The costs of publication of this article were defrayed, in part, by the payment of page charges. This article must, therefore, be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
EA3502, Laboratoire de Nutrition, Service de Nutrition Hôtel Dieu, 1, place du Parvis Notre Dame, 75181 Paris Cedex 04, France. E-mail: email@example.com
Objective: Adipocytes secrete a series of acute phase proteins including serum amyloid A (SAA); the link with metabolic status is unknown. We studied the variations of expression of the SAA gene in adipose and liver tissues and of SAA serum levels, as well as their relationships with metabolic features during weight loss.
Research Methods and Procedures: Plasmatic variations of SAA, inflammatory markers (high sensitivity C-reactive protein, interleukin-6, fibrinogen, and orosomucoid), and adipokines (adiponectin, leptin) were studied in 60 morbidly obese patients before and after gastric surgery. For 10 subjects, SAA mRNA expression was measured at baseline in subcutaneous white adipose tissue (scWAT) and visceral white adipose tissue (vWAT) and in the liver. The evolution of SAA mRNA expression was also studied after surgery in scWAT.
Results: SAA serum concentration displayed a significant reduction 3 months after surgery and remained stable beyond 6 months. mRNA expression of inducible SAA isoforms (SAA 1 and 2) in scWAT was higher than in vWAT (p = 0.01) and the liver (p < 0.01) and correlated significantly with BMI, SAA, and high sensitivity C-reactive protein serum concentrations but not with metabolic markers (glucose, insulin, lipid parameters, adiponectin). SAA serum level and its variation during weight loss significantly correlated with adiposity markers (BMI and adipocyte volume, p < 0.01) and inflammatory markers but not with variations of metabolic parameters. The variations of SAA expression in scWAT after surgery correlated with changes in BMI and SAA protein serum levels (p < 0.05).
Discussion: SAA can be considered as a marker of adiposity-induced low-grade inflammation but not of the metabolic status of obese subjects.
Adipose tissue is a multifunctional organ with a role not only in fat storage but also in the production of numerous molecules including the so-called “adipokines” (1) involved in the regulation of energy homeostasis, glucose, lipid metabolism, and inflammation (2, 3, 4). Adipokines contribute to the low-grade inflammation state observed in obese subjects and possibly to the development of obesity-related comorbidities such as insulin resistance, metabolic syndrome, and atherogenesis (5, 6, 7, 8). Several studies have shown that weight loss is associated with an improvement of the inflammatory profile both in the circulation (9) and in the subcutaneous white adipose tissue (scWAT)1 itself (10). The relative contribution of adipocytes and non-adipose cells of adipose tissue to the production of inflammatory molecules is debated. A wide proportion of these inflammatory molecules, notably those modulated by weight variation, are expressed in the stroma vascular fraction. We (10) and others (11) have shown that serum amyloid A (SAA) is among the molecules preferentially expressed and secreted by human mature adipocytes. An excess of SAA secretion by adipose cells (11, 12) may be responsible for increased circulating SAA in obese subjects compared with normal weight subjects (12, 13) and could contribute to some obesity-related complications such as atherosclerosis and thrombosis (14, 15). Several observations suggest that SAA could be implicated in glucose and lipid metabolism (16, 17, 18, 19). After moderate weight loss, a decreased expression of SAA genes in scWAT of obese subjects is observed along with a diminution of serum SAA levels (11, 12).
To our knowledge, the relationship between SAA and metabolic status in humans has not yet been studied. The aim of this study was to analyze the link among SAA, lipid, and glucose homeostasis at baseline and during acute weight loss in a large group of subjects. We chose the model of gastric surgery in morbidly obese subjects where major weight loss is associated with improvement of glucose metabolism and of inflammatory biomarkers.
Here we studied 1) the relationships between SAA circulating levels and markers of inflammation and of insulin or lipid metabolisms at baseline and after surgically induced weight loss and 2) the relationships between SAA gene expression in liver and adipose tissues and the metabolic status of obese subjects.
Research Methods and Procedures
Subjects and Study Design
This study enrolled a total of 60 unrelated morbidly obese subjects (BMI ≥ 40 kg/m2) prospectively recruited between 2002 and 2004 at the Department of Nutrition of the Hôtel-Dieu Hospital (Paris, France) to study the variations of adiposity signals and of inflammatory markers after gastric surgery as previously described (20). Subjects were weight stable for at least 3 months before surgery (laparoscopic adjustable gastric banding or gastric bypass) performed at the Department of Surgery of Hôtel-Dieu hospital. Patients were excluded if they had evidence of current acute or chronic inflammation or infectious diseases. In all patients, clinical and biochemical parameters were assessed at baseline and at 1 year after gastric surgery (mean follow-up period: 13.2 ± 0.4 months; Table 1).
Table 1. . Clinical and biochemical characteristics in morbidly obese subjects before and 1 year after gastric surgery
Data represent means ± SE. p values were obtained using a Student's t test. QUICKI, Quantitative Insulin Sensitivity Check Index; SAA, serum amyloid A; TNF-α, tumor necrosis factor α IL, interleukin; hsCRP, high sensitivity C-reactive protein.
Difference between value after surgery and value before surgery.
Comparison between obese subjects before and after weight loss:
p < 0.0001 and
p < 0.01.
Fat mass was available in only 48 subjects.
Adipocyte volume was measured after surgery in only 19 patients.
For a subset of 20 patients, all of the clinical and biochemical data were collected at 3, 6, and 12 months after surgery. Body composition was measured in 48 subjects by DXA (QDR 2000; Hologic France, Massy, France). The weight of the remaining 12 subjects (>150 kg) was too high for DXA evaluation.
Visceral white adipose tissue (vWAT), scWAT, and liver specimens were obtained at baseline during the surgical intervention in 10 of the morbidly obese individuals (Table 2). Three months after gastric surgery, additional scWAT samples were obtained, after overnight fasting, by needle biopsy from the periumbilical area under local anesthesia (1% xylocaine). We checked that biopsy performance (surgical vs. needle biopsy) did not affect the reverse transcriptase-polymerase chain reaction (RT-PCR) results. The adipose tissue specimens were immediately frozen into liquid nitrogen and stored at −80 °C for total RNA preparation.
Table 2. . Clinical and biochemical characteristics of 10 obese subjects before and 3 months after gastric surgery
Data represent mean ± SE. p values were obtained using a Student's t test. QUICKI, Quantitative Insulin Sensitivity Check Index; SAA, serum amyloid A; TNF, tumor necrosis factor; IL, interleukin; hsCRP, high sensitivity C-reactive protein.
Difference between value after surgery and value before surgery.
Venous blood samples were collected in the fasting state for determination of routine biochemical parameters and evaluation of circulating serum SAA, leptin, adiponectin, interleukin 6 (IL-6), tumor necrosis factor-α (TNF-α), high sensitivity C-reactive protein (hsCRP), fibrinogen, and orosomucoid. In addition, a control group was used including 14 unrelated normal weight women (weight: 55.5 ± 1.6 kg; age: 40.6 ± 2.3 years; BMI 20.52 ± 0.39 kg/m2) who were volunteers. None of the subjects had a familial or personal history of diabetes or were taking medication.
All clinical studies were approved by the Ethics Committees of Hôtel-Dieu (Paris). Informed consent was granted by all obese and normal weight subjects.
Total RNA Extraction
Total RNA was prepared using the RNeasy total RNA Mini kit (Qiagen, Courtaboeuf, France), following the manufacturer's protocol. The concentration of total RNA was determined using Ultrospec 2000 spectrophotometer (Pharmacia Biotech, Piscataway, NJ), and the integrity of the RNA was assessed using a 2100 Bioanalyzer (Agilent Technologies, Massy, France).
Real-time Quantitative PCR
We assessed gene expression changes by reverse transcription and real-time PCR, as described elsewhere (21). We used 18S ribosomal RNA (Ribosomal RNA Control TaqMan Assay kit; Applied Biosystems, Foster City, CA) as a normalization control. The primers and TaqMan probes for mRNA were also obtained from Applied Biosystems. These probes were labeled with a reporter dye (6-carboxyfluorescein) on the 5′ end. The probe for 18S ribosomal RNA was labeled with the reporter dyes VIC and 6-carboxy-tetramethylrhodamine on the 5′ end and the 3′ end, respectively. For each primer and probe pair, a standard curve was obtained using serial dilutions of human adipose tissue cDNA before mRNA quantification.
Blood samples were collected after an overnight fast of 12 hours. Glycemia was measured enzymatically. Serum insulin concentrations were measured using a commercial IRMA kit (Bi-INSULINE IRMA; cisBio International, Bagnols-sur-Cèze, France). SAA concentrations were measured using the enzyme-linked immunosorbant assay Cytoscreen Immunoassay kit (BioSource International, Camarilla, CA) that interacts with the inducible SAA1 and SAA2 isoforms. The sensitivity was 5 μg/mL. Intra-assay and inter-assay coefficients of variation (CV) were 7.7% and 10.8%, respectively, at 80 μg/mL. Serum leptin and adiponectin were determined using a radioimmunoassay kit from Linco Research (St. Louis, MO) according to the manufacturer's recommendations. The sensitivities were 0.5 and 0.8 μg/mL for leptin and adiponectin, respectively. Intra-assay and inter-assay CVs were <4% and 9% for leptin and adiponectin, respectively. Serums levels of IL-6 and TNF-α were measured by an ultrasensitive enzyme-linked immunosorbant assay system (Quantikine US; R&D System Europe Ltd., Abingdon, United Kingdom). The sensitivities of these assays were <0.04 and 0.12 pg/mL for IL-6 and TNF-α, respectively. Intra-assay and inter-assay CVs were <8% for IL-6 and 8.8% and 16%, respectively, for TNF-α. hsCRP and orosomucoid were measured with an IMMAGE automatic immunoassay system (Beckmann –Coulter, Fullerton, CA). The sensitivities were 0.02 and 35 mg/dL, respectively. Intra-assay and inter-assay CVs were <5% and 7.5%, respectively, for hsCRP and 4% and 6% for orosomucoid. Serum levels of ferritin and transferrin were measured with an Olympus AU 600 (Olympus Diagnostica, Lismeehan, Ireland). The sensitivities were 10 μg/L and 0.75 g/L, respectively. Intra-assay and inter-assay CVs were <2.3% and 3.4%, respectively, for ferritin and 0.6% and 1.0% for transferrin. Serum levels of fibrinogen were measured with a Star Diagnostica Stago by Fibri prest (Parsippany, NJ).
Insulin Sensitivity Calculation
Insulin sensitivity was evaluated using the quantitative insulin sensitivity check index (QUICKI) method, which has been shown to be well correlated with the hyperinsulinemic-euglycemic clamp method, considered as the reference method. Calculations were performed from fasting glucose and insulin as described elsewhere (22).
Data are expressed as means ± SE. The Shapiro-Wilk test was used to test the Gaussian distribution of variables under study. Skewed variables were log-transformed to normalize their distribution. Differences in means of clinical, metabolic, and biochemical characteristics were determined by paired Student's t test for data within the obese group before vs. after surgery and ANOVA for data between normal weight and obese groups. Pearson's correlation coefficients were calculated to assess relationships between clinical and biochemical parameters before and after surgery.
For gene expression as analyzed by real-time quantitative PCR, differences (ratios before/after surgery; vWAT/scWAT; liver/scWAT) were compared with one and tested using the Wilcoxon rank test. The correlations between mRNA levels and the clinical and biochemical parameters were examined by the non-parametric Spearman's rank correlation test. A p value <0.05 was considered significant. Statistical analysis was performed with JMP statistics software (SAS Institute, Cary, NC).
Clinical and Biological Parameters at Baseline
Table 1 shows the characteristics of obese subjects at baseline. Obese subjects had higher circulating concentrations of insulin, leptin, hsCRP, IL-6, and TNF-α and lower circulating concentrations of adiponectin (mean insulin, 5.0 ± 0.4 μU/mL; mean leptin, 11.3 ± 1.0 ng/mL; mean hsCRP, 0.1 ± 0.04 mg/dL; mean IL-6, 0.8 ± 0.1 pg/mL; mean TNFα, 1.0 ± 0.05 pg/mL; mean adiponectin, 11.5 ± 1.1 μg/mL; p < 0.001) compared with normal weight subjects. SAA circulating levels were significantly higher in obese (range: 2.6 to 162.7 μg/mL) than in normal weight subjects (mean, 7.1 ± 1.6 μg/mL; range: 2.4 to 20.4 μg/mL; p < 0.0001).
In morbidly obese subjects, SAA circulating concentrations were correlated positively with BMI (r = 0.35, p = 0.009) and adipocyte cell volume (r = 0.46, p = 0.003 adjusted for BMI) and negatively with triglycerides (r = −0.29, p = 0.03). This last correlation disappeared after adjustment for BMI. Basal SAA circulating levels did not correlate with glycemia, insulin, QUICKI, high-density lipoprotein-cholesterol, and total cholesterol. In contrast, after adjustment for BMI, SAA strongly correlated with IL-6 (r = 0.56, p ≤ 0.0001), hsCRP (r = 0.65, p ≤ 0.0001), fibrinogen (r = 0.39, p = 0.009), and orosomucoid (r = 0.54, p ≤ 0.0001). No correlation was observed between SAA and TNF-α ferritin, transferrin, leptin, and adiponectin.
Effect of Weight Loss on SAA and Inflammatory Proteins
The changes after gastric surgery in clinical and biological parameters as well as in markers of inflammatory profile are shown in Table 1.
Surgery resulted in drastic weight loss, with a significant decrease in BMI associated with a significant improvement in glucose homeostasis evaluated by fasting glycemia, insulinemia, and the QUICKI (Table 1). The weight loss reduced the average level of SAA by 19% (p < 0.0001). One year after surgery, mean levels of circulating leptin and IL-6 were decreased, whereas mean levels of adiponectin were increased. Other acute phase response proteins such as hsCRP, fibrinogen, and orosomucoid decreased with weight loss, whereas ferritin and transferrin did not change significantly.
Changes in SAA concentrations correlated with the variation in BMI (r = 0.35, p = 0.006) and with changes in the inflammatory markers, hsCRP (r = 0.36, p = 0.005), IL-6 (r = 0.51, p ≤ 0.0001), and orosomucoid (r = 0.30, p = 0.03; Figure 1). Adjustment for BMI did not change the correlations. The variation in serum SAA correlated negatively with the variation in triglycerides (r = −0.37 p = 0.009), but the correlation disappeared after adjustment for BMI variation. No correlation was found with other parameters of lipid (total cholesterol and high-density lipoprotein-cholesterol) or glucose metabolism (fasting glycemia or insulin and QUICKI).
In a subset of 20 subjects, we explored the effect of weight loss 3, 6, and 12 months after surgery (Figure 2). BMI decreased continuously from 50.0 ± 1.7 to 41.1 ± 2.6 kg/m2 at 3 months, 40.2 ± 1.6 kg/m2 at 6 months, and 36.3 ± 1.5 kg/m2 at 12 months (p < 0.001; Figure 2A). As shown in Figure 2B, SAA levels displayed an initial strong reduction in the first 3 months postoperatively and remained stable after 6 months. SAA concentration decreased from 36.3 ± 3.6 μg/mL, by 55.1% to 16.3 ± 1.4 μg/mL at 3 months and by 64.1% to 13.0 ± 1.0 μg/mL at 6 months (p < 0.001). Similarly, the reduction in hsCRP levels was most pronounced in the first 3 months (−42.4%, p < 0.05; Figure 2D). In contrast to SAA and hsCRP, plasma IL-6 levels remained stable for the first 3 months after surgery and started to decrease significantly at 12 months by 42.5% from 2.47 ± 0.17 to 1.42 ± 0.07 pg/mL (Figure 2C).
SAA mRNA Expression in the scWAT, vWAT, and Liver
We compared the level of expression of SAA in scWAT and vWAT, as well as in the liver, of 10 morbidly obese subjects and analyzed their relationships with clinical and biochemical parameters.
mRNA expression of the three SAA isoforms was higher in scWAT than in vWAT (vWAT/scWAT ratios: 0.76 ± 0.07, p = 0.03; 0.24 ± 0.04, p = 0.01; and 0.29 ± 0.08, p = 0.01, respectively, for SAA1, SAA2, and SAA4). The mRNA expression of the inducible SAA1 and SAA2 mRNA isoforms was higher in scWAT than in the liver (liver/scWAT ratios: 0.35 ± 0.17, p = 0.03; and 0.33 ± 0.14, p = 0.003 for SAA1 and SAA2, respectively). In contrast, the constitutive isoform SAA4 was highly expressed in the liver (liver/scWAT ratio: 61.2 ± 15.3, p = 0.002). No correlation was found between SAA mRNA expression in the liver and BMI.
No significant relationship was found between liver expression of the two inducible SAA isoforms and SAA circulating level (ρ = 0.55, p = 0.1 and ρ = 0.56, p = 0.1, respectively, for SAA1 and SAA2). Trends toward correlations were found between mRNA expression of all isoforms and hsCRP (ρ = 0.58, p = 0.04 for SAA1; ρ = 0.46, p = 0.08 for SAA2; and ρ = 0.57, p = 0.06 for SAA4).
In scWAT expression of SAA, inducible isoforms (SAA1 and SAA2) correlated with weight, BMI, SAA circulating levels, and hsCRP (Table 3). mRNA expression of the constitutive isoform SAA4 correlated only with hsCRP. No correlation was found between SAA mRNA expression and metabolic parameters.
Table 3. . Correlation between SAA expression in scWAT and clinical and biochemical parameters evaluated with Spearman's correlation coefficient
scWAT SAA1 expression
scWAT SAA2 expression
scWAT SAA4 expression
SAA, serum amyloid A; scWAT, subcutaneous white adipose tissue; hsCRP, high sensitivity C-reactive protein.
0.61 p = 0.04
0.63 p = 0.03
0.50 p = 0.1
0.75 p = 0.007
0.75 p = 0.007
0.60 p = 0.06
SAA circulating levels
0.72 p = 0.01
0.58 p = 0.05
0.43 p = 0.18
0.63 p = 0.03
0.8 p = 0.003
0.77 p = 0.005
Effect of Weight Loss on SAA mRNA Expression in Adipose Tissue and Relationships with Variations in Metabolic Status
The decrease of SAA circulating levels was maximal 3 months after surgery. We examined the changes in SAA expression at this time. Weight loss was also associated with a significant reduction of SAA mRNA isoform expression (p < 0.001 for SAA2 and SAA4 and p = 0.05 for SAA1) as expected (Figure 3). The decrease in SAA mRNA expression in scWAT correlated with the decrease in BMI (0.58, p = 0.09; 0.75, p = 0.01; and 0.70, p = 0.05, respectively, for SAA1, SAA2, and SAA4 isoforms). SAA mRNA levels of inducible isoforms correlated with SAA circulating levels (0.53, p = 0.05; and 0.75, p = 0.01, respectively, for SAA1 and SAA2). No significant correlation was found between the changes in SAA mRNA expression and the variations in metabolic parameters especially markers of insulin resistance.
In this study, we examined in morbidly obese subjects the relationship between SAA and metabolic parameters during surgically induced weight loss associated with a major improvement in insulin resistance. Toward this aim, we studied the relationships among SAA circulating levels, SAA expression in adipose and hepatic tissues, and the markers of metabolic status.
In keeping with previous reports, morbid obesity was associated with high levels of circulating SAA and other inflammatory biomarkers such as IL-6, TNF-α, and acute phase proteins, hsCRP, fibrinogen, and orosomucoid (9, 12, 13). SAA circulating levels were 4-fold higher in obese subjects compared with normal weight subjects, and 10% of the very obese subjects had SAA levels >100 μg/mL, despite the absence of any cause of clinically evident inflammation. The same observation was made with hsCRP; 25% of the obese subjects in our cohort had levels >1.56 mg/dL. SAA correlated mainly with hsCRP, orosomucoid, fibrinogen, and IL-6. SAA is usually described as an acute phase response protein that increases >1000-fold after an acute injury. It can now be considered as a reliable marker of low-grade inflammation observed in human obesity.
In this study, we showed that a 1-year drastic weight loss resulted in a sustained improvement in the low-grade inflammatory state, associated with a dramatic decrease of serum SAA. The measure of SAA level might be useful to follow the evolution of the low-grade inflammation after surgery. These data are in agreement with our previous study on the effect of a moderate and short-term weight loss (very low calorie diet) on the inflammatory profile within adipose tissue (10, 12). In addition, we characterized the dynamic variation in acute phase proteins during a 1-year follow-up. These measurements revealed different patterns of changes. While BMI decreased continuously during the first year after surgery, SAA serum levels decreased during the first 3 months and remained stable 6 months after surgery. The same trend was observed for hsCRP variations. The decrease in SAA levels remained correlated with the decrease in hsCRP, orosomucoid and IL-6 1 year after surgery. The dynamic of serum acute phase proteins was studied recently by two groups. Holdstock et al. (23) observed a reduction of CRP, IL-6, and SAA after gastric bypass and a relation between CRP and insulin sensitivity. Another study showed that lipopolysaccharide binding protein and CRP increased at 3 months after surgery, whereas α1 acid-glycoprotein decreased (24). According to the authors, sustained elevation of lipopolysaccharide binding protein and CRP could be explained by an enhanced metabolic stress response or an aggravation of a non-alcoholic fatty liver disease. These results are in apparent contradiction with our observations. However, in our cohort of obese subjects, we observed a major improvement in hepatic parameters characterizing non-alcoholic fatty liver disease in the first 3 months after surgery (data not shown). This could be consistent with the major decrease in liver-produced acute phase proteins, including SAA. SAA is, indeed, thought to be mainly produced by liver cells.
The origin of increased circulating levels of SAA in morbid obesity is debated. They may result from an increased hepatic production, possibly stimulated by IL-6 or TNF-α released from expanded fat mass, as described for hsCRP or orosomucoid (9, 25). We provided additional information arguing for a contribution of inducible SAA isoforms derived from scWAT to the increased serum SAA level observed. Indeed, we and others have shown that human mature adipocytes produce SAA (11, 12). The expression of inducible isoforms SAA1 and SAA2 is higher in adipose tissue, especially in scWAT, compared with liver. Further experiments focusing on SAA secretion in the different sites of SAA expression are now required. In our study, it was not feasible because of the small amount of adipose tissue and liver samples.
In this study, the expression of SAA1 and SAA2 in scWAT strongly correlated with BMI and SAA serum levels not only at baseline but also with the weight variation, whereas SAA1 and SAA2 expression in liver was not related to BMI and SAA serum levels. Although based on correlation analysis, these findings may suggest a contribution of expanded subcutaneous adipose tissue to the increased circulating levels of SAA. Furthermore, the adipocyte cell volume and SAA circulating levels were highly correlated, independently of BMI, validating the strong relationships between adipose tissue enlargement and SAA.
In the liver, the gene expression of SAA is induced by inflammatory cytokines such as IL-6, TNF-α, or IL-1 (26), all molecules expressed in human adipose tissue. SAA3 expression is up-regulated by TNF-α and lipopolysaccharide in 3T3-L1 adipocytes (16). Until now, the regulation of SAA expression in human adipose tissue has not been studied. We found no correlation between SAA circulating levels or SAA adipose tissue expression with TNF-α at baseline or after weight loss. However, the local regulation of SAA secretion by adipose-derived cytokines such as TNF-α, IL-6, and the inflammatory inductor lipopolysaccharide needs to be examined in scWAT. We cannot exclude the possibility that local production of TNF-α, a powerful regulator within adipose tissue (2), could contribute to the induction of SAA.
Finally, during the acute phase response, the main local functions of SAA consist in the recruitment of macrophage and the remodeling of extracellular matrix at the site of injury to repair the lesions. Macrophage cells infiltrate enlarged adipose tissue (10, 27, 28, 29, 30). We showed that weight loss is associated with a reduced number of macrophages within scWAT and that the improvement in the inflammatory profile, especially in the SAA level, is related to the variation in macrophagic infiltration (31). Measures of SAA reflect the level of systemic low-grade inflammation and maybe the importance of local inflammation within the scWAT. Whether SAA could have local effects on adipose tissue extracellular matrix remodeling and macrophage aggregation in adipose tissue is an open question.
The metabolic consequences of increased SAA circulating levels remain unclear. SAA is known to influence lipoprotein and glucose metabolism. During acute inflammation, SAA is implicated in the high-density lipoprotein decrease and in its dysfunction by replacing apoA-I as the major high-density lipoprotein apolipoprotein (32). However, the role of SAA in the context of the low-grade inflammation state is not established. One study showed a correlation between circulating SAA and atherogenic remnant-like lipoproteins in type 2 diabetic patients with coronary heart disease (33). In our study, no dependence other than with BMI was found between SAA circulating levels and lipid parameters, in agreement with a previous publication (18).
A series of studies suggest that SAA could be implicated in insulin signaling processes and glucose metabolism. In lean rats, it was shown that SAA3 expression is up-regulated by hyperglycemia (16). The Tanis protein, described as a putative receptor for SAA, is deregulated in diabetic rats (17). Tanis mRNA expression in human skeletal muscle and adipose tissue is positively correlated with circulating SAA (19). We analyzed the relationships between SAA and insulin sensitivity index in this context of major weight loss associated with a drastic improvement of insulin sensitivity (34, 35, 36). Neither SAA circulating levels nor SAA expression in adipose tissues or liver correlated with parameters of glucose metabolism and insulin sensitivity at baseline or after weight loss in a large cohort of 60 patients. Finally, we observed that the expression of inducible SAA isoforms was higher in scWAT than in vWAT, the latter being classically associated with the insulin resistance process. All these observations suggest the absence of relationships between SAA and insulin sensitivity in morbidly obese women. These results do not confirm previous findings of a correlation between SAA serum levels and fasting insulin level (11). The differences between the patient cohorts could explain these contradictory data. Indeed, our study enrolled mainly women, with a mean BMI = 48 kg/m2, whereas the Swedish group contained mainly men, with lower BMI (mean = 37 kg/m2), probably with a different distribution of adipose tissues and a different metabolic status. Furthermore, we used the index QUICKI, a combination of fasting glucose and insulin, which is the best index for the assessment of insulin sensitivity. QUICKI is close to the hyperinsulinemic-euglycemic glucose clamp (37), which is the gold standard for measuring insulin sensitivity but which is difficult to perform in a large population of morbidly obese subjects.
The panel of markers of inflammation produced by adipose tissue is increasing (1). Some of them have clearly shown a physiological role by informing the central nervous system of the state of energy reserves (i.e., leptin), by participating in the communication between peripheral organs (e.g., adiponectin, IL-6), or by acting locally through a paracrine/autocrine effect (TNF-α, vascular endothelial growth factor, metallothionein). Some of these molecules are associated with the development of obesity-related complications such as insulin resistance (IL-6) or hemostasis alteration (plasminogen activator inhibitor-1). In this complex context, the role of SAA remains to be clarified, especially regarding the links with obesity-related complications, but we show here that, in morbidly obese subjects, expression of inducible SAA isoforms in scWAT contributes to plasmatic variation in SAA and to a low-grade inflammation state.
In summary, SAA serum levels and SAA expression in adipose tissue during weight loss are linked to adiposity and inflammatory markers but not to variations in metabolic parameters. These results suggest that, at least in morbidly obese subjects, SAA is a biomarker of adiposity but not of insulin resistance. In clinical practice, SAA, like hsCRP, is a consistent marker to follow the early modification of low-grade inflammation during weight variations and subsequent risk of potential complications linked to chronic increased SAA such as atherosclerosis.
This work was supported by the Direction de la Recherche Clinique/Assistance Publique-Hopitaux de Paris and the Programme Hospitalier de Recherche Clinique A0R (AOP 02, 076, CRC) and by a Clinical Research Contract with Alfediam. Funding of the EA 3502 (2002/2003) team was provided by AFERO (French association for the study of obesity), the Benjamin Delessert Institute, and Paris 6 University (BQR, Bonus Quality Research). C.P. received a grant from Alfediam/Merck-Lipha.
Nonstandard abbreviations: scWAT, subcutaneous white adipose tissue; SAA, serum amyloid A; vWAT, visceral white adipose tissue; IL, interleukin; TNF, tumor necrosis factor; hsCRP, high sensitivity C-reactive protein; RT-PCR, reverse transcriptase-polymerase chain reaction; CV, coefficient of variation; QUICKI, quantitative insulin sensitivity check index.