Serum Fetuin-A concentration in PCOS and healthy women
Table 1 summarizes the demographic, anthropometric, and metabolic parameters and sex hormone levels of all women in the current study. The distribution of Fetuin-A concentrations in healthy women is shown in Figure 1A. We found that circulating Fetuin-A concentration ranged from 196.6 to 418.2 µg/L for most normal women (95 %). PCOS patients had higher circulating Fetuin-A levels than healthy women (Figure 1B, Table 1). PCOS patients were divided into obese/overweight (ob/ow) and lean groups by BMI < 25kg/m2 or ≥ 25kg/m2, respectively. We found that the levels of Fetuin-A in obese/overweight group were significantly higher than those in the lean group (Figure 1C). When control subjects were divided into overweight and lean groups, there was no significant difference in serum Fetuin-A (313.5 ± 60.6 vs. 317.4 ± 62.7µg/L), suggesting that there was no significant relationship between serum Fetuin-A level and overweight in normal women, In addition, serum Fetuin-A was divided into three tertiles (tertile 1, < 322.6 µg/L; tertile 2, 322.6-419.6 µg/L; tertile 3, >419.6 µg/L). The odds of developing PCOS were calculated by logistic regression analysis. In the tertile 2 and 3 of blood Fetuin-A, the odds ratios of developing PCOS were higher than tertile 1 (95% CI 1.28 - 5.14 for tertile 2 and 95% CI 11.8 – 113.5 for tertile 3; vs. tertile 1, both p < 0.01; Figure 1D).
Table 1 Main clinical features and circulating Fetuin-A levels in PCOS and control subjects
Characteristics
|
Controls
(n=85)
|
PCOS
(n=122)
|
P-value
|
Age (yr)
|
26.0 ± 3.4
|
25. 3 ± 3.4
|
0.114
|
BMI (kg/m2)
|
21.6 ± 2.9
|
24.4 ± 4.4
|
< 0.001
|
WHR
|
0.81 ± 0.08
|
0.88 ± 0.10
|
< 0.001
|
SBP (mmHg)
|
113 ± 8
|
128 ± 8
|
< 0.001
|
DBP (mmHg)
|
75 ± 7
|
77 ± 7
|
0.045
|
TG (mmol/L)
|
0.92 (066-1.34)
|
1.37(1.01-2.48)
|
< 0.001
|
TC (mmol/L)
|
3.94 ± 0.76
|
4.64 ± 0.93
|
< 0.001
|
HDL-C (mmol/L)
|
1.23 ± 0.27
|
1.18 ± 0.33
|
0.278
|
LDL-C (mmol/L)
|
2.18 ± 0.58
|
2.89 ± 0.58
|
< 0.001
|
FBG (mmol/L)
|
4.77 ± 0.46
|
5.06 ± 0.91
|
< 0.01
|
FIns (mU/L)
|
7.40 (5.55-9.17)
|
18.00 (9.30-27.08)
|
< 0.001
|
HOMA-IR
|
15.0 (1.15-2.04)
|
3.85 (1.97-5.83)
|
< 0.001
|
FSH (IU/L)
|
5.92 ± 2.00
|
5.78 ± 1.36
|
0.574
|
LH (IU/L)
|
6.07 ± 5.14
|
8.39 ± 4.94
|
< 0.001
|
T (nmol/L)
|
1.01 ± 0.68
|
1.44 ± 0.65
|
< 0.001
|
DHEA-S (μmol/L)
|
4.50 ± 1.86
|
6.58 ± 2.93
|
< 0.001
|
Fetuin-A (μg/L)
|
313.8 ± 60.5
|
437.9 ± 119.3
|
< 0.001
|
Values are given as mean ± SD or median (inter quartile range). BMI, Body mass index; WHR, Waist hip ratio; SBP, Systolic blood pressure; DBP, Diastolic blood pressure; FBG, fasting blood glucose; FIns, fasting plasma insulin; TG, Triglyceride; TC, Total cholesterol; HDL-C, High-density lipoprotein cholesterol; LDL-C, Low-density lipoprotein cholesterol; HOMA-IR, homeostasis model assessment of insulin resistance; FSH, follicle-stimulating hormone; LH, luteinizing hormone; T, testosterone; DHEA-S, dehydroepiandrosterone sulfate.
Serum Fetuin-A level and its association with other parameters in the study population
Next, we investigated the relationship between the levels of circulating Fetuin-A and various other parameters. Serum Fetuin-A was positively correlated with BMI, WHR, TG, TC, LDL-C, HOMA-IR, LH, T, and DHEA-S (Table 2). Moreover, as previously reported, we also found a significant correlation between T and HOMA-IR (r = 0.351, p <0.01). Next we performed a multiple stepwise regression to determine variables that had independent associations with serum Fetuin-A. The results showed that only WHR, TG, HOMA-IR, and DHEA-S were independent predictors of the levels of circulating Fetuin-A (Table 2). The multiple regression equation was Y lg10(Fetuin-A) = 2.11 + 0.012 XHOMA-IR + 0.401XWHR + 0.017X TG+ 0.009X DHEA-S .
Table 2 Correlation analysis of variables associated with circulating Fetuin-Aa levels in the study populations
|
Simple
|
|
Multiple
|
Variable
|
r
|
p
|
|
b
|
p
|
Age (years)
|
-0.124
|
0.076
|
|
------
|
------
|
BMI (kg/m2)
|
0.294**
|
< 0.001
|
|
------
|
------
|
SBP (mmHg)
|
0.076
|
0.279
|
|
------
|
------
|
DBP (mmHg)
|
0.039
|
0.579
|
|
------
|
------
|
WHR
|
0.351**
|
< 0.001
|
|
0.401
|
< 0.001
|
TG (mmol/L)
|
0.293**
|
< 0.001
|
|
0.017
|
< 0.05
|
TC (mmol/l)
|
0.184**
|
< 0.01
|
|
------
|
------
|
HDL-C (mmol/L)
|
-0.106
|
0.128
|
|
------
|
------
|
LDL-C (mmol/L)
|
0.339**
|
< 0.001
|
|
------
|
------
|
HOMA-IR
|
0.511**
|
< 0.001
|
|
0.012
|
< 0.001
|
FSH (IU/L)
|
-0.079
|
0.259
|
|
------
|
------
|
LH (IU/L)
|
0.178*
|
< 0.05
|
|
------
|
------
|
T (nmol/L)
|
0.320**
|
< 0.001
|
|
------
|
------
|
DHEA-S (μmol/L)
|
0.330**
|
< 0.001
|
|
0.009
|
< 0.01
|
aVariables with a non-normal distribution were transformed by logarithm before analysis
Additionally, logistic regression analysis revealed that Fetuin-A was significantly related to PCOS, even after controlling for anthropometric variables, blood lipid and so on (Table 3).
Table 3 Association of circulating Fetuin-A with PCOS in fully adjusted models
Model adjustments
|
PCOS
|
OR
|
95%CI
|
P
|
Age, SBP, DBP
|
1.015
|
1.010-1.020
|
0.000
|
Age, SBP, DBP, BMI, WHR
|
1.014
|
1.009-1.019
|
0.000
|
Age, SBP, DBP, BMI, WHR, FBG, FIns
|
1.011
|
1.005-1.017
|
0.000
|
Age, SBP, DBP, BMI, WHR, FPG, FIns, Lipid profile
|
1.010
|
1.004-1.017
|
0.002
|
Age, SBP, DBP, BMI, WHR, FPG, FIns, Lipid profile, Hormone parameters
|
1.008
|
1.001-1.016
|
0.036
|
Results of binary logistic regression analysis are presented. 95% CI, confidence interval; OR, odds ratio. Data from two subject groups were pooled to calculate logistic regression.
ROC curve analysis
To explore the prediction of PCOS and IR by blood Fetuin-A, we performed receiver operating characteristic (ROC) curve analysis. The result showed that the area under the ROC curves for PCOS (AUCPCOS) was 0.82 with a specificity of 83.5 %, and sensitivity of 69.7 % (p < 0.01, Figure 2A), and AUCIR was 0.80 with a specificity of 81%, and sensitivity of 72.3 %. The best cut-off values for Fetuin-A to detect PCOS and IR were 366.3 µg/L for PCOS and 412. 6 µg/L for IR.