|
Variable
|
IQR in non-CBPs
|
Factor – biosensor associationa
|
Carbon black – biosensor associationa
|
Mediation effectb
|
|---|
|
Estimate
|
SE
|
P
|
Estimate
|
SE
|
P
|
PM
|
P perm
|
|---|
|
Carbon black exposure
| | | | |
−0.920 (c)c
|
0.198
|
< 0.0001
| | |
|
IL-6 (pg/ml)
|
52.1
|
−0.134 (b)
|
0.040
|
0.0010
|
−0.480 (c’)
|
0.233
|
0.041
|
0.478
|
0.014
|
|
TNF-α (pg/ml)
|
94.3
|
−0.422 (b)
|
0.078
|
< 0.0001
|
−0.152 (c’)
|
0.232
|
0.51
|
0.835
|
< 0.002
|
|
MIP-1β (ng/ml)
|
1.05
|
−0.130 (b)
|
0.044
|
0.0033
|
−0.600 (c’)
|
0.222
|
0.0076
|
0.348
|
0.02
|
|
IL-1β (pg/ml)
|
6.2
|
−0.090 (b)
|
0.030
|
0.0027
|
−0.798 (c’)
|
0.198
|
< 0.0001
|
0.133
|
0.03
|
|
CRP (mg/L)
|
0.75
|
−0.055 (b)
|
0.028
|
0.0487
|
−0.812 (c’)
|
0.204
|
0.0001
|
0.117
|
0.082
|
- CBP Carbon black packer, IQR Inter-quartile range, SE Standard error, GLM Generalized linear model
- aGLM was used to assess the association (c’) between carbon black exposure and biosensor PC1 with age, overweight and obesity, current smoking status, packyears, passage of cells, and cytokines or chemokines (b) included as covariate for adjustment. Non-transformed data of cytokine and chemokine levels were used in GLM. c’ and b should be referred to Fig. 1
- bThe proportion mediated effect size that quantifies the proportion of a total effect mediated was calculated using the following equation: (c-c’) / c. The database was permuted for 500 times to generate a null distribution of c-c’ (Fig. 1b). Pperm was calculated as the number of permuted databases generating a c-c’ that is less than observed value divided by 500. c should be referred to Fig. 1
- cGLM was used to assess the association (c) between carbon black exposure and biosensor PC1 in 106 non-CBPs and 82 CBPs with adjustment for age, overweight and obesity, current smoking status, and packyears, and passage of cells. No cytokines or chemokines were included in this model. c should be referred to Fig. 1
