Arch Toxicol (2011) 85:11891193 DOI 10.

1007/s00204-011-0669-2

R E G U L A T O RY T O X I CO L O G Y

Quantitative cancer risk assessment for ethylene oxide inhalation in occupational settings
Ciriaco Valdez-Flores Robert L. Sielken Jr M. Jane Teta

Received: 1 February 2011 / Accepted: 3 February 2011 / Published online: 24 February 2011 Springer-Verlag 2011

Abstract The estimated occupational ethylene oxide (EO) exposure concentrations corresponding to speciWed extra risks are calculated for lymphoid mortality as the most appropriate endpoint, despite the lack of a statistically signiWcant exposureresponse relationship. These estimated concentrations are for occupational exposures 40 years of occupational inhalation exposure to EO from age 20 to age 60 years. The estimated occupational inhalation exposure concentrations (ppm) corresponding to speciWed extra risks of lymphoid mortality to age 70 years in a population of male and female EO workers are based on Cox proportional hazards models of the most recent updated epidemiology cohort mortality studies of EO workers and a standard life-table calculation. An occupational exposure at an inhalation concentration of 2.77 ppm EO is estimated to result in an extra risk of lymphoid mortality of 4 in 10,000 (0.0004) in the combined worker population of men and women from the two studies. The corresponding estimated concentration decreases slightly to 2.27 ppm when based on only the men in the updated cohorts combined. The diVerence in these estimates reXects the diVer-

ence between combining all of the available data or focusing on only the men and excluding the women who did not show an increase in lymphoid mortality with EO inhalation exposure. The results of sensitivity analyses using other mortality endpoints (all lymphohematopoietic tissue cancers, leukemia) support the choice of lymphoid tumor mortality for estimation of extra risk. Keywords Ethylene oxide Excess cancer risk calculation for occupational inhalation exposure NIOSH and UCC epidemiology data Cox proportional hazards models Lymphoid mortality Abbreviations Dow Dow Chemical Company EO Ethylene oxide IARC International Agency for Research on Cancer NIOSH National Institute for Occupational Safety and Health SMR Standardized mortality ratio UCC Union Carbide Corporation

Electronic supplementary material The online version of this article (doi:10.1007/s00204-011-0669-2) contains supplementary material, which is available to authorized users. C. Valdez-Flores R. L. Sielken Jr (&) Sielken & Associates Consulting Inc., 3833 South Texas Avenue, Suite 230, Bryan, TX 77802, USA e-mail: SielkenAssoc@aol.com M. Jane Teta Exponent Health Sciences, 8 Dogwood Court, Middlebury, CT 06762, USA e-mail: jteta@exponent.com

Introduction There is a large body of published occupational studies involving ethylene oxide (EO). Despite the extensive investigations of EO worker cohorts, the results do not indicate a consistent pattern of excess for any speciWc type of cancer (Valdez-Flores et al. 2010; Swaen et al. 2009). In 2008, the International Agency for Research on Cancer (IARC) classiWed the human data as limited. There has been some evidence in exposureresponse analyses related to lymphohematopoietic cancers (especially lymphoid cancers, a grouping suggested by the

123

1190

Arch Toxicol (2011) 85:11891193

authors of a study of sterilant workers conducted by the National Institute for Occupational Safety and Health (NIOSH) (Steenland et al. 2004)). Lymphoid tumors (a grouping of non-Hodgkin lymphoma, multiple myeloma, and lymphocytic leukemia) in men and in men and women combined showed a statistically signiWcantly increased hazard rate in the highest quintile of cumulative exposure compared to the lowest quintile in categorical analyses, although the Cox proportional hazards modeling on the continuous exposure scale showed no increasing trends (Valdez-Flores et al. 2010). Women by themselves had an estimated non-statistically signiWcant negative slope in the Cox proportional hazards modeling. Furthermore, the Swaen et al. study of Union Carbide Corporation (UCC) manufacturing workers showed no evidence of lymphoid mortality excesses or exposure response (Swaen et al. 2009). (Since UCC was acquired by the Dow Chemical Company in 2001, the cohort is denoted UCC/Dow). These two mortality cohort studies (NIOSH and UCC/ Dow) have suYcient size and length of follow-up and provide adequate exposure information to quantify an exposureresponse relationship for EO inhalation. The NIOSH cohort was updated by Steenland et al. (2004) with 11 more years of follow-up. The UCC/Dow cohort was updated by Swaen et al. (2009) with 15 more years of follow-up. These updated cohorts currently provide the best available human epidemiology data on ethylene oxide inhalation and the risk

of cancer. Some characteristics of these updated cohorts are given in Table 1. The updated NIOSH cohort includes approximately 18,000 male and female sterilant workers with an average cumulative EO exposure of 27 ppm-years and an average follow-up of 26 years. The updated UCC/ Dow cohort includes approximately 2,000 men in manufacturing with an average cumulative EO exposure of 67 ppmyears and an average follow-up of 37 years. Thus, these two cohorts have long-term average follow-up, estimates of cumulative exposure for each individual worker, and follow-up initiated in early years when exposures were much higher. Earlier forms of these cohorts (Steenland et al. 1991; Teta et al. 1993) have been used in previous risk assessments (Stayner et al. 1993; Teta et al. 1999; Kirman et al. 2004). In 2004, Kirman et al. focused on leukemia and the potential nonlinearity in the extrapolation of the exposure response relationship to low doses (Kirman et al. 2004). At the Aspen Toxicology Forum in 2009 (Sielken 2009) and in the subsequent Valdez-Flores et al. (2010) publication, the authors used the updated cohorts to conduct an updated environmental quantitative cancer risk assessment. Part of that update was to expand the risk assessment from consideration of leukemia mortality to a broader set of 12 potential endpoints to discern the one most appropriate for exposureresponse analyses. The epidemiology data were assessed using cumulative external concentration

Table 1 Some characteristics of the updated NIOSH and UCC/Dow epidemiological cohorts of workers exposed to ethylene oxide: number of worker deaths for each endpoint by study and sex for workers with exposure information Endpoints NIOSH and UCC/Dow Total Lymphohematopoietic tissue Lymphoid tumors
a

NIOSH Female 37 26 12 9 5 7 15 8 8 102 17 13 1,283 9,859 13% 261,038 26.5 Total 74 53 30 13 10 10 25 14 13 103 36 24 2,749 17,493 16% 450,906 25.8 Male 37 27 18 4 5 3 10 6 5 1 19 11 1,466 7,634 19% 189,868 24.9 Female 37 26 12 9 5 7 15 8 8 102 17 13 1,283 9,859 13% 261,038 26.5

UCC/Dow Male 27 17 12 3 2 7 11 13 12 0 15 10 1,051 2,063 51% 75,306 36.5

Male 64 44 30 7 7 10 21 19 17 1 34 21 2,517 9,697 26% 265,174 27.3

101 70 42 16 12 17 36 27 25 103 51 34 3,800 19,556 19% 526,212 26.9

Non-Hodgkin lymphoma Multiple myeloma Lymphocytic leukemia Myeloid leukemia Leukemia Central nervous system Malignant brain neoplasm Breast cancer Pancreas cancer Stomach cancer All causes All workers % deceased Number of person-years Average follow-up (person-years)
a

Lymphoid tumors are deWned herein in the same manner as in Steenland et al. (2004); namely, non-Hodgkin lymphoma, multiple myeloma, or lymphocytic leukemia

123

Arch Toxicol (2011) 85:11891193

1191

(ppm-years) as the dose measure. Cumulative exposures were calculated as described by Swaen et al. (2009) for the UCC cohort and by Stayner et al. (1993) for the NIOSH cohorts. The focus of the quantitative cancer risk assessment in Valdez-Flores et al. (2010) was on potential risks due to environmental inhalation exposures. In the current paper, using the most up-to-date epidemiology data and using similar methodology to our published environmental risk assessment (Valdez-Flores et al. 2010), the focus shifts to estimated extra risks associated with occupational inhalation exposures.

Methods In Valdez-Flores et al. (2010), there were twelve endpoints (lymphohematopoietic tissue, lymphoid tumors, non-Hodgkin lymphoma, multiple myeloma, lymphocytic leukemia, myeloid leukemia, leukemia, central nervous system, malignant brain neoplasm, breast cancer, pancreas cancer, and stomach cancer) and six subcohorts (NIOSH men, UCC/Dow men, NIOSH and UCC/Dow men combined, NIOSH women, NIOSH men and women combined, and NIOSH and UCC/Dow men and women combined). (There was no UCC/Dow female subcohort). None of the SMRs for any combination of the 12 cancer endpoints and 6 subcohorts analyzed were statistically signiWcantly greater than one. In Valdez-Flores et al. (2010), the Wtted Cox proportional hazards models with cumulative EO exposure did not have statistically signiWcant positive slopes for any of these combinations, and the lack of increasing trends was corroborated by categorical analyses. Consistent with these statistical analyses and EO being at worst a weak carcinogen (Vogel and Nivard 1998; Albertini 2009 Aspen Toxicology Forum), there is no clear target cancer endpoint. Nevertheless, lymphoid mortality seems to be the most appropriate cancer endpoint for risk characterization, given the increased Wndings in highest exposure group of the largest study in categorical analyses (Valdez-Flores et al. 2010).

The estimated occupational exposure concentrations (ppm) corresponding to speciWed extra risks to age 70 years are calculated herein for lymphoid mortality as a worst case despite the lack of a statistically signiWcant exposure response relationship. These estimated concentrations are for occupational exposuresspeciWcally, 40 years of occupational inhalation exposure to EO from age 20 to age 60 years. The estimated occupational exposure concentrations (ppm) corresponding to speciWed extra risks of lymphoid mortality in a population of men and women combined are based on Cox proportional hazards models of the most recent updated NIOSH and UCC/Dow epidemiology data and a standard life-table calculation (see ValdezFlores et al. 2010 for additional modeling details and the electronic supplementary material for details of the lifetable calculations).

Results: quantitative risk estimates for lymphoid mortality Tables 25 present the results of diVerent calculations of the estimated occupational exposure concentrations (ppm) corresponding to diVerent speciWed extra risks of potential interest (ranging between 1 in 100,000 and 4 in 1,000). The life-table calculations are for an inference population of male and female workers. The only diVerences between the diVerent life-table calculations are background hazard rates and the slopes in the Cox proportional hazards models. The slope depends on the worker cohort (gender and study) used in the Cox modeling and the type of cancer (lymphoid, lymphohematopoietic tissue, or leukemia). In Table 2, based on the men and women combined and the updated NIOSH and UCC/Dow cohorts combined, an occupational exposure at an inhalation concentration of 2.77 ppm ethylene oxide for up to 40 years starting at age 20 years is estimated to result in an extra risk of lymphoid mortality of 4 in 10,000 (0.0004). The corresponding estimated concentration decreases slightly to 2.27 ppm when based on only the men in the updated NIOSH and UCC/

Table 2 Standard life-table calculations of the estimated occupational exposure concentrations (ppm) corresponding to speciWed extra risks to age 70 years of lymphoid mortality for males and females combined

based on Cox proportional hazards models of the updated NIOSH and UCC/Dow epidemiology data: 40 years of occupational exposure to ethylene oxide from age 20 to age 60 years

Occupational exposure concentrations (ppm) corresponding to speciWed extra risks of lymphoid mortality for males & females Cohort used to estimate Cox proportional hazards model Gender Study Extra risk for males & females 1 in 1,000 6.58a 5.38 1 in 10,000 0.712 0.582 1 in 100,000 0.072 0.058 4 in 1,000 21.35 17.46 4 in 10,000 2.77 2.27 4 in 100,000 0.286 0.234

Males & females Males only

a

NIOSH & UCC/Dow NIOSH & UCC/Dow

ppm

123

1192 Table 3 Standard life-table calculations of the estimated occupational exposure concentrations (ppm) corresponding to speciWed extra risks of lymphoid mortality to 70 years for males and females combined based on Cox proportional hazards models of the updated NIOSH and

Arch Toxicol (2011) 85:11891193 UCC/Dow epidemiology data: 40 years of occupational exposure to ethylene oxide from age 20 to age 60 years: sensitivity analyses: alternative epidemiology cohorts

Occupational exposure concentrations (ppm) corresponding to speciWed extra risk of lymphoid mortality for males & females Cohort used to estimate Cox proportional hazards model Gender Study Extra risk for males & females 1 in 1,000 6.58a 5.38 3.63 3.24 1 in 10,000 0.712 0.582 0.393 0.351 1 in 100,000 0.072 0.058 0.040 0.035 4 in 1,000 21.35 17.46 11.79 10.52 4 in 10,000 2.77 2.27 1.53 1.37 4 in 100,000 0.286 0.234 0.158 0.141

Males & females Males only Males & females Males only Females only Males only
a

NIOSH & UCC/Dow NIOSH & UCC/Dow NIOSH NIOSH NIOSH UCC/Dow

Non-positive slope: estimated risk does not increase with exposure Non-positive slope: estimated risk does not increase with exposure

ppm

Table 4 Standard life-table calculations of the estimated occupational exposure concentrations (ppm) corresponding to speciWed extra risks of lymphohematopoietic tissue mortality to 70 years for males and females combined based on Cox proportional hazards models of the

updated NIOSH and UCC/Dow epidemiology data: 40 years of occupational exposure to ethylene oxide from age 20 to age 60 years: sensitivity analyses: alternative target organ: lymphohematopoietic tissue

Occupational exposure concentrations (ppm) corresponding to speciWed extra risk of lymphohematopoietic tissue mortality for males & females Cohort used to estimate Cox proportional hazards model Gender Study Extra risk for males & females 1 in 1,000 8.86a 5.78 4.15 3.31 1 in 10,000 0.949 0.620 0.445 0.354 1 in 100,000 0.096 0.062 0.045 0.036 4 in 1,000 29.35 19.17 13.76 10.95 4 in 10,000 3.71 2.42 1.74 1.38 4 in 100,000 0.382 0.249 0.179 0.142

Males & females Males only Males & females Males only Females only Males only
a

NIOSH & UCC/Dow NIOSH & UCC/Dow NIOSH NIOSH NIOSH UCC/Dow

Non-positive slope: estimated risk does not increase with exposure Non-positive slope: estimated risk does not increase with exposure

ppm

Dow cohorts combined. The diVerence in these estimates reXects the diVerence between combining all of the available data or focusing on only the men and excluding the NIOSH women who did not show an increase in lymphoid mortality with ethylene oxide inhalation exposure.

Discussion The updated NIOSH and UCC/Dow EO epidemiology data are the most appropriate data for quantitative exposure response assessment. Lymphoid tumors are the most appropriate target organ. However, results should be considered worst case, given the absence of consistency between men and women and the inconsistency between epidemiology studies. Furthermore, in Valdez-Flores et al. (2010), none of the SMRs were statistically signiWcantly greater than one

for any combination of the 12 cancer endpoints and 6 subcohorts analyzed, and no evidence of a positive cumulative exposureresponse relationship was found. In particular, in Valdez-Flores et al. (2010), the Wtted Cox proportional hazards models with cumulative EO exposure as the dose metric did not have statistically signiWcant positive slopes, and the lack of increasing trends was corroborated by categorical analyses. In the current exposureresponse assessment, we conducted sensitivity analyses using various combinations of gender and study cohort and two other mortality endpoints, lymphohematopoietic tissue cancers and leukemia. Table 3 shows the sensitivity analysis with regard to the estimated ppm levels corresponding to the diVerent combinations of gender and study cohort included in the modeling. For extra risks of 4 in 10,000, the range of positive estimates is 1.372.77 ppm. For both the

123

Arch Toxicol (2011) 85:11891193 Table 5 Standard life-table calculations of the estimated occupational exposure concentrations (ppm) corresponding to speciWed extra risks of leukemia mortality to 70 years for males and females combined based on Cox proportional hazards models of the updated NIOSH and

1193 UCC/Dow epidemiology data: 40 years of occupational exposure to ethylene oxide from age 20 to age 60 years: sensitivity analyses: alternative target organ: leukemia

Occupational exposure concentrations (ppm) corresponding to speciWed extra risk of leukemia mortality for males & females Cohort used to estimate Cox proportional hazards model Gender Study Extra risk for males & females 1 in 1,000 39.42a 12.19 33.23 9.15 60.17 1 in 10,000 4.68 1.45 3.94 1.09 7.14 1 in 100,000 0.477 0.148 0.402 0.111 0.728 4 in 1,000 108.80 33.65 91.71 25.25 166.08 4 in 10,000 17.58 5.44 14.81 4.08 26.83 4 in 100,000 1.90 0.59 1.60 0.44 2.89

Males & females Males only Males & females Males only Females only Males only
a

NIOSH & UCC/Dow NIOSH & UCC/Dow NIOSH NIOSH NIOSH UCC/Dow

Non-positive slope: estimated risk does not increase with exposure

ppm

women in the NIOSH cohort and the men in the UCC/ Dow cohort, there was no estimated increase in lymphoid mortality with ethylene oxide inhalation exposure. Tables 4 and 5 show the impact on the estimated ppm levels if lymphoid mortality was changed to lymphohematopoietic tissue mortality or leukemia mortality, respectively. For both endpoints, exposure concentrations associated with extra risk of 4 in 10,000 are larger than those calculated for lymphoid tumor mortality. For leukemia, the exposure concentrations are substantially greater. These Wndings support the choice of lymphoid tumor mortality. In the estimation of the occupational EO inhalation exposure concentrations (ppm) corresponding to speciWed extra risks of mortality, we have used the most appropriate endpoint based on the most recent epidemiological data suitable for exposureresponse analyses. In addition to the reWned analyses of the separate cohorts, statistical power of the Wtted models has been increased by analyzing the combined cohorts.
Acknowledgments This manuscript is based on a December 16, 2010, slide presentation requested by the ScientiWc Committee on Occupational Exposure Limits (SCOEL). This work was Wnancially supported by the CEFIC Ethylene Oxide & Derivatives Sector Group, Avenue Van Nieuwenhuyse 4, Box 2, B-1160 Brussels. The conclusions are those of the authors. The epidemiological data for the updated NIOSH study were obtained from the National Institute for Occupational Safety and Health through a Freedom of Information Act request (May 13, 2004). The epidemiological data for the updated UCC study were obtained under a November 19, 2008, conWdentiality agreement between Union Carbide Corporation and Sielken & Associates Consulting, Inc. ConXict of interest The authors have served as EO consultants to the American Chemistry Council (ACC) and the European Chemical Industry Council (CEFIC). Statements of fact and opinion are those of the authors.

References
Albertini RJ (2009) Ethylene oxide low potency mutagen low potency carcinogen. Presentation: ethylene oxide session, Toxicology Forum, Aspen Colorado IARC Monographs on the Evaluation of Carcinogenic Risks to Humans (2008) 1,3-butadiene, ethylene oxide and vinyl halides (Vinyl Fluoride, Vinyl Chloride and Vinyl Bromide), vol 97. World Health Organization. International Agency for Research on Cancer. Lyon, France Kirman CR, Sweeney LM, Teta MJ, Sielken RL Jr, Valdez-Flores C, Albertini RJ, Gargas ML (2004) Addressing nonlinearity in the exposure-response relationship for a genotoxic carcinogen: cancer potency estimates for ethylene oxide. Risk Anal 24(5):11651183 Sielken RL Jr (2009) Use of pooled epidemiology data in quantitative risk assessment. Presentation: ethylene oxide session, Toxicology Forum, Aspen Colorado Stayner L, Steenland K, Greife A, Hornung R, Hayes RB, Nowlin S, Morawetz J, Ringenburg V, Elliot L, Halperin W (1993) Exposure-response analysis of cancer mortality in a cohort of workers exposed to ethylene oxide. Am J Epidemiol 138:787798 Steenland K, Stayner L, Griefe A, Halperin B, Hayes R, Hornung R, Nowlin S (1991) A cohort mortality study of workers exposed to ethylene oxide. N Eng J Med 324(20):14021407 Steenland K, Stayner L, Deddens J (2004) Mortality analyses in a cohort of 18 235 ethylene oxide exposed workers: follow up extended from 1987 to 1998. Occup Environ Med 61:27 Swaen GM, Burns C, Teta JM, Bodner K, Keenan D, Bodnar CM (2009) Mortality study update of ethylene oxide workers in chemical manufacturing: a 15 year update. J Occup Environ Med 51(6):714723 Teta MJ, Benson LO, Vitale JN (1993) Mortality study of ethylene oxide workers in chemical manufacturing: a mortality study. Br J Ind Med 50:704709 Teta MJ, Sielken RL Jr, Valdez-Flores C (1999) Ethylene oxide cancer risk assessment based on epidemiologic data: application of revised regulatory guidelines. Risk Anal 19:11351155 Valdez-Flores C, Sielken RL Jr, Teta MJ (2010) Quantitative cancer risk assessment based on NIOSH and UCC epidemiological data for workers exposed to ethylene oxide. Regul Tox Pharmacol 56:312320 Vogel EW, Nivard M (1998) Heritable and cancer risks of exposures to anticancer drugs: inter-species comparisons of covalent deoxyribonucleic acid-binding agents. Mutat Res 400:509540

123

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.