3.1. Population characteristics
The characteristics of the participants are displayed in Table 1. Altogether, 4,030 individuals were included, of whom 1,076 (26.7%) diagnosed with MetS. The average age of the included population was 47.8 ± 17.2 years old. 43.4% participants were non-Hispanic white, and 20.9% were non-Hispanic black. More than half (88.8%) attained a high school education or higher. Furthermore, the median values for BMI, waist circumference, HDL, FPG, and TG in the total population were 29.3 kg/m², 99.5 cm, 51.9 mg/dL, 109.0 mg/dL, and 131.0 mg/dL. Age, gender, race, income, education, exercise and smoking status differed significantly between individuals with and without MetS (all P<0.05). Participants with MetS exhibit characteristics including senior age, female, non-Hispanic white ethnicity, smoking habits, lower income, and education levels, as well as less engagement in sporting activity.
Table 1. Characteristics of participants in the study, NHANES, 2007–2012 (N = 4,030).
(Line 222)
Characteristic
|
Overall
|
MetS
|
P value
|
No
|
Yes
|
No. of participants, n (%)
|
4,030 (100%)
|
2,954 (73.3%)
|
1,076 (26.7%)
|
-
|
Age (mean ± SD, years)
|
47.8 (17.2)
|
45.1 (17.0)
|
55.3 (15.4)
|
<0.001
|
< 60 years old
|
2,864 (71.1%)
|
2,276 (79.5%)
|
588 (20.5%)
|
|
≥ 60 years old
|
1,166 (28.9%)
|
678 (58.1%)
|
488 (41.9%)
|
|
Race
|
|
|
|
<0.001
|
Mexican American
|
640 (15.9%)
|
456 (15.4%)
|
184 (17.1%)
|
|
Non-Hispanic Black
|
843 (20.9%)
|
624 (21.1%)
|
219 (20.4%)
|
|
Non-Hispanic White
|
1,751 (43.4%)
|
1,252 (42.4%)
|
499 (46.4%)
|
|
Other Hispanic
|
433 (10.7%)
|
315 (10.7%)
|
118 (11.0%)
|
|
Other Race
|
363 (9.01%)
|
307 (10.4%)
|
56 (5.20%)
|
|
Gender
|
|
|
|
0.007
|
Male
|
2,064 (51.2%)
|
1,551 (52.5%)
|
513 (47.7%)
|
|
Female
|
1,966 (48.8%)
|
1,403 (47.5%)
|
563 (52.3%)
|
|
Income
|
|
|
|
<0.001
|
$0 to $19,999
|
1,050 (26.1%)
|
749 (25.4%)
|
301 (28.0%)
|
|
$20,000 to $44,999
|
1,351 (33.5%)
|
958 (32.4%)
|
393 (36.5%)
|
|
$45,000 to $74,999
|
695 (17.2%)
|
513 (17.4%)
|
182 (16.9%)
|
|
≥ $75,000
|
934 (23.2%)
|
734 (24.8%)
|
200 (18.6%)
|
|
Education
|
|
|
|
<0.001
|
Below high school
|
454 (11.3%)
|
292 (9.9%)
|
162 (15.1%)
|
|
High school
|
1,522 (37.8%)
|
1,061 (35.9%)
|
461 (42.8%)
|
|
College and above
|
2,054 (51.0%)
|
1,601 (54.2%)
|
453 (42.1%)
|
|
Energy Intake
(mean ± SD, kcal)
|
2,005 (952)
|
2,053 (974)
|
1,874 (877)
|
<0.001
|
Log-transformed Creatinine (mean ± SD, mg/dL)
|
4.58 (0.74)
|
4.58 (0.76)
|
4.60 (0.67)
|
0.448
|
Smoking status
|
|
|
|
<0.001
|
no
|
2,237 (55.5%)
|
1,698 (57.5%)
|
539 (50.1%)
|
|
yes
|
1,793 (44.5%)
|
1,256 (42.5%)
|
537 (49.9%)
|
|
Extra Exercise
|
|
|
|
<0.001
|
no
|
3,130 (77.7%)
|
2,155 (73.0%)
|
975 (90.6%)
|
|
yes
|
900 (22.3%)
|
799 (27.0%)
|
101 (9.4%)
|
|
BMI (mean ± SD, kg/m²)
|
29.3 (6.74)
|
27.7 (5.98)
|
33.6 (6.84)
|
<0.001
|
Waist (mean ± SD, cm)
|
99.5 (16.2)
|
95.2 (14.7)
|
111 (14.5)
|
<0.001
|
HDL (mean ± SD, mg/dL)
|
51.9 (15.4)
|
54.7 (15.4)
|
44.2 (12.6)
|
<0.001
|
FPG (mean ± SD, mg/dL)
|
109.0 (35.3)
|
100.0 (25.6)
|
123.0 (43.4)
|
<0.001
|
TG (mean ± SD, mg/dL)
|
131.0 (103.0)
|
99.5 (54.6)
|
182.0 (138.0)
|
<0.001
|
Diabetes
|
|
|
|
<0.001
|
no
|
487 (12.1%)
|
184 (6.2%)
|
303 (28.2%)
|
|
yes
|
3,464 (86.0%)
|
2,722 (92.1%)
|
742 (69.0%)
|
|
Borderline
|
77 (1.9%)
|
47 (1.6%)
|
30 (2.8%)
|
|
NHANES: National Health and Nutrition Examination; BMI: body mass index; HDL: High-density lipoprotein; FPG, Fasting glucose; TG, Triglycerides.
T-test and chi-square tests were used to compare the characteristics between the MetS and non-MetS groups.
3.2. Distribution and correlation of 15 chemicals
Figure 2A and Table S1 provides details of the 15 compounds' distribution. All 15 chemicals were detectable in over 80% of the participants. Notably, among pesticides, 2,5-DCP, among PAHs, 1-NAP, and among phthalates, MEP exhibited the highest concentrations. The correlation heat map for the 15 chemicals is shown in Figure S2. There existed a robust positive correlation between chemicals within the same category, with correlation coefficients notably high at 0.77 between 2,4-DCP and 2,5-DCP, and ranging from 0.56 to 0.96 for PAHs. Within the phthalates group, MCPP and MCOP showed the strongest correlation, reaching an index of correlation of 0.73. The correlation between the groups was less pronounced, with the correlation coefficients primarily falling between 0.2 and 0.4.
3.2. Exploring associations of individual chemical exposures with MetS using weighted generalized linear regression model
We employed a weighted generalized regression model to explore the associations between single-chemical exposure and MetS. Our analysis revealed a noteworthy association between 2-PHEN, a constituent of the PAH group, and an increased risk of MetS (OR: 1.37, 95% CI: 1.19-1.59) (Figure 2B). And the association was consistent when adjusting for different set of covariates (Figure S3).
3.4. Identifying key chemicals associated with MetS through variable selection methods
Considering the high correlations among most chemicals, we performed dimensionality reduction techniques using LASSO regression and BMA models. We identified statistically significant associations of the 1-PYR, 1-PHEN, and 2-PHEN and the risk of MetS. Notably, 2-PHEN was positively associated with MetS (OR (95% CI): 2.91 (1.95, 4.36)). Conversely, 1-PYR and 1-PHEN showed significant negative associations with MetS. Furthermore, employing the BMA model, we selected eight variables in the analysis. The regression results further supported the positive associations of 2-PHEN and 2-NAP with MetS, while revealing a negative association between 1-PYR and MetS (Figure 3).
3.5. Investigating the association between mixed chemical exposure and MetS using an exposome model
Table 2 demonstrates that a positive relationship between the WQS index of the mixture and the MetS existed in the entire sample (OR: 1.25 (95%CI: 1.04, 1.51)). The mixed exposure had the highest weight assigned to 2-PHEN, MEOHP, 2-NAP and 2,5-DCP. When exploring the associations of different chemical category mixtures with Mets, we found that PAHs were linked to a higher incidence of MetS (OR: 1.20 (95%CI: 1.05, 1.37), Table 2). The highest weights of 2-PHEN and 2-NAP were also reflected in the results of the separate analysis within the PAH group (Figure 4). When set in the negative direction, the associations of 15 chemicals as well as individual classes of chemicals with MetS were not statistically significant.
Table 2. Results of WQS model for association between mixture exposure and MetS.
Groups Direction
|
OR
|
95% CI
|
P value
|
Total
|
Positive
|
1.25
|
(1.04,1.51)
|
0.017
|
Negative
|
0.97
|
(0.80,1.19)
|
0.802
|
Pesticides
|
Positive
|
1.07
|
(0.97,1.18)
|
0.153
|
Negative
|
1.01
|
(0.92,1.11)
|
0.895
|
PAHs
|
Positive
|
1.20
|
(1.05,1.37)
|
0.008
|
Negative
|
1.02
|
(0.89,1.15)
|
0.810
|
Phthalates
|
Positive
|
0.98
|
(0.86,1.12)
|
0.800
|
Negative
|
0.94
|
(0.80,1.11)
|
0.493
|
All models were adjusted for age, gender, race, education, income, exercise, smoking status and creatinine.
We performed subgroup analyses for age and gender to examine the stability of the results. In subgroup analyses stratified by gender, the weighted linear regression model demonstrated that 2-PHEN increased the risk of MetS for both males and females (Figure S4). The LASSO regression model and the BMA model jointly revealed that 2-PHEN as an essential variable and positively related to the MetS in males and females (Figure S5-S6). The WQS results demonstrated that only in the female subgroup, the mixture statistically significantly associated with the MetS (OR=1.40, 95% CI: 1.08-1.83) (Table S2). 2-PHEN, 2,5-DCP, 2-NAP, and MEP contributed the most to the results (Figure S11). While in the age-stratified analyses, the results of the weighted linear regression model both showed significant positive associations among 2-PHEN and MetS (Figure S4). Consensus results from the variable selection models supported 2-PHEN being a crucial influence on the MetS (Figure S7-S8). Additionally, the WQS results were not statistically significant in either group (Table S2).