After excluding 62 participants without SUA data, 256 PCOS patients were included in this study. The mean (±SD) of SUA was 376.84 ± 87.95 μmol/L for all subjects and their media (IQR) of age was 27.5 (24.3-31.0) years old. The prevalence rate of hyperuricemia was 56.3%, with the mean (±SD) of SUA 438.91 ± 58.95 μmol/L. Patients with hyperuricemia had higher BMI and VAI (Table 1), and with the increase of BMI and VAI, the level of SUA and the prevalence of hyperuricemia also increased (Table 2, Figure 1).
Characteristics of study population
The detailed anthropometric and metabolic characteristics of females with PCOS categorized by the existence of hyperuricemia were shown in Table 1. Compared with the normouricemia group, the group with hyperuricemia had significantly greater values for SBP (116±11 vs. 120 ± 13, P=0.013), TC (4.94 ±0.90 vs. 5.19 ±0.92, p=0.028), TG (1.26 [0.86 - 1.84] vs. 1.61 [1.17 - 2.19], P<0.001), LDL-c (2.62 [2.19 - 3.18] vs. 2.85 [2.56 - 3.44], P=0.001), and creatinine (52.29 ± 8.33 vs. 55.57 ± 9.26, p=0.004) and lower values for HDL-c (1.32 [1.12 - 1.57] vs. 1.18 [1.03 - 1.33], P<0.001) and eGFR (150.96 ±27.74 vs. 141.44 ±31.12, p=0.011). The obesity indices such as BMI (25.50 ± 4.74 vs. 29.64 ± 4.61, P<0.001), WHR (0.85±0.07 vs.0.88±0.06, P=0.001), WC (84.9 ±12.5 vs. 93.8 ± 11.3, P<0.001), HC (99.3± 9.6 vs. 106.8±9.6, P<0.001), BFP (34.23 ±7.35 vs. 39.57 ±6.51, P<0.001) and VAI (1.88 (1.18 - 2.79) vs. 2.54 (1.81 - 3.73), P<0.001)) were higher in the hyperuricemia group than that in the normouricemia group. However, there was no significant differences in age, DBP, and T between two groups.
Besides, in order to further study the relationship between obesity and other indicators, subjects were categorized into three groups (normal weight, overweight and obesity) in Table 2. As the degree of obesity increased, the level of SUA (312.6 ± 71.16, 372.57 ± 84.86, 410.93 ± 78.76, respectively, p<0.001) and the prevalence rate of hyperuricemia also increased (p<0.001). In addition, subjects with higher BMI were more likely to be older and occasional smoker and had significantly higher levels of SBP, WC, HC, WHR, BFP, VAI (1.45 (0.86 - 2.08), 2.19 (1.42 - 3.46), 2.65 (1.93 - 3.77), P<0.001, respectively), TG, HDL-c and LDL-c. However, there was no significant difference in DBP, TC, creatinine, eGFR and T.
Figure 1 showed the distributions of SUA levels and prevalence of hyperuricemia according to tertiles of VAI in patients with PCOS. As the tertiles of VAI increased, the level of SUA (p<0.001) and the prevalence rate of hyperuricemia also increased (2.6%, 21.3%, 22.4%, respectively, p<0.001).
Correlations of SUA level with clinical characteristics
We performed Pearson’s correlation to investigate the correlation between SUA with clinical characteristics. We found that there were significant association between the obesity indexes (BMI, BFP and WHR), lipid profiles (log (TG), log (LDL-c) and log (HDL-c)), SBP, and log (VAI) (r=0.346, p<0.001) with SUA. However, further stepwise linear regression analysis indicated that SUA level were only positively associated with BMI (β= 0.325, p < 0.001) and Log (VAI) (β = 0.243, p < 0.001) in PCOS women (Table 3).
Association of VAI and SUA level
In addition, multivariate linear logistic regression analysis was used to further assess the association between VAI and SUA level (Table 4). In model 1 with adjustment for age and occasional drinking, VAI was significantly associated with SUA level, and the coefficient (95% CI) was 16.52(10.18-22.85, P<0.001). In model 2 with further adjustment for SBP, DBP, and eGFR, the significant association of VAI with SUA level remained and the coefficient (95% CI) was 14.74 (8.45-21.03, P<0.001). Further, after adjusted for additional TC, LDL-c, T, and BMI in model 3, VAI was still significantly associated with the SUA level, and the coefficient (95% CI) was 9.20 (2.85-15.56, P=0.005).
Association of VAI and hyperuricemia
Multivariate logistic regression analysis was performed to explore the association between VAI and hyperuricemia. The following three models were performed with same adjustments as those in multivariable linear regression analyses. In model 1, VAI were significantly associated with hyperuricemia, and the adjusted OR (95% CI) was 1.56 (1.26 - 1.92, P<0.001). In model 2, the associations of VAI with hyperuricemia remained significant, with the adjusted OR (95% CI) of 1.53 (1.24 - 1.89, P<0.001). In model 3, the significant associations between VAI and hyperuricemia still existed, and the adjusted OR (95% CI) was 1.32 (1.05-1.65, P=0.018) (Table 4).