or insulin. Five patients were taking lipid-lowering agents and 6 patients were receiving hormone replacement therapy.
View this table:
TABLE 1. Clinical characteristics of the subjects at the time of screening1
The purpose, nature, and potential risks of the study were explained by a physician, and written informed consent was obtained from each patient. The experimental protocol was approved by the ethics committee of the Hôtel-Dieu hospital, Paris. The trial has been registered in the public trials registry at http://www.clinicaltrials.gov with the following identification: clinicaltrials.gov ID NCT0037.
Dietary follow-up
Before they started the study, the patients were followed on a regular basis. Then, each patient entered a run-in period of 2 mo. Food consumption was individually assessed by a registered dietitian, and each subject received individual counseling. The subjects were asked to follow their usual diet recommendation more strictly with consumption of 55% of their caloric intake as carbohydrates, 15% as protein, and 30% as lipids. Patients were asked to complete a 7-d food diary just before the start of the treatment period. They were recommended to keep their initial caloric intake and nutrient proportions constant throughout the study. To determine compliance with the dietary recommendations, the patients were asked to keep another food diary to be completed the last 7 d of each treatment period. Even if this method (7-d food diary) of measuring food intake might slightly underestimate true calorie intake, the same method was used before and after treatments, and hence the results can be compared. All records were analyzed by a registered dietitian using the computer software program PROFILE DOSSIER V3 (AuditConseil en Informatique Médicale, Bourges, France), with a dietary database made up of 400 foods representative of the French diet as described previously (26).
Study design
The patients were randomly allocated to 2 mo of 3 g/d of either fish oil (containing 1.8 g n–3 PUFAs: 1.08 g eicosapentaenoic acid and 0.72 g docosahexaenoic acid) or placebo (paraffin oil) in a double-blind parallel study design. Fish oil and placebo capsules were provided by Pierre Fabre Médicament (Castre, France). The plasma fatty acid profile was measured at the beginning and at the end of the treatment periods to assess compliance.
At the beginning and at the end of the trial, the subjects were hospitalized for 2 d after they had fasted overnight. During the first day, fasting blood samples were collected to determine plasma concentrations of HbA1c, glucose, insulin, lipids, plasminogen activator inhibitor factor-1 (PAI-1), and several systemic adipokines [leptin, adiponectin, serum amyloid A, interleukin-6, and tumor necrosis factor- (TNF-)] Three blood samples were taken while the subjects were in the fasting state, at 5-min intervals, to measure homeostasis model assessment (HOMA). Body lean and fat mass distributions were measured with a total-body dual-energy X-ray absorptiometry scanner (Hologic QDR-2000; Hologic Inc, Waltham, MA) as described previously (27). Subcutaneous and visceral fat areas were determined by a single-slice computerized tomography (CT) scan (Philips Brillance 16; MedImage System Inc, Memphis, TN) at the L4–L5 disc space level. The second day, clamp studies and a fat biopsy were performed. After the subjects had fasted for 12 h, a sample of abdominal subcutaneous adipose tissue was obtained by needle biopsy with a 14-gauge needle and a 30-mL syringe under local anesthesia with lidocaine 10% without epinephrine. One-half of the biopsy sample was used immediately to measure adipocyte diameter (in a 10-µL sample) and to perform in vitro adipocyte culture. The other half was immediately frozen at –80 °C for subsequent RNA extraction and real-time reverse-transcriptase polymerase chain reaction (RT-PCR) analysis.
Insulin-sensitivity measurements
Estimation of pancreatic β-cell function (insulin secretion) and insulin sensitivity were calculated from repeated fasting plasma insulin and glucose measurements by using HOMA/CIGMA software (28). Moreover, to obtain a better estimate of whole-body insulin sensitivity, a euglycemic hyperinsulinemic clamp study was performed before and after each treatment period as described (26). For technical reasons, only 5 patients in each group were studied. After we administered a priming dose of insulin, the infusion rate was maintained at 6 mU · kg–1 · min–1 for 180 min. Blood samples were withdrawn every 5 min to adjust the glucose infusion and retain plasma glucose at 5.5 mmol/L, then at 10-min intervals for the last 30 min once a steady state had been obtained. Insulin sensitivity was calculated as the amount of exogenous glucose needed to keep euglycemia during the steady state (mg·min–1 · kg–1).
Adipocyte morphology and culture
Adipocytes from subcutaneous periumbilical adipose tissue were immediately isolated by collagenase digestion. For cell size measurements, adipocyte suspensions were then visualized under a light microscope attached to a camera and computer interface. Adipocyte diameters were measured by using PERFECT IMAGE software (Numeris, Orsay, France). Mean diameter was defined as the median value for the distribution of adipocyte diameters of 250 cells. For cell culture, isolated adipocytes were maintained at a density of 5–10 x103 cells/mL in Dulbecco's modified Eagle's medium supplemented with 1% fetal calf serum, 2% bovine serum albumin, and antibiotics for 2 d. The medium was changed daily. After 24 and 48 h, the culture medium was aspirated and frozen at –20 °C for measurement of the different proteins released in the medium.
RNA extraction and amplification
Total RNA from subcutaneous adipose tissue (50–80 mg of frozen tissue) was obtained by using the Rneasy Lipid Tissue MiniKit (Qiagen, Courtaboeuf, France) according to the manufacturer's recommendations. RNA quantity and integrity were measured by using an Agilent 2100 Bioanalyzer (Agilent Technologies, Massy, France). For the microarray study, 1 µg of total RNA was amplified by using the Message-Amp aRNA kit (Ambion, Austin, TX). This amplification procedure is well validated and it has been shown that it does not distort the relative abundance of individual messenger RNAs within an RNA population (29).
Microarray analysis
To define new targets of fish oil metabolic effects, a microarray analysis was performed from adipose tissue samples collected before and after fish oil treatment only. Amplified RNA (10 µg) was labeled by using the CyScribe Post-Labeling Kit (Amersham Biosciences, Munich, Germany) developed for the generation of Cy3- and Cy5-labeled first-strand cDNA probes by the post-labeling (amino allyl) method. The resulting fluorescent cDNA populations were hybridized to an 800-gene PIQOR cDNA Microarray. These 800 genes were involved in the major metabolic pathways and their transcriptional control (lipogenesis, lipolysis, insulin signaling, fatty acid transport and oxidation, glucose metabolism, thermogenesis, and energy metabolism). Known transcription factors, nuclear receptors, and cofactors involved (or suspected to be involved) in the hormonal and nutritional control of intermediate metabolism were also included. We also included all genes previously found to be regulated by insulin during a 3-h euglycemic hyperinsulinemic clamp (29).
Type 1 hybridizations were performed by using the cDNA obtained at baseline (Cy3 labeling) and at 2 mo (Cy5 labeling) in 7 patients in the fish oil group according to the recommended protocol (Internet: http://www.memorec.com/). The 7 slides were scanned with a FLA8000GR microarray scanner (Fuji, Raytest, France). The images were analyzed with GENEPIX PRO 4.1 software (www.moleculardevices.com). Normalization of the signal intensities between Cy3 and Cy5 was performed by using BioConductor (Internet: http://www.bioconductor.org/). Flagged spots and spots with fluorescence intensity below 2.5-fold above the background for both dyes were not taken into account. The log2 (Cy5/Cy3) ratio of the other spots was calculated. To compare results from the different subjects, data from each slide were normalized in log-space to have a mean of zero by using CLUSTER 3.0 software (30). Only the spots that were present at 80% were recovered. Genes with significant changes in expression were identified by using the significance analysis of microarrays (SAM) procedure (available at Internet: http://www-stat.stanford.edu/tibs/SAM/) (31). In total, 450 Unigene clusters were also analyzed by use of SAS statistical software (version 8.0; SAS Institute Inc, Cary, NC) to identify differentially expressed genes.
Quantification of mRNA by using real-time RT-PCR
First-strand cDNA was synthesized from 500 ng of total RNA obtained from adipose tissue biopsy samples with 100 units of Superscript II (Invitrogen, Cergy Pontoise, France) by using oligo (dT) primers and random hexamers (Promega, Charbonnières, France). Real-time RT-PCR was performed on a Light-Cycler instrument (Internet: http://www.roche-applied-science.com/). The cDNA was amplified by using a TaqMan probe approach in a glass capillary tube in a final volume of 10 µL reaction mix containing 2.5 µL, 100-fold diluted cDNA, 1x LightCycler-FastStart DNA Master Hybridization probes (Roche Diagnostics, Meylan, France), 3 mmol/L of MgCl2, and the specific forward, reverse, and Taqman-specific probes (TIB Mol-Biol Syntheselabor, Berlin, Germany). PCR was performed in 50 cycles with 48 s at 95 °C and 10 s at 95 °C followed by 40 s at 60 °C and 60 s at 60 °C. The specificity of amplification was determined by melting curve analysis. The quantification of 18S ribosomal RNA was used for sample normalization.
Biochemical assays
Plasma glucose was measured by the glucose oxidase method (Beckman Fullerton, Palo Alto, CA). Plasma insulin was determined by radioimmunoassay (Linco Research, Inc, St Charles, MO). Plasma triacylglycerol and free fatty acids were measured with Biomérieux kits (Marcy l'Etoile, France); total cholesterol, HDL, and LDL cholesterol were measured with Labintest kits (Aix-en-Provence, France). Plasma PAI-1 was measured with Chromolize/PAI-1 kits (Biopool International, Umea, Sweden). Leptin and adiponectin were determined by using radioimmunoassay kits from Linco Research. Interleukin-6 and TNF- were measured by using enzyme-linked immunosorbent assay kits from R&D Systems Inc (Minneapolis, MN). Plasma serum amyloid A was measured by enzyme-linked immunoabsorbent assay Cyto-screen immunoassay kits (BioSource International, Camarilla, CA). Concentrations of fatty acids in plasma phospholipids were chromatographed as methyl esters on a 30-m fused-silica column and were detected by electron-impact ionization mass spectroscopy.
Statistical methods
The effects of the 2 treatments, fish oil and placebo, were compared by analysis of covariance (ANCOVA) by using the baseline value and treatment as fixed covariates and the 2-mo measurements as the dependent variable. When a significant difference between groups was found, the effect of each treatment in each group between the beginning and the end of treatment was compared further 2 by 2 by use of Student's paired t tests. Concerning the expression of some inflammatory-related genes measured by real-time RT-PCR, a treatment-by-time interaction was found by using a multivariate analysis of repeated measurements. Consequently, the effect of each treatment was analyzed separately in each group by Student's t test for paired data. Relations between variables were analyzed by the nonparametric Spearman's correlation test, and r coefficients are provided. The Kolmogorov-Smirnov test was used to test for a Gaussian distribution. Statistical analysis was performed with XLSTAT (version 2007; Addinsoft, Paris, France), JMP IN (version 5; SAS Institute Inc, Cary, NC), or GraphPad Prism (GraphPad Software Inc, San Diego, CA). A P value < 0.05 was considered significant. Data are expressed as means ± SEMs.
RESULTS
Patient compliance
Patients followed the fish oil and placebo treatments without any reported difficulty. The treatments were well tolerated, and the patients had no complaints or side effects. According to self-report, the subject's lifestyle was unchanged throughout the study. We evaluated the n–3 fatty acid composition of plasma phospholipids. As shown in Table 2, the concentration of both eicosapentaenoic (20:5n-3) and docosahexaenoic (22:6n-3) acids in plasma phospholipids was significantly increased in the fish oil group but not in the placebo group. Accordingly, the n–6:n–3 ratio was significantly decreased by fish oil treatment compared with placebo. These results reflected acceptable compliance with the fish oil treatment. One patient was excluded at the end of the study because of noncompliance with the fish oil treatment. The results of this patient were excluded.
View this table:
TABLE 2. Concentrations of n–3 fatty acids in plasma phospholipids before and after 2 mo of either placebo or fish oil treatment1
Baseline characteristics
There were no detectable significant differences in the clinical and biological variables measured between the patients in the 2 treatment groups at baseline. However, some significant differences were detected in the gene expression values of some inflammation–related markers (measured by real-time RT-PCR). In the 26 overweight women with type 2 diabetes included in the study, significant correlations were found between adipocyte markers (adipocyte diameter and whole fat mass) and the main adipokines (plasma leptin and adiponectin) as well as plasma atherogenic factors (PAI-1, insulin, and triacylglycerol; Table 3). There was no significant correlation between adipocyte diameter and the systemic cytokines (plasma TNF- or plasma interleukin-6). As shown in Table 3, adipocyte diameter (and fat mass percentage) was also correlated with some atherogenic (cysteine protease cathepsin S, or CTSS) and inflammation-related genes (the chemoattractant gene plasminogen activator urokinase receptor, or PLAUR; the macrophage surface marker CD11b; and the macrophage phagocytic activity marker CD68).
View this table:
TABLE 3. Correlations between baseline values of adipose tissue gene expression of some adipokines (leptin and adiponectin) and inflammatory markers on one hand and both adiposity markers and several clinical variables on the other hand1
Effects of intervention
Dietary control, body weight, and adiposity
There was no significant difference between the groups in daily intake of total energy, intake of macronutrients, or fatty acid composition, as estimated by 7-d dietary records (Table 4). Weekly fish intake did not differ significantly between the 2 treatment groups. Similarly, total body weight remained unchanged. However, whole-body fat mass, evaluated by dual-energy X-ray absorptiometry, was reduced by the fish oil treatment compared with placebo when analyzed by ANCOVA with adjustment for baseline (P = 0.02). This reduction in fat mass was mainly due to a decrease in trunk fat (P = 0.04; Table 4). The trunk region covers the part from the shoulders to the hip joints and represents the whole fat mass minus fat in the limbs (arms and legs) and the head. However, a single-slice CT scan at the L4–L5 level could not detect a significant difference in the area of subcutaneous or visceral fat at this specific level between the 2 treatments.
View this table:
TABLE 4. Dietary record data and metabolic and clinical variables at baseline and after 2 mo of either placebo or fish oil treatment1
Adipocyte diameter was also reduced significantly (P = 0.002) by fish oil treatment compared with placebo after adjustment for baseline values according to the ANCOVA. A 6% reduction was observed in the fish oil group (P < 0.0003) but not in the placebo group (Table 4).
Glucose homeostasis
Fasting plasma glucose, insulin, and HbA1c were not significantly influenced by fish oil treatment compared with placebo (Table 4). Similarly, insulin secretion (HOMA-B) and sensitivity (HOMA-S) determined by HOMA remained unchanged in all subjects (n = 26 patients). There was no significant difference in the glucose disposal rate, as assessed by euglycemic hyperinsulminemic clamp in a subgroup (n = 5 patients per treatment) of subjects studied (Table 4) that could be regarded as a pilot study because of the small n. These data show that dietary fish oil supplementation did not deteriorate or ameliorate plasma glucose control and insulin sensitivity in this group of women with type 2 diabetes.
Lipid homeostasis
Plasma triacylglycerol was significantly lower after 2 mo in the fish oil group than in the placebo group (ANCOVA, P < 0.03; Table 4). A 12% reduction (between baseline and 2 mo, P = 0.03) was found in the fish oil but not the placebo group. Plasma total cholesterol, LDL cholesterol, and HDL cholesterol were not significantly affected by fish oil treatment compared with placebo. The atherogenic index was calculated as the logarithm of the ratio of plasma triacylglycerol to HDL cholesterol (32). The atherogenic index was lowered by 2 mo of fish oil treatment compared with placebo (ANCOVA, P = 0.03; Table 4). Plasma free fatty acids tended to decrease after 2 mo of fish oil treatment compared with placebo, without reaching statistical significance (ANCOVA, P = 0.09). Positive correlations were found between adipocyte diameter and fat mass on one hand and plasma triacylglycerol and PAI-1 at the end of 2 mo of fish oil treatment on the other hand (data not shown).
Adipokines, atherogenic factors, and inflammatory factors
PAI-1 was the circulating factor that showed the greatest change after the fish oil treatment. When 2 mo values were adjusted by baseline values, PAI-1 was found to be lower after fish oil treatment than after placebo (ANCOVA, P = 0.01). In contrast, plasma concentrations of leptin, interleukin-6, TNF-, and serum amyloid A were not significantly changed after 2 mo of fish oil treatment compared with placebo (Table 4). Although fish oil treatment did not significantly increase adiponectin concentrations when compared with placebo treatment (ANCOVA, P = 0.14), plasma adiponectin concentrations were negatively correlated with the atherogenic index values (r = –0.44, P = 0.015).
To determine whether fish oil treatment intrinsically alters adipose cell secretory capacity, we measured adipose-secreted factors in a culture medium of adipose cells isolated from adipose tissue biopsies before and after 2 mo of the 2 treatments. Whatever the treatment, levels of PAI-1 activity, leptin, adiponectin, interleukin-6, and TNF- in the media harvested from cells obtained at baseline and after 2 mo were not significantly different (data not shown).
Gene expression analysis in adipose tissue
The gene expression of a subset of adipose tissue-secreted proteins was analyzed by real-time RT-PCR in the 2 treatment groups before and after 2 mo of fish oil treatment (n = 7) or placebo (n = 7). Leptin and adiponectin gene expression (the main adipocyte-secreted hormones) was not significantly affected in either group. Although the PAI-1 circulating concentration was decreased by the fish oil treatment, no significant changes could be detected in adipose tissue PAI-1 gene expression (data not shown).
A series of genes encoding inflammation-related factors was also analyzed by real-time RT-PCR. The choice in selecting these genes was based first on the preliminary data of a cDNA microarray analysis of adipose tissue from fish oil-treated patients only (baseline samples were plotted against 2-mo samples on the same slide). A detailed description of these genes along with their average fold changes are given in Table 5. Among these genes, we selected the gene encoding the matrix metalloprotease MMP9, which was significantly down-regulated by fish oil treatment. The MMPs are involved in matrix modeling and are required for macrophage infiltration. Moreover, the same gene (MM9) and other inflammation-related genes were found to be decreased by 2 mo of dietary fish oil supplementation in long-term insulin-resistant sucrose-fed rats (M Guerre-Millo, N Naour, Y Lombardo, K Clement, and S Rizkalla, unpublished observations, 2007). These genes included the cysteine protease cathepsin S, or CTSS; the chemoattractant gene plasminogen activator urokinase receptor, or PLAUR; the macrophage surface markers CD11b and CD18, which also make up the integrin; and the macrophage phagocytic activity marker CD68.
View this table:
TABLE 5. Genes with differential regulation in the subcutaneous fat after 2 mo of fish oil treatment1
Therefore, we analyzed the same genes by real-time RT-PCR in the 2 treatment groups before and after 2 mo of treatment. A treatment effect (fish oil versus placebo, P = 0.02) was found for MMP9 with a significant decrease when comparing 2 mo with baseline (P < 0.05). Because of a significant (P < 0.05) treatment by time interaction, some genes were analyzed separately in each group and were found to be decreased or to tend to decrease by fish oil (2 mo versus baseline) but not by placebo (Figure 1): CTSS (P = 0.13), PLAUR (P < 0.05), CD11b (P < 0.05), CD18 (P = 0.13), and CD68 (P = 0.11). By contrast, no significant changes in the monocyte chemotactic protein MCP-1 were observed.
FIGURE 1.. Effect of 2 mo of treatment with fish oil or placebo on the expression of inflammation-related genes in subcutaneous adipose tissue. MMP9 gene expression showed a treatment effect (fish oil versus placebo, P = 0.02). There was also a significant difference between the groups at baseline for MMP9. Because of the presence of a significant treatment by time interaction (using a multivariate analysis of repeated measurements) for CTSS, CD68, PLAUR, CD11b, and CD18, values between baseline and 2 mo were compared separately in each group by use of Student's t test for paired data. Data are mean ± SEM for baseline (open bars) and 2 mo (black bars). AU, arbitrary units. *P < 0.05, #P = 0.13 for CTSS and 0.11 for CD68 versus baseline.
DISCUSSION
In the women with type 2 diabetes in the present study, 2 mo of a moderate dose of fish oil, which is commonly recommended in France, induced improvements in adiposity markers and atherogenic factors. These effects were seen with a decrease in plasma triacylglycerol even in nonhypertriacylglycerolemic subjects. Previously, however, this decrease had only been shown in hypertriacylglycerolemic subjects (4, 5, 26). The intervention had no significant effect on total cholesterol or LDL cholesterol. By contrast, HDL cholesterol increased in the fish oil group, in agreement with the results of a few previous studies (33-35). The elevated concentrations of HDL cholesterol and lower concentrations of triacylglycerol, and hence the reduced atherogenic index, induced by 2 mo of fish oil might play an important role in protecting patients with type 2 diabetes against the development of premature atherosclerosis (36, 37).
Another marker of atherogenic risk, the plasma PAI-1 concentration (38), was also significantly reduced by fish oil treatment. In the literature, the effects of n–3 PUFAs on PAI-1 activity or antigen concentrations are divided. In vitro, eicosapentaenoic acid significantly increases PAI-1 secretion from cultured endothelial cells (39).The results of several studies in healthy subjects have been controversial on the favorable or unfavorable effects of n–3 fatty acids on PAI-1 activity (40, 41). In contrast with our data, in subjects with type 2 diabetes, another study showed that PAI-1 activity was increased with n–3 PUFA supplements (42). This difference might be due to the high doses of n–3 PUFAs (3 g) used in the later studies, which resulted in a deterioration in fasting blood glucose concentrations and HbA1c that might in turn have upregulated PAI-1. The decreased plasma PAI-1 concentrations caused by fish oil treatment in our study were not associated with reduced PAI-1 gene expression in subcutaneous adipose tissue. This dissociation between plasma PAI-1 and PAI-1 gene expression agrees with the results of another study showing that a low-calorie diet reduced plasma PAI-1 but not PAI-1 expression in subcutaneous adipose tissue (43). Changes in PAI-1 gene expression in other adipose depots, most notably in visceral adipose tissue, or in other tissues might contribute to the modification of PAI-1 circulating concentrations.
Another interesting aspect of the effects of fish oil shown in the present study is its ability to reduce adiposity. This study is the first intervention study to demonstrate a beneficial effect on adiposity markers in diabetic patients. Although body weight did not differ significantly between the 2 treatment groups, both total fat mass and adipocyte size in subcutaneous abdominal adipose tissue were significantly reduced by 3.5% and 6%, respectively, after fish oil treatment compared with placebo. However, visceral and subcutaneous adipose tissue at only one level, L4–L5, measured with a single-slice CT scan, did not differ significantly between groups. This is not surprising, because the decrease in fat mass after 2 mo of treatment was due mainly to the difference in whole trunk fat and not only to the abdominal fat at this single level. It was recently shown that the use of 8-slice multidetector CT gives better results than does a single-slice CT scan (44). Therefore, the difference between visceral and subcutaneous adipose tissue could not be accurately calculated in our study. The findings of the present study on adiposity agree with studies in rodents. In humans, however, indirect evidence has been found in only 2 studies. One study showed that dietary counseling to substitute dietary saturated fat with polyunsaturated fat in a mixed group of subjects with type 2 diabetes, obese subjects, and healthy subjects resulted in a decrease in subcutaneous, but not abdominal or visceral, fat area (45). Another epidemiologic study showed a correlation between adipocyte size and the n–6 and n–3 fatty acid content in subcutaneous abdominal adipose tissue in a group of overweight patients who had undergone abdominal surgery (46). The present study points out the direct beneficial role of fish oil to lower adiposity in humans.
Several mechanisms can be considered to understand why fish oil treatment compared with placebo induced a selective decrease in fat mass. One possible factor could be decreased energy intake. However, energy intake in the present study did not differ significantly during the 2 treatments. Even if the food diary method often leads to an underestimation of energy intake, the same method was used by the same subjects before and after treatment and thus can be compared. This is strengthened by the absence of a significant change in body weight. Alternatively, decreased fat absorption and increased gastric emptying and gut hormone responses could be implicated in the observed results. Robertson et al (47) showed that ingestion of a meal rich in n–3 fatty acids resulted in rapid gastric emptying in humans. The authors attributed these results to the fact that n–3 fatty acids resulted in both a slower release of cholecystokinin and lower postprandial glucagon-like peptide-1, both of which have been implicated in the nutrient-induced slowing of gastric emptying. Unfortunately, we did not measure gastric emptying in our study. Additional factors, however, might be considered, such as increased lipolysis, which has been previously shown in rat models (7). However, the application of these results to our study at this stage (2 mo of fish oil treatment) is doubtful, because the microarray analysis for the fish oil group in the present study did not show significant modifications in the concerned enzymes.
Another possible mechanism might be enhanced energy expenditure. Recently, Mostad et al (48) showed that n–3 fatty acids alter the proportion of carbohydrate and fat utilization without changing total resting energy expenditure as assessed by indirect calorimetry. There was a relative shift in substrate utilization, because the fish oil group oxidized more fat than carbohydrate after 9 wk than did the placebo group. This hypothesis is likely to be implicated in the results of the present study.
Whatever the mechanism, decreasing total adiposity might increase insulin sensitivity and reduce cardiovascular disease risk factors as has been found in rodents (8, 49, 50). In the present patients with type 2 diabetes, insulin sensitivity and plasma glucose control (plasma glucose and Hb1Ac) were neither deteriorated nor ameliorated by moderate doses of n–3 PUFAs (1.8 g/d). In contrast, the recent study of Mostad et al (48) showed decreased insulin sensitivity in patients with type 2 diabetes after high doses of n–3 PUFAs (5.9 g/d), which agrees with previous results showing the deleterious effects of high doses of fish oil on plasma glucose control (4, 5). Therefore, moderate doses of n–3 PUFAs did not reduce insulin sensitivity (mainly measured by HOMA, and in a pilot clamp study) or plasma glucose control in patients with type 2 diabetes.
The relation between decreasing adiposity and the beneficial effect of fish oil on cardiovascular disease risks could be mediated by a decrease in the inflammatory markers secreted by adipose tissue (51). The current study was not able to demonstrate any alteration in concentrations of systemic inflammatory factors in response to fish oil treatment in patients with type 2 diabetes. Nor was there an intrinsic influence of fish oil treatment on the secretory capacity of isolated adipocytes in vitro. By contrast, fish oil treatment significantly reduced or tended to reduce the expression of some inflammation-related genes in the subcutaneous adipose tissue of patients with type 2 diabetes. Interestingly, adipose expression of some of the chemotactic genes (PLAUR) and markers for macrophage surface (CD11b) and phagocytic activity (CD68) were positively correlated with adiposity markers (adipocyte size and whole body fat mass percentage) at baseline. The parallelism between the down-regulation of these genes and the reduction in adiposity and adipocyte diameter by fish oil treatment suggests a positive relation between adipose cell size and adipose tissue inflammation, which agrees with recent observations in obese subjects (17, 52). The results of the present study in women with type 2 diabetes are strengthened by 2 studies published in rodents during the preparation of the manuscript. Those studies showed that 6 wk of n–3 PUFA supplementation prevents adipose tissue inflammation induced by a high-fat diet (53) and that endogenously biosynthesized n–3 PUFAs in transgenic mice reduce inflammation and tissue injury in colitis (54). Additionally, in a recent study from our laboratory, we showed that the presence of fish oil in the diet of long-term insulin-resistant, sucrose-fed rats decreased adipocyte diameter (8) and significantly reduced several inflammation-related genes: MMP9, CTSS, PLAUR, CD11b and TNF- (Guerre-Millo M, Naour N, LombardoY, Clement K, Rizkalla S, unpublished observations, 2007). These studies suggest that reducing adiposity with fish oil could decrease adipose tissue inflammation and macrophage infiltration. This hypothesis warrants further evaluation, in particular, the modification of these biomolecules at the protein level and the histochemical determination of macrophage infiltration of adipose tissue during long-term fish oil treatment.
In conclusion, 2 mo of a moderate dose of fish oil in women with controlled type 2 diabetes decreased adiposity and the expression of some inflammation-related genes in adipose tissue, without any beneficial or deleterious effect on whole-body insulin sensitivity. Additional beneficial effects of this treatment on cardiovascular disease risk factors could be attributed to a decreased atherogenic index and PAI-1 concentration in keeping with a triacylglycerol-lowering capacity. The beneficial effects of n–3 PUFAs can be linked to local blunting of adipose tissue inflammation. More long-term studies are needed to identify and understand the relation between fish oil and risk factors in diabetes.
ACKNOWLEDGMENTS
We thank the medical and nonmedical staff of the Department of Diabetes for their assistance. We are grateful to Aude Rigoire for dietary counseling of the patients and dietary data analysis. We express our gratitude to Nicolas Zamaria for the opportunity to perform the fatty acids profile in plasma phospholipids in his laboratory, Amine Duval for CT scan realization, Josette Boillot for technical assistance during the different stages of this study, and Marmar Kabir and Florence Combes for gene data management. We also thank Pierre Fabre Médicament (Castre, France) for offering the fish oil and placebo capsules.
The contributions of the authors were as follows—SWR: conceived the overall study, was responsible for planning the study, and drafted the manuscript; MK: was responsible for adipokine and cytokine analysis, adipocyte culture in vitro, microarray and real-time RT-PCR analysis, and an important part of the manuscript writing; GS and VP: the follow-up of patients, performing the clamp procedures, blood glucose and lipid measurements, and the statistical analysis; NN: the gene expression of the inflammation-related genes and their statistical analysis; HV, EM, and SR: contributed to designing the microarray analysis; MG-M and KC: contributed to evaluating the results, designing the gene expression section, and participating in manuscript writing; AQ-B: determining adipocyte size and advising on adipocyte culture design in vitro. All contributors helped with the revision of the paper. None of the authors had a conflict of interest.
REFERENCES