Characteristics of participants and metals distribution
We included 3324 adults aged 18 years or older in our research. The demographic characteristics of the study subjects with low TOFAT or high TOFAT were listed in Table 1. Overall, gender, BMI, race, smoking, physical activity, energy intake, carbohydrate intake, and protein intake were statistically significant between low TOFAT and high TOFAT participants. Participants with high TOFA were more likely to be female than those with low TOFAT. These subjects were also more likely to have an abnormal BMI and less time for physical activity. A significant difference has existed in SII between the low TOFAT group and the high TOFAT group (Table 1). As shown in Table S1, we also reported the demographic information of males and females. BMI, smoking, physical activity, and the intake of fat were different across TOFAT status in males and females. The geometric mean (GM) and quartiles of metal concentrations were shown in Table 2. The correlation analysis between the log-transform level of urinary metals was reported in Fig.S1.
Table 1 Characteristics of study population in NHANES 2011–2018 (n = 3,324).
Baseline characteristics
|
Total (n=3324)
|
TOFAT
|
p
|
Low (n=1662)
|
High (n=1662)
|
Age (%)
|
|
|
|
0.855
|
<45 year
|
2194 (66.0)
|
1100 (66.2)
|
1094 (65.8)
|
|
≥45 year
|
1130 (34.0)
|
562 (33.8)
|
568 (34.2)
|
|
Gender (%)
|
|
|
|
<0.001
|
Male
|
1676 (50.4)
|
926 (55.7)
|
750 (45.1)
|
|
Female
|
1648 (49.6)
|
736 (44.3)
|
912 (54.9)
|
|
BMl (%)
|
|
|
|
<0.001
|
Normal
|
1012 (30.4)
|
560 (33.7)
|
452 (27.2)
|
|
Abnormal
|
2312 (69.6)
|
1102 (66.3)
|
1210 (72.8)
|
|
Race(%)
|
|
|
|
<0.001
|
Mexican American
|
546 (16.5)
|
249 (15.0)
|
297 (17.9)
|
|
Other Hispanic
|
349 (10.5)
|
149 (9.0)
|
200 (12.0)
|
|
Non-Hispanic White
|
1184 (35.6)
|
526 (31.6)
|
658 (39.6)
|
|
Non-Hispanic Black
|
665 (20.0)
|
431 (25.9)
|
234 (14.1)
|
|
Non-Hispanic other
|
580 (17.4)
|
307 (18.5)
|
273 (16.4)
|
|
Education level (%)
|
|
|
|
0.660
|
Less than high school degree
|
613 (18.4)
|
298 (17.9)
|
315 (19.0)
|
|
High school degree
|
775 (23.3)
|
384 (23.1)
|
391 (23.5)
|
|
More than high school degree
|
1936 (58.2)
|
980 (59.0)
|
956 (57.5)
|
|
Drinking status (%)
|
|
|
|
0.533
|
Never drinker
|
480 (14.4)
|
242 (14.6)
|
238 (14.3)
|
|
Ever drinker
|
356 (10.8)
|
1252 (75.3)
|
1236 (74.4)
|
|
Current drinker
|
2488 (74.8)
|
168 (10.1)
|
188 (11.3)
|
|
Smoking Status (%)
|
|
|
|
0.018
|
Never smoker
|
2070 (62.3)
|
1056 (63.5)
|
1014 (61.0)
|
|
Ever smoker
|
542 (16.3)
|
323 (19.4)
|
389 (23.4)
|
|
Current smoker
|
712 (21.4)
|
283 (17.0)
|
259 (15.6)
|
|
Physical activity (%)
|
|
|
|
0.002
|
<100min MVPA
|
2312 (69.6)
|
1115 (67.1)
|
1197 (72.0)
|
|
≥100 min MVPA
|
1012 (30.4)
|
547 (32.9)
|
465 (28.0)
|
|
Sedentary (%)
|
|
|
|
0.297
|
<360min
|
1583 (47.6)
|
807 (48.6)
|
776 (46.7)
|
|
≥360min
|
1741 (52.4)
|
855 (51.4)
|
886 (53.3)
|
|
Nutrient Intakes
|
|
|
|
|
Energy (%)
|
|
|
|
0.001
|
<2050 kcal
|
1635 (49.2)
|
764 (46.0)
|
871 (52.5)
|
|
2050-3050 kcal
|
1088 (32.7)
|
587 (35.3)
|
501 (30.1)
|
|
≥3051 kcal
|
601 (18.1)
|
311 (18.7)
|
290 (17.4)
|
|
Protein (%)
|
|
|
|
0.019
|
<51.5 gm
|
719 (21.6)
|
333 (20.0)
|
386 (23.2)
|
|
51.5-91.5 gm
|
1356 (40.8)
|
670 (40.3)
|
686 (41.3)
|
|
≥91.6 gm
|
1249 (37.6)
|
659 (39.7)
|
590 (35.5)
|
|
Carbohydrate (%)
|
|
|
|
0.001
|
<282.0 gm
|
2117 (63.7)
|
1009 (60.7)
|
1108 (66.7)
|
|
282.0-457.5 gm
|
932 (28.0)
|
511 (30.7)
|
421 (25.3)
|
|
≥457.6 gm
|
275 (8.3)
|
142 (8.6)
|
133 (8.0)
|
|
Fat (%)
|
|
|
|
0.061
|
<45.5 gm
|
615 (18.5)
|
281 (16.9)
|
334 (20.1)
|
|
45.5-101.5 gm
|
1683 (50.6)
|
858 (51.6)
|
825 (49.6)
|
|
≥101.6 gm
|
1026 (30.9)
|
523 (31.5)
|
503 (30.3)
|
|
SII (103/mL)
|
444 (312, 617)
|
312 (249, 375)
|
617 (517, 765)
|
<0.001
|
Associations between metal concentrations and TOFAT
The associations between the level of metals and TOFAT were shown in Table 3. As shown in model 3, after adjusting for covariates, 9 metals (As, Cd, Co, Cs, Hg, Mo, Mn, Pb, and Sb) had negative and significantly decreased trends of OR for TOFAT (all P for trend < 0.05), while Ba had an increased trend of OR for TOFAT (P for trend < 0.05). Compared with participants in the lowest quartile of As, Cd, Co, Cs, Hg, Mo, Mn, and Pb, participants in the highest quartile showed 51.2% (95%CI: 0.388, 0.614), 46.7% (95%CI: 0.406, 0.700), 39.3% (95%CI: 0.481, 0.767), 45.9% (95%CI: 0.433, 0.676), 20.1% (95%CI: 0.642, 0.995), 42.4% (95%CI: 0.462, 0.718), and 56.0% (95%CI: 0.344, 0.561) decreased of TOFAT, respectively. In addition, RCS analyses showed significant nonlinear associations between Ba, Cd, Co, Cs, and TOFAT (with P-values for the nonlinearity of <0.05). Non-linear associations were consistent between males and females at different metal concentrations (Fig.2).
Table 2 Geometric means and quartile of urinary metal concentrations in NHANES (n = 3324).
Urinary metals
(mg/g Cr)
|
Low TOFAT
|
High TOFAT
|
GM
|
Q1
|
Q2
|
Q3
|
Q4
|
GM
|
Q1
|
Q2
|
Q3
|
Q4
|
As
|
0.079
|
<0.038
|
0.038-0.065
|
0.065-0.141
|
>0.141
|
0.064
|
<0.033
|
0.033-0.053
|
0.053-0.105
|
>0.105
|
Ba
|
1.072
|
<0.625
|
0.625-1.093
|
1.093-1.870
|
>1.870
|
1.110
|
<0.593
|
0.593-1.165
|
1.165-2.095
|
>2.095
|
Cd
|
0.155
|
<0.081
|
0.081-0.143
|
0.143-0.296
|
>0.296
|
0.166
|
<0.089
|
0.089-0.163
|
0.163-0.297
|
>0.297
|
Co
|
0.369
|
<0.242
|
0.242-0.350
|
0.350-0.522
|
>0.522
|
0.388
|
<0.242
|
0.242-0.3785
|
0.379-0.585
|
>0.585
|
Cs
|
4.056
|
<2.859
|
2.859-3.9566
|
3.957-5.7013
|
>5.701
|
4.008
|
<2.941
|
2.941-3.900
|
3.900-5.371
|
>5.371
|
Hg
|
0.003
|
<0.001
|
0.001-0.003
|
0.003-0.005
|
>0.005
|
0.002
|
<0.001
|
0.001-0.002
|
0.002-0.005
|
>0.005
|
Mo
|
35.509
|
<24.095
|
24.095-35.750
|
35.750-52.706
|
>52.706
|
34.297
|
<23.300
|
23.300-34.454
|
34.454-50.190
|
>50.190
|
Mn
|
0.126
|
<0.072
|
0.072-0.116
|
0.116-0.200
|
>0.200
|
0.119
|
<0.067
|
0.067-0.110
|
0.110-0.1972
|
>0.197
|
Pb
|
0.317
|
<0.188
|
0.188-0.305
|
0.305-0.516
|
>0.516
|
0.288
|
<0.175
|
0.175-0.281
|
0.281-0.448
|
>0.448
|
Sb
|
0.051
|
<0.031
|
0.031-0.047
|
0.047-0.073
|
>0.072
|
0.050
|
<0.031
|
0.031-0.046
|
0.046-0.071
|
>0.071
|
Sn
|
0.400
|
<0.208
|
0.208-0.368
|
0.368-0.693
|
>0.693
|
0.483
|
<0.244
|
0.244-0.427
|
0.427-0.859
|
>0.859
|
Tl
|
0.162
|
<0.110
|
0.110-0.158
|
0.158-0.228
|
>0.228
|
0.159
|
<0.113
|
0.113-0.157
|
0.157-0.224
|
>0.224
|
Tu
|
0.064
|
<0.036
|
0.036-0.061
|
0.061-0.106
|
>0.106
|
0.063
|
<0.035
|
0.035-0.059
|
0.059-0.105
|
>0.105
|
GM: geometric mean.
Table 3 The logistic regression of metals and TOFAT in NHANES.
Metals
|
Model
|
Q1
|
Q2
|
Q3
|
Q4
|
P for trend
|
|
|
|
OR (95%CI)
|
OR (95%CI)
|
OR (95%CI)
|
|
As
|
Model 1
|
Ref.
|
0.749 (0.623, 0.900)
|
0.733 (0.607, 0.884)
|
0.531 (0.435, 0.650)
|
<0.001
|
|
Model 2
|
Ref.
|
0.686 (0.564, 0.834)
|
0.680 (0.555, 0.834)
|
0.490 (0.393, 0.611)
|
<0.001
|
|
Model 3
|
Ref.
|
0.686 (0.562, 0.838)
|
0.671 (0.545, 0.827)
|
0.488 (0.388, 0.614)
|
<0.001
|
Ba
|
Model 1
|
Ref.
|
0.809 (0.664, 0.985)
|
0.900 (0.742, 1.091)
|
1.174 (0.961, 1.435)
|
0.057
|
|
Model 2
|
Ref.
|
0.817 (0.662, 1.007)
|
0.914 (0.742, 1.127)
|
1.241 (0.996, 1.546)
|
0.024
|
|
Model 3
|
Ref.
|
0.815 (0.658, 1.009)
|
0.916 (0.741, 1.133)
|
1.240 (0.990, 1.552)
|
0.027
|
Cd
|
Model 1
|
Ref.
|
1.245 (1.046, 1.483)
|
1.430 (1.187, 1.723)
|
1.089 (0.887, 1.338)
|
0.043
|
|
Model 2
|
Ref.
|
0.950 (0.787, 1.148)
|
0.911 (0.738, 1.125)
|
0.553 (0.428, 0.713)
|
<0.001
|
|
Model 3
|
Ref.
|
0.928 (0.765, 1.125)
|
0.866 (0.695, 1.080)
|
0.533 (0.406, 0.700)
|
<0.001
|
Co
|
Model 1
|
Ref.
|
0.846 (0.700, 1.022)
|
0.989 (0.818, 1.195)
|
1.304 (1.074, 1.583)
|
0.004
|
|
Model 2
|
Ref.
|
0.740 (0.603, 0.908)
|
0.732 (0.593, 0.904)
|
0.736 (0.585, 0.926)
|
0.010
|
|
Model 3
|
Ref.
|
0.751 (0.610, 0.924)
|
0.751 (0.605, 0.931)
|
0.758 (0.599, 0.959)
|
0.024
|
Cs
|
Model 1
|
Ref.
|
1.171 (0.973, 1.409)
|
1.153 (0.954, 1.392)
|
0.887 (0.731, 1.078)
|
0.336
|
|
Model 2
|
Ref.
|
1.033 (0.847, 1.260)
|
0.872 (0.707, 1.076)
|
0.616 (0.490, 0.774)
|
<0.001
|
|
Model 3
|
Ref.
|
1.024 (0.837, 1.254)
|
0.863 (0.696, 1.071)
|
0.607 (0.481, 0.767)
|
<0.001
|
Hg
|
Model 1
|
Ref.
|
0.976 (0.808, 1.179)
|
0.888 (0.734, 1.074)
|
0.778 (0.640, 0.947)
|
0.008
|
|
Model 2
|
Ref.
|
0.857 (0.701, 1.047)
|
0.714 (0.581, 0.877)
|
0.562 (0.453, 0.699)
|
<0.001
|
|
Model 3
|
Ref.
|
0.853 (0.695, 1.046)
|
0.698 (0.565, 0.861)
|
0.541 (0.433, 0.676)
|
<0.001
|
Mo
|
Model 1
|
Ref.
|
0.989 (0.821, 1.192)
|
0.925 (0.765, 1.117)
|
0.859 (0.706, 1.045)
|
0.102
|
|
Model 2
|
Ref.
|
0.965 (0.792, 1.174)
|
0.865 (0.706, 1.058)
|
0.778 (0.629, 0.964)
|
0.013
|
|
Model 3
|
Ref.
|
0.970 (0.795, 1.185)
|
0.884 (0.719, 1.086)
|
0.799 (0.642, 0.995)
|
0.032
|
Mn
|
Model 1
|
Ref.
|
0.820 (0.678, 0.993)
|
0.799 (0.659, 0.967)
|
0.855 (0.704, 1.038)
|
0.103
|
|
Model 2
|
Ref.
|
0.674 (0.550, 0.827)
|
0.610 (0.495, 0.752)
|
0.585 (0.471, 0.726)
|
<0.001
|
|
Model 3
|
Ref.
|
0.667 (0.543, 0.821)
|
0.600 (0.484, 0.742)
|
0.576 (0.462, 0.718)
|
<0.001
|
Pb
|
Model 1
|
Ref.
|
0.873 (0.732, 1.040)
|
0.864 (0.715, 1.044)
|
0.632 (0.515, 0.775)
|
<0.001
|
|
Model 2
|
Ref.
|
0.753 (0.623, 0.908)
|
0.664 (0.538, 0.819)
|
0.446 (0.353, 0.563)
|
<0.001
|
|
Model 3
|
Ref.
|
0.739 (0.609, 0.898)
|
0.650 (0.522, 0.808)
|
0.440 (0.344, 0.561)
|
<0.001
|
Sb
|
Model 1
|
Ref.
|
1.076 (0.888, 1.303)
|
1.015 (0.838, 1.230)
|
0.939 (0.775, 1.138)
|
0.435
|
|
Model 2
|
Ref.
|
0.951 (0.777, 1.165)
|
0.823 (0.671, 1.008)
|
0.778 (0.634, 0.954)
|
0.006
|
|
Model 3
|
Ref.
|
0.964 (0.785, 1.185)
|
0.867 (0.704, 1.067)
|
0.823 (0.668, 1.015)
|
0.041
|
Model 1: no covariates were adjusted; Model 2: age, gender, and race were adjusted; Model 3: age, gender, race, education, smoking, drinking, physical activity, sedentary, and nutrient intakes were adjusted.
Associations between metal concentrations and SII
Fig.3 showed the relationship between metal concentrations and SII in the total, male, and female populations. In the total population, We found a significant positive connection between Cd, Co, Tu, and SII in the no covariate-adjusted models (model 1)[β = 54.48 (95%CI: 28.12, 80.84); β = 104.33 (95%CI: 67.59, 141.06); β = 35.90 (95%CI: 8.20, 63.60)] and the fully adjusted model (model 3) [β = 34.67 (95%CI: 1.65, 67.70); β = 54.49 (95%CI: 12.20, 96.78); β = 29.46 (95%CI: 1.67, 57.24)]. We also found that the reduction of SII was 28.86-fold (95%CI: -54.13, -3.60) and 62.37-fold (95%CI: -85.99, -38.74) for each unit increase in As and Hg concentrations. In males, the metal with which we found a statistically significant positive relationship with SII was only Cd [β = 78.96 (95%CI: 30.70, 127.23)], while in females, that were Co [β = 82.05 (95%CI: 27.51, 136.58)] and Sn [β = 40.89 (95%CI: 3.82, 77.96)]. In males, the metals negatively associated with SII were As and Hg, while in females they were Hg and Pb.
Associations between SII and TOFAT
Table S3 showed the associations between SII and TOFAT based on the multivariate logistic regression model. In the total, male, and female populations, SII always had a positive trend with TOFAT (P for trend < 0.05). In the total population, subjects in the highest quartile had a 93.1% increase in TOFAT compared to subjects in the lowest quartile of SII (Model 3), whereas TOFAT increased by 20.1% and 86.7% for males and females, respectively.
Mediation analyses
Based on the above results, the mediation analyses were performed. In the total population, after adjusting for covariates, SII had significant mediated effects on the associations of Co with TOFAT, and the beta coefficient of mediating effect was 0.8% (95%CI: 0.2, 0.5%; P = 0.020). The mediating effects of SII were not statistically significant on the associations of As, Cd, and Hg with TOFAT (all P > 0.05). Moreover, the direct effect coefficients of the associations between As, Co, Cd, Hg and TOFAT were -0.7% (95%CI: -13.7%, -1.5%), -3.4% (95%CI: -7.8%, 1.0%), -12.1% (95%CI: -23.9%, -3.1%), and -44.0% (95%CI: -48.6%, -31.3%), respectively (Fig.4). In male, SII was not found to have a mediating role in the relationship between As, Cd, Hg and TOFAT (Fig.S2). However, in females, SII was found to have a significant mediated effect on the associations of Co with TOFAT [βindirect = 1.2% (95%CI: 0.3%, 2.2%); P = 0.016] (Fig.S3).