Table IV.

Exposure-response relationship between air concentrations of PM_2.5 and cytokine levels

	Diesel engine exhaust exposure categoryb
		Diesel engine exhaust exposed workers
Cytokinesa	Control workers (n=46)	1st tertile (n=14)	2nd tertile (n=12)	3rd tertile (n=15)	Correlation (all subjects)c		Correlation (exposed only)c		Ptrend (FDR) (all subjects)d	Ptrend (FDR) (exposed only)d
					r	P	r	P
IL-1βe (pg/ml)	1.2 (1.2–7.5)	2.8 (1.2–72.5)	1.15 (1.2–6.6)	1.15 (1.2–6.2)	−0.13	0.23	−0.24	0.14	-	-
IL-6 (pg/ml)	7.2 (5.2–12.8)	7.4 (6.3–36.3)	7.0 (2.4–23.1)	7.4 (5.3–9.2)	0.04	0.73	−0.20	0.20	0.31 (0.62)	0.12 (0.30)
IL-8 (pg/ml)	11.9 (8.5–18.1)	12.2 (8.9–26.7)	11.3 (9.9–16.6)	9.4* (8.4–11.9)	−0.16	0.14	−0.51	0.001	0.053 (0.15)	0.0023 (0.009)
MIP-1β (pg/ml)	71.1 (31.4–130.8)	67.8 (23.0–236.4)	50.0* (12.9–180.4)	30.8* (8.9–59.4)	−0.43	<0.001	−0.43	0.005	<0.001 (<0.001)	0.001 (0.005)
MCP-1 (pg/ml)	106.7 (59.3–203.9)	118.7 (83.1–328.5)	101.9 (43.3–156.1)	90.0 (27.9–138.4)	−0.21	0.047	−0.43	0.005	0.15 (0.33)	0.009 (0.03)
TNF-αe (pg/ml)	0.6 (0.6–5.1)	0.6 (0.6–59.0)	0.6 (0.6–4.1)	0.6 (0.6–4.5)	−0.12	0.28	−0.27	0.09	-	-

^a,Data are shown as unadjusted medians (10^th-90^th percentiles).

^b,Unadjusted concentrations are presented to characterize PM_2.5 tertile levels for DEE exposed workers. The median (range) PM_2.5 (μg/m³) levels for the three categories: 1^st tertile, (296, 203–338 μg/m³), 2^nd tertile, (385, 346–396 μg/m³), 3^rd tertile, (408, 397–536 μg/m³).

^c,Spearman correlation coefficients for cytokines and PM_2.5 exposure categorized as 0 (controls), 1 (1^st tertile), 2 (2^nd tertile), and 3 (3^rd tertile) in analyses including all subjects, and 1 (1^st tertile), 2 (2^nd tertile), and 3 (3^rd tertile) in analyses including exposed subjects only.

^d,P value using the ln transformation of the dependent variable and category of PM_2.5 as an ordinal variable taking on values of 0, 1, 2, and 3 (0: controls, 1: 1^st tertile, 2: 2^nd tertile, 3: 3^rd tertile) in analyses including all subjects and 1, 2, and 3 (1: 1^st tertile, 2: 2^nd tertile, 3: 3^rd tertile) in analyses including exposed subjects only, adjusted for age, BMI, smoking status, current alcohol use and recent infection. FDR: Multiple comparisons were adjusted for by applying the false discovery rate (FDR) using the Benjamini-Hochberg method.

^e,P_trend for IL-1β and TNF-α not analyzed using linear regression models, because the concentrations of IL-1β from 51 samples (58%) and concentrations of TNF-α from 59 samples (67%) were lower than the limit of detection.

^*P value <0.05 compared to controls, adjusted for age, BMI, smoking status, current alcohol use and current infection.