Obesity is becoming a global epidemic, and the prevalence of class II and III obesity is constantly increasing. Insulin resistance is common in obese patients and is a cornerstone in the pathophysiology of metabolic syndrome 1. Cardiovascular diseases are the primary cause of morbidity and mortality in obesity. Each 5 kg/m² increase above a body mass index (BMI) of 25 kg/m² results in a 40% increase in cardiovascular mortality 2. A dyslipidaemia is often present and represents a major cardiovascular risk factor. This typical dyslipidaemia is characterised by the quartet: hypertriglyceridaemia, reduction in HDL-cholesterol (HDL-c), increased small and dense low-density lipoproteins (LDL) and post-prandial hyperlipidaemia 3. A number of studies have shown that low plasma HDL-c levels are an independent cardiovascular risk factor 4.

However, obese patients are a very heterogeneous population. For instance, the new concept of a metabolically healthy obese (MHO) group has emerged 5. This subset of obese individuals, who could represent up to 30% of the obese population 6, is characterised by high insulin sensitivity, no sign of hypertension, normal lipid, inflammation and hormonal profiles and a low incidence of cardiovascular diseases 5. Although still discussed, visceral adiposity appears to be more closely linked to insulin resistance, atherogenic dyslipidaemia, decreased blood adiponectin level and inflammation than subcutaneous adiposity 78. Nevertheless, the criteria used to distinguish between MHO and at-risk obese individuals are poorly defined and the factors and mechanisms that could explain these different metabolic profiles, particularly HDL-c level, remain unclear.

Adipocytokines secreted by adipose tissue, such as adiponectin, leptin and resistin, play major roles in the regulation of energy metabolism and/or insulin sensitivity 910 and could directly or indirectly modulate lipid metabolism 11. Likewise, the ghrelin hormone, produced predominantly by the stomach, is involved in energy balance regulation, insulin sensitivity and adiposity 1213. However, no study has yet investigated the link between HDL-c level and hormonal profiles in different subsets of obese patients.

In the present study, our main objective was to investigate the potential link between HDL-c levels and hormonal profiles (insulin, adipocytokines, ghrelin) in different subsets of obese patients: 1) the metabolically abnormal obese (MAO) with metabolic syndrome and the typical metabolic dyslipidaemia; 2) the metabolically healthy obese (MHO) with no metabolic syndrome and with a normal lipid profile; and 3) the low HDL-c obese (LHO) with no metabolic syndrome but with an isolated low HDL-c level; in comparison to 4) healthy and lean control subjects.

The anthropometric, body composition, biochemical and hormonal characteristics of the three groups of obese patients and the control group are presented in Table 1. There was no difference in age or gender ratio between the obese groups and the control group.

In this study, we have shown that specific hormone profiles were associated with HDL-c levels in the three subsets of obese patients we investigated: metabolically healthy obese (MHO) patients, metabolically abnormal obese (MAO) patients and isolated low-HDL-c obese (LHO) patients. Our main results show a strong and positive correlation between HDL-c and unacylated ghrelin levels in the LHO group, and between HDL-c and adiponectin levels in the MAO group, independently of insulin resistance and hs-CRP levels in both cases.

The first subset of our obese patients corresponded to MHO patients described in the literature, which is thought to represent between 20-30% of the obese population 6. Their lipid and apoprotein profiles were indistinguishable from our non-obese control group. Despite the difference in insulin resistance between MHO and control groups, we confirmed the high insulin sensitivity of this group compared to MAO and LHO obese groups 6. High adiponectin concentrations have been associated with the MHO phenotype 6. We also documented higher serum adiponectin levels in our MHO group compared to the MAO group, but these two groups displayed the same hormonal profiles regarding adipose tissue-derived leptin or resistin and stomach derived ghrelin.

The second subset corresponded to the MAO patients, a group that displayed the typical dyslipidaemia seen in insulin resistant patients. This pathophysiology is widely explained by the accumulation of triglyceride-rich lipoproteins (TRL) from the liver (TRL-apoB-100 or VLDL) and from the intestine (TRL-apoB-48 or chylomicrons). This accumulation of TRL has been attributed to the overproduction of VLDL 3 and chylomicrons 15 associated with a defective TRL removal due to the reduction of lipoprotein lipase (LPL) activity mainly in the post-prandial state 16 in a context of competition between VLDL and chylomicrons for this common, saturable, removal pathways, to abnormalities in the apoprotein composition of TRL 17 and to a defect in the hepatic uptake of TRL 16. The increase in the plasma resident time of the TRL would enhance cholesterol ester transfer protein (CETP) exchange of triglycerides and cholesteryl ester between HDL, and LDL on one hand and TRL on the other hand leading to the increase of HDL catabolism and the increase formation of small and dense LDL 316. In this study, we have confirmed that a high level of insulin resistance is a driving parameter in the pathophysiology of the dyslipidaemia seen in the MAO group compared to the MOA and LHO groups and found the typical lipid profile: increased plasma triglycerides, apoB-100, apoB-48, apoC-III and apoC-II, and decreased HDL-c and apoA-I. The same total apoC-III/apoC-II ratios between the subgroups of obese patients suggest that this parallel increase in apoproteins having opposite functional effects in the TRL clearing process is most likely secondary to the increase of plasma TRL, as we have previously reported for type 2 diabetic patients 18. In our multiple regression analysis, adiponectin levels were strongly and positively correlated with the HDL-c levels and were negatively correlated with TG levels independently of BMI, insulin resistance and hs-CRP levels. In our study, this negative association between adiponectin and TG levels was shared by the control and the MHO and MAO groups but not by the LHO group. This difference could be caused by the complexity of the factors regulating TG metabolism. Indeed, a kinetic study has shown that adiponectin, but not leptin or resistin, was the most significant predictor for plasma VLDL apoB concentration, independently of both insulin resistance and size of adipose tissue compartments in subjects with a large range of BMI (from 22 to 35 kg/m²) but that plasma VLDL apoB kinetic was controlled differently, with adiponectin and total body fat regulating catabolism and insulin resistance regulating hepatic secretion 19. Adiponectin levels explained 59% of the variance of HDL-c, thus showing a tight link between these two parameters. A positive correlation between plasma adiponectin and HDL-c has already been found in several studies that have included non diabetic 20, diabetic 21 and obese patients 22. Moreover, two kinetic studies have found similar results showing a strong negative correlation between adiponectin and apoA-I FCR (fractional catabolic rate) 2223, although in one study, this only occurred in the non-obese group 22. This negative correlation has been shown to be independent of BMI, body fat distribution and insulin sensitivity, which suggests that adiponectin may have a direct role in HDL metabolism 23. The inverse relationship shown in vivo between adiponectin and plasma hepatic lipase activity in non diabetic and diabetic patients could be a link between plasma adiponectin levels and HDL metabolism 24.

The third subset of our obese patients, characterised by isolated low-HDL-c levels (LHO) represented an intermediate obese population between the MHO and MAO groups. Despite a higher BMI that explained the significantly higher levels of serum leptin, the insulin sensitivity of this group is in between the two other groups. A positive association has been shown between BMI and triglyceride concentration, which is stronger for visceral than for subcutaneous abdominal adipose tissue 25, but we did not assess these two parameters. Despite higher BMI, LHO patients displayed normal triglyceride levels. Their higher BMI could reflect a better expandability of adipose tissue in the LHO group, permitting a safer storage of fat and avoiding metabolic syndrome 26. Indeed, the lipid profile of LHO group was comparable to that of the MHO group, with the exception of FFA, HDL-c and apoA-I levels. We only found a negative correlation between leptin and triglyceride levels in the LHO group. Their normal triglyceride concentrations could be due to a better insulin and leptin sensitivity in this group compared to the MAO group. Indeed, leptin has been negatively associated with hepatic VLDL-TG secretion rate independently of plasma FFA concentrations, possibly by stimulating hepatic fatty acid oxidation and decreasing de novo lipogenesis 27. Hs-CRP levels were significantly higher in the LHO group compared to the two other obese groups and could partly explain the low HDL-c levels. Indeed, the metabolism, plasma level and apolipoprotein contents of HDL can be strongly impaired in acute and chronic inflammation, including obesity 28. Nevertheless, multiple regression analysis showed that only unacylated ghrelin levels were positively correlated with HDL-c levels, which explained 61% of the variance of HDL-c, thus demonstrating a tight link between these two parameters. The association between plasma ghrelin levels and insulin resistance is debated. Whereas low levels of total plasma ghrelin have been associated with insulin resistance 29, the same negative association has been shown between plasma total or deacylated ghrelin and insulin resistance but a positive association between plasma acylated ghrelin or acylated to deacylated ghrelin ratio and insulin resistance 13. In the present study, the significant association between low serum unacylated ghrelin (we did not measure serum acylated ghrelin for technical reasons) and low HDL-c levels could be partly explained by the same pathophysiology of the MAO group. However, our LHO group did not display hypertriglyceridaemia. An explanation for this could be an increased in post-prandial response in triglyceride levels that was not seen in our fasting measurements. Alternatively, serum ghrelin could act more directly in HDL metabolism by interacting with a species of HDL associated with paraoxonase and apolipoprotein J 30, but further studies are needed to characterize this mechanism. Several HDL turnover studies have been performed to elucidate the mechanism that results in low HDL-c levels, and have included subjects with normal TG levels, but the results are contradictory. On the one hand, Le et al found decreased apoA-I production rate and normal apoA-I FCR in low-HDL-c and normal TG subjects compared to normal HDL-c and normal TG subjects 31. On the other hand, Brinton et al and Gylling H et al found increased apoA-I FCR in low-HDL-c and normal TG subjects 3233 (as it has been shown in these and other studies in low-HDL-c and high TG subjects 3234), but no difference in apoA-I production rate compared to normal HDL-c and normal TG subjects 3233. HDL-c levels are under considerable genetic control with heritability estimates of up to 80% 12. Among the genetic variants of genes involved in HDL metabolism, a polymorphism (g.-1062 G > C) identified in the promoter region of the ghrelin gene has been independently associated with serum HDL-c levels in a population of type 2 diabetic Korean patients 35. However, another study found no association between five SNPs in the ghrelin gene and HDL-c levels in a Canadian population 36. Further genetic studies are needed.

In conclusion, this study shows that the HDL-c levels are associated with a specific plasma hormonal profile in different subsets of class II and III obese patients. The low HDL-c level in the metabolically abnormal obese (MAO) patients was mainly driven by insulin resistance and was associated with a low level of serum adiponectin. The isolated low-HDL-c obese (LHO) patients represented a very interesting population with a level of insulin sensitivity that was between those of the MAO group and the metabolically healthy obese (MHO) group. These patients also showed a high level of hs-CRP and both of these factors could partly explain the low HDL-c levels associated with a direct or indirect role of serum ghrelin levels. Further studies are needed to investigate the direct or indirect role of ghrelin and adiponectin, and to better assess the importance of the genetic background or hormonal profile in HDL-c metabolism of obese patients.

Apo: Apoprotein; BFM: Body fat mass; BMI: Body mass index; DXA: Dual energy X-ray absorptiometry; FCR: Fractional catabolic rate; FFA: Free fatty acids; FFM: Fat free mass; HDL-c: High-density lipoprotein-cholesterol; HOMA-IR: Homeostasis model assessment-insulin resistance; hs-CRP: High-sensitivity C-reactive protein; LDL-c: Low-density lipoprotein-cholesterol; LHO: Low HDL-c obese; MAO: Metabolically abnormal obese; MHO: Metabolically healthy obese; REE: Resting energy expenditure; TG: Triglycerides; TRL: Triglyceride-rich lipoproteins; VLDL: Very low-density lipoprotein.

The authors declare that they have no competing interests.

JPN, MM, RV: conception and design, data collection, data analysis, manuscript writing. SB, AML, BB, ABL, ND, RG, CM, AN, HP, BV: data collection, data analysis. JG: statistical analysis. All authors read and approved the final manuscript.