1.1 Sex-specific trajectories of serum UA
We first performed group-based trajectory model (GBTM) analysis to derive serum UA trajectories by sex. Four trajectories were identified for both sexes for whom they shared similar patterns (Figure 1): “low-stable” trajectory 1 (n=783 men; n=1143 women), “moderate–moderate increasing” trajectory 2 (n=2794 men; n=3266 women), “moderate high–moderate increasing” trajectory 3 (n=1660 men; n=1464 women), and “high-increasing” trajectory 4 (n=341 men; n=193 women). The “low-stable” trajectory 1 group had the lowest baseline serum UA level, and the trajectory pattern over time appeared to remain unchanged. The “high-increasing” trajectory 4 group had the highest baseline serum UA level with a rapid elevation over time, characterized by an abnormal baseline level in the presence of hyperuricemia in both sexes. The remaining two trajectories were intermediate to “low-stable” and “high-increasing” trajectories. The “moderate–moderate increasing” trajectory group had a baseline serum UA level higher than the “low-stable” trajectory 1 group but at a normal level, accompanied by a weak increase in trajectory pattern over time. The “moderate high–moderate increasing” trajectory 3 group had a higher baseline serum UA level with a moderate increase in change than the “low-stable” trajectory 1 group.
1.2 Baseline characteristics by sex-specific trajectories of serum UA
The serum UA level at baseline for the four trajectories was, on average, 278.23 ± 37.53, 355.03 ± 36.91, 432.56 ± 46.25, and 520.18 ± 62.61 μmol/L for men and 209.96 ± 28.84, 267.31 ± 31.84, 328.41 ± 38.62, and 411.19 ± 60.88 μmol/L for women, respectively (Table 1). Significant differences among serum UA trajectory groups for both men and women were consistently observed in most parameters, including age, BMI, SBP, DBP, WC, HC, TC, TG, serum UA, hypertension, and abdominal obesity; however, women experienced a more extensive impact on AST and ALT (Table 1). Men in trajectory 3 and 4 groups were younger than those in the remaining two trajectory groups, while women in the trajectory 4 group were the oldest among those in the four trajectory groups (Table 1). The trajectory 4 group had the highest frequency of BMI, DBP, SBP, WC, HC, TC, TG, and serum UA values for both sexes (Table 1). Surprisingly, the percentages of hypertension and abdominal obesity were the highest in the trajectory 2 group for men and trajectory 1 group for women, but not in the trajectory 4 group (Table 1).
1.3 Association between serum UA trajectories and new-onset NAFLD
From 2019 to 2021, 840 and 408 participants in this study were diagnosed with new-onset NAFLD among men and women, respectively (Table 2). During the 3-year follow-up period, 840 (15.06%) men and 408 (6.72%) women developed NAFLD. Next, we assessed the association between serum UA trajectories and NAFLD risk by sex. For men, compared with the trajectory 1 group, trajectory 2, 3, and 4 groups were consistently and significantly associated with an increased risk of NAFLD in all models, characterized by a gradual increase from trajectory 2 to 4 groups. Similar results were observed for women. Moreover, the risk of NAFLD was the highest in the “high-increasing” group for men, whereas for women the risk of NAFLD in the “high-increasing” group was comparable with that of the “moderate high–moderate increasing” group. Specifically, in Model 3 of adjusting covariates with regarding to age, BMI, SBP, DBP, WHR, AST, ALT, TC, TG, and serum UA, the OR with 95% CI for male NAFLD in trajectories 2–4 versus trajectory 1 was 1.43 (1.07–1.94), 2.23 (1.52–3.3), and 2.99 (1.7–5.24), whereas for female NAFLD it was 1.37 (0.93–2.03), 2.37 (1.45–3.92), and 2.37 (1.04–5.33), respectively.
1.4 Subgroup analyses for trajectories of serum UA and incident NAFLD
As hypertension and abdominal obesity were widely reported to have an intimate relationship with the risk of NAFLD (15-17), We further performed a subgroup analysis to detect their specific impact (Table 3) and the numbers of participants in subgroup analyses were presented in Supplementary Table S1. In non-overweight (BMI < 24) people, a dose–response relationship between serum UA trajectories and NAFLD incidence was observed in both sexes, and the “high-increasing” trajectory groups had the highest risk of incident NAFLD. In people with 24 ≤ BMI < 28, a similar pattern of association between serum UA trajectories and NAFLD risk was observed in men. However, in women with 24 ≤ BMI < 28, compared with the trajectory 1 group, only the trajectory 3 group had the significant high risk. In people with BMI ≥ 28, there was no significant association between the serum UA changing trajectory and the risk of NAFLD in both sexes. Likewise, a dose–response association for serum UA trajectories in incident NAFLD in non-hypertensive patients was also observed in both sexes. However, for men and women with hypertension, compared with the trajectory 1 group, the trajectory 2, 3, and 4 groups did not exhibit a significantly increased risk of NAFLD but showed a nominally increased trend risk (Table 3).
1.5 Sensitivity analysis
We also performed sensitivity analyses to confirm the relationship between serum UA trajectories and the risk of NAFLD. We stratified serum UA levels into quartile categories at baseline. When the serum UA levels in quartile categories at baseline were further adjusted, the associations of UA trajectories 2, 3, and 4 with NAFLD risk remained significant in men, whereas the associations of UA trajectories 3 and 4 remained statistically and nominally significant in women, respectively (Supplementary Table S2). Moreover, this risk association were not obviously attenuated in both genders, illustrated by the similar OR values before and after adjustment, underlining the important role of UA trajectories associated with NAFLD (Supplementary Table S2). Moreover, this risk association was not attenuated in either sex, as illustrated by the similar OR values before and after adjustment, underlining the important role of UA trajectories associated with NAFLD (Supplementary Table S2). Next, we reassessed the results by deleting the missing data. Consistent with previous findings, high serum UA trajectories remained at risk of NAFLD incidence in both sexes (Supplementary Table S3), suggesting that missing data probably did not influence the association between trajectories and risk of NAFLD.