Anthropometric and biochemical parameters and serum fetuin-B levels in healthy and MetS women
In the current cross-sectional study, the distribution range of serum fetuin-B concentrations was from 1.08 to 10.30 mg/L for most healthy women (83.2%) (Fig. 1A). The main clinic features and metabolic parameters of the 377 participants (average age, 37.9 ± 15.6 years) were shown in Table 1. Compared with the healthy controls, MetS women had a significant increase in serum fetuin-B concentration (8.03 ± 3.75mg / L for MetS vs. 6.01 ± 3.94 mg/L for healthy women) (p < 0.001, Table 1 and Fig. 1B ). In MetS women, age, BMI, FAT%, WHR, blood pressure (BP), blood lipid (including TG, TC, LDL-C, and FFA), FBG, 2h-BG, FIns, 2-h insulin after glucose overload (2h-Ins), HbA1c, HOMA-IR, VAI, LAP significantly raised, while HDL-C was lower compared with the healthy women ( all p < 0.001, Table 1). In addition, there was a statistically significant increase in serum fetuin-B levels of overweight/obese (OW/OB) women (n=161, BMI ≥ 24 kg/m2) than the lean group (n=216, BMI < 24 kg/m2; 7.52 ± 4.01 vs. 6.68 ± 3.91 mg/L, p < 0.05; Fig. 1C). To explore the association of serum fetuin-B with IR, we divided all subjects into IR (HOMA-IR > 3) and no-IR (HOMA-IR ≤ 3) group and found serum fetuin-B levels were markedly elevated in IR women than those in no-IR women (7.92 ± 4.04 vs. 6.33 ± 3.95 mg/L, p < 0.01; Fig. 1D).
Association of fetuin-Bwith other variables
Linear correlation analysis showed that there was a significant positive correlation between fetuin-B with obesity- and lipid-related parameters (BMI, WHR, fat, TG, lap, VAI), and glucose-related parameters (HbA1c%, FBG, 2 h-BG, fIns, 2 h-INS, HOMA-IR) (p < 0.05 or p < 0.01) in all subjects (Table 2). We further adjusted age and WHR, and found that there was still a relationship between fetuin-B and TG, BMI, fat %, FIns, 2 h-INS, HOMA-IR, LAP as well as VAI, but no correlation between fetuin-B and HDL-C and FFA (Table 2). Moreover, we demonstrated that TG and WHR were independently related factors of serum fetuin-B (Fig. 1E). The multiple regression equation was Y fetuin-B= 4.381+ 0.674XTG + 1.731XWHR (R2=0.063).
The relationship between serum fetuin-B and MetS
A logistic regression analysis was applied to the data, which demonstrated that serum fetuin-B were related to MetS (OR, 1.150; 95% CI, 1.086 - 1.217; p < 0.01). This relationship persisted even after Age, BMI, FAT%, HbA1c%, insulin, TC, LDL, FFA, and possible confounding factors were controlled (Table 3). Meanwhile, a significant linear trend and independent correlation were found between serum fetuin-B and the MetS by the row mean score and the Cochran–Armitage trend test (Additional file 1: Table 1). Moreover, according to MetS components, we divided the mean levels of serum fetuin-B into six grades. Fig. 1F showed that with increase MetS components, serum fetuin-B increased progressively (p for trend < 0.05). Individuals with 0, 1, 2, 3, 4 and 5 component of the MetS had increased serum fetuin-B levels of 5.75 ± 3.49, 6.03 ± 4.00, 6.29 ± 4.38, 7.75 ± 3.39, 8.08 ± 4.05 and 8.78 ± 3.97mg/L. Furthermore, we divided serum fetuin-B into three tertiles (tertile 1, ≤ 5.49 mg/L; tertile 2, 5.49-8.58mg/L; tertile 3, > 8.58mg/L). The odds ratio was calculated as an estimate of developing MetS. As shown in Fig. 1G, the risk of developing MetS in the tertile 2 and tertile 3 were obviously increased than tertile 1 (OR, 2.07; 95% CI, 1.25 - 3.43 for tertile 2; OR, 2.96; 95% CI, 1.78 - 4.94 for tertile 3; vs. tertile 1, all p < 0.01).
Receiver operating characteristic (ROC) curve analysis
We explored the predictive value of serum Fetuin-B on MetS by ROC curve. The results demonstrated that the areas under the ROC curves for MetS (AUCMetS) were 0.64 (p < 0.01) with a sensitivity of 91.1% and specificity of 65.9 % (Fig. 2). The best cut-off value for the serum fetuin-B to predict MetS was 3.87 mg/L.
Effects of hyperglycemia and hyperinsulinemia on serum Fetuin-B
To explore the impacts of glucose load on the circulating levels of fetuin-B, an OGTT was conducted in healthy women and those in MetS. As shown in Fig. 3A, serum fetuin-B in healthy subjects increased significantly after glucose challenges compared with the basal value, to a peak at 30 minutes (from 3.34 ± 2.96 to 10.52 ± 5.35mg/L) and retained to the end of the experiment (Fig. 3A). However, in patients with MetS, glucose load had not caused any significant changes in serum fetuin-B (Fig. 3A). In MetS women, the area under the curve for fetuin-B (AUCf) was increased significantly relative to healthy women (Fig. 3A). Then, we performed an EHC in patients with MetS and healthy women to further explore the factors affecting the secretion of serum fetuin-B (Fig. 3B). In response to hyperinsulinemia, serum fetuin-B significantly increased in MetS individuals, whereas there was no change in healthy women (Fig. 3C). During the stable-state of the EHC, serum fetuin-B in MetS women was significantly increased as compared to the baseline(6.03 ± 4.15 vs. 9.74 ± 4.45mg/L, p < 0.01, Fig. 3D). Meanwhile, the women with MetS had a lower M-values than those of healthy controls (4.47 ± 1.88 vs. 10.23 ± 2.79 mg/kg/min; p <0.01, Fig. 3C). The above results indicate that patients with MetS had an IR, and the secretion of serum fetuin-B in vivo might be regulated by circulating insulin levels.
Effects of GLP-1RA intervention on serum Fetuin-B concentration
Twenty-four patients with MetS participated in the GLP-1RA intervention study for six months (Fig. 4A). The main clinic features and metabolic parameters for pre- and post-treatment were shown in Table 4. After three months of GLP-1RA treatment, the markers of lipid metabolism and obesity (BMI, FAT%, TG, TC, HDL-C, LDL-C, and LAP ) and the parameters of glucose metabolism and IR (HbA1c, FIns, HOMA-IR ) were significantly ameliorated compared with those in pre-treatment in these patients with MetS (p < 0.01 or p < 0.05). Furthermore, after treatment for six months, FBG, 2h-BG, VAI also declined significantly than those of pre-treatment(Table 4, p < 0.01). Importantly, serum fetuin-B exhibited a noticeable decline from 10.67 ± 4.87 at pre-treatment to 8.90 ± 3.45 for post-treatment 3 months, and finally to 7.38 ± 2.74 mg/L for post-treatment 6 months (Fig. 4B). Meanwhile, we found that blood glucose at 120min and the area under the curve for glucose (AUCg, 16.68 ± 2.79 vs. 19.46 ± 4.74 mmol/h/L, p < 0.05;) during the OGTT were significantly lower than that before GLP-1RA intervention (Fig. 4C). When compared with pre-treatment, the M-values during the EHC were significantly increased at both 3 and 6 months post-treatment (4.39 ± 1.30 and 4.66 ± 1.53 vs. 3.29 ± 0.82 ; all p < 0.01; Fig. 4D). Therefore, these data further confirm that fetuin-B levels decreased in vivo with the improvement of IR.