Circulatory metabolites trigger ex vivo arterial endothelial cell dysfunction in population chronically exposed to diesel exhaust

Table 4 Mediation effect of serum cGMP level on the association between diesel exhaust exposure and biosensor responses in 124 DETs and 123 non-DETs

Mediator	Biosensor	Mediator biosensor asso^a			Diesel biosensor asso^a			Mediation effect^b
Mediator	Biosensor	Est (b)	SE	Raw P	Est (c')	SE	Raw P	PM	P perm
Individual mediator
cGMP	Biosensor PC1	− 0.669	0.251	0.008	0.204	0.119	0.087	0.232	0.02
	CCL5	− 0.555	0.228	0.016	0.210	0.106	0.049	0.196	0.01
	ICAM	− 0.616	0.294	0.037	0.167	0.136	0.222	0.253	0.055
	SELP	− 0.460	0.207	0.027	0.144	0.096	0.136	0.228	0.01
	VCAM	− 0.858	0.367	0.020	0.283	0.170	0.097	0.218	0.025
Tail DNA %^c	Biosensor PC1	0.062	0.030	0.042	0.110	0.138	0.423	0.528	0.035
	CCL5	0.074	0.027	0.007	0.119	0.122	0.332	0.553	0.005
	ICAM	0.058	0.035	0.102	0.038	0.159	0.811	0.752	0.055
	SELP	0.033	0.025	0.182	0.088	0.112	0.432	0.427	0.07
	VCAM	0.109	0.045	0.016	0.093	0.202	0.645	0.699	0.02
Double mediator^c
cGMP	Biosensor PC1	− 0.607	0.274	0.028	0.078	0.137	0.572	0.667	0.015
Tail DNA %	Biosensor PC1	0.051	0.030	0.095
cGMP	CCL5	− 0.589	0.248	0.018	0.085	0.122	0.485	0.679	< 0.005
Tail DNA %	CCL5	0.065	0.027	0.019
cGMP	ICAM	− 0.565	0.324	0.082	0.005	0.160	0.977	0.970	0.025
Tail DNA %	ICAM	0.049	0.036	0.172
cGMP	SELP	− 0.487	0.227	0.033	0.061	0.112	0.587	0.607	0.015
Tail DNA %	SELP	0.025	0.025	0.311
cGMP	VCAM	− 0.799	0.410	0.052	0.047	0.202	0.814	0.846	0.005
Tail DNA %	VCAM	0.096	0.045	0.035

^aInternal dose of diesel exhaust exposure was defined as urinary PC extracted from six urinary metabolites which has shown an excellent association with diesel exhaust exposure status. This PC was converted into an ordered categorical variable with three values (0, 1, 2) with each group having same number of subjects to calculate the estimate for the dose–response. Generalized linear model was used to assess association (c') between internal dose of diesel exhaust exposure and ex vivo biosensor PC with age, obesity, internal of for cigarette smoke exposure, passage of cells, and mediators (e.g., natural-log transformed serum metabolite levels or tail DNA %, b) as covariate for adjustment
^bThe proportion mediated effect size that quantifies the proportion of a total effect mediated was calculated using the following equation: ab / (ab + c'). The database was permuted for 200 times to generate a null distribution of a*b. P_perm was calculated as the number of permuted databases generating an a*b that is greater than observed value divided by 200. For analyses involving double mediators, sum of a*b was calculated
^cAny analyses involving Tail DNA% were conducted in 105 DETs and 119 non-DETs. Metabolite biosensor association was quantified with every 10% increase in tail DNA %

ISSN: 1743-8977