An evidence-based analysis of epidemiologic associations between lymphatic and hematopoietic cancers and occupational exposure to gasoline

Abstract

The presence of benzene in motor gasoline has been a health concern for potential increased risk of acute myelogenous leukemia and perhaps other lymphatic/hematopoietic cancers for approximately 40 years. Because of the widespread and increasing use of gasoline by consumers and the high exposure potential of occupational cohorts, a thorough understanding of this issue is important. The current study utilizes an evidence-based approach to examine whether or not the available epidemiologic studies demonstrate a strong and consistent association between occupational exposure to gasoline and lymphatic/hematopoietic cancers. Among 67 epidemiologic studies initially identified, 54 were ranked according to specific criteria relating to the relevance and robustness of each study for answering the research question. The 30 highest-ranked studies were sorted into three tiers of evidence and were analyzed for strength, specificity, consistency, temporality, dose-response trends and coherence. Meta statistics were also calculated for each general and specific lymphatic/hematopoietic cancer category with adequate data. The evidence-based analysis did not confirm any strong and consistent association between occupational exposure to gasoline and lymphatic/hematopoietic cancers based on the epidemiologic studies available to date. These epidemiologic findings, combined with the evidence showing relatively low occupational benzene vapor exposures associated with gasoline formulations during the last three decades, suggest that current motor gasoline formulations are not associated with increased lymphatic/hematopoietic cancer risks related to benzene.

Keywords

Gasoline cancer leukemia benzene

Introduction

The widespread use of motor gasoline over the past several decades has led to ubiquitous airborne exposures to gasoline vapor components in the workplace and in background air.¹ Benzene, a known cause of acute myelogenous leukemia (AML) in humans, has been a consistent component of gasoline vapors historically, although changes in refining processes and formulations in recent years have reduced gasoline’s benzene content. Benzene exposures continue to be the primary focus of public health protection measures for gasoline.^2

–5 Airborne concentrations of benzene, in recent years, can range from about 0.1–2.0 ppb in rural areas to 6.0–20 ppb in major US cities.⁶

Although many studies have now confirmed the strong, specific, consistent and coherent association between high benzene exposure and AML in humans, there have been weak suggestions about associations between benzene exposure and other lymphatic/hematopoietic cancers in epidemiologic studies.⁷ In particular, associations with non-Hodgkin lymphoma (NHL), multiple myeloma (MM) and forms of leukemia other than AML have been reported, but the shortcomings in the studies have been significant.^7,8 Recently, Kane and Newton⁹ published a review and meta-analysis of occupational exposure to gasoline and the risk of NHL and reported no increased risk. The two most important differences between our study and Kane and Newton’s study are: (1) the inclusion of over 10 different lymphatic/hematopoietic cancers; and (2) the use of a systematic, evidence-based approach as described generally by Guzelian et al.¹⁰ An evidence-based approach provides the ability to answer a well-framed analytical question that takes into account exposure and toxicology.

The purpose of this study is to evaluate the published epidemiologic findings concerning occupational exposures to gasoline and any possible associated risk of lymphatic/hematopoietic cancers using a systematic, evidence-based approach. This evidence-based approach involves framing an appropriate research question, identifying all potentially relevant literature, sorting and ranking the studies according to quality and relevance using objective criteria and analyzing the study findings in evidence tiers according to study rank. The final conclusions are drawn based on rigorous evaluation of the findings with respect to strength, specificity, consistency, temporality, dose-response trends and coherence. To our knowledge, this is the first time this approach has been applied to address the question of gasoline carcinogenicity.

Composition of motor gasoline and its vapors

Motor gasoline (CAS 86290-81-5) is a complex, volatile mixture of over 500 saturated and unsaturated hydrocarbons produced by the fractional distillation of crude oil that may vary in presence and proportion depending on the crude oil source and the refining process. Of these hydrocarbons, approximately 150 are regularly detectable.¹¹ Gasoline displays both temporal and geographic variability in its composition. Gasoline components also vary depending on additives that may be included to meet certain performance or regulatory specifications. Modern gasoline contains, on average, approximately 33% aromatic compounds, 57.5% paraffins (alkanes) and 9.3% olefins (unsaturated hydrocarbons) and often small amounts of alcohols, ethers, detergents, corrosion inhibitors, antioxidants and other oxygenates.^1,12 A thorough explanation of the production and composition of motor gasoline can be found in the Petroleum High Production Volume Testing Group’s submission to the USEPA concerning gasoline blending streams under the Petroleum High Production Volume (HPV) assessment program.¹²

Between 1972 and 1991, the average benzene concentration in American gasoline ranged from 0.8% to 3.18%. Historical benzene concentrations in gasoline from the European community and other countries have often been greater than 2%.^1,13,14 For example, a 2002-study reported that the mean benzene content of gasoline distributed in Europe over the last three decades ranged from 1.0% to 4.0%, with values reported as high as 8.6%.¹ The United States Reformulated Gasoline (RFG) program affects approximately one third of all gasoline sold in the United States and stipulates that no batch of RFG can have a benzene concentration of greater than 1% by volume.² New USEPA regulations in effect since January 1, 2011 further reduce the allowable benzene concentration to 0.62% by volume on all gasoline (reformulated and conventional) nationwide.¹⁵

Although liquid gasoline is predominantly comprised of volatile organic compounds, the relative proportions of hydrocarbon components in gasoline vapors (the airborne exposure component), not surprisingly, differ from those present in the liquid. Table 1 shows average gasoline vapor concentrations and variance reported for gasoline sold in the United States or Europe during the 1970s and 1980s. These measurements indicate that the vast majority of gasoline vapors by volume (>85% v/v) are comprised of alkane or alkene compounds with three to six carbons, while the aromatic content is relatively low (e.g. <3% v/v). With increasing use of oxygenates (e.g. ethanol and methyl tert-butyl ether or MTBE), the aromatic content of gasoline vapor becomes slightly diluted, and lowers the relative benzene concentrations in the liquid and vapor forms.

Table 1.

Gasoline vapor compositions (vol%) for United States and European gasoline in the 1970s and 1980s.^a

Component	United States^b		Europe^c
Component	Average	SD	Average	Range
C3: n-Propane	0.8	1.1	1.6	0–4.8
C4: Non-iso	39.2	–	34.9	15.2–44.5
C4: n-Butane	38.1	5.7	–	–
C4: Iso-butane	5.2	1.9	15.5	1.8–23.5
C4: Iso-butylene	1.1	1.5	–	–
C5: Non-iso	11.7	–	15.4	6.4–26.8
C5: Iso-pentane	22.9	6.1	23.3	16.7–30.4
C5: n-Pentane	7	4.0	–	–
C5: Cyclopentane	0.7	0.7	–	–
C5: 2,3-Dimethylbutane	0.7	0.5	–	–
C5: 2-Methyl-1-butene	1.6	2.1	–	–
C5: 2-Methyl-2-butene	1.7	1.8	–	–
C6: Non-aromatic	6.5	–	6.8	1.3–18.1
C6: n-Hexane	1.5	0.9	–	–
C6: 2-Methylpentane	2.1	1.3	–	–
C6: 3-Methylpentane	1.6	0.9	–	–
C6: Methylcyclopentane	1.3	0.4	–	–
C6: Benzene	0.7	0.4	0.7	0.1–2.3
C7: 2,3-Dimethylpentane	0.7	0.6	0.9	0.1–1.7
C7: Toluene	1.8	1.3	0.8	0.1–2.2
C8: 2,2,4-Trimethylpentane	0.5	0.5	0.2	0–0.4
C8: Xylenes	0.5	0.6	0.3	0–0.1
MTBE	–	–	5.4	0–5.8

MTBE: methyl tert-butyl ether.

^a“–” represents Values not present (for MTBE in US gasoline samples) or not specifically quantified (for European gasolines). Values for nonaromatic, non-iso components represent the sum of other major components for the same carbon number.

^bValues for 95 samples of Shell gasoline from the 1970s as reported by Halder et al.¹⁶

^cValues reported for 10 samples of European gasoline from 1984 to 1985 by CONCAWE.¹⁷

Occupational exposure to gasoline vapors

The concentration of airborne gasoline vapors in the workplace depends largely on the frequency and duration of uncontrolled transfer events that result in gasoline vapor release into the worker’s breathing zone.¹⁸ Because of the well-recognized volatility and flammability of gasoline vapors, the vast majority of workplace exposures (primarily at refineries) are inherently limited by engineering controls and procedures in place to limit explosion and fire risks. Much of the process for refining gasoline, for example, occurs in virtually closed systems within the refinery that incorporate vapor recovery and/or other controls (e.g. ventilation stacks and flares) that avoid vapor accumulation at explosive concentrations. Such control systems and procedures have been shown to limit most refinery worker benzene exposures to relatively low levels (e.g. <1 mg/m³), except for those involved in certain sampling, transfer/storage and maintenance procedures that may involve relatively brief but unavoidable exposures to gasoline vapors.^18

–22

Many studies have examined the various job types and tasks that involve higher exposures to gasoline vapors in the workplace.^{16,17,23

–33} A summary of temporal trends in gasoline vapor and benzene exposures for selected job groups with higher full-shift measurements at European petroleum production and distribution facilities is provided in Table 2.^—17,32,33 Each of the studied job types involves daily exposures to higher concentrations of gasoline vapors during loading or other transfer procedures, and two trends are readily apparent. First, for most of these job types, a decreasing temporal trend exists that generally reduced the mean occupational exposure to gasoline and/or benzene vapors after about 1985, in some instances by >5- to 10-fold without the use of vapor recovery systems. Second, separate measurements collected during the use of vapor recovery systems indicate reduced work shift exposures by about two- to fivefold. These data also suggest that since the 1980s, the average full-shift airborne benzene concentrations for these high-end gasoline exposure jobs have been below 2.2 mg/m³, in compliance with the current occupational exposure limit of 3 mg/m³.³⁴

Table 2.

Full-shift gasoline and benzene vapor exposure measurements (mg/m³) with and without vapor recovery for selected European petroleum production or distribution job groups and time intervals reported by CONCAWE.^a

	1984–1985^b		1993–1998^c		1999–2001^d
Job group	Gasoline mean (n)	Benzene mean (n)	Gasoline mean (n)	Benzene mean (n)	Gasoline mean (n)	Benzene mean (n)
Road tanker drivers, top loading	118 (63)	1.7 (142)	98 (64)	2.1 (69)	47 (33)	0.6 (33)
Road tanker drivers, bottom loading, no vapor recovery	66 (34)	1.2 (34)	55 (120)	0.82 (223)	–	–
With vapor recovery	–	–	19 (28)	0.37 (137)	–	–
Rail car operators, top loading, no vapor recovery	85 (32)	1.5 (32)	34 (22)	1.34 (69)	–	–
With vapor recovery	–	–	7.5 (8)	0.23 (8)	–	–
Service station attendants, no vapor recovery	29 (13)	0.34 (13)	19 (333)	0.25 (417)	14 (26)	0.3 (34)
With vapor recovery	–	–	–	–	7.7 (7)	0.1 (7)
Deck crew, open loading	118 (8)	1.5 (8)	78 (35)	0.56 (41)	–	–
Jetty staff	120 (21)	1.5 (21)	11 (16)	0.37 (46)	3.5 (4)	0.1 (4)
Refinery production, onsite operators	53 (62)	0.88 (62)	9 (34)	0.22 (97)	–	–
Refinery production, off-site operators	66 (27)	0.97 (27)	16 (293)	0.32 (321)	12 (6)	0.2 (6)

^a“–” represent values not reported, or the authors identified quality control issues that likely biased these samples high.

^bCONCAWE.¹⁷

^cCONCAWE.³²

^dCONCAWE.³³

A summary of historical changes in gasoline tanker truck driver exposures in the United States as reported by Smith et al.²³ is illustrated in Table 3. These changes help to explain the continually decreasing exposures to gasoline/benzene vapor in many tasks performed by distribution workers. As gasoline usage for private vehicle use expanded between the 1950s and 1980s, changes in the equipment and procedures were instituted in order to reduce vapor exposures and increase delivery efficiency. Specifically, increases in tanker truck volume and in transfer pump capacity reduced fill times, while more frequent use of spill controls, submerged pipe loading and tank venting in later years actually led to reduced worker exposures, despite the higher gasoline volumes delivered. These findings, combined with environmental regulations on gasoline reformulation that led to lower benzene content in gasoline, provide an understanding of why gasoline/benzene vapor exposures to delivery personnel since the 1980s are likely much lower than those that occurred in the 1950s and earlier.²³

Table 3.

Historical changes effecting tanker driver exposures to gasoline in the United States as identified by Smith et al.²³

	Changes in tasks, equipment, or procedures affecting exposure levels
Time period	Fill time (min)	Tanker capacity (gal)	Pump capacity (gal/min)	Spill controls	Route time (min)	Tank venting
Before 1950	40	2000	50	Limited	50	Rare
1950–1964	40	6000	150	Limited	50	Limited
1965–1974	30	8000	250	Common	50	Common
1975–1985	20	10,000	500	Common	70	Common

Adapted from Smith et al.²³

Methods

Evidence-based approach

Based on the authors’ past experience reviewing the literature concerning occupational exposures to petroleum products, the following research question was framed and considered to be answerable by rigorous evaluation of the available studies:

Does the available epidemiologic literature identify a strong, specific, consistent and coherent association between occupational exposure to gasoline and an increased risk of lymphatic and/or hematopoietic cancers in humans?

A thorough search of the published epidemiologic literature for studies on cancer mortality or incidence and occupational gasoline or oil refinery exposures led to initial inclusion of 67 separate epidemiologic reports. Of these, 13 studies were excluded because they were either not relevant to occupational gasoline exposures or refinery exposures, which were made irrelevant by follow-up studies, or were reviews that did not contribute new information (Appendix Table A1).^{9,35
–37,39,40,42,44

–49} A total of 54 studies were thus identified for evidence-based ranking on quality and relevance to the research question and are included in Appendix Table A2.

Each of the 54 studies was ranked according to the following scheme:

Strength of study design: Cohort or nested case–control within a cohort study was ranked at two; other case–control studies were ranked at one.

Target cancer robustness: For total lymphatic/hematopoietic cancers or the largest relevant subset analyzed, studies with fewer than 10 cases were ranked at zero; 10–20 cases were ranked at one; 20–40 cases were ranked at two; 40–80 cases were ranked at 3; and more than 80 cases were ranked at four.

Gasoline exposure score: Studies examining petroleum workers not specifically including fuel distribution were ranked at one; one additional point was assigned for including a job exposure matrix that incorporated gasoline exposure measurements for a rank of two; a separate evaluation of job types that were explicitly identified as having elevated gasoline exposure was assigned the rank of three, a total rank of four was given for studies examining gasoline distribution workers specifically.

Dose–response score: Studies with no dose–response measures were assigned a rank of zero; analysis by duration of exposure only was ranked at one; analysis by a job exposure matrix was ranked at two.

Latency and confounding control score: Studies with no analysis of temporality (latency) between exposure and disease were ranked at zero; those with analysis of temporality were ranked at one; those with additional confounder analysis were ranked at two.

A summary of the individual ranks assigned for each of the 54 studies is provided in the Appendix Table A2. In some cases, the same study was ranked more than once because it contained data on separate cohorts (e.g. a refinery cohort and a distribution worker cohort). The highest total rank assigned to any study was 13 and the lowest was three. Studies with a total rank of 11–13 were assigned to Tier 1 (six studies; see Table 4); those with a total rank of 9–10 were assigned to Tier 2 (13 studies; see Table 5), and those with a total rank of 7 or 8 were assigned to Tier 3 (11 studies; see Table 6). Studies with a total rank of 3–6 (24 studies, mainly small cohort and case–control studies) were not considered further in the evidence-based analysis.

Table 4.

Summary of key findings on lymphohematopoietic cancers in Tier 1 studies.^a

	Summary statistic for most relevant gasoline-exposed groups by selected disease category
References	L&H Total (ICD 200-209)	Sarcomas (ICD 200)	HD (ICD 201)	NHL (ICD 200, 202)	MM (ICD 203)	Leukemias (ICD 204-207)	AML (ICD 205.0)	CML\| (ICD 205.1)	ALL (ICD 204.0)	CLL (ICD 204.1)	Other Leukemias (ICD 204-207)^b
Rushton and Romaniuk^50,c	NA	NA	NA	NA	NA	OR 1.004 (90/354)	OR 1.00 (90/354)	OR 1.004 (90/354)	NA	OR 0.98 (90/354)	NA
Wong and Raabe^51,d	NA	NA	NA	NA	SMR 93 (48/51.46)	NA	NA	NA	NA	NA	NA
Wong and Raabe^51,d	NA	NA	NA	NA	SMR 93 (48/51.46)	NA	NA	NA	NA	NA	NA
Wong^8,e	SMR 69^f (89/129.14)	NA	SMR 48 (6/12.62)	SMR 42^f (19/45.23)	SMR 79 (17/21.52)	SMR 80 (43/53.75)	SMR 117 (18/15.38)	NA	NA	NA	NA
Wong^8,g	NA	NA	NA	NA	RR 1.00 (11/55)	RR 1.00 (35/175)	RR 1.00 (13/65)	NA	NA	NA	NA
Sorahan^43,h	SMR 110 (165/150.4)	NA	SMR 124 (15/12.1)	SMR 100 (54/54.0)	SMR 96 (29/30.1)	SMR 124 (67/54.2)	SMR (myeloid) 117 (36/30.7)		SMR (lymphoid) 142 (23/16.2)		SMR 108(8/7.4)
Sorahan^43,h	SIR 102 (214/209.1)	NA	SIR 79 (11/13.9)	SIR 96 (84/87.1)	SIR 96 (38/39.4)	SIR 118 (81/68.7)	SIR (myeloid) 133 (42/31.6)		SIR (lymphoid) 101 (32/31.8)		SIR 132 (7/5.3)
Wong et al.^50,i	SMR 75.4 ^f (55/72.9)	SMR (Other lymphatic tissues) 91.9 (18/19.5)				SMR 89.1 (27/30.3)	NA	NA	NA	NA	NA
Wong et al.^50,j	SMR 60.5 ^f (34/50.2)	SMR (Other lymphatic tissues) 73.3 (11/15.0)				SMR 70.0 (16/22.9)	NA	NA	NA	NA	NA
Lewis et al.^61,k	SMR 114 (70/61.39)	SMR102 (6/5.9)	SMR 79 (3/3.8)	SMR107 (20/18.66)	SMR 179 ^f (18/10.05)	SMR 100 (23/22.98)	SMR 98 (5/5.11)	SMR 39 (1/2.57)	SMR 0 (0/1.11)	SMR 82 (4/4.92)	SMR 141 (13/9.21)

NA: not analyzed; NC: not calculated; L&H: lymphohematopoietic; HD: Hodgkin Disease; NHL: non-Hodgkin lymphoma; MM: multiple myeloma; AML: acute myelogenous leukemia; CML: chronic myelogenous leukemia; ALL: acute lymphocytic leukemia; CLL: chronic lymphocytic leukemia.

^aSummary statistics include SMR or SIR (standardized mortality or incidence ratio) with observed/expected in parentheses, and OR (odds ratio) or RR (relative risk) with cases/controls in parentheses.

^bIncludes all leukemias and aleukemias except ICD 205.0, 205.1, 204.0 and 204.1.

^cRushton and Romaniuk⁵⁰ UK cohort of marketing and distribution workers, OR for cumulative exposure (continuous).

^dWong and Raabe⁵¹ US, UK, Canada and Australia petroleum workers, distribution division only.

^eWong et al.⁴¹ US petroleum workers with gasoline exposure update of cohort with land and marine terminal workers.

^fSignificant different at p < 0.05.

^gWong et al.⁵² US distribution workers with gasoline exposure, land-based terminals only.

^hSorahan⁴³ UK cohort update, distribution subcohort only.

ⁱWong et al.⁴¹ US petroleum workers with gasoline exposure, land-based only, RR for cumulative exposure.

^jWong et al.⁵⁰ US distribution workers with gasoline exposure, marine-based terminals only.

^kLewis et al.³⁸ Canadian cohort, marketing/distribution and marine subcohorts combined.

Table 5.

Summary of key findings on lymphohematopoietic cancers in Tier 2 studies.^a

	Summary statistic for most relevant gasoline-exposed groups by selected disease category
References	L&H Total (ICD 200-209)	Sarcomas (ICD 200)	HD (ICD 201)	NHL (ICD 200, 202)	MM (ICD 203)	Leukemias (ICD 204-207)	AML (ICD 205.0)	CML (ICD 205.1)	ALL (ICD 204.0)	CLL (ICD 204.1)	Other Leukemias (ICD 204-207)^b
Gun et al.^53,c	NA	NA	NA	SMR 89 (48/54.0)	SMR 114 (16/14.0)	SMR 107 (34/31.9)	SMR 97 (8/8.3)	SMR 109 (5/4.6)	SMR 170 (3/1.8)	SMR 103 (12/11.7)	SMR 107 (6/5.6)
Gun et al.^53,c	NA	NA	NA	SIR 99 (22/22.2)	SIR 139 (12/8.6)	SIR 99 (18/18.2)	SIR 73 (6/8.2)	SIR 145 (4/2.8)	SIR 124 (2/1.6)	SIR 230 (3/1.3)	SIR 221 (5/2.3)
Sathiakumar et al.^54,d	NA	NA	NA	NA	NA	OR 1.4 (15/46)	OR (myelogenous) 0.83 (6/28)		OR (lymphocytic) 1.6 (5/14)		NA
Schnatter et al.^55,e	NA	NA	NA	NA	NA	OR 0.98 (14/55)	NA	NA	NA	NA	NA
Raabe et al.^56,f	SMR 133^g (63/47.3)	SMR 96 (8/8.3)	SMR 68 (3/4.4)	SMR 140 (22/15.7)	SMR 121 (9/7.5)	SMR 143 (28/19.6)	SMR 136 (8/5.9)	SMR 115 (3/2.6)	SMR 0 (0/1.1)	SMR 25 (1/4.0)	NA
Divine et al.^57,h	NA	NA	NA	SMR 88 (74/84.1)	SMR 101 (36/35.6)	SMR 131 (82/62.3)	SMR 129 (20/15.5)	SMR 105 (12/11.4)	SMR 101 (5/4.9)	SMR 80 (15/18.6)	SMR 252^g (30/11.9)
Tsai et al.^58,i	SMR 96 (103/107.3)	NA	SMR 70 (4/5.7)	SMR 97 (40/41.2)	NA	SMR 101 (41/40.6)	SMR 90 (11/12.16)	SMR 66 (3/4.53)	SMR 166 (3/1.81)	SMR 114 (9/7.88)	SMR 116 (15/12.83)
Wong et al.^59,j	SMR 118 (79/66.88)	SMR 85 (8/9.42)	SMR 61 (3/4.95)	SMR 97 (26/26.75)	SMR 96 (14/14.54)	SMR 137 (37/27.1)	SMR 147 (12/8.15)	SMR 131 (5/3.81)	SMR 0 (0/1.23)	SMR 19 (1/5.39)	NA
Huebner et al.^60,k	SMR 129 (135/104.5)	NA	SMR 20 (1/4.95)	SMR 118 (44/37.62)	SMR 126 (24/19.03)	SMR 154 ^g (64/41.47)	NA	SMR 170 (9/5.36)	SMR 211 (4/1.89)	SMR 183 ^g (15/8.2)	NA
Lewis et al.^61,l	SMR 122 ^g (116/95.2)	SMR 114 (13/11.4)	SMR (ICD 202-203) 110 (45/41.0)			SMR 150 ^g (56/37.3)	NA	NA	NA	NA	NA
Lewis et al.^38,m	SMR 84 (57/67.91)	SMR 100 (7/5.97)	SMR 25 (1/3.96)	SMR 109 (22/20.2)	SMR 70 (8/11.48)	SMR 72 (19/26.3)	SMR 73 (4/5.51)	SMR 108 (3/2.78)	SMR 0 (0/1.24)	SMR 85 (5/5.92)	SMR 67 (11/16.37)
Gamble et al.^62,n	SMR 113 (98/86.8)	SMR 118 (13/11.0)	NA	SMR (ICD 202, 203) 137 ^g (46/33.6)		SMR 90 (35/39.1)	NA	NA	NA	NA	NA
Wong and Raabe⁵¹)^o	NA	NA	NA	NA	SMR 92 (145/158)	NA	NA	NA	NA	NA	NA
Sorahan⁴³p	SMR104 (256/246.3)	NA	SMR 67 (14/20.1)	SMR 114 (100/87.6)	SMR 95 (46/48.7)	SMR 107 (96/89.9)	SMR (myeloid) 107 (54/50.4)		SMR (lymphoid) 101 (27/26.6)		SMR 117 (15/12.8)
Sorahan⁴⁴p	SIR 100 (340/340.0)	NA	SIR 94 (23/24.5)	SIR 100 (141/140.6)	SIR 98 (62/63.1)	SIR 102 (114/111.8)	SIR (myeloid) 103 (53/51.7)		SIR (lymphoid) 92 (47/51.3)		SIR 159 (14/8.8)

NA: not analyzed; NC: not calculated.; L&H: lymphohematopoietic; HD: Hodgkin Disease; NHL: non-Hodgkin lymphoma; MM: multiple myeloma; AML: acute myelogenous leukemia; CML: chronic myelogenous leukemia; ALL: acute lymphocytic leukemia; CLL: chronic lymphocytic leukemia.

^bIncludes all leukemias and aleukemias except ICD 205.0, 205.1, 204.0 and 204.1.

^cGun et al.⁵³ Australian petroleum industry cohort.

^dSathiakumar et al.⁵⁴ US petroleum worker cohort, results for production-related, marketing division.

^eSchnatter et al.⁵⁵ Canadian distribution worker cohort, OR for cumulative benzene exposure, model 2.

^fRaabe et al.⁵⁶ US refinery worker cohort (Beaumont, TX), males only.

^gSignificant different at p < 0.05.

^hDivine et al.⁵⁷ Houston, TX refining, research & petrochemical workers.

ⁱTsai et al.⁵⁸ US refinery and chemical worker cohort (Houston, TX), male employees.

^jWong et al. ⁵⁹ US refinery worker cohort (Beaumont, TX), male employees.

^kHuebner et al.⁶⁰ US refinery/petrochemical worker cohort (Baton Rouge/Baytown).

^lLewis et al.⁶¹ US refinery/petrochemical worker cohort, active/terminated (Baton Rouge, Baytown, Bayway).

^mLewis et al.³⁸ Canadian cohort, refinery subcohort only.

ⁿGamble et al.⁶², US refinery/petrochemical plant retirees (Baton Rouge, Baytown, Bayway).

^oWong and Raabe⁵¹ US, UK, Canada and Australia petroleum workers, refinery subcohort only.

^pSorahan⁴³ UK cohort update, refinery subcohort only.

Table 6.

Summary of key findings on lymphohematopoietic cancers in Tier 3 studies.^a

	Summary statistic for most relevant gasoline-exposed groups by selected disease category
References^b	L&H Total (ICD 200-209)	Sarcomas (ICD 200)	HD (ICD 201)	NHL (ICD 200, 202)	MM (ICD 203)	Leukemias (ICD 204-207)	AML (ICD 205.0)	CML (ICD 205.1)	ALL (ICD 204.0)	CLL (ICD 204.1)	Other Leukemias (ICD 204-207)^c
Satin^63,d	SMR 106 (133/125.3)	SMR 55 (12/18.4)	SMR 147 (18/12.2)	SMR 71 (30/42.2)	SMR 98 (19/19.4)	SMR 105 (56/53.2)	SMR 83 (10/15.9)	SMR 84 (6/7.1)	SMR 260^e (8/3.1)	SMR 30^e (3/10.0)	SMR 170 (29/17.1)
Satin^65,f	SMR 104 (131/124.7)	SMR 120 (22/18.3)	SMR 95 (8/8.42)	SMR 104 (41/39.34)	SMR 159^e (29/18.22)	SMR 94 (47/50.2)	SMR 114 (17/14.92)	SMR 45 (3/6.71)	SMR 140 (4/2.86)	SMR 93 (9/9.66)	SMR 87 (14/16.05)
Honda^64,g	SMR 114 (104/91.2)	SMR 89 (14/15.7)	SMR 106 (9/8.5)	NA	NA	SMR 123 (46/37.4)	NA	NA	NA	NA	NA
Dryver^66,h	NA	NA	NA	OR 1.46^e (78/89)	NA	NA	NA	NA	NA	NA	NA
Jarvholm et al.^67,i	NA	Lymphomas (ICD 200-202) SIR 42 (1/2.38)			NA	SIR 0 (0/?)	NA	NA	NA	NA	NA
Jarvholm et al.^67,j	NA	Lymphomas (ICD 200-202) SIR 120 (2/1.67)			NA	SIR 400 (3/0.75)	NA	NA	NA	NA	NA
Blair e t al.^68,k	NA	NA	NA	OR 1.0 (265/511)	NA	NA	NA	NA	NA	NA	NA
Eriksson et al.^69,l	NA	NA	NA	NA	RR 0.86 (6/7)	NA	NA	NA	NA	NA	NA
Schumacher and Delzell ^70,m	NA	NA	NA	OR 0.97 (2/2)	NA	NA	NA	NA	NA	NA	NA
Holly et al.^72,n	NA	NA	NA	OR 0.97 (50/58)	NA	NA	NA	NA	NA	NA	NA
Huebner et al.^71,o	SIR 118 (55/46.56)	SIR 97 (8/8.3)	SIR 154 (4/2.59)	SIR 106 (22/20.64)	SIR 139 (9/6.48)	SIR 129 (17/13.16)	NA	SIR 181 (4/2.21)	SIR 299 (2/0.67)	SIR 122 (5/4.1)	SIR 131 (9/6.88)
Lynge et al.^74,p	NA	NA	SIR 98 (10/10.24)	SIR 107 (37/34.6)	SIR 56 (9/15.99)	SIR 89 (26/29.16)	SIR 142 (13/9.14)	NA	NA	SIR 79 (8/10.16)	SIR 51 (5/9.87)

^aSummary statistics include SMR or SIR (standardized mortality or incidence ratio) with observed/expected in parentheses and OR (odds ratio) or RR (relative risk) with cases/controls in parentheses.

^bAcute nonlymphocytic leukemia reported in lieu of AML.

^cIncludes all leukemias and aleukemias except ICD 205.0, 205.1, 204.0 and 204.1.

^dSatin et al.⁶³ cohort mortality study of US oil refinery workers in Port Arthur, Texas, males only.

^eSignificant different at p < 0.05.

^fSatin et al.⁶⁵ cohort mortality study of US petroleum refinery workers in California (Richmond and El Segundo), males only.

^gHonda et al.⁶⁴ cohort mortality study in white males at a US petrochemical plant in Illinois, overall cohort.

^hDryver et al.⁶⁶ population-based case–control of NHL in southern Sweden; OR for self-reported gasoline exposures.

ⁱJarvholm et al.⁶⁷ cohort study of cancer incidence in Swedish petrochemical industry workers; distribution workers only.

^jJarvholm et al.⁶⁷ cohort study of cancer incidence in Swedish petrochemical industry workers; refinery workers only.

^kBlair et al.⁶⁸ population-based case–control study of NHL in selected US cities; OR for occupations involving gasoline and diesel exhausts.

^lEriksson and Karlsson⁶⁹ population-based case–control of MM in Sweden; OR for petrol station/garage workers.

^mSchumacher and Delzell⁷⁰ population-based case–control of NHL in North Carolina; OR for whites with industry exposure to fuels.

ⁿHolly et al.⁷² population-based case–control of NHL in homosexual men in a US city; OR for 10+ years self-reported gasoline exposure, all subjects.

^oHuebner et al.⁷¹ cohort study of H&L incidence in a US petrochemical plant in Baton Rouge, LA.

^pLynge et al.⁷³ population-based cohort study of cancer incidence among service station workers in Scandinavian countries; males only.

The reported findings for lymphatic/hematopoietic cancers in each study were summarized by evidence tier (Tables 4–6) and a meta-statistic was derived for each general and specific disease category, by tier (Table 7). Mixed effects models were fit for the data in each tier for each cancer endpoint as described in DerSimonian and Laird.⁹⁶ When there were too few risk estimates to fit a mixed effects model (e.g. high number of iterations for the restricted maximum likelihood method), fixed effects models were fit using inverse variance weights. Statistical analyses were conducted using SAS software (Version 9.3, SAS Institute Inc., Cary, North Carolina, USA).

Table 7.

Meta statistics for hematopoietic and lymphatic cancers, by evidence tier.

	Tier 1			Tier 2			Tier 3
		95% Confidence interval			95% Confidence interval			95% Confidence interval
Disease category	RR	Lower bound	Upper bound	RR	Lower bound	Upper bound	RR	Lower bound	Upper bound
Total hematopoietic and lymphatic (ICD 200-209)	65.8^a	25.2	106	101^b	89.9	112	93.3^b	59.4	127
Total leukemia (ICD 204-207)	100^a	99.8	100.2	102^b	91.5	113	101^b	84.2	118
Acute myelogenous leukemia (ICD 205.0)	100^a	99.3	100.7	109^a	87.5	131	109^b	59.9	159
Chronic myelogenous leukemia (ICD 205.1)	99.9^a	89.6	110	107^a	84.5	130	65.1^a	20.6	110
Multiple myeloma (ICD 203)	82.4^b	66.0	98.8	95.7^b	86.0	105	105^b	23.3	187
Non-hodgkins lymphoma (ICD 200, 202)	65.8^a	25.2	106.4	101^b	90.4	112	99.4^b	83.5	115
Acute lymphocytic leukemia (ICD 204.0)	–	–	–	131^a	85.7	176	192^a	105	279
Chronic lymphocytic leukemia (ICD 204.1)	97.5^b	60.4	134	68.1^b	26.6	110	56.5^a	21.2	91.8
Other leukemia (ICD 204-207)	124^a	91.7	156	97.3^b	26.2	168	92.1^b	10.7	173

^aFixed effects model.

^bMixed effects model.

Results

Tier 1 study findings that focus most specifically on more robust studies of gasoline distribution workers are summarized in Table 4. Tier 1 includes nine cohorts from six different studies. The most robust study in Tier 1 is the cancer mortality and incidence study of UK petroleum distribution workers;⁴³ it reported no significant changes in any of the lymphatic/hematopoietic cancers, and analysis of total leukemia and acute non-lymphocytic leukemia was not significantly elevated among more exposed job types, such as operators and drivers. Lewis et al.³⁸ identified a significant excess of MM deaths (Standardized Mortality Ratio (SMR), 179) among Canadian petroleum marketing and distribution workers (land and marine cohorts combined); the MM excess was concentrated among employees with 25–34 years of employment, but total lymphatic/hematopoietic cancers and other subtypes were not significantly different from expected. Wong and Raabe⁵¹ found no excess MM in a large multinational cohort of petroleum workers in distribution jobs. Wong et al.⁵² reported significantly lower than expected SMR values for total lymphatic/hematopoietic cancers among US gasoline distribution workers, and Wong et al.⁴¹ reported significantly lower total lymphatic/hematopoietic cancers and NHL in US petroleum workers with gasoline exposure jobs. Case–control analyses of leukemia by Rushton and Romaniuk⁵⁰(UK cohort) and leukemia and MM by Wong et al.⁴¹ (US cohort) did not identify significant changes associated with gasoline exposure using multiple job exposure matrices.

Tier 2 study findings that include most of the larger oil refinery and petrochemical plant cohorts with plausible gasoline exposure are summarized in Table 5. Tier 2 includes 14 cohorts from 13 different studies. The most robust studies are the UK petroleum refinery worker cohort and studies of US refineries and petrochemical plants; no significant changes were found for lymphatic/hematopoietic cancer mortality or incidence, and analysis of leukemia by job type revealed no significant associations. Relatively few statistically significant findings were reported for the other cohort studies and no significant findings were reported in the case–control analyses. Divine et al.⁵⁷ identified a significant excess of ‘other leukemias’ (SMR 252) in the Houston, Texas refinery, although total leukemias and more prevalent lymphatic/hematopoietic disease subtypes such as AML, chronic myelogenous leukemia, acute lymphocytic leukemia (ALL) and chronic lymphocytic leukemia (CLL) were not significantly changed, nor were they related to higher exposure job types. Raabe et al.⁵⁶ reported a significant increase in lymphatic/hematopoietic cancers (SMR 133) in the Beaumont, Texas refinery, although no significant changes were reported for specific cancer subtypes. Lewis et al.⁶¹ reported significant increases in lymphatic/hematopoietic cancers (SMR 122) and leukemias (SMR 150) in US refinery cohorts in Louisiana (Baton Rouge), Texas (Baytown) and New Jersey (Bayway), and related analyses identified increased leukemias (SMR 154) and excess CLL (SMR 183).⁶⁰ Gamble et al.⁶² reported excess NHL and MM (SMR 137).⁶² These authors reported that the significant excesses appeared to be largely attributable to mortality patterns at one refinery (Baton Rouge) and among individuals hired before 1950, when industrial hygiene practices and engineering controls to limit benzene exposures were less rigorous.^60
–62 Raabe et al.⁵⁶ reported similar temporal trends regarding the total lymphatic/hematopoietic cancer excess in the Beaumont, Texas refinery cohort.

Tier 3 study findings are presented in Table 6, and include refinery cohorts and population-based studies with plausible gasoline exposure. Tier three includes 12 cohorts from 11 different studies. Overall, there were relatively few statistically significant findings. Satin et al.⁶³ reported no significant excess of leukemias (SMR 105) in the Port Arthur refinery cohort, but found significantly increased ALL (SMR 260) and significantly decreased CLL (SMR 30). Satin et al.⁵⁹ reported significantly increased mortality from MM (SMR 159) in workers from two California refineries. The Swedish case–control study by Dryver et al.⁶⁶ reported significantly greater self-reported gasoline exposures among NHL cases, but three smaller case–control studies found no significant difference.^68,70,72

Table 7 provides meta-statistics for the general and specific lymphatic/hematopoietic cancer categories for Tiers 1, 2 and 3. The Tier 3 meta-relative risk estimate for ALL was elevated and statistically significant (RR: 192, 95% confidence interval (CI): 105–279).^62,54,55 This value was based on three studies and was estimated through a fixed effects model. The Tier 2 RR for ALL was not significantly elevated (RR: 131, 95% CI: 85.7–176). This value was based on eight cohorts in seven studies and was also estimated through a fixed effect model.^{38,53,56

–59} None of the other meta-relative risks was statistically significant.

Discussion

The findings presented in Tables 4 to 7 suggest that there is no strong, specific, consistent and coherent association between occupational gasoline exposure and lymphatic/hematopoietic cancers, based on the most robust and rigorous epidemiological studies that have been published. Although the ranking was inherently subjective in nature, it allowed us to stratify the studies in a transparent manner. To our knowledge, ours is the first study to utilize an evidence-based approach to assist in drawing conclusions about potential relationships between gasoline and human lymphatic/hematopoietic cancer risks.

The highest-ranked studies of Tier 1 (Table 4) were the most telling, in that lymphohematopoietic cancers were not significantly elevated among the more robust petroleum distribution worker studies. Only one study identified a statistically significant increased risk of MM (SMR 179, 18 observed/10 expected) that was based on relatively small numbers.³⁸ This increase was not found in other Tier 1 studies and the meta-statistic for MM was not elevated (RR: 82.4, 95% CI: 66.0–98.8). Similarly, there was no strong or consistent increase in total lymphatic/hematopoietic cancers or any specific subtype (Table 7). All of the meta-relative risk estimates were below 130 for Tier 1 studies, and none of the estimates were statistically significant for any of the three evidence tiers, other than ALL in Tier 3. The ALL estimate was based on 16 people across three cohorts. Based on this small sampling of workers and the lower magnitude and specificity of gasoline exposures in Tier 3, the isolated significant increase in ALL risk appears inconsistent and is not supported by the stronger weight of evidence in Tiers 1 and 2.

Many studies of larger populations of oil refinery and petrochemical plant workers have been published to date, but the frequency and duration of gasoline vapor exposures is expected to be much less than for distribution workers. Among these studies identified in Tier 2 (Table 5) and Tier 3 (Table 6), there again are few statistically significant increases in lymphatic/hematopoietic cancers and/or specific subtypes. Reported increases in leukemias among US refinery populations in Tier 2 appear to be isolated to certain regions (e.g. Baton Rouge, Louisiana, and Beaumont, Texas), and occurred mainly among those hired before 1950, before more rigorous benzene exposure controls commenced. For example, in Baton Rouge, statistically significant elevations are seen in the pre-1950 hire group for all leukemia (SMR 1.95) and CLL subtypes (SMR 2.81)⁵³ and an SMR of 154 was observed in Beumont, Texas in employees with 40+ years since the date of first employment.^59,60. Overall, there is no consistent, statistically significant increase in lymphatic/hematopoietic cancers and/or specific subtypes in Tiers 2 and 3, and the meta-statistics (Table 7) calculated for each tier further support the absence of a biologically meaningful cancer risk based on epidemiologic studies expected to be most relevant for assessing gasoline-related cancer risks.

Benzene is the only commonly detectable component of gasoline vapors that is demonstrated to cause cancer in humans above certain cumulative doses. The vast majority of gasoline vapor components are alkane and alkene hydrocarbons, none of which have any known carcinogenic hazard to humans. Nevertheless, the prevalent uses of gasoline in occupational settings and the recognition of benzene carcinogenicity have raised sufficient public health concerns to spur dozens of epidemiologic studies to determine whether or not elevated leukemia rates observed in workers with high benzene exposures are seen among those exposed to gasoline. While some oil refinery studies identified significant increases in leukemias, the prominent occurrence among workers hired before 1950 suggests that improved industrial hygiene and engineering controls implemented in subsequent decades may have reduced or eliminated any increased risk. Furthermore, as noted by Wong et al.⁸, exposures to benzene among distribution workers were relatively low (3–4 ppm) even in the 1950s, as compared to classic benzene exposed cohorts such as the Pliofilm cohort. Since gasoline formulations over the past three decades have been modified to contain lower concentrations of benzene, the potential risk of benzene-related cancer associated with gasoline-exposed populations should be decreasing. It should be noted that studies included in Tier 1 of this analysis included gasoline exposed cohorts that encompassed a period of time greater than five decades (1947–2003) and no apparent time-related differences in disease occurrence were evident in our meta-analysis or in the individual studies. Indeed, OSHA recently reported that most of the occupations that have the highest potential for exposure to gasoline since the 1980s have full-shift airborne benzene concentrations for workers that are below the current US occupational exposure limits of 3 mg/m³.³⁴

Each of the individual studies and meta–analysis techniques have limitations. The evidence-based ranking approach is expected to assist in sorting the most relevant studies as the basis for drawing valid conclusions regarding the research question. In many of the studies, however, the population subset with regular and/or appreciably elevated exposures to gasoline vapors is likely to be small. Additionally, many of the cohorts included in Tier 2 and 3 studies may have had little gasoline exposure, leading to potential dilution of results. However, it is expected that the specificity and robustness of Tier 1 studies may attenuate such dilution. The meta-analysis approach, too, may be limited, in that it combines SMRs from various populations. An analysis of heterogeneity across the studies assists in characterizing how well a group of studies can be considered repeated measures. Although heterogeneity estimates were not calculated, a recent review and meta-analysis of gasoline and NHL risk looked at most of the same studies evaluated in the current study and did not identify significant heterogeneity issues.⁹ Nonetheless, mixed effects models were utilized in this meta-analysis to account for heterogeneity. In addition, some of the studies included in the current analysis may have been somewhat redundant (e.g. such as when authors reported combined-cohort findings for selected cancers, as in the 1997 MM analysis by Wong and Raabe.⁵¹). Furthermore, as discussed above, gasoline is a complex mixture that has changed through time and has differences worldwide. Although the specific types of gasoline that individuals included in this analysis were exposed to varied inter- and intracohort, we believe that the meta-analysis approach utilized in this article is the best way to approach questions involving mixtures of chemicals. Despite these limitations, the transparent evidence-based approach allows readers to draw their own conclusions based on the reported findings of individual studies, with or without the aid of the meta-analysis.

Past reviews and meta-analyses concerning gasoline carcinogenicity in humans have not identified any strong or consistent associations with hematolymphopoietic cancers. Kane and Newton⁹ performed a rigorous meta-analysis on occupational gasoline exposure and NHL that included many of the studies reported in the current study. They identified a meta risk estimate of 1.02, comparable to our meta-statistic of 101 (95% CI: 90. 4-112) for Tier 2, the most robust evidence tier that is dominated by large refinery worker studies.⁹ The comprehensive review of gasoline carcinogenicity performed by Dement et al.⁹⁷ summarized and reviewed several cohort and case control studies, some of which were included in our analysis. The authors focused primarily on hematopoietic cancers and kidney cancer. They determined that cohort studies did not demonstrate statistically significant associations between gasoline exposure and leukemia. A more recent review by Jamall and Willhite³⁵ focused on benzene exposures arising from gasoline. The authors evaluated gasoline exposures using benzene as a surrogate for deriving risk-based clean-up levels for gasoline impacted soil and groundwater. They reviewed several epidemiological and toxicological studies, and concluded that cancer potency factors and theoretical cancer risks should not be used to dictate clean-up levels for gasoline. They also concluded that gasoline should be treated as a complex mixture, whether in response to clean-up levels or to questions regarding carcinogenicity. Finally, IARC, USEPA and ACGIH either do not consider gasoline to be a carcinogen or state that there is insufficient evidence to determine its carcinogenicity.^3,5,15,98,99

In conclusion, an evidence-based approach was utilized to evaluate whether or not convincing epidemiologic evidence exists for a strong, specific, consistent and coherent association between occupational gasoline exposure and lymphatic/hematopoietic cancers. While isolated associations have been observed, the most relevant and robust studies available to date do not demonstrate any such association between gasoline exposure and the general and specific cancer types evaluated. If benzene is the sole carcinogenic component of gasoline vapors, the available studies suggest that lower benzene vapor exposures from gasoline formulated during the past three decades may ultimately demonstrate that current occupational exposures are virtually safe with respect to lymphoid and hematopoietic cancer risks. Additional studies focusing on gasoline-exposed worker populations should be completed to further assess this possibility.

Footnotes

Authors’ Note

After the current analysis was in press, Schnatter et al. (2012) published additional relevant data on AML among gasoline-exposed workers in the Journal of National Cancer Institute (104:1724-1737). They reported no increase in AML risk for occupationally exposed gasoline workers in a cohort that would have been classified under Tier 2 in our analysis. Adding these findings would further strengthen our finding of no strong or consistent association between gasoline exposure and increased AML risk.

All of the authors are employed by ChemRisk, a consulting firm that provides scientific advice to the government, corporations, law firms, and various scientific/professional organizations. ChemRisk has been engaged by entities involved in the petroleum industry. This paper was prepared and written exclusively by the authors without review or comment by employees or counsel for those entities. It is likely that this work will be relied upon in industrial hygiene research, risk assessment research, and litigation. Some of the authors may be called upon to serve as expert witnesses.

Appendix

Table A2.

Ranking of relevant studies

Study authors/characteristics	Ranks assigned based on study design and population characteristics					Total rank
Study authors/characteristics	Strength of design	Robustness for L&H cancers	Specificity to gasoline	Dose-response rigor	Confounding control rigor	Total rank
Rushton and Romaniuk⁵⁰. Nested Case–Control, UK petroleum distribution workers; n = 91 leukemia deaths.	2	4	4	2	1	13
Wong and Raabe⁵¹. Cohort, US, UK, Canadian, Australian petroleum worker meta-analysis for MM deaths, n = 158 exp L&H cancer deaths overall, n = 51 exp distribution workers only.	2	4	4	2	1	13
Wong et al.⁴¹. Cohort, US petroleum workers exposed to gasoline; n = 129 exp L&H cancer deaths	2	4	4	2	1	13
Wong et al.⁴¹. Nested Case–control, US petroleum workers exposed to gasoline. N = 59 L&H cancer deaths.	2	3	4	2	1	12
Sorahan⁴³. Cohort, UK petroleum distribution workers, n =150 exp L&H deaths; n = 209 exp L&H incidence	2	4	4	1	1	12
Wong et al.⁵². US petroleum distribution workers, n = 73 exp L&H deaths	2	3	4	2	1	12
Lewis et al.³⁸. Cohort, US petroleum distribution (land and marine); n = 61 exp L&H deaths	2	3	4	1	1	11
Gun et al.⁵³ Cohort, Australian petroleum workers; n = 100 incident and 49 deaths from L&H.	2	4	2	1	1	10
Sathiakumar et al.⁵⁴ Nested Case–control, US petroleum workers; n = 69 leukemia cases.	2	3	2	2	1	10
Schnatter et al.⁵⁵ Nested Case–control, Canadian petroleum distribution workers; n = 14 leukemia deaths.	2	1	4	1	1	9
Raabe et al.⁵⁶. Cohort, US refinery workers (Beaumont, TX); n = 49 exp L&H deaths.	2	3	2	1	1	9
Divine et al.⁵⁷. Cohort, US refinery & petrochemical workers (Texas); n > 100 exp L&H deaths.	2	4	1	1	1	9
Tsai et al.⁵⁸ Cohort, US refinery and chemical workers (Deer Park, TX); n = 96 exp L&H deaths.	2	4	1	1	1	9
Wong et al.⁵⁹ Cohort, US refinery workers (Beaumont, TX); n = 69 exp L&H deaths.	2	4	1	1	1	9
Huebner et al.⁶⁰. Cohort, US refinery and petrochemical workers (Baton Rouge, Baytown); n = 105 exp L&H deaths.	2	4	1	1	1	9
Lewis et al.⁶¹. Cohort, US refinery/petrochemical workers (Baton Rouge, Bayway, Baytown; active/terminated); n = 95 exp L&H deaths.	2	4	1	1	1	9
Gamble et al.⁶² Cohort, US refinery/petrochemical retirees (Baton Rouge, Baytown, Bayway; retirees); n = 113 exp L&H deaths.	2	4	1	1	1	9
Satin et al.⁶³ Cohort, US refinery workers (Port Arthur, TX); n = 135 exp L&H deaths.	2	4	1	0	1	8
Honda et al.⁶⁴. Cohort, US Petroleum manufacturing plant in Illinois; n = 91 exp L&H deaths.	2	4	1	0	1	8
Satin et al.⁶⁵ Cohort, US refinery workers (Richmond & El Segundo, CA); n = 126 exp L&H deaths.	2	4	1	0	1	8
Dryver et al.⁶⁶ Case–control, regional hospital cases in Sweden; n = 859 NHL cases.	1	4	1	1	1	8
Jarvholm et al.⁶⁷ Cohort, Swedish petroleum workers; n = 14 incident L&H cases.	2	1	2	1	1	7
Blair et al.⁶⁸ Case–control, US male hospital cases; n = 622 NHL cases and 1245 controls.	1	4	1	1	0	7
Eriksson and Karlsson⁶⁹. Case–control, selected counties in Sweden; n = 275 MM cases.	1	4	1	1	0	7
Schumacher and Delzell⁷⁰. Case–control, US textile workers; n = 501 NHL deaths and 569 controls.	1	4	1	1	0	7
Huebner et al.⁷¹ Cohort, US petrochemical workers (Baton Rouge); n = 47 exp L&H incidents.	2	3	1	0	1	7
Holly et al.⁷² Case–control, US homosexual males; n = 312 NHL cases and 420 controls.	1	4	1	1	0	7
Lynge et al.⁷³ Cohort, Nordic population-based cohort of service station workers, n = 119 exp L&H incidents	2	4	1	0	0	7
Collingwood et al.⁷⁴ Cohort, US refinery workers (Paulsboro, NJ); n = 36 exp L&H deaths.	2	2	1	0	1	6
Tsai et al.⁷⁵ Cohort, US refinery/petrochemical workers (Baton Rouge, LA); n =18 exp L&H deaths.	2	1	1	1	1	6
Wong et al.⁷⁶. Cohort, US refinery (Torrance, CA); n = 19 exp L&H deaths.	2	1	1	1	1	6
Tsai et al. (2004). Cohort, US refinery and petrochemical workers (Norco and Geismar, LA); n = 25 incident L&H cases.	2	2	1	0	1	6
Linet et al.⁷⁷ Case–control, US MM cases; n = 100 white cases.	1	4	1	0	0	6
Lindquist et al.⁷⁸ Case–control, regional hospital cases in Sweden; n = 125 acute leukemia cases.	1	4	1	0	0	6
Morris et al.⁷⁹, Case–control, US multi-region cases; n = 698 MM cases.	1	4	1	0	0	6
Cuzick and De Stavola⁸⁰ Case–control, UK regional hospital cases; n = 399 MM cases	1	4	1	0	0	6
Flodin et al.⁸¹ Case–control, regional hospital cases in Sweden; n = 131 MM cases.	1	4	1	0	0	6
Persson and Fredrikson⁸² Case–control, Swedish cancer registry; n = 199 NHL cases and 479 referents.	1	4	1	0	0	6
Franceschi et al.⁸³ Case–control, Italian hospital cases; n = 208 NHL cases and 401 controls	1	4	1	0	0	6
Nadon et al.⁸⁴. Case–control, Montreal hospital cases; n = 206 NHL cases and 2599 controls.	1	4	1	0	0	6
Karunanayake et al.⁸⁵ Case–control, Canadian male cases; n = 513 NHL cases and 1506 controls.	1	4	1	0	0	6
t'Mannetje et al.⁸⁶ Case control, New Zealand cancer registry cases; n = 291 NHL cases and 471 controls.	1	4	1	0	0	6
Orsi et al.⁸⁷ Case–control, French hospital cases; n = 824 lymphoid cancer cases and 752 controls.	1	4	1	0	0	6
Bernard et al.⁸⁸ Case–control, UK hospital cases; n = 285 lymphoid cancer cases and 285 controls.	1	4	1	0	0	6
Hardell et al.⁸⁹ Case–control, Swedish hospital cases; n = 105 NHL cases and 335 controls.	1	4	1	0	0	6
Zheng et al.⁹⁰ Case–control, US male hospital cases; n = 555 NHL/CLL cases and 2380 controls.	1	4	1	0	0	6
Brandt et al.⁹¹ Case–Control, hospital consults in Sweden; n = 50 ANLL cases.	1	3	1	0	0	5
Pukkala⁹² Cohort, Finnish oil refinery workers; n = 24 exp L&H incident cases.	2	2	1	0	0	5
Consonni et al. (2002). Cohort, Italian refinery workers; n = 8.4 exp L&H deaths.	2	0	1	1	1	5
Lagorio et al.⁹³ Cohort, Italian service station workers, n = 7.3 exp L&H	2	0	1	1	0	4
McCraw et al.⁹⁴ Cohort, US refiner workers (Wood River, IL); n = 10 exp L&H deaths.	2	1	1	0	0	4
Jakobsson et al.⁹⁵ Cohort, Swedish population-based cohort of service station workers, n = 2.8 exp L&H	2	0	1	0	0	3

L&H: lymphohematopoietic; MM: multiple myeloma.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

References

Verma

des Tombe

. Benzene in gasoline and crude oil: occupational and environmental implications. AIHA J (Fairfax, Va) 2002; 63: 225–230.

HEI. The health effects institute. program summary: research on oxygenates added to gasoline, http://www.healtheffects.org/Pubs/oxyprog.htm (2010). accessed 7/7/12.

IARC. International Agency for the Research on Cancer. Occupational exposures in petroleum refining, crude oil and major petroleum fuels. Vol. 45. IARC Monographs. International Agency for Research on Cancer, Lyon, 1989.

ACGIH. American Conference of Governmental Industrial Hygienists. Industrial Ventilation. A manual of recommended practices for design. 26th ed. American Conference of Governmental Industrial Hygienists, Cincinnati, OH, 2007, p. 6–6.

United States Environmental Protection Agency (USEPA). Evaluation of the carcinogenicity of unleaded gasoline. Office of Health and Environmental Assessment, Environmental Protection Agency, Washington, D.C., 1987.

McCarthy

Hafner

Montzka

. Background concentrations of 18 air toxins for Noth America. J Air Waste Manage Assoc 2006; 56: 3–11.

Galbraith

Gross

Paustenbach

. Benzene and human health: a historical review and appraisal of associations with various diseases. Crit Rev Toxicol 2010; 40: 1–46.

Wong

. A critique of the exposure assessment in the epidemiologic study of benzene-exposed workers in China conducted by the Chinese academy of preventive medicine and the US national cancer institute. Regul Toxicol Pharmacol 1999; 30: 259–267.

Kane

Newton

. Occupational exposure to gasoline and the risk of non-Hodgkin lymphoma: a review and meta-analysis of the literature. Canc Epidemiol 2010; 34: 516–522.

10.

Guzelian

Victoroff

Halmes

James

Guzelian

. Evidence-based toxicology: a comprehensive framework for causation. Hum Exp Toxicol 2005; 24: 161–201.

11.

ACGIH. American Conference of Governmental Industrial Hygienists. Benzene CAS number: 71-42-2. American Conference of Governmental Industrial Hygienists, Cincinnati, OH, 2001.

12.

The American Petroleum Institute. (API). Petroleum HPV testing group. Gasoline blending streams category Assessment. Environmental Protection Agency, Washington, D.C., 2008.

13.

Egeghy

Tornero-Velez

Rappaport

. Environmental and biological monitoring of benzene during self-service automobile refueling. Environ Health Perspect 2000; 108: 1195–1202.

14.

Schnatter

. Petroleum workers studies and benzene risk assessment. J Toxicol Environ Health A 2000; 61: 433–437.

15.

USEPA. Regulations of fuel and fuel additives: reformulated gasoline requirements for former severe nonattainment areas under the 1-hour ozone standard that were redesignated to attainment for the 1-hour standard prior to its revocation, and which are current nonattainment areas for the 8-hour ozone standard. 40 CFR Part 80. In Federal Register, 2006.

16.

Halder

Van Gorp

Hatoum

Warne

. Gasoline vapor exposures. Part I. Characterization of workplace exposures. Am Ind Hyg Assoc J 1986; 47: 164–172.

17.

CONCAWE. A survey of exposures to gasoline vapour. Report No. 4/87, 1987.

18.

Gaffney

Panko

Unice

Burns

Kreider

Gelatt

Occupational exposure to benzene at the ExxonMobil refinery in Baytown, TX (1978-2006). J Expo Sci Environ Epidemiol 2011; 21: 169–185.

19.

Widner

Gaffney

Panko

Unice

Burns

Kreider

Airborne concentrations of benzene for dock workers at the ExxonMobil refinery and chemical plant, Baton Rouge, Louisiana, USA (1977–2005). Scand J Work Environ Health 2011; 37: 147–158.

20.

Kreider

Unice

Panko

Burns

Paustenbach

Benzene exposure in refinery workers: ExxonMobil Joliet, Illinois, USA (1977-2006). Toxicol Ind Health 2010; 26: 671–690.

21.

Panko

Gaffney

Burns

Unice

Kreider

Booher

Occupational exposure to benzene at the ExxonMobil refinery at Baton Rouge, Louisiana (1977-2005). J Occup Environ Hyg 2009; 6: 517–529.

22.

Wong

. Recent findings and new initiatives for epidemiologic research on benzene. J Toxicol Environ Health A 2000; 61: 457–466.

23.

Smith

Hammond

Wong

Health effects of gasoline exposure. I. Exposure assessment for U. S. distribution workers. Environ Health Perspect 1993; 101(Suppl 6): 13–21.

24.

Runion

. Occupational exposures to potentially hazardous agents in the petroleum industry. Occup Med 1988; 3: 431–444.

25.

Lewis

Bell

Cordingley

Pearlman

Rushton

. Retrospective estimation of exposure to benzene in a leukaemia case-control study of petroleum marketing and distribution workers in the United Kingdom. Occup Environ Med 1997; 54: 167–175.

26.

Kearney

Dunham

. Gasoline vapor exposures at a high volume service station. Am Ind Hyg Assoc J 1986; 47: 535–539.

27.

McDermott

Vos

. Service station attendants' exposure to benzene and gasoline vapors. Am Ind Hyg Assoc J 1979; 40: 315–321.

28.

Hakkola

Saarinen

. Customer exposure to gasoline vapors during refueling at service stations. Appl Occup Environ Hyg 2000; 15: 677–680.

29.

Verma

Cheng

Shaw

Verma

Julian

A simultaneous job- and task-based exposure evaluation of petroleum tanker drivers to benzene and total hydrocarbons. J Occup Environ Hyg 2004; 1: 725–737.

30.

Saarinen

Hakkola

Kangas

. Comparison of tanker drivers' occupational exposures before and after the installation of a vapour recovery system. J Environ Monit 2000; 2: 662–665.

31.

Hakkola

Saarinen

. Exposure of tanker drivers to gasoline and some of its components. Ann Occup Hyg 1996; 40: 1–10.

32.

CONCAWE. A review of European gasoline exposure data for the period 1993-1998. Report No. 2/00, 2000.

33.

CONCAWE. Method for monitoring exposure to gasoline vapour in air-revision 2002. Report No. 8/08, 2002.

34.

Occupational Safety and Health Standards (OSHA). Toxic and hazardous substances. 1910. 1028 App A. Substance safety data sheet, benzene. OSHA, 2011.

35.

Jamall

Willhite

. Is benzene exposure from gasoline carcinogenic? J Environ Monit 2008; 10: 176–187.

36.

Miller

Fransman

Heederik

Hurley

Kromhout

Fitzsimons

. A review of the data quality and comparability of case-control studies of low-level exposure to benzene in the petroleum industry. Int Arch Occup Environ Health 2010; 83: 69–76.

37.

Schnatter

Theriault

Katz

Thompson

Donaleski

Murray

. A retrospective mortality study within operating segments of a petroleum company. Am J Ind Med 1992; 22: 209–229.

38.

Lewis

Schnatter

Katz

Thompson

Murray

Jorgensen

Updated mortality among diverse operating segments of a petroleum company. Occup Environ Med 2000; 57(9): 595–604.

39.

Schnatter

Katz

Nicolich

Theriault

. A retrospective mortality study among Canadian petroleum marketing and distribution workers. Environ Health Perspect 1993; 101(Suppl 6): 85–99.

40.

Raabe

Wong

. Leukemia mortality by cell type in petroleum workers with potential exposure to benzene. Environ Health Perspect 1996; 104(Suppl 6): 1381–1392.

41.

Wong

Trent

Harris

. Nested case-control study of leukaemia, multiple myeloma, and kidney cancer in a cohort of petroleum workers exposed to gasoline. Occup Environ Med 1999; 56: 217–221.

42.

Rushton

Alderson

. Epidemiological survey of oil distribution centres in Britain. Br J Ind Med 1983; 40: 330–339.

43.

Sorahan

. Mortality of UK oil refinery and petroleum distribution workers, 1951-2003. Occup Med (Lond.) 2007; 57: 177–185.

44.

Rushton

L. A

39-year follow-up of the U. K. oil refinery and distribution center studies: results for kidney cancer and leukemia. Environ Health Perspect 1993; 101(Suppl 6): 77–84.

45.

Rushton

. Further follow up of mortality in a United Kingdom oil distribution centre cohort. Br J Ind Med 1993; 50: 561–569.

46.

Rushton

. Further follow up of mortality in a United Kingdom oil refinery cohort. Br J Ind Med 1993; 50: 549–560.

47.

Wong

Raabe

. Critical review of cancer epidemiology in petroleum industry employees, with a quantitative meta-analysis by cancer site. Am J Ind Med 1989; 15: 283–310.

48.

Divine

Hartman

. Update of a study of crude oil production workers 1946-94. Occup Environ Med 2000; 57: 411–417.

49.

Schwartz

. Proportionate mortality ratio analysis of automobile mechanics and gasoline service station workers in New Hampshire. Am J Ind Med 1987; 12: 91–99.

50.

Rushton

Romaniuk

. A case-control study to investigate the risk of leukaemia associated with exposure to benzene in petroleum marketing and distribution workers in the United Kingdom. Occup Environ Med 1997; 54: 152–166.

51.

Wong

Raabe

. Multiple myeloma and benzene exposure in a multinational cohort of more than 250,000 petroleum workers. Regul Toxicol Pharmacol 1997; 26: 188–199.

52.

Wong

Harris

Smith

. Health effects of gasoline exposure. II. Mortality patterns of distribution workers in the United States. Environ Health Perspect 1993; 101(Suppl 6): 63–76.

53.

Gun

Pratt

Ryan

Roder

. Update of mortality and cancer incidence in the Australian petroleum industry cohort. Occup Environ Med 2006; 63: 476–481.

54.

Sathiakumar

Delzell

Cole

Brill

Frisch

Spivey

. A case-control study of leukemia among petroleum workers. J Occup Environ Med 1995; 37: 1269–1277.

55.

Schnatter

Armstrong

Nicolich

Thompson

Katz

Huebner

Lymphohaematopoietic malignancies and quantitative estimates of exposure to benzene in Canadian petroleum distribution workers. Occup Environ Med 1996; 53: 773–781.

56.

Raabe

Collingwood

Wong

An updated mortality study of workers at a petroleum refinery in Beaumont, Texas. Am J Ind Med 1998; 33: 61–81.

57.

Divine

Hartman

Wendt

. Update of the Texaco mortality study 1947-93: part II. Analyses of specific causes of death for white men employed in refining, research, and petrochemicals. Occup Environ Med 1999; 56: 174–180.

58.

Tsai

Ahmed

Wendt

Foster

Donnelly

Strawmyer

. A 56-year mortality follow-up of Texas petroleum refinery and chemical employees, 1948-2003. J Occup Environ Med 2007; 49: 557–567.

59.

Wong

Harris

Rosamilia

Raabe

. An updated mortality study of workers at a petroleum refinery in Beaumont, Texas, 1945 to 1996. J Occup Environ Med 2001; 43: 384–401.

60.

Huebner

Wojcik

Rosamilia

Jorgensen

Milano

. Mortality updates (1970-1997) of two refinery/petrochemical plant cohorts at Baton Rouge, Louisiana, and Baytown, Texas. J Occup Environ Med 2004; 46: 1229–1245.

61.

Lewis

Gamble

Jorgensen

Mortality among three refinery/petrochemical plant cohorts. I. 1970 to 1982 active/terminated workers. J Occup Environ Med 2000; 42: 721–729.

62.

Gamble

Lewis

Jorgensen

Mortality among three refinery/petrochemical plant cohorts. II. Retirees. J Occup Environ Med 2000; 42: 730–736.

63.

Satin

Wong

Yuan

Bailey

Newton

Wen

A 50-year mortality follow-up of a large cohort of oil refinery workers in Texas. J Occup Environ Med 1996; 38: 492–506.

64.

Honda

Delzell

Cole

. An updated study of mortality among workers at a petroleum manufacturing plant. J Occup Environ Med 1995; 37: 194–200.

65.

Satin

Bailey

Newton

Ross

Wong

. Updated epidemiological study of workers at two California petroleum refineries, 1950-95. Occup Environ Med 2002; 59: 248–256.

66.

Dryver

Brandt

Kauppinen

Olsson

. Occupational exposures and non-Hodgkin's lymphoma in Southern Sweden. Int J Occup Environ Health 2004; 10: 13–21.

67.

Jarvholm

Mellblom

Norrman

Nilsson

Nordlinder

. Cancer incidence of workers in the Swedish petroleum industry. Occup Environ Med 1997; 54: 686–691.

68.

Blair

Linos

Stewart

Burmeister

Gibson

Everett

Evaluation of risks for non-Hodgkin's lymphoma by occupation and industry exposures from a case-control study. Am J Ind Med 1993; 23: 301–312.

69.

Eriksson

Karlsson

. Occupational and other environmental factors and multiple myeloma: a population based case-control study. Br J Ind Med 1992; 49: 95–103.

70.

Schumacher

Delzell

. A death-certificate case-control study of non-Hodgkin's lymphoma and occupation in men in North Carolina. Am J Ind Med 1988; 13: 317–330.

71.

Huebner

Chen

Friedlander

Jorgensen

Bhojani

Incidence of lymphohaematopoietic malignancies in a petrochemical industry cohort: 1983-94 follow up. Occup Environ Med 2000; 57: 605–614.

72.

Holly

Lele

Bracci

. Non-Hodgkin's lymphoma in homosexual men in the San Francisco Bay Area: occupational, chemical, and environmental exposures. J Acquir Immune Defic Syndr Hum Retrovirol 1997; 15: 223–231.

73.

Lynge

Andersen

Nilsson

Barlow

Pukkala

Nordlinder

Risk of cancer and exposure to gasoline vapors. Am J Epidemiol 1997; 145: 449–458.

74.

Collingwood

Raabe

Wong

. An updated cohort mortality study of workers at a northeastern United States petroleum refinery. Int Arch Occup Environ Health 1996; 68: 277–288.

75.

Tsai

Wendt

Cardarelli

Fraser

. A mortality and morbidity study of refinery and petrochemical employees in Louisiana. Occup Environ Med 2003; 60: 627–633.

76.

Wong

Harris

Rosamilia

Raabe

. Updated mortality study of workers at a petroleum refinery in Torrance, California, 1959 to 1997. J Occup Environ Med 2001a; 43: 1089–1102.

77.

Linet

Harlow

McLaughlin

. A case-control study of multiple myeloma in whites: chronic antigenic stimulation, occupation, and drug use. Canc Res 1987; 47: 2978–2981.

78.

Lindquist

Nilsson

Eklund

Gahrton

. Acute leukemia in professional drivers exposed to gasoline and diesel. Eur J Haematol 1991; 47: 98–103.

79.

Morris

Koepsell

Daling

Taylor

Lyon

Swanson

Toxic substance exposure and multiple myeloma: a case-control study. J Natl Canc Inst 1986; 76: 987–994.

80.

Cuzick

De Stavola

. Multiple myeloma: a case control study. Br J Canc 1988; 57: 516–520.

81.

Flodin

Fredriksson

Persson

. Multiple myeloma and engine exhausts, fresh wood, and creosote: a case-referent study. Am J Ind Med 1987; 12: 519–529.

82.

Persson

Fredrikson

. Some risk factors for non-Hodgkin's lymphoma. Int J Occup Med Environ Health 1999; 12: 135–142.

83.

Franceschi

Serraino

Carbone

Talamini

La Vecchia

. Dietary factors and non-Hodgkin's lymphoma: a case-control study in the northeastern part of Italy. Nutr Canc 1989; 12: 333–341.

84.

Nadon

Siemiatycki

Dewar

Krewski

Gerin

. Cancer risk due to occupational exposure to polycyclic aromatic hydrocarbons. Am J Ind Med 1995; 28: 303–324.

85.

Karunanayake

McDuffie

Dosman

Spinelli

Pahwa

Occupational exposures and non-Hodgkin's lymphoma: Canadian case-control study. Environ Health 2008; 7: 44.

86.

t'Mannetje

Dryson

Walls

McLean

Maule

Cheng

High risk occupations for non-Hodgkin's lymphoma in New Zealand: case-control study. Occup Environ Med 2008; 65: 354–363.

87.

Orsi

Troussard

Monnereau

Berthou

Fenaux

Marit

Occupation and lymphoid malignancies: results from a French case-control study. J Occup Environ Med 2007; 49: 1339–1350.

88.

Bernard

Cartwright

Bird

Richards

Lauder

Roberts

. Aetiologic factors in lymphoid malignancies: a case-control epidemiological study. Leuk Res 1984; 8: 681–689.

89.

Hardell

Eriksson

Degerman

. Exposure to phenoxyacetic acids, chlorophenols, or organic solvents in relation to histopathology, stage and anatomical localization of non-Hodkin's lymphoma. Canc Res 1994; 54: 2386–2389.

90.

Zheng

Blair

Zhang

Weisenburger

Zahm

. Occupation and risk of non-Hodgkin's lymphoma and chronic lymphocytic leukemia. J Occup Environ Med 2002; 44: 469–474.

91.

Brandt

Nilsson

Mitelman

. Occupational exposure to petroleum products in men with acute non-lymphocytic leukaemia. Br Med J 1978; 1: 553.

92.

Pukkala

. Cancer incidence among Finnish oil refinery workers, 1971-1994. J Occup Environ Med 1998; 40: 675–679.

93.

Lagorio

Fuselli

Iavarone

Vanacore

Carere

. Exposure to benzene of service station employees and composition of benzene. Med Lav 1994; 85: 412–421.

94.

McCraw

Joyner

Cole

. Excess leukemia in a refinery population. J Occup Med 1985; 27: 220–222.

95.

Jakobsson

Ahlbom

Bellander

Lundberg

. Acute myeloid leukemia among petrol station attendants. Arch Environ Health 1993; 48: 255–259.

96.

DerSimonian

Laird

. Meta-analysis in clinical trials. Controlled Clin Trials 1986; 7: 177–188.

97.

Dement

Hensley

Gitelman

. Carcinogenicity of gasoline: a review of epidemiological evidence. Ann NY Acad Sci 1997; 837: 53–76.

98.

USEPA. Fuel trends report: Gasoline 1995-2005. Report for the Compliance and Innovative Strategies Division. Office of Transportation and Air Quality. Report no. EPA420-R-08-002, 2008.

99.

ACGIH. American Conference of Governmental Industrial Hygienists. TLVs and BEIs based on the documentation of the threshold limit values for chemical substances and physical agents and biological exposure indicies, 2011.